(in alphabetical order by presenters last name)
● Role of Socio-Economic & Demographic
Factors in Preventing Infant Deaths: Historical
Evidence Using Annual Data from Pakistan
Faisal Abbas, Ph.D.
Presented by: Faisal Abbas, Ph.D., Junior
Researcher, Economic & Technological Change,
Centre for Development Research, Walter Flex
Street 3, Bonn, 53113, DE, Phone: +004-9176-
2073-1590; Email: fabbas@uni-bonn.de
Research Objective: The study aims at answering the following questions: What factors (economic, demographic, social and political) determines the improvement of the health status of infants in
Pakistan? What role health spending and unemployment can play in explaining infant mortality in historical context? Whether the improved health status has causal relation with public health care spending and prevailing persistent income inequality? What is the likely effect of policy variables like mothers’ education, immunization of children, urban population growth in improving health status?
Study Design: This study is unique in the sense that it is first of its kind that uses the time series data and applied cointegration analysis in a developing country context to draw attention towards the long run relationship between health status and some important policy variables. The study employs the Augmented Dickey Fuller (ADF) and Philip Perron (PP) test to examine the issue of stationarity and unit root hypothesis. Johansen full information maximum likelihood multivariate cointegration approach is used to determine the factors responsible in explaining the long run variations in infant mortality and life expectancy in
Pakistan. Engle-Granger (EG) causality test is employed to see the direction of causality between health status, public health care spending and other variables of policy relevance. As Johansen methodology provide long run estimates therefore following general to specific modeling approach, a vector error correction model (VECM) is applied to capture the short run dynamics. An orthogonalized impulse response analysis (IRA) is carried out to see the impact of shocks (i.e. innovation accounting) to health status and its probable impact is analyzed on other variables of interest.
Population Studied: It is a country level study using annual data from Pakistan.
Principal Findings: The trace statistics strongly rejects the null hypothesis that there is no cointegration between variables (i.e. r = 0), but do not reject the hypothesis that there is one cointegrating relationship (i.e. r = 1).It is therefore concluded that our model has one cointegrating vector (i.e., a unique long-run equilibrium relationship exists). The maximal Eigenvalue test also rejects the null of no cointegration and hence it is also in line with the results of the trace statistics.
Total fertility rate is positively affecting infant mortality rate. It means that an increasing fertility has a threatening affect on infant survival chances.
The availability of health personnel per capita also affecting infant mortality negatively indicating that an increase in health personnel especially in rural areas, increase the access to health facilities and has a positive affect on infant survival chances.
Public health expenditures are also affecting infant mortality negatively, meaning that increasing public health interventions especially for mother and child care and shifting emphasis from preventive to curative care will bring a shift in infant mortality rate.The sign of unemployment coefficient is a priori and the value is low because in a traditional society like Pakistan working women concept is still premature especially in rural areas where almost 70 percent of the population is residing.
Conclusion: The analysis helps in better resource allocation and target intervention, for better population health, in cost effective manner.
Implications for Policy, Delivery or Practice:
Helps in formulating better policies that help to prevent aviodable deaths of infants by allocating resources where it is needed and secondly also helps to build a better link between improved health status, health spending and growth nexus in the long run.
Theme: Child Health
● Barriers & Consequences of Care Transitions:
Who Dropped the Ball?
Amy Abbott, Ph.D., R.N.; Kevin Fuji, Pharm.D.;
Joan Norris, Ph.D., R.N.; Kimberly Galt, Pharm.D.,
Ph.D.
Presented by: Amy Abbott, Ph.D., R.N., Assistant
Professor, School of Nursing, Creighton University,
2500 California Plaza Office 193A, Omaha, NE
68178, Phone: (402) 280-2055; Email: aabbott@creighton.edu
Research Objective: Care transitions are a set of actions designed to ensure the coordination and continuity of care as patients transfer from: facility to facility, facility to home, or within a facility. A gap exists in what is known about these processes from admission through post-hospital care. The study’s objective was to identify barriers and consequences in care transition processes to inform professionals within the community and to develop strategies aimed at minimizing negative outcomes.
Study Design: Using a mixed methods design, snowball sampling was used to invite professionals from hospital and community facilities, as well as patients to participate in focus groups about their understanding and experiences in care transition processes. Qualitative themes emerged from the focus groups with each participant completing a questionnaire to gather quantitative data.
Population Studied: Professionals from hospital and community facilities included: nurses, pharmacists, physicians, occupational therapists, respiratory therapists, social workers, case managers, and representatives of rehabilitation services. Patients had experiences representing transitions across all levels of care.
Principal Findings: Professionals from four hospitals (n=13), five community facilities (n=6), and patients (n=19) participated in focus groups.
Professionals were in practice for an average of 16 years. The average age of patients was 51 years with 13 of them reporting that they or a family member had been hospitalized within the past year.
Barriers to effective care transitions were identified and categorized as: a) lack of coordination, b) inadequate time and/or staffing, c) decision delays, d) rushed or inadequate communication, e) placement difficulties, f) economic limitations, and g) patient-specific problems. Professionals and patients reported that these barriers led to negative consequences such as readmissions, extended hospital stays, treatment delays, lack of follow-up, poorly informed receiving facilities, and inadequate patient and family education.
Conclusion: The lack of clearly assigned care transition tasks presents a barrier to professionals who share in this responsibility. Similarly, patients do not have the necessary knowledge to integrate themselves into this process. Findings indicate that professionals within and between facilities do not know what each other is doing. This leads to poor coordination and numerous other costly and potentially harmful consequences.
Implications for Policy, Delivery or Practice: The need for a common vision and understanding between professionals, within and between facilities, and with patients and family caregivers is evident. Standards of care need to be developed and practiced consistently. Community-wide education and other strategies must be implemented by all professionals involved in transitioning patients across various levels of care.
This will aid in the reduction of barriers and negative consequences and promote the development and implementation of individualized strategies to facilitate efficient and safe transitioning practices.
Theme: Quality and Efficiency: Measurement
● Medicaid & SCHIP in the U.S. Insular Areas:
Review of Data Available to Assess the
Programs
Rebecca Abela; Linda Kohn; Susan Anthony;
Gerardine Brennan
Presented by: Rebecca Abela, Health Policy
Analyst, U.S. Government Accountability Office,
200 West Adams, Suite 700, Chicago, IL 60606,
Phone: (312) 220-7673; Email: AbelaR@gao.gov
Research Objective: Federal Medicaid and SCHIP funding is available to the five largest U.S. insular areas of American Samoa, the Commonwealth of the Northern Mariana Islands, Guam, Puerto Rico, and the U.S. Virgin Islands. However, Medicaid and
SCHIP operate and are funded differently in the insular areas than in the States. For example,
Federal matching Medicaid funds are capped to the insular areas, and SCHIP allotments are distributed based on different criteria than in the States.
Further, the Medicaid and SCHIP programs of the insular areas differ in terms of their eligibility criteria, coverage goals, and data reporting ability. For example, in some of the areas' SCHIP funds are used to continue health care coverage for children already enrolled in Medicaid. The most recent
SCHIP reauthorization discussion could involve changes to insular areas SCHIP programs.
However, limited knowledge exists about the extent to which data are available to asses the programs and about actual insular area funding needs. Based on congressional interest, this study sought to determine what Medicaid, SCHIP, and related demographic data are available about the areas to inform policy makers.
Study Design: We interviewed officials from the
Centers for Medicare and Medicaid Services, the
Census Bureau, and Medicaid Directors and
Statistics officials from each insular area.
Population Studied: Medicaid and SCHIP programs in five U.S. insular areas.
Principal Findings: At the time the abstract was submitted all findings were preliminary; research will be finalized by March 2009. We expect to report on the extent to which reliable program data, including data on program expenditures and enrollment, are available for the insular areas. We may also report on the extent to which reliable and consistent income and population data, including the data used to determine SCHIP allotments, are available for each insular area.
Conclusion: We may provide conclusions about the types of program and demographic data that are available for policy makers to use when assessing Medicaid and SCHIP in the insular areas.
Implications for Policy, Delivery or Practice: At the time we conducted our work, legislators were considering SCHIP reauthorization, including how to align state needs with federal allotments.
Previous SCHIP reauthorization proposals also included discussion about Medicaid funding in the insular areas. This work could help policy makers understand the data available to inform their assessments of insular area Medicaid and SCHIP programs and considerations for future program funding.
Funding Source(s): GAO
● Effects of Fiscal Incentives on Antidepressant
Adherence
Stephen Able, Ph.D.; Nicole Engel-Nitz, Ph.D.; Yiyu
Fang, Ph.D; Daniel Ball, Dr.P.H.
Presented by: Stephen Able, Ph.D., Associate
Senior Outcomes Scientist, Eli Lilly and Company,
Lilly Corporate Center, D.C. 4123, Indianapolis, IN
46285, Phone: (317) 433-4006; Email: sable@lilly.com
Research Objective: Previous research has shown antidepressant medication adherence is associated with patient characteristics, physician specialty, and practice patterns. Fiscal incentives stemming from health plan benefit designs have also been hypothesized to impact compliance and persistence. In this study, we examined how the fiscal elements of patients’ health plan benefit design affected compliance and persistence.
Study Design: We used a large claims database to conduct a retrospective analysis of compliance and persistence among patients filling at least 1 prescription for antidepressants duloxetine, venlafaxine XR, escitalopram, sertralene, paroxetine, fluoxetine, citalopram, and/or bupropion between 10/1/2004 and 9/30/2005. Compliance was measured as a dichotomous variable with value of 1 if the medication possession ratio (days covered by drug in the index treatment regimen during follow-up period divided by 365) was >80%;
0 otherwise. Persistence was measured as time to discontinuation (45-day gap in treatment with index antidepressant). Predictors of compliance were assessed using logistic regression, while persistence was assessed with Cox Proportional
Hazards model. Benefit design-related variables included in the models included patient co-pay level, formulary tier status, deductible, and benefit cap.
Population Studied: Health plan members age 18-
64 with continuous medical and pharmacy coverage
12 months pre- and post-index date (first recorded prescription date for the index drug) and with a recorded depression diagnosis code within 6 months of the initial index drug prescription were included in the study (n=126,889). Patients with diagnosis codes for bipolar disorder or schizophrenia within 365 days of the index date were excluded.
Principal Findings: After adjusting for patient characteristics, provider specialty, line of therapy and initial dose, the patients’ co-pay was significantly related to compliance (OR=0.993;
95%CI=0.990-0.995); a $10 dollar increase in copay was associated with a 7% decrease in the odds of compliance. Other predictors of lower compliance included African-American race, index drug initiation as second or higher line of therapy or at subtherapeutic dosing levels, psychotropic polypharmacy, and comorbid pain diagnoses.
Predictors of higher compliance included age greater than 45, education beyond high school, income > $100K, and comorbid anxiety. Similar results were obtained in models incorporating formulary tier status rather than co-pay level.
Assessment of persistence found that patients initiating a Tier 3 level medication had a 9% higher hazard of discontinuing their index antidepressant compared with patients initiating a Tier 1 medication
(HR=1.091; 95% CI: 1.025-1.161).
Conclusion: The co-pay level associated with an antidepressant impacts compliance with and persistence to depression treatment. When possible, comparisons of adherence to alternative antidepressant medications should be adjusted for elements of benefit design along with patient characteristics and physician treatment patterns.
Implications for Policy, Delivery or Practice:
Health plan benefit design should consider the potential implications of copayments on medication compliance/persistence, consistent with Institute of
Medicine recommendations that health plans remove barriers to effective and appropriate treatments that may be created by copayments or other coverage policies. As this study demonstrates, a higher copayment level may be associated with lower compliance to antidepressant medications.
Funding Source(s): Eli Lilly and Company
● A Positive Deviance Approach to Improving the Quality of Psychosocial Care for Cancer
Patients
Kathleen Abrahamson, Ph.D., R.N.; Nadine Najjar,
M.S.; Katherine Schilling, M.L.S., Ed.D.; Lori Losee,
M.A.; Caroline Carney-Doebbeling, M.D., M.S.
Presented by: Kathleen Abrahamson, Ph.D., R.N.,
Post Doctoral Research Fellow, Center of
Excellence in Implementing Evidence Based
Practice, Department of Veterans Affairs-Health
Services Research & Development, 1481 West
10th Street, Indianapolis, IN 46202, Phone: (765)
430-8784; Email: kaabraha@iupui.edu
Research Objective: Psychosocial distress is common in cancer patients. Approximately 35-45% of U.S. cancer patients report significant emotional distress. This percentage likely underestimates the
true incidence of cancer related distress. Although common, distress is frequently under-diagnosed and poorly managed in the current U.S. healthcare system. Though practice guidelines for the management of cancer related distress exist, evidence shows these guidelines are rarely followed in practice. This project uses a positive deviance approach to identify those cancer care centers that are able, through organizational innovation, to provide recommended levels of psychosocial support. Positive deviance uncovers successful, yet uncommon, organizational behaviors which are potentially accessible to other providers working within similar resource constrained environments. This approach has not previously been applied to psychosocial care in cancer.
Study Design: Cancer related psychosocial distress influences patient treatment decisions, compliance with treatment, and overall quality of life; therefore it is important to understand how psychosocial services are delivered. Cancer care providers at a variety of organizational levels ranging from administration to direct care will be queried to assess the barriers, resources, and current practices of organizations in the provision of psychosocial care. Structured interviews are scheduled with academic, tertiary, and community cancer clinics and hospitals. Interviews address the availability of mental health professionals, number and type of referrals, reimbursement issues, and current pathways for the delivery of psychosocial care to cancer patients. In order to delve further into organizational processes, the interviews also include questions regarding inter-organizational communication, leadership style, and openness to diverse perspectives within the care delivery setting.
Population Studied: Because innovation arises in a wide variety of organizations, our sample of 40 cancer care centers includes small, regional providers in addition to centers which have been identified as national leaders in cancer care.
Interviews are conducted with three individuals from differing organizational strata within each center.
The unit of analysis for this project is the organization.
Principal Findings: We will present preliminary data on the identification of innovative processes of care which allow for the provision of recommended psychosocial care to a wide range of cancer patients in a cost-effective manner.
Conclusion: Positive deviance identifies those clinical settings that are able to provide for patient needs within resource constrained environments, thus providing practitioners with cost effective organizational strategies which are potentially transferable to wide variety of settings and patients.
Organizational processes identified within the positive deviant centers can be used to design psychosocial care implementation projects in the cancer care setting.
Implications for Policy, Delivery or Practice:
Consequences of not providing psychosocial care to cancer patients are serious and expensive.
Unresolved distress can lead to excess utilization, poor decision making regarding treatment, as well as non-compliance with the recommended treatment plan (non-adherence to, termination of, or never starting treatment). The yet unfulfilled challenge is the discovery of an interventional approach at the organizational level which benefits a heterogeneous cancer population in a cost effective manner. Positive deviance approaches is an important first step in defining successful implementation processes.
Funding Source(s): Regenstrief Foundation
Theme: Quality and Efficiency: Organized
Processes
● Organizational & Market Predictors of
Multidimensional Process Measures of Hospital
Quality: A Multi-Level Analysis
Azza AbuDagga, Ph.D.; Robert Weech-Maldonado,
Ph.D.; Rhonda BeLue, Ph.D.
Presented by: Azza AbuDagga, Ph.D., Graduate
Student, Health Policy & Administration, Penn State
University, University Park, PA 16802, Email: ama220@psu.edu
Research Objective: The robustness of patient assessments of care as measures of health care quality has been established empirically. However, the literature in this area is plagued by several weaknesses: lacking theoretical foundations; being mostly based on simple analyses; being limited to a single or few organizations. These weaknesses are implicated in the inconsistent findings on predictors of patient assessments of care. This study capitalized on the 2002 Picker/NRC survey of
Patients’ Evaluation of Performance in California
(PEP-C) to quantify and partition overall true variations in patient assessments of care into within-hospital, between-hospital, and betweenmarket components and employed predictive models to account for these variations. The conceptual framework for this study derived from
Donabedian’s Structure-Process-Outcome framework, whereby Picker’s multidimensional domains of patient assessments of care – coordination of care, continuity/transition of care, physical comfort, emotional support, information & education, involvement of family & friends, and respect for patient preferences – were conceptualized as measures of process quality.
Study Design: Using PEP-C, American Hospital
Association (AHA) survey, and Area Resource File
(ARF) data, this study examined patient assessments in a sample of 24,887 medical/surgical patients from 173 short-term hospitals in 46 counties in CA. It employed threelevel hierarchical linear models to decompose variations in the domains of process quality across
patients, hospitals, and markets. Patient, organizational, and market predictors were introduced, at their respective levels, in a sequential model-building approach to explain these variations.
Population Studied: The sample represented 80% of counties in CA and accounted for 47% of eligible hospitals, 51% of hospital discharges, and 54% of the licensed beds in the state.
Principal Findings: Variations in all domains of process quality were abound. Strikingly, however, most of the overall variations (95% - 99%) were predominately within-hospitals (i.e., due to differences among patients at the hospital level), rather than between-hospitals or between-markets.
Patient demographic/insurance type characteristics accounted for only up to 13% of variations at the within-hospital level. At most 3.4% of the true overall variations existed between hospitals and a maximum of 1% of these variations were attributable to differences across hospital markets.
Interestingly, despite the relatively small betweenhospital and between-market variations, net of patient characteristics, organizational and market level characteristics predicted a sizable amount of these variations: up to 61% of between-hospital variances and up to 99% of the between-market variances in process quality measures.
Conclusion: The findings of this study suggest that most of the variations in patient assessments of care exist within the patient-health-care-provider relationship. On the other hand, relevant organizational and market predictors need to be accounted for upon making comparisons across hospitals in quality report cards.
Implications for Policy, Delivery or Practice:
This study supports the relevance of patientcentered measures of quality, as opposed to the traditional mortality and morbidity measures, for elucidating sources of variations in patient outcomes across organizations and markets.
Patient-centered measures can be derived from multiple data sources, including reporting systems for patients’ experiences, such as NRC Picker surveys and the Hospital-CAHPS program.
Furthermore, these measures can be readily used to answer policy-specific questions.
● Organizational & Market Predictors of
Multidimensional Process Measures of Hospital
Quality: A Multi-Level Analysis
Azza AbuDagga, Ph.D.; Robert Weech-Maldonado,
Ph.D.; Rhonda BeLue, Ph.D.
Presented by: Azza AbuDagga, Ph.D. Email: ama220@psu.edu
Research Objective: The robustness of patient assessments of care as measures of health care quality has been established empirically. However, the literature in this area is plagued by several weaknesses: lacking theoretical foundations; being mostly based on simple analyses; being limited to a single or few organizations. These weaknesses are implicated in the inconsistent findings on predictors of patient assessments of care. This study capitalized on the 2002 Picker/NRC survey of
Patients’ Evaluation of Performance in California
(PEP-C) to quantify and partition overall true variations in patient assessments of care into within-hospital, between-hospital, and betweenmarket components and employed predictive models to account for these variations. The conceptual framework for this study derived from
Donabedian’s Structure-Process-Outcome framework, whereby Picker’s multidimensional domains of patient assessments of care – coordination of care, continuity/transition of care, physical comfort, emotional support, information & education, involvement of family & friends, and respect for patient preferences – were conceptualized as measures of process quality.
Study Design: Using PEP-C, American Hospital
Association (AHA) survey, and Area Resource File
(ARF) data, this study examined patient assessments in a sample of 24,887 medical/surgical patients from 173 short-term hospitals in 46 counties in CA. It employed threelevel hierarchical linear models to decompose variations in the domains of process quality across patients, hospitals, and markets. Patient, organizational, and market predictors were introduced, at their respective levels, in a sequential model-building approach to explain these variations.
Population Studied: The sample represented 80% of counties in CA and accounted for 47% of eligible hospitals, 51% of hospital discharges, and 54% of the licensed beds in the state.
Principal Findings: Variations in all domains of process quality were abound. Strikingly, however, most of the overall variations (95% - 99%) were predominately within-hospitals (i.e., due to differences among patients at the hospital level), rather than between-hospitals or between-markets.
Patient demographic/insurance type characteristics accounted for only up to 13% of variations at the within-hospital level. At most 3.4% of the true overall variations existed between hospitals and a maximum of 1% of these variations were attributable to differences across hospital markets.
Interestingly, despite the relatively small betweenhospital and between-market variations, net of patient characteristics, organizational and market level characteristics predicted a sizable amount of these variations: up to 61% of between-hospital variances and up to 99% of the between-market variances in process quality measures.
Conclusion: The findings of this study suggest that most of the variations in patient assessments of care exist within the patient-health-care-provider relationship. On the other hand, relevant organizational and market predictors need to be
accounted for upon making comparisons across hospitals in quality report cards.
Implications for Policy, Delivery or Practice:
This study supports the relevance of patientcentered measures of quality, as opposed to the traditional mortality and morbidity measures, for elucidating sources of variations in patient outcomes across organizations and markets.
Patient-centered measures can be derived from multiple data sources, including reporting systems for patients’ experiences, such as NRC Picker surveys and the Hospital-CAHPS program.
Furthermore, these measures can be readily used to answer policy-specific questions.
Theme: Quality and Efficiency: Measurement
● School Beverage Vending Machines: Content
& Advertising
Anna Adachi-Mejia, Ph.D.; Meghan Longacre,
Ph.D.; Linda Titus-Ernstoff, Ph.D.; Michael Beach,
Ph.D.; Susan Martin, B.S.; Madeline Dalton, Ph.D.
Presented by: Anna Adachi-Mejia, Ph.D.,
Research Assistant Professor, Pediatrics,
Dartmouth Medical School, HB 7465, One Medical
Center Drive, Lebanon, NH 03756, Phone: (603)
653-0789; Email: anna.adachimejia@dartmouth.edu
Research Objective: Many studies have established a strong positive association between sugar-sweetened beverage consumption and bodymass index among children. Studies have also established a link between television advertising and youth consumption of advertised foods. The purpose of this study was to describe beverage vending machine content and advertising present on school vending machines.
Study Design: We enrolled 32 public schools in
Vermont and New Hampshire. Between October
2007 and May 2008, trained coders counted the number of beverage vending machines in each school, and described content (number of slots, type and quantity of beverages available) and advertising on school vending machines.
Population Studied: Schools in predominantly suburban and rural towns with grades ranging from
K-12. Most schools (80.5%) were high schools
(grades 9-12 only).
Principal Findings: All but one school had a vending machine. On average, there were 4.6 beverage vending machines per school. There was a positive correlation between the number of vending machines and the number of students enrolled. Coders found 149 vending machines, of which three were empty and 23 did not contain beverages, leaving a sample size of 123 beverage vending machines. The three vending machines containing soda and the nine containing diet soda were located in high schools. Other types of beverages (sugar-sweetened, other diet, plain water, flavored water, juice, flavored and unflavored milk) were present in vending machines at all grade levels. The most frequently occurring beverages overall were flavored water (76.4% of machines) and plain water (67.5%). A little over half of the vending machines contained sugar-sweetened beverages (57.7%), and fewer machines contained juice (35%) or milk. Most (83.7%) of vending machines featured advertisements. The most common advertisements were for Dasani (51.6%).
Every school had at least one advertisement on their beverage vending machines. Advertisements for sugar-sweetened beverages appeared on onequarter of the vending machines (25.2%).
Conclusion: Water is predominantly stocked in school vending machines. Although soda appears to be scarce in school vending machines, sugarsweetened beverages are stocked in vending machines at all grade levels. School vending machines feature beverage advertisements.
Implications for Policy, Delivery or Practice:
Policymakers should be aware that even though schools have considerable amounts of water and less soda in their vending machines, sugarsweetened beverages are still finding their way into the schools. Students have the potential of beverage advertising exposure through vending machines.
Funding Source(s): RWJF
● The Effect of Breast & Cervical Cancer
Prevention & Treatment Act (BCCPTA) on
Timing of Disenrollment from Medicaid in
Georgia
E. Kathleen Adams, Ph.D.; Li-Nien, Chien, M.S.,
M.P.H.
Presented by: E. Kathleen Adams, Ph.D.,
Professor, Health Policy & Management, Emory
University, 1518 Clifton Road Northeast, Atlanta,
GA 30322, Phone: (404) 727-9198; Email: eadam01@sph.emory.edu
Research Objective: To examine the effect of the
Breast and Cervical Cancer Prevention and
Treatment Act (BCCPTA) on timing of disenrollment from Medicaid.
Study Design: The BCCPTA allowed states to extend Medicaid coverage to uninsured women who are screened and diagnosed with breast and/or cervical cancer under the National Breast and
Cervical Cancer Early Detection Program site and found in need of treatment. The State of Georgia chose to cover women regardless of their screening site starting in July 2001. Data from the Georgia
Comprehensive Cancer Registry (GCCR) from
1999 to 2004 linked to Medicaid enrollment files were used to identify new breast (N=2,840), cervical
(N=558) and five control cancer (colorectal, bladder, melanomas of the skin, non Hodgkin lymphomas and thyroid, N=926) Medicaid enrollees. The incidence rate of disenrollment, measured from the month of enrollment in Medicaid to the month of
disenrollment from Medicaid was compared pre and post BCCPTA. To account for unobserved differences among subjects between pre and post
BCCPTA, the difference-in-difference approach was employed. Since the time-to-disenrollment varied for each subject and was not observed for some, the hazard function was used to count for right censoring.
Population Studied: New cases of breast, cervical and five control cancers in Georgia enrolled in
Medicaid up to 3 months before or after their cancer diagnosis as found in the GCCR.
Principal Findings: The incidence rate of disenrollment (per 100 person-month) for breast, cervical and control cancers enrollees were 3.17,
5.14 and 5.33 in the pre BCCPTA period and 1.59,
2.85 and 6.84 in the post BCCPTA period. The differences between pre and post BCCPTA decreased 50% for women with breast and cervical cancers while it increased 30% for women with control cancers. The difference-in-difference estimation of the net effect of BCCPTA indicated that the likelihood of disenrolling from Medicaid for breast and cervical cancer cases decreased significantly (Hazard Ratio=.50, p<.001 and Hazard
Ratio=.60, p<.01 respectively). The mean estimated marginal effects of BCCPTA indicated that the Act decreased the rate of disenrollment of women with breast cancer by 1.9 of 100/month and by 2.4 of
100/month for cervical cancer cases.
Conclusion: The Georgia BCCPTA program significantly decreased the probability of women with breast and cervical cancer disenrolling from
Medicaid. The data from Georgia is limited since it is from a single state and cannot be generalized while the effect from Georgia is stronger because it covered more women than states with more restricted coverage.
Implications for Policy, Delivery or Practice:
Medicaid is regarded as a safety net for those lowincome patients. It is especially important for cancer patients to remain enrolled as they need time to complete their treatment. A stable insurance coverage leads to more successful care and better survival. The BCCPTA is a policy approach for addressing the challenges of the uninsured facing serious illness. The BCCPTA effect from Georgia has been shown how well the BCCPTA program works to improve the health of those uninsured women with breast and/or cervical cancers. This legislation also represents a potential model for further Medicaid expansions to uninsured people with other cancers.
Funding Source(s): American Cancer Society
(ACS)
Theme: Medicaid, SCHIP and State Health Reform
● A Qualitative Analysis of Bathing among
Older Persons
Sangeeta Ahluwalia, Ph.D.; Thomas Gill, M.D.;
Dorothy Baker, Ph.D.; Terri Fried, M.D.
Presented by: Sangeeta Ahluwalia, Ph.D.,
Postdoctoral Fellow, Section on Geriatrics, Yale
School of Medicine, 20 York Street, New Haven, CT
06511, Phone: (510) 673-7531; Email: sangeeta.ahluwalia@yale.edu
Research Objective: Among older persons, bathing disability is highly prevalent and associated with several adverse outcomes, including long-term nursing home admission and home aide services.
Because bathing is a complex and highly personal self-care task, effective interventions to forestall or prevent disability will depend on a robust understanding of older persons’ perspectives of their bathing experiences. We sought to describe the bathing experiences, preferences and goals of older persons.
Study Design: Qualitative cross-sectional study of
23 community-living persons aged 78 and older with a range of bathing ability. In-depth, semistructured interviews were conducted in the participant’s home. Questions addressed the meaning and purpose of bathing, frequency of bathing, use and preferences for different types of bath aids, concerns about bathing, thoughts about future changes in bathing, and wishes about bathing.
Population Studied: Community-dwelling older persons
Principal Findings: Four key themes emerged that illustrate the complexity in identifying and characterizing bathing disability and in older persons’ responses to bathing disability. These included: 1) traditional quantitative assessments may not capture the full extent of an individual’s level of disability; 2) older persons often receive and use bath aids through self-assessment rather than through a systematic, formal assessment of bathing need; 3) the preferences for assistance with bathing vary considerably according to older persons’ distinct goals regarding bathing; and 4) older persons adjust to disability by accepting a restricted form of bathing and demonstrate an inability to envisage how their bathing function could be preserved.
Conclusion: Our results underscore the limitations of current strategies to assess bathing disability and highlight several challenges to developing effective interventions. Asking about modifications in bathing may enhance disability assessment and, in turn, facilitate the identification of at risk older persons who might benefit from early intervention. In addition, resetting older persons’ expectations about future decline may help them work towards maintaining or improving their bathing ability.
Implications for Policy, Delivery or Practice:
Bathing interventions must incorporate older persons' preferences for and attitude towards independence in order to be successful. Providers should work with older persons to identify concrete steps and goals around bathing in order to help individuals envision the possibility of forestalling
disability and work towards achieving functional independence.
Finally, quantitative disability assessments should be enhanced to identify task modification among older persons characterized as independent bathers who may actually have underlying difficulty or disability.
Funding Source(s): NIA
Theme: Long Term Care
● Health, Illness & Healing: The Female Veteran
Experience
Sarah Aktepy, B.S., L.P.N.
Presented by: Sarah Aktepy, B.S., L.P.N.,
Research Assistant, Department of
Sociology/Health Services Research &
Development, Indiana University Purdue University
Indianapolis/Roudebush VA Medical Center, 425
University Boulevard, Indianapolis, IN 46202,
Phone: (757) 773-6409; Email: saktepy@iupui.edu
Research Objective: This research project’s main objective is to use qualitative interviews to understand the experiences of women veterans who are currently utilizing services within the VA system from their own perspectives and the meanings they associate with their experiences.
Also, to determine what VA service providers think are the most important needs of the target population. The primary research questions are:
What are the needs and experiences described by women veterans? What are VA service provider’s perceptions about the health care, counseling, and other related service needs of OEF/OIF women veterans? To address these questions, this project has four specific aims: 1.to identify what the health care, counseling, and other related service needs are of women veterans and describe their experiences while utilizing VA services, 2.to understand and describe how women veterans returning form Iraq and Afghanistan perceive cultural expectations and ideologies surrounding the military and the VA system, and 3.to explore VA service provider’s perceptions of what women veterans needs are and what strengths and barriers exist in utilizing these services.
Study Design: Private, in-person semi-structured interviews will be conducted. All interviews will be audio recorded and transcribed. Transcriptions will be completed within one week of conducting the interviews. Sentence by sentence coding will be used to determine emergent themes. Themes and gathered information from interviews will be used to guide the interviews that follow, allowing emergent themes to direct interview questions. Data analyses of generated interviews will be done using
Kathy Charmaz’s grounded theory techniques.
Memos that I develop from the transcribed data will be used to do a complete analysis. Female veteran’s experiences will be reported in the findings section of the project and supported with narratives. Additionally, the reported service needs will be presented in question and answer format with a recommendations section based on study findings. These methods were chosen due to the exploratory nature of the pilot study and to fully capture the lived experience of the target population.
Population Studied: This study will enroll a total of
20 participants. The project will include 15
OEF/OIF women veterans and 5 key informant VA service providers. The qualitative interviews will be audio recorded and last 1-2 hours. They will consist of: Key informants that are VA service providers recommended by OEF/OIF women veterans, and are providers within the VA system that mainly serve the women veteran population.
Women OEF/OIF veterans, varying in age, race/ethnicity, and level of education, occupation, service-connected disabled or non-service connected disabled, deployed or non-deployed history, but have utilized services within the VA system. The first phase of the study will include interviews of all 15 OEF/OIF women veterans.
Selection of 5 key informants will be based on data gathered from first phase interviews. All key informant interviews will be completed in the second phase of this study. Phase 2 of this study will focus on what VA services key informants perceive OEF/OIF women veterans need and their recommendations for improving the VA system serving the target population.
Principal Findings: This project is a thesis study currently in progress. This study has received IRB approval from Indiana Univeristy Purdue Univeristy
Indianapolis and R&D approval from the
Roudebush VAMC in Indianapolis. Currently, interviews are being conducted. This study will be completed prior to the conference.
Conclusion: This research project aims to contribute to sociology of gender and health by examining the impact of military culture on the experiences of women who seek health care, counseling, and other related services within the VA system. According to the Department of Veterans
Affairs’ Office of Policy and Planning, women make up approximately 5% of VA patients and will comprise nearly 10% of the total veteran user population by the year 2010. Currently, the VA system offers a continuum of services for male and female veterans. However, women returning from
Iraq and Afghanistan are showing health conditions sequel to combat exposure and other types of trauma, not identified in previous conflicts.
Implications for Policy, Delivery or Practice:
Responses generated by in-depth interviews, is critical in understanding what the unique needs are of the fastest growing segment within the veteran population. The research is designed to empower women by affording them the opportunity to reflect on their experiences and determine their needs in the context within their own lives as women veterans. The knowledge gained from this study
may help others and may improve VA services in general. In addition, the results of this study will help the VA provide the best possible services for women veterans. This study promises to contribute to theories about how gender and health employ the construction of identity in the face of war, illness, and a male dominant institution.
Theme: Military and Veterans Health Care
● Self Perceived Quality of Life (SP-QOL) in an
Impoverished Mexican-American Community in
Texas along the U.S.-Mexican Border
Sartaj Alam, M.S.; Belinda Reininger, Dr.P.H.;
Mohammad Rahbar, Ph. D.
Presented by: Sartaj Alam, M.S., Graduate
Research Assistant, Biostatistics, University of
Texas School of Public Health., 1200 Herman
Presslar Street, Houston, TX 77030, Phone: (713)
500-7907; Email: Sartaj.Alam@uth.tmc.edu
Research Objective: This study examined the SP-
QOL in an impoverished, minority-majority neighborhood in the Lower Rio Grande Valley. By exploring the association between life status, perceptions, and QOL outcomes, the study aim was to inform program planning, outreach activities and community-based participatory action research and suggest interventions. . In addition to scholars of
Hispanic health and QOL, the results of this study should prove useful to policymakers and the advocacy agencies and academic institutions that promote development, implementation, and analysis.
Study Design: A simple random sample of 402 households were administered a survey via a trained outreach worker. Participants were able to conduct the interview in their language of choice, with over 95% selecting Spanish. The neighborhood represented a defined geographical area containing a defined population of interest, with every household having a known, nonzero, and equal chance of being included in the sample. The response rate was 80% (320/402). All data collection protocols had prior approval from the
Committee for the Protection of Human Subjects
(CPHS) of the University of Texas.
The survey instrument, based on the work of M.J.
Sirgy et al, assessed levels of satisfaction with various life domains and specific local government, non-profit services and neighborhood social capital.
SP-QOL measures were assessed using the
“Delighted-Terrible” scale and were treated as continuous variables.
Population Studied: Lower Rio Grande Valley in
Texas along the U.S.-Mexican Border.
Principal Findings: Linear regression procedures developed statistical models including subjective and objective variables. Demographic variables like
Age and Sex were not significant statistically and were kept in the model as confounding variables.
Subjective factors which were more effective at predicting outcome and significantly contributed positively to the model included: satisfaction with
‘Personal Health’ (p<.001), ‘Community’ (p<.001),
‘Friends’ (p<.001), and ‘Education’ (p<.05).
Therefore, participants who reported an increase in satisfaction in any of these variables also reported greater SP-QOL. Similarly, objective variables which positively contributed significantly are Health
(p<0.001), Education Level (p<0.01), and
Opportunities to Exercise/ recreation (p<0.01).
None of the factors contributed negatively towards the outcome. One of the most conspicuous results was the strong interaction between subjective and objective terms. Education Level strongly modified the effect (interaction effect) of ‘Satisfaction with
Education’ in its relationship to the outcome
(p<0.05). ‘Satisfaction with Friends’ also showed strong interaction with ‘Opportunities to Exercise /
Recreation’ vis-à-vis SP-QOL (p<.001).
Conclusion: Analysis of the survey results showed subjective outcomes such as SP-QOL and other terms based on perceptions do not always match up with objective life status. The predominantly
Mexican-American participants expressed generally higher levels of SP-QOL with various life domains than would seem justified by their objective health situation and living conditions.
Implications for Policy, Delivery or Practice: By exploring the association between life status, perceptions, and QOL outcomes, the study aim was to inform program planning, outreach activities and community-based participatory action research and suggest interventions.
Funding Source(s): US Department of Health and
Human Services.
● Redesigning Health Care Processes Through
Focused Clinics: Comparing Initiatives in
Belgium
Veronica Alas, M.P.H., M.A.
Presented by: Veronica Alas, M.P.H., M.A., Ph.D.
Candidate, The Heller School for Social Policy &
Management, Brandeis University, 415 South
Street, Waltham, MA 02453, Phone: (707) 953-
0615; Email: gvalas@brandeis.edu
Research Objective: Escalating health care costs and overuse of technology is motivating Belgium to seek ways in which to redesign their hospitals to improve quality and reduce inefficiencies. Focused or specialized care is an approach for improving health care performance due to the high volume and repetition of performing procedures. Yet little is known about how hospitals that use focus are organized and the processes that are used to deliver care. This study seeks to explore different types of focus, or specialized care, for patients undergoing unilateral primary hip arthroplasty surgery in three Belgian hospitals. In particular, the research objectives are to: 1) identify the characteristics of the organization design and the
processes used to focus services for patients undergoing unilateral primary hip arthroplasty surgery in hospitals; and 2) identify different types of focus strategies used for patients undergoing unilateral primary hip arthroplasty surgery in hospitals.
Study Design: This study utilized a comparative case study design to explore focused care in three
Belgian hospitals performing unilateral primary hip arthroplasty surgery. Data collection techniques included: semi-structured interviews with clinical providers, document review, and non-participant observations. Two approaches, pattern-matching and cross-case synthesis, were used for the analysis of the data guided by three conceptual frameworks from organization management and strategy: the focused factory, strategic service vision and the service triangle.
Population Studied: Clinical providers caring for patients undergoing unilateral primary hip arthroplasty surgery in hospitals and administrators
(n=20) were included in the sample. These providers were experts and key stakeholders in the hip arthroplasty care process.
Principal Findings: Findings for this study are pending and will be available prior to the start of the
Annual Research Meeting on June 28, 2009.
Conclusion: Conclusions for this study are pending and will be available prior to the start of the Annual
Research Meeting on June 28, 2009.
Implications for Policy, Delivery or Practice:
Focus is an approach for redesigning the delivery of care in hospitals. The findings from this study will provide greater understanding of different models of focus in health care. It will offer insight into ways in which hospitals organize their health care delivery by using focus, and may be a means by which other hospitals can do the same to enhance their competitive advantage in the health care market.
● ADRCs & Single Entry Points: A Growing
National Trend in Long Term Care
Lisa Alecxih, M.P.A.
Presented by: Lisa Alecxih, M.P.A., Vice
President, The Lewin Group, 3130 Fairview Park
Drive, Falls Church, VA 22042, Phone: (703) 269-
5542; Email: lisa.alecxih@lewin.com
Research Objective: In 2003, AoA and CMS launched the Aging and Disability Resource Center
(ADRC) grant initiative. The ADRC program seeks to empower individuals to make informed choices about long-term support options and to streamline access for consumers to long-term support services through Single Entry Point and/or No Wrong Door systems. ADRCs work with partners to streamline all the critical functions involved in accessing public programs so that consumers and their families receive needed support in the most simple, timely, and efficient manner as possible. These critical functions include preliminary eligibility screening, comprehensive medical or functional assessment, programmatic and financial eligibility determination, and ultimately service delivery. This study sought to document the range of program activity over the demonstration period (2003-2008), identify the preliminary outcomes, success and challenges of the ADRC program in its first five years, and to measure the progress of the grantees toward streamlining access to long term services and supports.
Study Design: Researchers made site visits to
ADRC sites in 26 states, conducted telephone interviews, reviewed program materials, and analyzed qualitative and quantitative data reported by ADRC grantees to their federal funders to discover the successes and challenges they experienced in designing and implementing their
ADRCs.
Population Studied: The ADRC programs included in this study are state agencies and local organizations that assist aging adults, individuals with disabilities and their families to access long term services and supports.
Principal Findings: After five years, ADRC programs have made progress toward achieving three broad goals of streamlining consumer access to long term supports and services: 1) improving the ease with which people with disabilities and older adults access information and assistance; 2) helping consumers make informed decisions about services through the provision of options counseling; and 3) improving the efficiency and timeliness of the application process for public programs. They have used a variety of strategies to achieve these goals, such as coordination and standardization of processes across local organizations involved in the long term service system, cross-training staff across partnering organizations, providing online access to program applications, developing web-based searchable resource databases, shortening or simplifying application forms for public programs, co-locating
Medicaid staff with I&R staff and options counselors, and integrating management information systems.
Conclusion: Over the course of this five year demonstration, ADRCs across the country have developed different models, including highly centralized Single Entry Point models and "No
Wrong Door" (network or coalition) models to accomplish the following outcomes: Better coordination of aging and disability service systems at the state and local levels; Raised visibility among consumers and professionals about the full range of options that are available in the community; Raised the professional standard for providing objective information, assistance, and long term service options counseling; Empowered people to make informed decisions about their long term supports;
Served as convenient entry points for all public and private long term-care programs and support services.
Implications for Policy, Delivery or Practice:
Long term care policy makers should recognize
ADRCs and other Single Entry Point models as a growing national trend. As the grant period in most state ends, ADRC programs continue to grow and expand across the country. As of January 2009, there are over 180 programs operating in 48 states.
24 of the original grantee states have allocated state funding and/or passed legislation calling for the continuation or expansion of their ADRC programs. Several states that never received federal funding for ADRCs have now implemented programs (including Connecticut, New York,
Deleware). The ADRC program provides states with an opportunity to effectively maximize and use their long term support resources for providers and consumers in a single coordinated serviced delivery system, but strong partnerships across agencies and networks at the state and local levels are essential to making the ADRC program work. In order to effectively streamline access to the long term care system, partnerships between the aging and disability service networks and between these networks and Medicaid are critical.
Funding Source(s): Administration on Aging
Theme: Long Term Care
● The Experience of Diagnosis & Treatment of
Localized Prostate Cancer among Gay Men
Don Allensworth-Davies, M.Sc.; Jack Clark, Ph.D.
Presented by: Don Allensworth-Davies, M.Sc.,
Research Manager, Data Coordinating Center,
Boston University School of Public Health, 801
Massachusetts Avenue, Crosstown-3, Boston, MA
02118, Phone: (617) 638-5816; Email: ddavies7@bu.edu
Research Objective: Increasingly research demonstrates that prostate cancer and its treatment affects men in multiple ways. Urinary, bowel, and sexual side effects of treatment may affect daily activities, intimate relationships, and self image, while the uncertainties of treatment effectiveness may cause misgivings about cancer control, treatment decisions, and regret. However, recent research has largely ignored the role of sexual orientation in shaping men’s experiences, from interactions with doctors when deciding a course of treatment to the challenges faced in long-term survival. The purpose of this qualitative study is to explore the experiences of gay prostate cancer survivors in order to develop a quality of life conceptual framework inclusive of this population.
Study Design: Open-ended, in-depth interviews conducted in-person and by phone with 24 gay and
3 straight men treated for localized prostate cancer.
Men are sampled based on age, race, and partnership status to provide initial contrasts; 3 straight men are included to help identify distinctive features of the gay prostate cancer experience. All interviews are recorded, transcribed and coded collaboratively to ensure consistency. Men were asked to describe their experiences with prostate cancer diagnosis, treatment and healthcare providers. Analysis focuses on constant comparison and frequency of themes across groups or within a particular group using a grounded theory approach and a constructivist perspective.
Population Studied: All men are age 50 or older, reside in the United States and have received prostate cancer treatment at least 12-months previously.
Principal Findings: Gay men may face unique challenges relating to 1) disclosing one’s sexual orientation to health care providers and 2) support during post-treatment recovery. Gay men may feel a lack of control in the disclosure of their sexual orientation and concern that their orientation may be shared with their providers without their knowledge, and to their disadvantage. Partnered gay men also report greater emotional support and help with everyday tasks during treatment recovery than unpartnered gay men.
Conclusion: These results suggest that gay patients may face an added burden in deciding whether or not to reveal their sexual orientation, risking either homophobia from their healthcare providers or the sharing of their orientation without their knowledge. There are also few gay community-based support groups for gay men recovering from prostate cancer; many gay men are reluctant to attend other support groups for fear of isolation and a lack of understanding from other group members. Partnered gay men benefit from having an alternative source of support in their relationships.
Implications for Policy, Delivery or Practice:
Misgivings about disclosure and withholding of information about sexual identity may inhibit open and candid discussion of treatment preferences.
This is important because: 1) shared decision making may lead to better adjustment to treatment outcomes; and 2) clear communication may avoid inappropriate treatment recommendations. Finally, there is a need for gay community-based support groups for prostate cancer. While some organizations have begun to create support groups for gay men, such groups are still not available in many areas of the country.
Theme: Gender & Health
● Using Capacity Planning Models to Estimate
Organizational Costs of HIV Care Within the
United States Department of Veterans Affairs
Henry Anaya, Ph.D.; Matthew Goetz, M.D.; Uday
Karmarkar, Ph.D.; Steven Asch, M.D., M.P.H.
Presented by: Henry Anaya, Ph.D., Health
Research Scientist, Center for the Study of
Healthcare Provider Behavior, VA Greater Los
Angeles Health Services Research & Development
Center of Excellence, U.S. Department of Veterans
Affairs, 11301 Wilshire Boulevard, Los Angeles, CA
90073, Phone: (310) 478-3711 ext. 48488; Email: henry.anaya@va.gov
Research Objective: Organizations must have the ability to estimate costs associated with current and future actions. For healthcare organizations, factors affecting costs are especially complex and multifaceted, due to issues such as patient flow, staffing levels, lab costs, etc. Recognizing the costs associated with specific outcomes assists managers in controlling costs associated with specific conditions. To perform an evaluation of the organizational costs associated with increases in
HIV screening rates, based on expertly-chosen criterion. As policy-driven researchers, it is not enough that we substantiate that a given HIV quality improvement research project is effective; it must also be elucidated clearly to non-research facility managers for whom costs associated with proposed changes are paramount. Business case modeling provides us that link.
Study Design: Models were constructed using MS
Excel. Inputs were derived from providers familiar with patient care factors. We estimated first-year costs in specific categories under two scenarios: 1) increasing screening rates from 0-5%, 2) increasing from 0-10%. Endpoints included antiretroviral and drug costs, patient flow estimates, testing/treating
HIV patients, changes in staff, pharmacy and lab costs.
Principal Findings: For HIV caseload increases, facilities incur a variety of quarter-to-quarter expenses. Expenses fluctuate over time, depending on the category. Explanations for these differences are provided.
Conclusion: Business case modeling is an effective, powerful (and potentially simple) method for analyzing the effects of contemporaneous policy decisions on future organizational impacts, fiscal and otherwise.
Implications for Policy, Delivery or Practice: The effectiveness and use of cost modeling has important implications for estimating the costs and organizational impacts associated with changes in care delivery.
Funding Source(s): VA
● Qualitative Assessment of Implementing
Routine Rapid HIV Testing
Henry Anaya, Ph.D.; Barbara Bokhour, Ph.D.;
Jamie Feld, B.A.; Joya Golden, B.A.; Herschel
Knapp, M.S.S.W., Ph.D.
Presented by: Henry Anaya, Ph.D., Health
Research Scientist, Center for the Study of
Healthcare Provider Behavior, VA Greater Los
Angeles Health Services Research & Development
Center of Excellence, U.S. Department of Veterans
Affairs, 11301 Wilshire Boulevard Mail Code: 111G,
Los Angeles, CA 90073, Phone: (310) 478-3711 ext. 48488; Email: Henry.Anaya@va.gov
Research Objective: Translating health services research findings into sustainable practice poses widely recognized implementation challenges.
Through a randomized controlled trial (RCT) we found that nurse-based HIV rapid testing (NRT) is more acceptable to patients than current venipuncture and increases receipt of test results.
On the basis of these findings we implemented
NRT at two VAMCs within the VA Greater Los
Angeles Healthcare System (GLA). To ascertain the effectiveness of the implementation, we conducted both formative and process evaluations.
Study Design: Staff at 2 VAMCs were trained to administer RT. Prior to implementation at site 1, we conducted a formative evaluation. At site 2 we conducted a process evaluation of ongoing implementation activities. For both evaluations we conducted semi-structured qualitative interviews of pre-identified key informants. In addition, we employed modified snowball sampling, resulting in
9 manager and 24 front-line practitioner interviews.
Field notes were analyzed for qualitative themes.
Principal Findings: Distinct themes emerged regarding barriers/facilitators to NRT.
Barriers included: clinical workload/staffing as insufficient for uptake of routine versus risk-based testing, lack of congruence with perceived responsibilities and roles in administrating NRT, bureaucratic delay for inclusion of NRT in nursing scope of practice, MD preference for blood draws.
Facilitators included: tailored staff trainings, specific to departmental mission/logistics, patient education activities/publicity, identification of local champions dedicated to NRT, potential expansion of RT training to include LVNs/LVPs.
Conclusion: Formative findings indicate staff concern regarding adequacy of training and incorporating NRT into workflow. Process findings indicate 1) concerns over training could be alleviated, but that workload/staffing concerns remained a barrier 2) expanding training to include
LVNs/LVPs may mitigate this constraint 3) postimplementation administration of NRT was highly variable by provider motivation and service area.
Community care, substance abuse and walk-in were identified as preferred clinics for NRT.
Implications for Policy, Delivery or Practice:
These qualitative findings regarding NRT can be used as a guide to implementing future routine HIV testing activities in accordance with CDC recommendations. Formative and process evaluations in health services research can reveal unforeseen barriers/facilitators and aid in dissemination of research findings.
Funding Source(s): VA
● Implementing an HIV Rapid Testing
Intervention for Homeless Veterans in Shelter
Settings Within Los Angeles County
Henry Anaya, Ph.D.; Jamie, Feld, B.A.; Tuyen
Hoang, Ph.D.; Herschel Knapp, M.S.S.W., Ph.D.;
Steven Asch, M.D., M.P.H.
Presented by: Henry Anaya, Ph.D., Health
Research Scientist, Health Services Research &
Development,Quality Enhancement Research
Initiative- HIV/ Hepatitis, U.S. Department of
Veterans Affairs, 11301 Wilshire Boulevard Mail
Code: 111G, Los Angeles, CA 90073, Phone: (310)
478-3711 ext. 48488; Email: henry.anaya@va.gov
Research Objective: Veterans are overrepresented among the homeless, a group at increased risk for HIV. Many homeless veterans do not routinely access the VA for healthcare; strategies for linking homeless veterans to HIV screening, testing and other health services are needed. To implement and evaluate an HIV rapid testing intervention in collaboration with the US
Department of Veteran’s Affairs and the Los
Angeles County Department of Health to 1) increase HIV testing and receipt of results, and 2) increase veteran access to VA homeless services.
Study Design: Ninety-seven veterans recruited in 9 shelters as part of the Los Angeles Homeless
Services Authority (LAHSA) emergency shelter program were randomized (2-to-1 ratio) into one of two interventional arms: Rapid test (RT) arm (onsite HIV rapid testing/referral to VA homeless program) and referral arm (VA referral only).
Endpoints included 1) rates of HIV testing and receipt of results and 2) rate of visit to VA homeless program. Eligibility criteria included: past military service, no previous HIV testing, past year, no knowledge of HIV status.
Principal Findings: Recruiters approached 2664 individuals; 136 (5.1%) were eligible. Ninety-seven
(71.3%) eligible patients accepted enrollment.
Testing rates were 100.0% in RT arm and 3.3% in
Referral arm (p<0.0001). Test result receipt rates were 98.5% in RT arm and 0.0% in Referral arm
(p<0.0001). There was no increase in visits to VA homeless programs in either arm. Prevalence/ rate of new incidence was 1.6%.
Conclusion: On-site rapid testing in homeless shelters is feasible and acceptable for testing hardto-reach and vulnerable populations. More robust outreach efforts are needed to improve linkage to care for other services.
Implications for Policy, Delivery or Practice:
Policymakers should consider HIV rapid testing at shelter sites when implementing routine HIV testing.
Funding Source(s): VA
● Assessing Organizational Readiness To
Implement Nurse-Based HIV Rapid Testing In
SUD Clinics
Henry Anaya, Ph.D.; Hildi Hagedorn, Ph.D.; S.
Randal Henry, Dr.P.H., M.P.H.; Hana Horns, B.A.;
Jamie Feld, B.A.; Joya Golden, B.A.
Presented by: Henry Anaya, Ph.D., Health Science
Researcher, Health Services Research and
Development, Quality Enhancement Research
Initiative, HIV/AIDS - Hepatitis, U.S. Department of
Veterans Affairs, 11301 Wilshire Boulevard Mail
Code: 111G, Los Angeles, CA 90073, Phone: (310)
478-3711 ext. 48488; Email: Henry.Anaya@va.gov
Research Objective: Despite the benefits of early
HIV identification, many persons at high risk for
HIV-infection have not been tested. Because substance use increases HIV transmission risk, substance use disorder (SUD) clinics are an important location for identifying HIV-infected patients. A developmental formative evaluation was conducted to guide refinement of an implementation strategy for Nurse-initiated HIV Rapid Testing
(NRT) at Veterans Administration (VA) SUD clinics.
Study Design: SUD clinic staff at the Minneapolis
VA Medical Center and the VA Greater Los Angeles
Healthcare Center (VA GLA) completed a survey to assess: 1) staff's perception of the evidence for
NRT testing in SUD clinics, and 2) organizational readiness to change. Questions were drawn from the Sales' Organizational Readiness for Evidencebased Health Care Interventions survey and the
Organizational Climate subscales of the Texas
Christian University Organizational Readiness for
Change scale. A subset of staff also completed semi-structured interviews assessing training needs, barriers and facilitators to implementation of
NRT.
Population Studied: Twenty-one staff members from the Minneapolis VA and 29 staff members from the VA GLA clinics completed the survey.
Survey participants were primarily RNs/LPNs/LVNs
(43.8%) and social workers (31.3%). Psychiatrists, psychologists, and addiction therapists were also represented. Eleven Minneapolis staff and 17 LA staff completed interviews.
Principal Findings: Staff at both sites agreed that
NRT in SUD clinics is supported by research evidence, clinical experience, and patient needs.
Minneapolis staff members were significantly more likely than LA staff to agree that NRT fits within the mission of SUD clinics. On Organizational Climate subscales, LA staff members scored significantly higher on scales assessing the shared understanding of the clinic mission, the perceived openness of communication channels, and a positive attitude toward new technology compared to Minneapolis staff members. LA staff also scored significantly lower on job related stress. Semistructured interviews of staff indicated that while staff support the idea of routine NRT, major barriers include limited resources (time and staff), staff resistance to taking on additional tasks, anxiety about communicating positive test results, and concerns about patient privacy and safety. Staff suggested that NRT could be facilitated by: 1) incorporating NRT into existing staff responsibilities,
2) training that emphasizes communication skills, 3) a private setting for pre-/post-test counseling, and
4) on-site/on-call mental health specialist to assist with delivering positive results.
Conclusion: Despite significant barriers, a majority of SUD staff expressed support for NRT and are willing to consider adopting NRT if supported by the organizational structure, appropriate policy modifications, and specific clinical practices.
Organizational context results indicate that LA clinics are more organizationally inclined to adopt and implement new practices and that the
Minneapolis clinic may require more extensive external facilitation to successfully integrate NRT into clinic procedures.
Implications for Policy, Delivery or Practice:
Successful implementation of NRT in SUD clinics would likely result in increased identification of HIVinfected patients, allowing them to access beneficial treatments. The results of this project highlight how developmental formative evaluation can contribute to the design of implementation strategies which address individual clinic needs and barriers.
Funding Source(s): VA
● Why Do Physician Group Practices Engage in
Quality Performance Measurements?
Douglas Anderson, M.H.A.
Presented by: Douglas Anderson, M.H.A., Student,
School of Public Policy, George Mason University,
4223 32nd Road South, Arlington, VA 22206,
Phone: (703) 627-2286; Email: dandersy@gmu.edu
Research Objective: The purpose of this study is to examine the use of voluntary quality performance measurements by medical group practices as a competitive strategy. Participation in multiple quality measurement programs is examined within a hierarchical framework developed from Porter's
Five Forces model of strategic behavior.
Study Design: This is a qualitative study. Porter’s model was conceptualized as a two-dimensional spectrum of organizational size, complexity, and resources; and of organizational focus from external macro to internal process concerns. It was hypothesized that increasing resources available to and complexity of medical group practices would determine the level and types of quality performance measurement program participation.
The increasing focus for medical groups would be the use of quality performance measurement programs as a competitive strategy of differentiation. Smaller groups would focus on external macro concerns of meeting emerging policy constraints, as represented by participation in the Medicare PQRI program. Larger, more resourced groups would focus on a broader array of quality measure programs, including internally developed standards to reduce variance and create demonstrable quality.
Population Studied: Data were collected from survey responses of seventy-six medical group practice administrators. Medical group practices in twenty-seven states were represented, ranging in size from less than five to more than fifty full time equivalent physicians. The medical practices were primarily physician owned, and over half had an electronic medical record system. Additional data were collected from in-depth interviews with two medical group practice administrators and two executives of healthcare associations that focus on medical group practice quality performance measurement activities and programs.
Principal Findings: Limited support was found for the hypothesis that size and resources are determinant factors in the level of participation in quality measurement programs by medical group practices. Support was not demonstrated that smaller groups would focus primarily on Medicare
PQRI to the exclusion of other voluntary quality measurement programs. Larger medical groups were not necessarily more likely to engage in internally developed quality programs. A progressive continuum of participation was partially supported by the data. Correlation with interview data revealed a preference of medical groups to focus primarily on internal operations and revenue enhancement in opposition to creating competitive advantages through the pursuit of quality measurement programs.
Conclusion: The conceptualization of internal competitive strategy as mechanism for organizational or service differentiation may mask more fundamental concerns for medical groups.
The model appears to be subsumed by the necessity for revenue-maximization behavior, especially for small medical group practices.
Instead of developing differentiations and competitive niches of healthcare services and treatments, internal strategies are developed to maximize organizational continuity and sustainability.
Implications for Policy, Delivery or Practice:
Health policy programs designed to measure and improve quality must be aligned with organizational needs to achieve changes in performance. Given the current fragmentation of medical group practices into primarily small and often single specialty organizations focused on internal operational issues, policies that focus on enhancing health outcomes need to provide incentives that achieve behavioral change while supporting the underlying sustainability issues facing medical group practice.
Theme: Organizational Performance and
Management
● Effects of Clinical Guidelines & Direct-to-
Consumer Advertising on Appropriate Use of a
Cervical Cancer Screening Test
Rebecca Anhang Price, M.S.
Presented by: Rebecca Anhang Price, M.S., Ph.D.
Candidate, Department of Health Care Policy,
Harvard Medical School, Harvard University, 180
Longwood Avenue, Suite 113, Boston, MA 02115,
Phone: (617) 642-6335; Email: ranhang@post.harvard.edu
Research Objective: Clinical guidelines and directto-consumer (DTC) advertising are two strategies used to influence the adoption of new health care technologies. Guidelines are issued primarily by professional groups, government agencies, and insurers to promote appropriate use; however, adherence to guidelines is often poor. DTC advertising is used primarily by manufacturers, who claim that consumer promotions increase the likelihood that patients receive appropriate care for under-diagnosed and under-treated conditions.
Critics counter that DTC may increase unnecessary and inappropriate use. This study evaluates the impact of clinical guidelines and a targeted directto-consumer advertising campaign on overall and appropriate use of the HPV DNA test, a cervical cancer screening tool.
Study Design: Data is from Medstat Group’s
MarketScan database of linked enrollment and medical claims. HPV DNA test use in markets in which there was DTC advertising is compared with use in a set of comparison markets in which there was not DTC advertising, before and after the advertising periods. The difference-in-differences method is used to account for secular trends in the use of HPV DNA tests. Associations between clinical guidelines and HPV DNA test use are assessed in both DTC and comparison cities.
Population Studied: Privately insured women ages 21 to 64 whose insurance claims were included in the MarketScan database for at least 12 consecutive months from January 2001 through
December 2005 (n = 500,000).
Principal Findings: Both the publication of clinical guidelines and the DTC advertising campaign were associated with increases in overall HPV DNA test use. DTC advertising was associated with a statistically significant increase in HPV DNA test use in both groups of DTC cities (+6.38 percent, p<0.0001 and +2.54 percent, p<0.0001). DTC advertising was associated with equal increases in the probability of appropriate and inappropriate use of the HPV DNA test in primary screening. Clinical guideline releases from the American College of
Obstetricians and Gynecologists, and by a cosponsored panel, were associated with greater increases in HPV DNA tests for appropriate primary screening among women age 30 and older than for inappropriate primary screening among women age less than 30 (ß=0.3347, p<0.05 and ß=0.4175, p<0.01).
Conclusion: DTC advertising appears to have led to increased use of HPV DNA tests, but not to differentially higher use of the test among women of a recommended age. Clinical guidelines were associated with greater increases in appropriate use than in inappropriate use.
Implications for Policy, Delivery or Practice:
Clinical guidelines have the potential to focus clinicians’ practices in the way intended, even in areas in which DTC advertising is broadcast.
Theme: Health Care Markets and Competition
● Ethnic Disparities in Patient Experiences of
Diabetes Care
Onyebuchi Arah, M.D., M.P.H., Ph.D.; Bastiaan
Roset, M.S.; Karina Schaake, M.A., M.Sc.; Diana
Delnoij, Ph.D.; Karien Stronks, Ph.D.
Presented by: Onyebuchi Arah, M.D., M.P.H.,
Ph.D., Department of Social Medicine, Academic
Medical Center, University of Amsterdam,
Meibergdreef 9, P.O. Box 22700, Amsterdam,
1100 DE, NL, Phone: +31205664892; Email: o.a.arah@amc.uva.nl
Research Objective: Increasingly, studies have found substantial ethnic/racial disparities in health and healthcare. The literature has tended to focus on ethnic/racial disparities in healthcare utilization, costs, quality, satisfaction, and more recently, experiences of care. Disparities in healthcare experiences have been documented in health plans, hospitals, and primary care but less so in care received for specific conditions. Given the rise in chronic conditions such as diabetes mellitus especially among minority ethnic groups in many societies, it is important to investigate ethnic differences in patient experiences of care for those chronic conditions. This study aimed to quantify ethnic disparities in patient experiences of diabetes care, accounting for the influence of patient education, language spoken at home as a measure of local language proficiency, and self-rated general health.
Study Design: The study was a cross-sectional, hierarchical study of diabetes patients nested within diabetes-care-networks in the Netherlands. In the
Netherlands, a diabetes-care-network is a practice or referral consortium of general practitioners, internists, nurses, dietitians, and other providers who manage the chronic care of diabetes patients within local communities. Patients were surveyed using the Consumer Quality Index instrument for diabetes mellitus developed and validated in the
Netherlands. The methodology of the validated survey instrument was based on that of the
Consumer Assessment of Healthcare Providers and
Systems instruments from the United States. The study outcomes were patient experiences based on
22 reported items clustered into six scales. Seven patient ethnicity groups were identified in the study, using the official definition of the Netherlands
Central Bureau of Statistics based on country of birth. We used varying-intercept-only multilevel regressions to estimate the association between patient ethnicity and their experiences, adjusting for age, sex, education, language spoken at home, and self-rated general health status.
Population Studied: We used data on 5,438 diabetes patients aged at least 18 years and who
responded to a cross-sectional survey of their experiences of diabetes care within 24 diabetescare-networks in the Netherlands in 2007.
Principal Findings: Overall, non-western ethnic minority patients reported worse patient experiences than the ethnic Dutch on most scales except ‘no language problems’. For instance, on the
‘communication with nurses’ scale, the regression coefficients for Moroccans, Turks,
Surinamese/Antilleans/Arubans and other non-
Westerners were respectively -0.341, -0.311, -0.110 and -0.396 compared to the ethnic Dutch
(P<0.001). These associations were attenuated but not largely accounted for by differences in education, language spoken at home, and selfrated general health. The disparities seemed concentrated among non-western ethnic minorities who were born outside the Netherlands (first generation) rather than those born in the
Netherlands (second generation).
Conclusion: Non-western ethnic minority patients had worse experiences of diabetes care than ethnic
Dutch, which could only for a small part be accounted for by differences in education, language, and general health status.
Implications for Policy, Delivery or Practice:
Policymakers and providers should realize that ethnic minorities might require culturally appropriate diabetes care so as to improve their experiences.
Disparities might aggravate diabetes burden among ethnic minorities if their less-than-optimal care experiences are ignored.
Funding Source(s): The Netherlands Organization for Health Research and Development
Theme: Disparities
● Investigating the Impact of Receiving
Recommended Diabetes Care on Patient
Experiences
Onyebuchi Arah, M.D., M.P.H., Ph.D.; Bastiaan
Roset, M.S.; Diana Delnoij, Ph.D.; Niek Klazinga,
M.D., Ph.D.; Karien Stronks, Ph.D.;
Presented by: Onyebuchi Arah, M.D., M.P.H.,
Ph.D., Assistant Professor, Department of Social
Medicine, Academic Medical Center, University of
Amsterdam, Meibergdreef 9, P.O. Box 22700,
Amsterdam, 1100 DE, NL, Phone: +31205664892;
Email: o.a.arah@amc.uva.nl
Research Objective: Healthcare systems increasingly assess technical quality and interpersonal care delivered to patients. This is particularly true for chronic conditions such as diabetes mellitus where patients who enjoy both better technical quality of care and interpersonal care seem to have better outcomes. The few studies looking at both domains have found mixed evidence of a positive correlation between technical quality of care and measures of interpersonal care such as satisfaction. This study adds to the evidence-base by investigating the associations between diabetes patients’ reports of receiving recommended care (technical quality) and their healthcare experience and ratings (interpersonal care).
Study Design: We conducted a cross-sectional survey of diabetes mellitus patients within 24 diabetes-care-networks in the Netherlands.
Diabetes-care-networks are akin to medical homes for diabetics. Patients were asked to report on their actual care experiences, specified guidelinerecommended preventive care aimed at diabetics, and their knowledge of diabetes complications and related lifestyle. The patient experience survey was based on the methodology of the Consumer
Assessment of Healthcare Providers and Systems used in the United States. We used multilevel regression to analyze the impact of nine measures of receiving recommended preventive care for diabetics on six scales of patient experience and three global ratings of general practitioner, nurses, and overall diabetes care.
Population Studied: We used data on 3,096 diabetes patients aged at least 18 years old, who received their chronic care within any of 24 diabetes-care-networks, and who responded to a cross-sectional survey of their care and experiences in the Netherlands in 2007.
Principal Findings: Only 47% of patients reported having had urine testing, 72% had foot examination, and almost 98% had their blood pressure checked.
About 95% of the patients had their A1C level checked in the preceding 12 months. Care aimed at lifestyle such as nutrition (37%), physical activity
(52%), and smoking (64%) were among the least received recommended care. Receiving foot examination, physical activity advice, smoking status check, eye examination, and A1C testing, but not nutritional advice, urine, cholesterol or blood pressure checks, were frequently associated with better patient experience and global ratings. For instance, those who received A1C testing rated their overall care 1.002 points higher (95% CI
0.726–1.278) on a scale of 0 to 10 than those who did not.
Conclusion: Diabetes patients who reported receiving recommended secondary preventive care
(better technical quality) were more likely to report better healthcare experience and ratings. This could be because their providers were better at both types of care, or the patients rewarded providers who delivered higher quality care with higher ratings, or both.
Implications for Policy, Delivery or Practice:
Diabetes can be a debilitating chronic disease if not well managed and its successful management partly hinges on securing patient compliance and trust. Ensuring good interpersonal care can go a long way in securing patient compliance which in turn might improve healthcare outcomes. Since measures of technical and interpersonal care are not perfectly correlated, both types of measures
should be monitored during performance assessment of providers and networks.
Funding Source(s): The Netherlands Organization for Health Research and Development
Theme: Prevention and Treatment of Chronic
Illnesses
● Understanding Providers’ Views Regarding the Delivery of Genomic Medicine at the VHA
Using Complexity Science Theory.
Nedal Arar, Ph.D.; Polly Noel, Ph.D.; Michael
Parchman, M.D.
Presented by: Nedal Arar, Ph.D., Associate
Professor, Medicine, University of Texas Health
Science Center at San Antonio/ Veterans Health
Administration, 7703 Floyd, San Antonio, TX 78229,
Phone: (210) 567-0075; Email: ararn@uthscsa.edu
Research Objective: To examine providers’ opinions of and attitudes toward the delivery of genomic medicine at the VHA through the lens of complex adaptive system (CAS) theory. Health care teams are a CAS, comprised of individuals who can learn, interconnect, self-organize, and interact with their environment in a way that demonstrates nonlinear dynamic behavior.
Study Design: Focus group and semi-structured interviews were conducted with 20 providers working in different units at our local VHA. The interviews focused on three main topics: (1) impact,
(2) facilitators and (3) barriers to the delivery of genomic medicine at the VHA. Interview materials were tape-recorded, transcribed and content analyzed using qualitative methods to examine characteristics of CAS.
Population Studied: VHA providers
Principal Findings: All participating providers perceived genomic medicine as an important area that will have positive impact on common diseases outcome. They perceived that the VHA has the necessary infrastructure (e.g., Electronic Health
Records) to foster the delivery of genomic services.
The majority of participants (80%) agreed that primary care providers will play a major role in delivering genomic services. Most providers (70%) raised concerns about the impact of using genomic services on the process of care: workload and workflow (self-organizing, non-linear properties).
Participants indicated that additional training for providers and patients, as well as system decision support will facilitate the delivery of genomic services (agents who learn). However, they identified three barriers, included (1) uncertainty about genomic findings (agents who learn); (2) coordination of care between primary care, specialists and genetic services
(interconnectedness of agents); and (3) ethical issues (e.g., privacy) associated with genomic information and services (interaction with environment).
Conclusion: Our findings highlighted several opportunities and challenges related to the delivery of genomic medicine at the VHA. Viewing the VHA as CAS will help in leveraging these opportunities and addressing the challenges. In this presentation, we will discuss our findings using the lens of CAS theory and suggest strategies that may enhance the delivery of genomic medicine to veterans.
Implications for Policy, Delivery or Practice:
Implementing genomic medicine will have broad implications for a large integrated health system like the VHA. Our work contributes to the ongoing efforts directed at improving the quality of health care for veterans with chronic common diseases.
Funding Source(s): VHA
● Preferences for Receiving Genetic Results: A
Comparison of Veterans & Non Veterans
Enrolled in Genetic Research
Nedal Arar, Ph.D.; Mickael Parchman, M.D.; Polly
Noel, Ph.D.
Presented by: Nedal Arar, Ph.D., Associate
Professor, Medicine, The University of Texas
Health Science Center at San Antonio/Veterans
Health Administration, TX 78229, Phone: (210)
567-0075; Email: ararn@uthscsa.edu
Research Objective: (1) to examine subjects’ preferences in receiving research results and (2) to identify challenges unique to veterans regarding participation in genomics research.
Study Design: Secondary analysis was performed on data collected in 2000-2004 from 1575 consent forms signed by participants enrolled in two genetic family studies (GFS) in San Antonio, The Family
Investigation of Nephropathy and Diabetes (FIND) and the Extended FIND. Participants included probands recruited from local dialysis units and their relatives. The consent form for these studies contained multiple-choice questions to examine participants’ preferences about receiving their (1) lab results, and (2) future genetic results. The FIND and EFIND databases had information on subjects’ demographic characteristics, and some selected clinical variables. We identified veterans using the
VHA’s centralized data repository. We compared veterans and non-veterans using Student’s t-test for continuous variables, and ?2 test for discrete variables. A logistic regression analyzed subjects’ preference for receiving their research results, controlling for other socio-demographic & clinical variables.
Population Studied: Veterans and nonvetans subjects participated in genetic research.
Principal Findings: The sample included 275
(18%) veterans, and 1247 (82%) non-veterans.
Most veteran participants (210, 76%) were relatives of the DN probands; only 65 (24%) were probands
(?2-test; P< 0.05). Recruitment of veteran probands was hard due to difficulties in contacting and enrolling family members as required by the studies
protocols. Our results indicated a strong desire among all participants 1451(94%) in getting their lab research results. Likewise, 1428 (91%) expressed interest in being informed about their future genetic results. There was no significant difference in veterans and non-veterans’ preference to disclosure of the research results (?2-test; P >
0.05). Regression analysis showed no significant relationship (P=0.449) between the outcome
(receiving research results) and veterans’ responses after controlling for demographics and educational levels.
Conclusion: Veterans are similar to non-veterans in their preferences in receiving their genetic research results. However, many veterans who receive care at VAMC are unable to contact family members, a crucial element in conducting GFS.
Implications for Policy, Delivery or Practice:
Participants value receiving their genetic research results. Offering genomic research results to participants should be based on well defined and structured plans to address issues such as study limitations, and interpretation of data.
Funding Source(s): VHA
Theme: Prevention and Treatment of Chronic
Illnesses
● Effective Strategies & Structures for
Children's Mental Health Services
Mary Armstrong, Ph.D.
Presented by: Mary Armstrong, Ph.D., Assistant
Professor, Child & Family Studies, University of
South Florida, 13301 Bruce B. Downs Boulevard,
Tampa, FL 33612, Phone: (813) 974-4601; Email: armstron@fmhi.usf.edu
Research Objective: To identify the financing strategies and structures that support effective children's systems of care
Study Design: 5-year national study that includes:
Identification of a hypothesized set of strategies and structures; Case studies of states and communities identified as having effective financing strategies;
Cross-analysis of case study findings; Synthesis of findings
Population Studied: States and communities with effective financing strategies and structures
Principal Findings: Core Financing Strategies: 1.
Identify current funding streams and strategies 2.
Identify other potential funding sources and strategies 3. Develop a strategic financing plan 4.
Ensure that strategies fund a comprehensive array of treatment services and supports
Implications for Policy, Delivery or Practice: Set of policy recommendations at the federal and state levels regarding the development of an effective financing
Funding Source(s): SAMHSA Center for Mental
Health Services and NIDAA
Theme: Behavioral Health
● Over-Utilization of Prenatal Services in Latin
American Private Healthcare Sector
Alejandro Arrieta, Ph.D.; Ariadna Garcia-Prado,
Ph.D.
Presented by: Alejandro Arrieta, Ph.D., Assistant
Professor, School of Public & Environmental Affairs,
Indiana University Purdue University Indianapolis,
801 West Michigan Street, Indianapolis, IN 46202,
Phone: (317) 274-2717; Email: arrietaa@iupui.edu
Research Objective: In this study, we focus on analyzing the effectiveness of the private sector in providing prenatal care as opposed to the public sector in some Latin American countries that have recently expanded the private sector as a result of their health care reforms.
Study Design: Many Latin American countries are responding to problems of coverage and quality in the public sector with the expansion of private health care. Although patients may expect to receive better quality in the private sector, there is evidence of private practitioners who perform significantly worse than public ones. This study focuses on comparing private sector performance with public sector regarding the provision on prenatal care services. Using multiple regression analysis, we compare the number of prenatal office visits of women in private and public facilities after controlling by insurance, and health and socioeconomic characteristics. Then, we investigate if the excess number of prenatal visits in the private sector was accompanied of better prenatal quality
(timing, better information, and tetanus vaccination) and better outcomes at delivery (e.g. newborn weight) that are known to be preventable with adequate prenatal care.
Population Studied: Women 15 to 49 years old, who had prenatal services in public and private facilities in post-reform years in Colombia, Peru and
Dominican Republic. Data was obtained from the
Demographic and Health Surveys.
Principal Findings: Prenatal visits at larger in the private sector, but after controlling by obstetric and socio-economic characteristics, the difference becomes non-significant for Colombia, a country that implemented a reform where the private sector played a more connected role in the public sector, and where the scope of health care regulation explicitly includes the private sector. Prenatal visits remained higher in Peru’s and Dominican
Republic’s private sectors. Private insurance played an important role in explaining over-utilization of prenatal visits in Peru, while in the Dominican
Republic the over-utilization was mainly driven by incentives given to private providers. However, in both cases, the excess number of prenatal visits did not guarantee a higher quality service when compared with the public sector. Most important, the over-utilization of prenatal services in those countries was ineffective in terms of better outcomes at delivery than the public sector.
Conclusion: While many patients believe that in the private sector they are going to receive better treatment, this study shows that private health care providers tends to over-utilize prenatal visits in countries that did not create an active role to private providers within the health care system, including lack of regulation. In those countries, over-utilization of prenatal visits did not guarantee either better prenatal quality services or better health outcomes at delivery.
Implications for Policy, Delivery or Practice:
Private and non-governmental providers are often an important source of health care in the Latin
American region. However, this sector is not integrated in the health system and often is neither regulated, nor subject to supervision. It is important to identify these factors to avoid over-provision of procedures, low quality and rising health costs, and set the and knowledge for its adequate regulation.
Funding Source(s): InterAmerican Development
Bank
Theme: Global Health
● Applying Group-Based Trajectories
Methodology to Measures of Patient Safety
Scott Ashwood, M.A. Martin Gaynor, Ph.D.; Daniel
Nagin, Ph.D.; Amelia Haviland, Ph.D.
Presented by: Scott Ashwood, M.A.,
Programmer/Analyst/Ph.D. Student, Public Policy,
RAND/Carnegie Mellon University, 4570 Fifth
Avenue, Suite 600, Pittsburgh, PA 15213, Email: ashwood@rand.org
Research Objective: Most of the research on hospital quality is either cross-sectional or identifies a single average trend. We used group-based trajectories models of hospital patient safety rates to determine whether there is evidence for distinct trends in patient safety among California hospitals.
In addition we were interested in whether hospital level characteristics are related to a hospital’s probability of following a particular trajectory.
Study Design: We generated hospital level patient safety indicator rates by applying the AHRQ Patient
Safety Indicator (PSI) software to California state hospital discharge data for the years 1997-2006.
Hospital level characteristics were added by merging on the financial report data provided by the
California OSHPD. We used group-based trajectories models to analyze hospital rates for each of 20 patient safety indicators over time. The group-based trajectories approach takes advantage of longitudinal data for individual hospitals to determine whether there is evidence of groups of hospitals following distinct trends in outcomes. In addition to estimating the number of groups, the form of the trajectory for each group is estimated.
As a result, groups of hospitals with constant rates can be distinguished from those with increasing or decreasing rates. It is then possible to determine whether hospital characteristics are related to a hospital’s trajectory.
Population Studied: General acute care hospitals in the state of California 1997-2006.
Principal Findings: While the patterns vary somewhat across the different indicators, in general there is evidence that most of the hospitals in
California follow trajectories with consistent, low rates. However, for each outcome, there are hospitals that are following more dynamic trends, with increasing or decreasing rates. For example, the average rate of Decubitus ulcer increased from
3.3 to 3.7 per 1,000 discharges from 1997-2006 in
California. There is evidence, however, for four distinct trajectories. 79.8% of hospitals had a constant rate of 1.8, and 13.8% had a slightly higher constant rate of 3.4. Two groups followed more dynamic paths with 3.4% experiencing an increase from a rate of 2.8 to 8.1 and 2.9% experiencing a decrease from 13.9 to 7.2. There appear to be three distinct trajectories among hospitals for selected infections due to medical care. Two trajectories are constant at 0.8 (84.7%) and 2.5 (13.3%). A third group (2%) experiences an increase from 0.9 to 7.2. For-profit hospitals are more likely to be in this third group.
Conclusion: There is evidence that there are distinct groups of hospitals following different trajectories of PSI rates over time. In addition, hospital level characteristics appear to predict group membership. Most hospitals appear to have low and constant rates with smaller groups experiencing more dynamic trajectories.
Implications for Policy, Delivery or Practice: The evidence presented here suggests that there is a rich, dynamic story to hospital patient safety. Since there appears to be a large group of hospitals with constant, low rates, it may be more useful to focus on the groups of hospitals that are actually improving or getting worse over time rather than to try to identify and study a small group of top performers in each year.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Measurement
● Maximizing the Potential Care Management
Savings for Selected Chronic Diseases
Ognian Asparouhov, Ph.D.; Donghui Wu, Ph.D.
Presented by: Ognian Asparouhov, Ph.D., Chief
Scientist, MEDai Inc., 4901 Vineland Road, Suite
450, Orlando, FL 32811, Phone: (321) 281-4559;
Email: oasparouhov@medai.com
Research Objective: The Chronic Impact Score
(CIS) provides a predictive modeling based score for identification of members that can produce the highest level of future savings potential by adherence to evidence-based guidelines for 16 major chronic diseases. The CIS solution provides case managers and other personnel the integrated reports/profiles needed to evaluate individual
members with the greatest opportunity for cost savings and/or high-cost care avoidance
Study Design: From multi-million insurance plan members, based their Evidence-Based Guidelines compliance status, and various conditions and costs constructed from insurance claims data, two large groups (compliance and noncompliance) of members with chronic conditions are matched based on their propensity socred. The Chronic
Impact Score(CIS) provides a predictive modeling based score for identification of members that can produce the highest level of future savings potential by adherence to evidence-based guidelines.
Currently this CIS provides potential savings assessment and actionable care guidelines for 16 chronic diseases: Asthma, CAD/MI, CHF, COPD,
CVA, Depression, Diabetes, Drug Management,
HIV, Hyperlipidemia, Low Back Pain, Migraine
Headache, Multiple Sclerosis, Renal Failure,
Rheumatoid Arthritis, Schizophrenia. The main emphasis of the model is to identify which guidelines have the greatest impact on savings potential, irrespective of the patient’s current compliance level. Patients who are now noncompliant have the potential to become compliant in the future year with proper management. However, those that are currently compliant will not necessarily remain compliant and also may require management attention. This distinction is paramount toward managing the true high-risk/high-impact patients. The model was developed to identify the most critical diseases and guidelines in an effort to focus precious medical management resources. Within the model, several guidelines show that moving from compliance to non-compliance is more costly than maintaining a failure status. To determine the percentage of savings opportunity, the model evaluated the expected noncompliant cost for a member and then adjusted that according to their guideline compliance.
Population Studied: Mulit-millon commercial plan members from multiple insurance plans with 24 month insurance claims data.
Principal Findings: The CIS model produces a potential cost savings for each member based on compliance to the specific disease guidelines, illness’ severity and acute / chronic costs. The presence of catastrophic co-morbidities, or highly complex clinical scenarios was factored in as variables that limit cost-savings potential.
Conclusion: The Chronic Impact Score provides a unique opportunity for managed care organizations and disease management companies to produce cost savings. The CIS solution provides case managers and other personnel the integrated reports/profiles needed to evaluate individual members with the greatest opportunity for cost savings and/or high-cost care avoidance.
Implications for Policy, Delivery or Practice:
Improve Chronic Disease Management effectiveness and ROI.
Theme: Prevention and Treatment of Chronic
Illness
● Emerging State Activities on Language
Access Services in Health Care
Melanie Au, M.P.P.; Erin Fries Taylor, Ph.D.
Presented by: Melanie Au, M.P.P., Mathematica
Policy Research, Inc., 600 Maryland Avenue, SW,
Suite 550, Washington, DC 20024, Phone: (202)
264-3459; Email: mau@mathematica-mpr.com
Research Objective: National interest in providing language access services in health care has increased in the past several years. This is particularly evident in recent state legislation and regulation that emphasize health plan responsibility in promoting language services. As interest in language access grows, the experiences of states that are forerunners in operationalizing and financing language services provided through health plans may offer insight into other states’ future activities. We profile legislation and other efforts in selected states to highlight challenges, successes, and implications for future policy and work on language access services.
Study Design: Qualitative case study including document review and interviews with over a dozen key informants.
Population Studied: We interviewed staff from national, state and private organizations. California,
Minnesota and New York—forerunners in policy and regulatory activity related to language access services—were a special focus of this case study.
Principal Findings: While attention to language access in health care has increased at the national level in recent years, concrete policies and activities have occurred almost exclusively at the state and local levels. California, for example, recently implemented statewide legislation requiring commercial health plans to provide language access services to members at all points of contact including the medical encounter; this represents the most comprehensive existing regulation for language access services. Each of the three states examined are at varying stages in their language access efforts and have experienced different challenges based on state-specific factors. Despite these differences, all states identified collaboration and involvement of interested stakeholders as a necessary facilitator to address challenges and drive successful efforts. States also reported actively learning from each other, using examples of legislation and regulation seen elsewhere to drive their own efforts. However, while state legislation and regulation are powerful ways to influence provision of language access services in publicly and privately sponsored health care, effective implementation efforts still require guidance and leadership by the state.
Conclusion: Some states are promoting language access services in health care through legislation
and regulation specific to health plans. The experiences of these states provide guidance to those looking to strengthen language access services statewide as well as give some insight on where other states may be headed.
Implications for Policy, Delivery or Practice:
While attention to language access has increased, proponents stress that important work remains. In order for language services to be effective, more information is needed on when such services are appropriate and what standards should be used to assess quality of services provided. For example, there are currently no federal and only a few state standards for certifying medical interpreters.
Developing standards for interpreter services could help improve the quality of communication during the medical encounter. Furthermore, once programs have been implemented, evaluation research will be needed to determine the impact of language access services on quality of health care for those with limited English proficiency.
Funding Source(s): AHRQ
Theme: Disparities
● Is the Doctor In? Recent Trends in Hours
Worked by Physicians in the U.S.
David Auerbach, Ph.D.; Douglas, Staiger, Ph.D.;
Peter Buerhaus, Ph.D.
Presented by: David Auerbach, Ph.D., Principal
Analyst, Health & Human Resources,
Congressional Budget Office, 106 East Indian
Spring Drive, Silver Spring, MD 20901, Phone:
(202) 226-2669; Email: david.auerbach@cbo.gov
Research Objective: Although it is well known that physicians work long hours, recent trends in the hours worked by physicians have not been analyzed.
Study Design: Using nationally representative workforce information from the US Bureau of the
Census Current Population Survey for 1976 through
2007, we examined trends in hours worked per week among all US Physicians, and examined whether recent trends differed by gender, age, and work setting. We also compared trends in hours worked for physicians located in metropolitan areas with high and low levels of physician fees, based on physician fee data collected by the General
Accounting Office in 2001.
Population Studied: Physicians identified in the
Current Population Survey from 1976 to 2007.
Principal Findings: After remaining stable for two decades, hours worked per week declined 7.1% in the last decade among physicians, from 54.6 hours per week in 1995-1997 to 50.7 hours per week in
2005-2007 (p<.001). Significant (p<.001) declines of between 5% and 10% in hours worked were observed over this period for male and female physicians, for physicians under and over age 40, for physicians who worked in hospital and nonhospital settings, and for physicians who were selfemployed or not. After adjusting for inflation, physician fees declined nationwide by nearly 25% between 1995 and 2006, coincident with the decline in physician hours. In 2001, physician hours had fallen below 51 hours per week in metropolitan areas with the lowest physician fees, while physician hours remained over 54 hours per week in metropolitan areas with the highest physician fees.
Conclusion: There was a steady decline in hours worked per week over the last decade for all physicians, which was temporally and geographically associated with lower physician fees. That decline is roughly equivalent to a reduction of 50,000 physicians, had hours worked per physician not changed.
Implications for Policy, Delivery or Practice: If these trends persist, and the observed associations causal, they will have important implications for the future supply of physician services and for health care delivery in general.
Theme: Health Care Workforce
● Changes in Access to Any & Adequate Mental
Health Care by Race/Ethnicity between 1990-
1992 & 2001-2003
Andrea Ault Brutus, M.P.A.
Presented by: Andrea Ault Brutus, M.P.A., Ph.D.
Candidate, Health Policy, Harvard University, 700
Huron Avenue, 8D, Cambridge, MA 02138, Phone:
(919) 364-4487; Email: aault@fas.harvard.edu
Research Objective: To estimate differences in rates of access to any and adequate mental health care between Whites and Blacks and Whites and
Latinos in 1990-1992 and 2001-2003, and assess how these disparities changed over time.
Study Design: Using data from the 1990-1992
National Comorbidity Survey (NCS) and the 2001-
2003 National Comorbidity Survey-Replication
(NCS-R), trends in disparities are assessed for use of any and adequate mental health care in the past
12 months between Whites and Blacks and Whites and Latinos. Any mental health care is defined as at least one visit to a provider in the general medical
(GM) or mental health specialty (MHS) sector in the past 12 months. Adequate mental health care is defined as four or more visits to providers in the GM or MHS sector plus medication use, or at least eight visits to a MHS provider with no medication use.
Analysis is stratified on having a 12-month mood and/or anxiety disorder. Rates of any care are presented for the GM sector, MHS sector, and both sectors combined. Rates of adequate care are presented for both sectors combined. White-Black and White-Latino comparisons are made within each time period using survey design adjusted chi square test. Difference-in-difference analysis is used to assess the magnitude of change in disparities over time.
Population Studied: Individuals in the continental
U.S. between the ages of 18-54 years old, have a
12-month mood and/or anxiety disorder, and have self-identified as White, Black, or Latino.
Principal Findings: In the GM sector, there was a significant increase in the White-Black disparity in any mental health care (15.9 percentage point increase, p<.01), but no significant change between
Latinos and Whites. For any mental health care in the MHS sector, there was a marginal White-Black disparity in Time 1, no disparity in Time 2, and no significant change in disparities. For Whites-
Latinos, disparities in the use of any mental health care in the MHS sector persisted over time, but with no significant change in disparities. For adequate care, White-Black disparities persisted in both time periods with no significant change in disparities.
There was White-Latino disparity in adequate care in Time 2 but not Time 1, and no significant change in disparities.
Conclusion: Except for the increase in White-Black disparities in any mental health care in the GM sector, disparities for Blacks and Latinos did not change over time. However, there were White-
Black and White-Latino disparities in any mental health care and adequate mental health care that either persisted over time or developed in Time 2.
Implications for Policy, Delivery or Practice:
Policies and programs targeted to communities of color may be needed in order to decrease racial/ethnic disparities in access to any and quality mental health care. The sector in which care is received should be considered when designing policies and programs aimed at eliminating these disparities.
Funding Source(s): Center for Minority Health and
Health Disparities, MacArthur Foundation,
Mathematica Policy Research, NIMH
● Association Between Utilization of
Mammogram Services & Number of Adult
Women in the Household: Explained by Health
Belief Model
Shamly Austin, M.H.A.
Presented by: Shamly Austin, M.H.A., Doctoral
Student/Graduate Research Assistant, Health
Services Administration, University of Alabama at
Birmingham, 1320, 11th Avenue South, #3,
Birmingham, AL 35205, Email: shamlyaugustine@gmail.com
Research Objective: To examine association between utilization of mammography services by an adult female, and number of adult females in her household.
Study Design: The Health Belief Model can be used to explain the association between utilization of mammography services by a female, and number of adult females in the household.
According to the model having more number of adult women in the household may prompt to discuss susceptibility, benefits, and barriers to mammography services. Having another adult woman to discuss health issues may prove as a cue for their health care utilization, which may lead to their utilization of mammography services. A cross-sectional study design is used. The data set is Behavioral Risk Factor Surveillance System
(BRFSS)2005. In the analysis, 10,860 females
>=18 years were included. The outcome variable was whether individual female had mammogram.
The independent variables were the number of adult women in the household, age, race, education, income, marital status, health insurance, health status, co-morbidities, have a personal physician, had social/emotional support. Descriptive statistics, chi-square, and logistic regression were used to analyze the utilization pattern.
Population Studied: Adult female respondents >=
18 years from BRFSS 2005
Principal Findings: Of 10,860 female respondents,
73% had mammogram, only 13% had more than one female in the household, 78% were <65 yrs,
55% were married, 61% had education more than high school, 89% had health insurance, and 81% had personal physician. The logistic regression found no association between an individual’s utilization of mammography services and number of adult females in her household (p=0.08) controlling for age, race, education, income, marital status, health insurance, health status, co-morbidities, having a personal physician, and having social/emotional support . However number of adult women in household, age, race, education, marital status, health insurance, health status, comorbidities, and having a personal physician were independently significant predictors of mammography utilization.
Conclusion: Having another adult female in the household did not predict a female’s use of mammography services. Other factors such as susceptibility (age, race), benefits (health status), and cues to action (marital status, education), barriers (income, health insurance, co morbidities) decided an individual’s utilization of mammography services.
Implications for Policy, Delivery or Practice:
This study reiterates the importance of factors that needs intervention in utilization of mammography services. It also adds to the literature that more number of adult females in a household is not a cue to utilization of mammography services.
● Medical Civic Action: Using Health Policy to
Stabilize Failing Nations
George Avery, Ph.D., M.P.A.
Presented by: George Avery, Ph.D., M.P.A.,
Assistant Professor, Health & Kinesiology, Purdue
University, 800 West Stadium Avenue, West
Lafayette, IN 47907, Phone: (765) 496-3330;
Email: gavery@purdue.edu
Research Objective: To explore the concept of medical civic action and to develop a model for the integration of medical and public health efforts into operations designed to stabilize failing nations and end insurgencies.
Study Design: The problem of governmental stability is examined from the point of social contract theory, identifying the strategic and operational targets for strategic. Historical examples of medical civic action programs (MEDCAP) conducted by the US military are examined, and a set of principles presented for ensuring that future programs are targetted at the strategic goal of stabilization.
Population Studied: Medical and public health civic action efforts conducted as part of US military stability operations.
Principal Findings: States fail and insurgencies arise when a government loses the trust of its population and the monopoly on legitimate use of force. Successful MEDCAP programs must be sustainable, meet local needs, develop local government capacity, and be part of an integrated political effort to create acceptence of the legitimacy of the state. MEDCAP efforts cannot be separated from security issues. The issues of neutrality and humanitarian space raised under Protocol I of the
Geneva Convention create a strong barrier to the success of a stabilization operation by creating opportunities for inadvertent legitimization of insurgents.
Conclusion: MEDCAP programs, like other military stabilization efforts, require a Clausewitzian mindset that is focused on the political goals of the effort, rather than the Jominian viewpoint traditionally held by the US military that focuses on the enemy. When the strategic political goals of the operation are clearly recognized, medical and public health civic action is a strong tool for the development of political legitimacy and the stabilization of failing nation-states.
Implications for Policy, Delivery or Practice:
This model expands on and provides a theoretical foundation for the inclusion of medical, public health, and humanitarian efforts as part of the toolkit for US military stability operations, as outlined by GEN David Petreaus in the wellpublicized changes to US counterinsurgency doctrine.
Funding Source(s): Department of Defense
Theme: Global Health
● Engaging Consumers in Health System
Improvement: Promising Strategies &
Implications
Holly Avey, Ph.D., M.P.H.; Heather Devlin, M.A.;
Karen Minyard, Ph.D.; Tina Anderson Smith,
M.P.H.; Natalie Towns, M.S.W.; Naima Wong,
Ph.D., M.P.H.
Presented by: Holly Avey, Ph.D., M.P.H., Senior
Research Associate, Georgia Health Policy Center,
Georgia State University, P.O. Box 3992, Atlanta,
GA 30302-3992, Phone: (404) 413-0291; Email: havey@gsu.edu
Research Objective: To understand consumer perspectives on health system improvement in a rural setting.
Study Design: A state health policy center and a regional healthcare access coalition engaged healthcare consumers using a community-based participatory research approach. A Freirian process of co-learning informed participatory study design, data collection and analysis. The consumer work group met regularly for over a year. Work group members produced recommendations for improvements to the local health system and also conducted several small-scale health improvement projects. Projects were documented through
PhotoVoice, a method that uses photography and storytelling to engage, empower and learn from community members about the local conditions that affect them. The study began by presenting the work group with local data on health status and chronic disease management. Reflecting on this data led work group members to conclude that challenges and stresses of daily life in their community contribute to a gap between health knowledge and healthy behaviors. In response, researchers provided information on the ten social determinants of health as defined by the World
Health Organization. The consumer group adopted this framework and selected three determinants as their focus for further work: access to healthcare, stress, and healthy living. Researchers presented the work group with examples of successful community health improvement models. Based on these examples, the group recommended local experimentation to identify promising interventions for their community. Consumer work group members formed three teams to design projects focused on increasing access to healthy food, decreasing stress among workers and empowering the homeless. The coalition provided mini-grants for these projects, along with coordination, data collection and technical assistance.
Population Studied: A highly diverse work group of 11 healthcare consumers living in a largely rural community included individuals with chronic disease, members of the faith community, workers in major area industries, and persons not participating in the workforce.
Principal Findings: Consumer work group members realized immediate benefits from a participatory engagement process and from undertaking projects of their own design. Benefits included individual empowerment, expanded networks and a better understanding of their local health system. The consumer engagement process generated insights and recommendations for improving local health systems. This group of consumers saw social determinants of health as part of their local health system and felt that the
ultimate goal of health system improvement must be culture change. They described the ideal culture as one where people are empowered, take responsibility for their own health, healthy options are easy options, and people routinely make healthy choices. They envisioned a community health clinic circled by walking paths, where patients could have access to farmers' market produce and healthy “fast food” and learn how to manage stress.
Conclusion: An interactive, participatory engagement process revealed the need for health system improvement strategies that incorporate both cultural and environmental change.
Implications for Policy, Delivery or Practice:
Insights from this study suggest that healthcare consumers want local health systems that move beyond health care delivery and account for conditions affecting their ability to make choices that maintain health and wellness.
Funding Source(s): Foundation, Community
Health Works
● Children in Upheaval: War Migration & Child
Health In Angola
Winfred Avogo, Ph.D.; Victor Agadjanian, Ph.D.
Presented by: Winfred Avogo, Ph.D., Assistant
Professor, Sociology & Anthropology, Illinois State
University, Campus Box 4660, Normal, IL 61790-
4660, Phone: (309) 438-5227; Email: wavogo@ilstu.edu
Research Objective: To investigate the short and long term impact of the effects of migration induced by war (war migration) and that motivated by other reasons besides war (non-war migration) on child health outcomes in Angola, a country of about 16.8 million people in South-Western Africa that has endured civil conflict for most of its independent existence.
Specifically, we test the assumption that although the antecedents and mechanisms of war migration and non-war migration are different, they do occur in the same contexts in settings of prolonged military conflicts that vary in intensity. Thus the two migration groups are expected to have different child health outcomes such as age adequate immunization and under-five mortality in the short and long term.
Study Design: Cross-sectional analysis of detailed data on birth and health histories of children underfive using discrete-time event history models. The unit of analysis is the person-year and each child is at risk of death for each year of migration until the child dies or is censored.
To study long term effects of migration on underfive mortality, we use negative binomial regression to model cumulative number of children under five years of age who died per respondent.
Population Studied: A 2004 representative survey of two peri-urban municipalities of Greater Luanda-
Samba and Viana. Samba is a long-established peri-urban community, close to the city center and was chosen as a community with a relatively small war migrant population, whereas Viana, a more distant and less urbanized suburb, was chosen because of a larger estimated share of war migrants.
The survey instrument gathered detailed data on birth histories (the month and year of birth and death if the child died) and the immunization status of the each child by age. Detailed migration histories were also collected. Respondents who migrated were asked to give the localities of their previous residence, the timing of migration and the reasons why they migrated from their previous locality. These responses where then classified by interviewers into six categories (war-related, economic, family, education, health or other). This data will be relied on to construct outcome measures (age adequate immunization and underfive mortality) and key predictors of this study (war migration, non-war migration and non-migration).
Several correlates of child health (such as age and gender, birth interval, mother’s age at birth and mother’s education etc.) exists in the data will be exploited for this analysis.
Principal Findings: Preliminary findings from discrete-time event history models show that in the short term, results indicate that net of other factors, non-war migration is associated with higher probabilities of age adequate immunization and lower probabilities of under-five mortality, compared to war migration.
In the long term, the effects of migration on child mortality and age adequate immunization are positive but not statistically significant.
Conclusion: We conclude that selection and disruptive mechanisms explain child health outcome differences between war and non-war migrants in the short-term.
Non-war migrants may be self selected based on unobserved distal factors such as the tendency to space births, avoidance of nutrient deficiency and personal illness control and on observed characteristics such as education and socioeconomic status. Whereas, child health outcomes of war migrants may have suffered from the disruptive effects of threats to personal security and inability to prepare before fleeing from war.
Implications for Policy, Delivery or Practice: Our findings and conclusions have implications for the attainment of the Millennium Development Goals
(MDGs), which enjoins all countries to reduce by the year 2015 two thirds of the childhood mortality rates that prevailed in 1990 (MDG Report 2006).
Post-war reproductive health policies can be shaped by making a distinction between economic and forced migration when it comes to child health outcomes, since the mechanisms of each of these processes are not only different but are on the high ascendancy throughout conflict areas.
Other wide ranging implications are being explored in the context of Angola.
Funding Source(s): Data collection partly funded by National Institute of Child Health and
Development (NIH/NICHD) of the USA
Theme: Child Health
● Children’s Mental Health Problems &
Caregiver Strain: Racial Differences & the Role of Referral Source
Melissa Azur, Ph.D.; Crystal Barksdale, Ph.D.
Presented by: Melissa Azur, Ph.D., Assistant
Scientist, Mental Health, Johns Hopkins Bloomberg
School of Public Health, 624 North Broadway,
HH801, Baltimore, MD 21205, Phone: (410) 502-
8506; Email: mazur@jhsph.edu
Research Objective: Caring for children with serious emotional and behavioral problems can be stressful. The strain involved in caring for these children may vary for different racial/ethnic groups and be influenced by the involvement of agencies such as juvenile justice. This study examines racial differences in caregiver strain and determines whether those differences vary depending upon how children enter into mental health services.
Study Design: Data are from the national evaluation longitudinal outcomes study of the
Children’s Mental Health Initiative (CMHI), a national mental health systems-of-care program.
The outcome variable was average caregiver strain
(range=0-5), as measured at baseline into services by the Caregiver Strain Questionnaire. Predictors include caregiver report of child race/ethnicity and source of referral into services (school, juvenile justice, child welfare, caregiver). We used random effects regression models to estimate caregiver strain as a function of race/ethnicity, referral source, and the interaction between race/ethnicity and referral source, adjusting for sociodemographic characteristics, functional impairment, behavior problems, and family resources.
Population Studied: The sample comprises ratings from 4510 caregivers of Caucasian, African
American, Hispanic, and multi-racial children, ages
5-18 years, who were referred into the CMHI between 1997 and 2005 due to a serious emotional or behavioral problem.
Principal Findings: There were no significant main effects for race on average caregiver strain; however, caregivers who referred their children into services themselves (ß=.23, 95% CI= .02, .21) and caregivers of children entering services from juvenile justice (ß=.23, 95% CI= .02, .21) reported significantly more strain than caregivers of children referred from schools. There was a significant interaction between race and referral source (ß= -
.19, 95% CI= .33, -.05) such that no significant differences in strain among caregivers of African
American and Caucasian children who were referred into services from school systems were detected; however, when children were referred into services from juvenile justice, caregivers of African
American children experienced significantly less strain than caregivers of Caucasian children.
Conclusion: Caregivers’ experience in caring for children with emotional and behavioral problems is affected by how the child enters into mental health treatment. Involvement with the juvenile justice system is particularly stressful for caregivers of
Caucasian children. The low levels of strain reported by caregivers of African American children could reflect strengths in managing a child with mental health problems, or a belief that the child does not have a problem that necessitates services.
Implications for Policy, Delivery or Practice:
Caregivers of Caucasian children in the juvenile justice system may need additional support and services. Caregivers of African American children may have strengths and coping mechanisms that can be incorporated into treatment plans. Service providers should carefully assess commitment to and perceptions of child’s need for mental health treatment.
Funding Source(s): NIMH
Theme: Child Health
● The Effects of Auxiliary Mental Health
Services on Children’s Emotional & Behavioral
Problems
Melissa Azur, Ph.D.; Lucas Godoy Garraza, M.S.;
Christine Walrath, Ph.D.; Crystal Barksdale, Ph.D.
Presented by: Melissa Azur, Ph.D., Assistant
Scientist, Mental Health, Johns Hopkins Bloomberg
School of Public Health, 624 North Broadway,
HH801, Baltimore, MD 21228, Phone: (410) 502-
8506; Email: mazur@jhsph.edu
Research Objective: Auxiliary services, such as recreational activities, respite care, and child and family support services are believed to enhance a child’s and family’s ability to cope, function, and progress toward recovery from mental health problems; however, little is known about the type of children who actually receive these services and how the services effect their mental health outcomes. This study examines the characteristics of children receiving auxiliary support services in community based mental health systems-of-care settings and determines whether these services are associated with improved behavioral outcomes.
Study Design: Data are from the national evaluation longitudinal outcomes study of the
Children’s Mental Health Initiative (CMHI), a national mental health systems-of-care program.
Data are collected at baseline and at 6 month intervals for 36 months. Outcome variables include caregiver report of internalizing and externalizing problems as measured by the Child Behavior
Checklist. Predictors include caregiver report of use of recreational and after-school activities, child support services (i.e., family preservation,
behavioral/therapeutic aide, physical health, day treatment, independent living), and family support services (i.e., transportation, respite care, flexible funds, family support). We used multilevel models to identify characteristics of children predictive of auxiliary service use, and then to estimate internalizing and externalizing problems as a function of these auxiliary services, adjusting for child and family level factors.
Population Studied: The sample comprises 2763 children, ages 5-17 years, who enrolled into mental health services with the CMHI between 1998 and
2005. Children who had complete data in at least one time point were included.
Principal Findings: Compared to Caucasian children, African American children entered into services with lower levels of internalizing problems.
The odds of any auxiliary services use decreased over time. African American children were significantly more likely to receive recreational/after school services (OR=1.53), while Hispanic children were significantly less likely to receive child support services (OR=.50. There were no gender differences in use of any of the auxiliary services, and there were no age or race/ethnicity differences in the use of family support services. There were no significant differences in internalizing or externalizing problems for children who received these services.
Conclusion: Receipt of auxiliary services can vary for minority children. The lack of significant differences in improvement in behavior problems for children who received child or family support services could suggest three things: (1) systems of care programs appropriately provide services to fit the individualized needs of children and their families, (2) other types of services may be more important in improving internalizing and externalizing behaviors, (3) auxiliary services may positively impact other areas of a child’s life, such as school performance, child and family functioning, and caregiver strain.
Implications for Policy, Delivery or Practice:
Case managers and providers should engage families in treatment planning and identify the resources and supports that best meet the needs of each child and family. Further dialogue regarding how auxiliary supports benefit children with mental health problems and their families is needed.
Funding Source(s): NIMH
Theme: Child Health
● Nurse Satisfaction & the Implementation of
Mandatory Overtime Regulations
Sung-Heui Bae, Ph.D., M.P.H., R.N.
Presented by: Sung-Heui Bae, Ph.D., M.P.H.,
R.N., Research Assistant Professor, School of
Nursing, University at Buffalo, The State University of New York, Kimball Tower Room 904, 3435 Main
Street, Buffalo, NY 14214, Phone: (716) 829-3218;
Email: sbae7@buffalo.edu
Research Objective: In the health sector, nurses appear to be the most visible group subject to overtime, that is, to be required to stay at work for an additional shift, especially in hospitals. More overtime hours have been associated with an increase in hospital-related mortality, nosocomial infections, and medication errors in patients and musculoskeletal problems in nurses. Nurse overtime has become an increasingly contentious issue in the U.S., leading to state and federal legislation to restrict mandatory overtime.
Mandatory overtime legislation prohibits a health care facility from requiring an employee to work more than their regularly scheduled hours except during a natural or other health care disaster that unexpectedly increases the need for health care personnel. However, how the mandatory overtime law affects nurse overtime hours (mandatory overtime, voluntary overtime, and on-call hours) has not been studied. There is also limited empirical evidence about whether these laws in turn increases nurse job satisfaction, which might improve nurse retention. This study examined the impact of implementation of mandatory overtime regulations on nurse overtime in hospitals in states that have implemented these regulations and how it affects nurse satisfaction.
Study Design: The current study used data from the 2004 national sample survey of registered nurses (NSSRN), which collected data on the actively licensed registered nurse population as of
March 2004. The data collection continued from
March 2004 through November 2004. As of March
2004, Maryland, Minnesota, New Jersey,
Washington, Maine, Oregon, Texas, and California had restrictions on the use of mandatory overtime for nurses. Overtime hours include three categories: paid on-call, paid as overtime, and mandatory/unscheduled overtime. Nurses were asked to rate their feelings about their nursing position on a 5-point Likert-type scale. Nurse demographic and employment characteristics were included in the model to control for their association with overtime hours and nurse satisfaction.
Differences in overtime hours across nurses with different demographic and employment characteristics were analyzed by ordinary leastsquares regression with sample weights.
Population Studied: The sample includes 35,724
RNs in U.S. hospitals. The data collection began in
March 2004 and continued through November
2004.
Principal Findings: Full-time nurses worked an average of 7.5 hours of overtime per week, while part-time nurses worked an average of 5.6 hours of overtime. Further analyses will explore relationships among mandatory overtime laws, nurse overtime, and nurse satisfaction.
Conclusion: In progress
Implications for Policy, Delivery or Practice:
Hospitals have used nurse overtime to manage fluctuations in patient census. This study provides
insight into how mandatory overtime laws are related to the actual use of nurse overtime including mandatory overtime, voluntary overtime, and on-call hours. It also suggests how this law indirectly affects nurse satisfaction, which may increase nurse retention. An Increasing number of states are considering mandatory overtime legislation. This study will provide healthcare managers with information about the underlying mechanisms involved in the implementation of mandatory overtime laws in hospitals.
Theme: Health Care Workforce
● The Relationship Between Financial
Dissatisfaction & Medication Adherence
Problems in the Treatment of Schizophrenia
Daniel Ball, Dr.P.H., M.B.A.; Kyle Grazier, Dr.P.H.,
M.S., M.P.H.; Peter Jacobson, J.D., M.P.H.; Baojin
Zhu, Ph.D., M.S.; Haya Ascher-Svanum, Ph.D.,
M.A.
Presented by: Daniel Ball, Dr.P.H., M.B.A.,
Research Advisor, Global Health Outcomes, Eli Lilly
& Company, Lilly Corporate Center, Indianapolis, IN
46285, Phone: (317) 277-6628; Email: ball_daniel_e@lilly.com
Research Objective: Schizophrenia is a chronic, serious mental illness affecting 1% of the population. Risk of relapse and hospitalization increases with antipsychotic prescription refill gaps of only ten days. Using primary, longitudinal data, this study of individuals with schizophrenia 1) determines the relationship between satisfaction with their current financial situation and medication adherence problems; and 2) assesses how adjusting for regimen complexity or patient, medication or diagnosis characteristics affects the association.
Study Design: The U.S. Schizophrenia Care and
Assessment Program (US-SCAP) was a nonrandomized, naturalistic, prospective study in which patients with schizophrenia-spectrum disorders were periodically assessed with standardized measures over 3 years. Data analyzed for this study were collected by patient interview and medical record abstraction at baseline and subsequent 6-month intervals. Psychotropic adherence and financial satisfaction were based upon patient self-reported categorical responses.
Bivariate analyses examined the incidence of adherence problems using predictor variables from a literature-based conceptual framework.
Multivariate regression models identified any remaining associations after controlling for other predictors.
Population Studied: Between 1997 & 2003, US-
SCAP enrolled 2,327 individuals diagnosed with schizophrenia, schizoaffective, or schizophreniform disorder who were 18 years or older. The study sample of 1,339 excluded individuals who received an injectable antipsychotic, resided in an institution, or lacked a psychotropic medication prescription.
Principal Findings: Adherence problems were reported in 30.3% of responses. Financial satisfaction ratings were as follows: 39.2% satisfied,
37.8% dissatisfied, and 23.0% mixed. Higher satisfaction was associated with fewer adherence problems -- 35.7% among dissatisfied, 32.4% among mixed, and 25.3% among satisfied respondents (p<0.001). Adherence problems were
24.1% more likely (p< 0.001) in regressions comparing mixed and dissatisfied respondents to satisfied respondents. Additional predictors with higher odds of adherence problems included younger age, non-sheltered employment, not completing college, being sad or blue, medication side effects, dosing > 2X daily, activity limitations, substance use and suicidal thoughts or attempts.
Predictors with lower odds of adherence problems included income below 125% of poverty, Medicaid only coverage, receiving assistance with taking medications, reporting that medication(s) improved thinking, being a high school graduate, availability of professional mental health assistance, and reporting very good or excellent health.
Conclusion: Financial dissatisfaction is significantly associated with lower medication adherence among individuals with schizophrenia even after controlling for numerous other characteristics.
Implications for Policy, Delivery or Practice: The association between financial dissatisfaction and adherence problems, combined with prior research linking adherence problems with increased hospitalization risk, suggests increased investment in programs to reduce financial dissatisfaction in this population may produce offsetting savings.
The counterintuitive income results may indicate that assistance be in a form other than additional payments. For example, rent subsidies can reduce housing expenses and offer a stable environment for providing wraparound services like money management or substance abuse counseling.
Because expenditures in one program may lead to savings elsewhere, coordination and integration of programs across mental health, substance abuse, housing and employment agencies is essential to reducing adherence problems in this population.
Finally, government and industry collaboration to coordinate patient assistance program requests concurrent with Medicare or Medicaid application may represent an opportunity to match prescription assistance to those with greater unmet needs.
Funding Source(s): Eli Lilly and Company
Theme: Behavioral Health
● Organization of Primary & Mental Health Care
& Diagnosed Depression among Women
Veterans with Chronic Physical Illness
Ranjana Banerjea, Ph.D.; Bevanne Bean-Mayberry,
M.D., M.P.H.; Patricia Finely, Dr.P.H., M.S.W.,
L.C.S.W.; Usha Sambamoorthi, Ph.D.
Presented by: Ranjana Banerjea, Ph.D., Health
Scientits, Department of Veteran Affairs, Center for
Healthcare Knowledge Management, 385 Tremont
Avenue, Mail Stop 129, East Orange, NJ 07018,
Phone: 973-676-1000 x12552; Email: ranjanabanerjea@gmail.com
Research Objective: The complexities of diagnosing and treating depression in the context of a co-occurring physical illness often require primary care providers to consult with mental health specialists. Our purpose is to analyze the association between organizational features related to general and women-specific practice structures and diagnosed depression among women veterans with chronic physical illnesses such as diabetes, heart disease or hypertension.
Study Design: Secondary data analyses of merged
Veteran Health Administration (VHA) and Medicare claims data for the fiscal years (FY) 2001 merged with the 1999 VHA Survey of Primary Care
Practices and VHA survey of Women Veterans
Health Programs and Practices. Veteran women using VHA clinics diagnosed with diabetes or heart disease or hypertension in FY2001 were identified with ICD-9-CM codes. Facility level variables
(practice structure) consisted of available mental health team providers within primary care, primary care within mental health care settings, integration of outpatient services for men and women, and separate mental health clinics for women. The dependent variable was a binary indicator for diagnosed depression during FY2001. We used a random-effects logistic regression model due to clustering of women veterans within facilities. Other individual-level independent variables included demographics, socio-economic characteristics and type of physical illness.
Population Studied: Veteran women from 112 facilities (N = 16,181). These facilities served on the average 144 women (range: 28 to 445 women).
Principal Findings: Overall, 28% of women were diagnosed with depression in FY2001. Across facilities the rates of diagnosed depression varied from a minimum of 6% to a maximum of 51%, and a median of 28%. Random effects logistic regression revealed a strong random effect in diagnosed depression due to facilities. Eleven percent facilities had separate mental health clinics for women. Facilities with a lower degree of integration of outpatient services for men and women were more likely to have prevalent depression (p < .01).
We found marginally significant association between primary care teams without psychiatrists and diagnosed depression (AOR = 1.11).
Conclusion: Our findings add to the growing body of literature of noting high rates of depression among women with diabetes, heart disease, and hypertension. A women-specific practice structure may lead to greater recognition of depression among women veterans. A facility-specific practice structure with regard to integration of mental health within primary or vice versa did not affect diagnosed rates of depression across facilities.
Implications for Policy, Delivery or Practice:
Higher rates and substantial variation of diagnosed depression across facilities among women with cardiovascular or those at risk for cardiovascular diseases suggest challenges to the healthcare delivery for women veterans. However, existing women-specific VHA organizational structures seem to be effective in detecting and diagnosing depression, a first step towards treatment planning and management.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● From Data to Action: Phase 1 - Responsibly
Engaging Physicians to Reduce Overuse
Taruna Banerjee, M.P.H.; Michael Bakerman, M.D.,
F.A.C.P.; Gregory Partridge, B.A.
Presented by: Taruna Banerjee, M.P.H., Program
Manager, Performance Management, Partners
Community Healthcare, Inc, 115 Fourth Avenue,
Needham, MA 02494, Phone: (781) 433-3029;
Email: tbanerjee@pachi.partners.org
Research Objective: To determine if primary care physicians and specialists can be successfully engaged in reducing overuse in clinical practice through a person-centered multi-faceted intervention.
Study Design: This is an observational study focusing on physician engagement in a process to identify areas of clinical overuse and encourage reduction in unwarranted variation in care. The success of physician engagement was evaluated using feedback from practice (RSO) leaders, results of outreach visits, attendance at physician meetings and physician involvement in program improvement. Episode Treatment Group (ETG) analysis identified potential areas of overuse that were then reviewed by an interdisciplinary team to select the behaviors felt to best reflect overuse.
Focused Medical Analytics was consulted to assist with variation analysis and physician engagement planning. A multi-faceted intervention was introduced including distribution of unblinded physician specific reports at physician practices tied to actionable recommendations, physician education regarding locally vetted practice guidelines, respectful physician outreach visits, and financial incentives.
Population Studied: Partners Healthcare is an integrated healthcare delivery network of over 1200
PCPs and 4000 specialists in greater Boston managing the care of over 500,000 patients in an incentive based P4P commercial contracting vehicle. The physician panel is divided into functional practice groups (RSO’s).
Principal Findings: Components of the program judged most successful in engaging physicians in overuse reduction were 1) engaging practice
leadership early on in the process and keeping them continuously updated on comparative progress, 2) presenting practice and physician specific unblinded data to RSO’s and practice physicians, 3) actively listening to and following up promptly on practitioner’s concerns, and 4) making actionable and practice specific recommendations.
The most compelling observation was the power of providing reliable actionable comparative individual practitioner data. Many had never seen their practice behaviors compared to respected, trusted colleagues. Tied to education regarding locally vetted recommended guidelines, practitioners were responsive to requests to change their practice patterns. The role of financial incentives was reported to be less significant. There was a natural progression through the stages of change so that evaluation of efficiency projects may require 18-36 months for meaningful evaluation.
Conclusion: In this project, the most successful approaches to engage generalist and specialist physicians to reduce overuse included: distributing valid and clinically meaningful individual and practice group data , involving practice leadership early and regularly, presenting unblinded physician level data in a non-judgmental and respectful fashion, being readily available to answer methodology questions and follow-up on practitioner concerns in a timely fashion.
Implications for Policy, Delivery or Practice:
Successful physician engagement on variation in clinical practice will result in standardization of care, decrease in overuse, and an increase in the value of care.
Theme: Quality and Efficiency: Policies &
Incentives
● The Relationship Between MRI Availability &
Low Back Pain Care Utilization
Jacqueline Baras, B.A.; Laurence Baker, Ph.D.
Presented by: Jacqueline Baras, B.A., Medical
Student, School of Medicine, Stanford University,
2052 Ashton Avenue, Menlo Park, CA 94025,
Phone: (727) 742-8229; Email: jbaras@stanford.edu
Research Objective: The rapid expansion in the number of magnetic resonance imaging (MRI) scanners in the United States has enabled more patients to receive cutting-edge imaging that can produce valuable diagnostic information. However, for patients with low back pain, the use of MRI is controversial. Spinal abnormalities detected by MRI often do not correlate with symptoms and can lead to additional, unnecessary interventions, including surgery, which in many patients is of uncertain efficacy. This paper investigates the relationship between the expanding MRI supply and the diagnosis and treatment of low back pain, providing new information about the broader effects of increased imaging availability in a relatively large group of patients.
Study Design: We use claims from a 20% sample of Medicare beneficiaries from 1998-2005 to develop cohorts of traditional Medicare patients with a new episode of care for low back pain. Each episode begins with an “index visit” in 1999, 2002, or 2004, three years for which data on MRI availability at the Metropolitan Statistical Area
(MSA) level are also available for the study. We assigned each patient to an MRI availability quartile based on the distribution of MRI units per one million population in their MSA of residence. We study the effect of expanding imaging availability using logistic regression models in which the key independent variable is the MRI availability quartile; the dependent variables are indicators of receipt of low back MRI or surgery within 30, 90, 180, or 365 days of the index visit; and controls include patient demographic and health characteristics as well as fixed effects for areas and years.
Population Studied: 831,794 episodes of treatment for low back pain, among fee-for-service
Medicare patients over age 65
Principal Findings: Among all episodes of care,
18.3% resulted in low back MRI and 3.6% resulted in low back surgery within one year of the index visit. Increases in MRI availability are statistically significantly associated with higher utilization of low back MRI. Patients in areas in the highest quartile of MRI availability, compared to patients in the lowest quartile, were 15.2% more likely to receive
MRI within 30 days of their index visit, and 9.6% more likely within 365 days (both p<.001).
Increases in MRI availability are also statistically significantly associated with higher probabilities of low back surgery receipt.
Conclusion: Diffusion of MRI equipment is associated with increased early use of low back
MRI, as well as subsequent use of low back surgery, both of which are discouraged for the majority of patients with new onset low back pain.
These results raise concerns that the widespread expansion of MRI may adversely impact quality of care for low back pain patients.
Implications for Policy, Delivery or Practice:
Policy efforts that assess the value of technology diffusion should consider both the monetary and non-monetary costs that increased MRI availability may have for particular patient groups. Programs that encourage more efficient MRI availability and utilization could be beneficial.
Funding Source(s): Stanford University
● Racial Disparities in Influenza Vaccination
Coverage among Michigan Long-Term Care
Residents: Understanding the Role of Within &
Between Facility Differences
Barbara Bardenheier, Ph.D., M.P.H., M.A.; Pascale
Wortley, M.D., M.P.H.; Faruque Ahmed, Ph.D.,
M.B.B.S.; Abigail Shefer, M.D.; Kevin Sullivan,
Ph.D., M.P.H., M.H.A.
Presented by: Barbara Bardenheier, Ph.D.,
M.P.H., M.A., Epidemiologist, Immunization
Services Division, Centers for Disease Control &
Prevention, 1600 Clifton Road, Northeast Mail Stop
E52, Atlanta, GA 30333, Phone: (404) 639-8789;
Email: bfb7@cdc.gov
Research Objective: In October, 2005 the Centers for Medicare and Medicaid Services (CMS) required that long-term care (LTC) facilities certified by CMS offer each resident annual influenza vaccination.
Although influenza vaccination is covered by
Medicare, racial disparities in receipt of the influenza vaccine remain. This study examines racial disparity in influenza vaccination among LTC residents within and between facilities.
Study Design: We analyzed data from the October
1, 2005 through March 31, 2006 Minimum Data Set
(MDS) for all residents in Michigan. Facility-level data from the Online Survey Certification and
Reporting (OSCAR) database were merged with
MDS.
Population Studied: Residents of 404 non-hospital based nursing homes in Michigan certified by CMS which included 61,150 (86.2%) whites and 9,818
(13.8%) blacks for whom vaccination status was known. The proportion with unknown status did not vary by race (30 and 31%). We chose not to include other minority groups as they are less than 2% of the LTC resident population.
Principal Findings: In Michigan, during the first year CMS mandated that the influenza vaccine be offered to all LTC residents 68.0% were vaccinated.
Among whites, 70.9% were vaccinated, 13.8% refused vaccination, and 11.9% were not offered vaccination; among blacks, 49.6% were vaccinated,
19.4% refused, and 25.3% were not offered vaccination. The black/white disparity in vaccination was 21.3 percentage points. Of the 404 LTCs, 304 had black and white residents (median vaccination coverage for whites 74.0%, and for blacks 60.1%; median facility difference of 5.4 percentage points);
99 facilities had only white residents (median vaccination coverage 85.0%), and 1 had only black residents (vaccination coverage 72.4%). Over twothirds (67.8%) of blacks were concentrated in 56 of
404 (13.9%) LTCs. These 56 LTCs had >100 black residents, and a median vaccination coverage of
48.1% for Blacks and 50.0% for whites. In the multilevel model including a random intercept and race, blacks were significantly less likely to be vaccinated (OR: 0.73 CI: 0.69 – 0.78 p <.001). After controlling for resident variables and a random intercept the gap narrowed but blacks were still significantly less likely to be vaccinated (OR: 0.78
CI: 0.73 – 0.84 p <.001). After controlling for resident and facility variables and a random intercept the gap narrowed but blacks were still significantly less likely to be vaccinated (OR: 0.81
CI: 0.76 – 0.87 p <.001). Significant variance between facilities remained in the final model
(u=0.16, p <.001).
Conclusion: The substantial racial disparity in influenza vaccination among LTC residents in
Michigan is attributable to disparities both between
LTC facilities and within facilities and persists after adjusting for confounders. Further analysis is needed to understand the relative importance of within and between facility disparities in offering and accepting vaccination.
Implications for Policy, Delivery or Practice:
Better understanding of systems attributes that result in disparities of influenza vaccination coverage as well as insight into individual decisions to refuse vaccine could be used to direct resources to the appropriate facilities and/or residents to increase influenza vaccination and reduce the observed disparity.
Funding Source(s): CDC
Theme: Disparities
● The Relationship Between Negative Parent
Attitudes Toward Treatment & Race: Does it
Vary Based on How Youth Enter Services?
Crystal Barksdale, Ph.D.; Melissa Azur, Ph.D.
Presented by: Crystal Barksdale, Ph.D.,
Postdoctoral Fellow, Mental Health, Johns Hopkins
Bloomberg School of Public Health, 624 North
Broadway HH 808, Baltimore, MD 21205, Phone:
(410) 502-9344; Email: cbarksda@jhsph.edu
Research Objective: The President’s New
Freedom Commission on Mental Health identifies the elimination of disparities in mental health services as an integral part to improving the health of the country. Parents’ attitudes about child mental health treatment are a critical factor in understanding why these disparities exist, particularly among racial and ethnic minority groups and may be influenced by how the family was referred into mental health treatment (e.g., from mental health agencies, juvenile justice, schools).
The purpose of this study is to determine the relationship between race/ethnicity and negative caregiver attitudes about mental health treatment in a sample of youth enrolled in community-based mental health services, and to determine whether this relationship varies by referral source.
Study Design: This is a cross-sectional study, which utilizes baseline data from the national evaluation of the Children’s Mental Health Initiative
(CMHI) systems-of-care program. Caregiver report of coercive/negative attitudes toward treatment was determined by the caregiver’s “yes” response to two questions: “Did you feel free to do what you wanted concerning the child’s treatment?” and “Did you feel like you had control over whether the child got treatment?” Caregiver report of referral source was categorized as school, juvenile justice system, child welfare system, and self/caregiver. The analyses were conducted using random effects logistic regression models.
Population Studied: This study examined the attitudes of 2,437 Caucasian, African American,
Hispanic caregivers of youth aged 5-18 years who have serious emotional and behavioral problems and are enrolled in a system-of-care program.
Principal Findings: African American caregivers were less likely to have coercive attitudes toward seeking treatment for their child compared to
Caucasian caregivers (OR=.18; 95%CI: .05-.64).
There was also a significant interaction between race and referral source (OR=7.66; 95%CI: 1.63-
36.00). Among families referred into services from child welfare agencies, African American parents/caregivers felt more coerced than
Caucasian parents/caregivers.
Conclusion: Racially diverse parents hold different attitudes about mental health services for their children. How the family enters into services plays a significant role in shaping attitudes, which subsequently, can affect youth treatment outcomes.
Families involved in child welfare services may have uniquely different experiences based on their race.
Implications for Policy, Delivery or Practice:
Working to improve parent’s attitudes about treatment may result in an increase in the number of children receiving needed services, and thus a reduction in disparities. In addition, increased effort should be given to address parent’s negative attitudes and increase parent engagement in mental health treatment, especially when the family is involved in the child welfare service sector.
Funding Source(s): NIMH
Theme: Child Health
● No Vacancy: Microsimulation Modeling of
Supply-Side Interventions to Meet Excess
Demand for Hospital Services During a
Pandemic Flu
Andrew Barnes, M.P.H., Ph.D.; Jerry Jacobson,
Ph.D.; Matt Solomon, Ph.D.; Heather Kun, Ph.D.
Presented by: Andrew Barnes, M.P.H., Ph.D.,
Senior Research Analyst, Research & Evaluation,
National Health Foundation, 515 South Figueroa
Street, Suite 1300, Los Angeles, CA 90071, Phone:
(213) 538-0707; Email: abarnes@nhfca.org
Research Objective: Growing concern over acute, unpredictable public health emergencies have led to disaster planning at the local healthcare service and governmental levels. Exogenous shocks to hospital systems, like a pandemic flu, may cause temporary surges in demand for hospital care in excess of current supply. However, quantifying excess demand has been absent from many preparedness efforts. To meet this need, the
Pandemic Flu Hospital Planning Model (the Model) was developed and is being tested in the Los
Angeles (LA) County hospital market.
Study Design: The Model is a stochastic, discreteevent microsimulation model of inpatient and emergency department care. In the Model, patients arrive at emergency departments requiring different levels of care or at inpatient wards needing scheduled or urgent services. Using patient-, hospital-, and geographic-level data, patients are routed to hospitals in the system according to historical utilization patterns, patient demographics, and bed type required. Model outputs include numbers of patients needing beds when none are available by week of the pandemic and in each regional market. Other variables estimated include the additional number of beds needed at each hospital to meet demand, numbers of ventilators needed, and inpatient occupancy rates.
Population Studied: This model examines the demand by Los Angeles County residents for hospital services at approximately 100 area general acute care hospitals.
Principal Findings: In severe flu simulations, current supplies of adult intensive care hospital beds are exhausted within the first four weeks of the pandemic and remain so weeks after the peak of the event. Excess demand for these beds continues to rise over the course of the pandemic as throughput of patients is slowed by higher numbers of sicker patients with longer lengths of stay. Pediatric acute and critical care beds also experience the large increases in demand. Finally, additional supply of critical care beds required to meet expected demand exceeds 20% of current levels.
Conclusion: Much of disaster planning to date has been based on expert opinions or findings from static, “out-of-the-box”, models. Dynamic microsimulation models, like ours, are useful tools for planners to determine how much and where additional supply is required to minimize individual, hospital, and social costs.
Implications for Policy, Delivery or Practice:
Health services and health economics research could benefit from more efforts to simulate patient flow through an interconnected system of services to plan for a variety of shocks besides Pandemic
Flu including hospital closures and other disasters such as floods and earthquakes.
Funding Source(s): CDC, Los Angeles County
Public Health Department
Theme: Public Health
● Predictors of Interest & Use of Behavioral
Medicine Services in an Urban Academic
Primary Care Outpatient Setting
Kelly Baron, Ph.D.; Emily Lattie, B.S.; Joyce Ho,
Ph.D.; Douglas Reifler, M.D.; David Mohr, Ph.D.
Presented by: Kelly Baron, Ph.D., Postdoctoral
Fellow, Institute for Healthcare Studies, Feinberg
School of Medicine, Northwestern University, 750
North Lakeshore Drive, 10th Floor, Chicago, IL
60611, Phone: (312) 503-5571; Email: kbaron@northwestern.edu
Research Objective: Behavioral medicine interventions in primary care have the potential to improve emotional and physical health in patients with chronic disease. The goal of this study was to evaluate predictors of interest and reported use of behavioral medicine counseling services among primary care outpatients, including counseling for diet, weight loss, smoking cessation, and pain management.
Study Design: Primary care patients were mailed questionnaires one week following an outpatient appointment in the general internal medicine clinic.
Patients indicated interest in various types of behavioral medicine services including: health/lifestyle (e.g. diet and exercise), smoking cessation, and pain management counseling by checking boxes on a list of possible counseling services. Self-rated health, average daily pain, and symptoms of depression and anxiety were assessed by self-report. Follow-up questionnaires were mailed 12 months later to assess attendance in behavioral medicine counseling services.
Depression was measured by the 8-item version of the Patient Health Questionnaire (PHQ-8) and anxiety was measured by the 7-item version of the
Generalized Anxiety Disorders Questionnaire
(GAD-7).
Population Studied: Participants included 658 primary care outpatients at a Midwestern urban academic medical center. Baseline surveys were collected from July-November, 2007 and time 2 surveys were collected from August-December,
2008. Average age was 50.99 years (SD= 15.4).
The sample was 70.1% female and 61.4%
Caucasian. Average BMI was 28.7 kg/m2
(SD=7.51), 15.9% reported smoking “every day” or
“some of the time”, and 28% reported average daily pain > 3 on a 10 point scale. Response rate was
19.7% at the first mailing and 61.2% respondents also completed the time 2 questionnaire.
Principal Findings: At baseline, interest in behavioral medicine counseling was 50.5% for health/lifestyle, 9.1% for smoking cessation, and
17.1% for pain management. In logistic regression models predicting interest in these services, female gender and higher BMI were associated with interest in health/lifestyle counseling; age, non-
Caucasian ethnicity, and anxiety were associated with interest in smoking cessation; poorer self-rated health and greater pain ratings were associated with interest in pain management counseling. At one year follow-up, attendance in behavioral medicine counseling was 6.1% for health/lifestyle,
2.4% for smoking cessation, and 3.6% for pain management. Interest in these services predicted attendance in health/lifestyle counseling and pain management. Depression at baseline was inversely associated with smoking cessation counseling attendance.
Conclusion: Interest in behavioral medicine services was high but reported attendance was low over the study period. The strongest predictors of interest were indices of severity of the condition, such as BMI and pain ratings. Interest at baseline was the strongest predictor of reported attendance over one year.
Implications for Policy, Delivery or Practice:
Identifying primary care patients interested in behavioral medicine counseling may help direct services to those with greater readiness to engage in health behavior change.
Funding Source(s): AHRQ
Theme: Prevention and Treatment of Chronic
Illness
● Evaluation of the Medical Assistance Program for the Poor in the Republic of Georgia: A
Regression Discontinuity Approach
Sebastian Bauhoff, M.P.A.; David Hotchkiss, Ph.D.;
Owen Smith, M.P.A.
Presented by: Sebastian Bauhoff, M.P.A., Ph.D.
Candidate in Health Policy, Health Care Policy,
Harvard University, 180 Longwood Avenue,
Cambridge, MA 02115, Email: bauhoff@fas.harvard.edu
Research Objective: To evaluate a medical insurance program for the poor in a low income setting (the country of Georgia) with regards to access, health impacts, expenditures and financial protection
Study Design: The research applies a regression discontinuity design framework to evaluate the impact of the program by comparing beneficiary and non-beneficiary households in the region of the eligibility threshold.
Population Studied: Low-income households who qualify for government assistance in Georgia. A dedicated survey was administered to approximately 3,500 households around the threshold, by sampling from the proxy means test administrative database.
Principal Findings: Results provide insights into program impact on key health sector objectives, including improvements in access, financial protection, and satisfaction among the poor. These reveal important implications for key program design features including the benefits package, eligiblity criteria, and the voucher mechanism.
Conclusion: Impact evaluation techniques provide an important evidence base which the Government of Georgia can use to refine and scale up the
Medical Assistance Program for the poor.
Implications for Policy, Delivery or Practice: The results inform the optimal design of health insurance in resource-poor settings, with regards to social safety nets and specific elements such as benefit design.
Funding Source(s): Government of Georgia
Theme: Coverage and Access
● Quantifying Cream-Skimming in the German
Social Health Insurance
Sebastian Bauhoff, M.P.A.
Presented by: Sebastian Bauhoff, M.P.A., Ph.D.
Candidate in Health Policy, Health Care Policy,
Harvard University, 180 Longwood Avenue,
Cambridge, MA 02115, Phone: (857) 928-3884;
Email: bauhoff@fas.harvard.edu
Research Objective: This audit study aims to empirically assess whether sickness funds in the
German Social Health Insurance discriminate against applicants from high-cost regions.
Study Design: The study uses a dedicated randomized experiment in which about 50 national sickness funds are presented with 30 applicant profiles from across Germany. The applicants differ only in their geographic location, so that any differences in response and response time by the funds is attributable to geographic selection.
Population Studied: About 60 large, national health insurance companies (sickness funds) in
Germany.
Principal Findings: Regional cost variations and imperfect risk adjustment can generate incentives for health plans to select based on geography.
Conclusion: Selection incentives can pose a challenge to equity and efficiency, even in a social health insurance system. Policy-makers have to trade-off between entrenching providers in costly regions with selection.
Implications for Policy, Delivery or Practice:
This study is a unique approach to causally identifying selection behavior. It also connects two strands of research, regional variations and health plan behavior. The findings have implications for health insurance regulation and risk adjustment methodologies. They also transfer to other settings, such as the US Medicare system.
Funding Source(s): Harvard Institute of
Quantitative Social Sciences
Theme: Health Care Markets and Competition
● How Does Proximity to Safety Net Resources
Affect Racial & Ethnic Disparities in Health
Care?
Gloria Bazzoli, Ph.D.; Lee Rivers Mobley, Ph.D.;
Hui-Min Hsieh, M.A.
Presented by: Gloria Bazzoli, Ph.D., Bon Secours
Professor, Health Administration, Virginia
Commonwealth University, P.O. Box 980203,
Richmond, VA 23298-0203, Phone: (804) 828-
5223; Email: gbazzoli@vcu.edu
Research Objective: Safety net hospitals (SNHs) play a critical role in the U.S. health system by providing health care services for vulnerable populations. SNHs also frequently offer a variety of social and outreach services, which could potentially reduce disparities among ethnic and racial minority groups. This may be especially true for minority individuals living near an SNH because they will have ready access to these important facilitating services. Our study examined whether proximity of uninsured and Medicaid patients to
SNHs affected their patterns of care, especially for non-Hispanic black and Hispanic individuals. In addition, we examined whether the closure or ownership conversion on nearby SNHs had especially detrimental effects on minority individuals.
Study Design: Hospital discharge data for the states of Arizona, California, Florida, and Wisconsin were examined from the early 1990s to 2003. We examined patterns of hospital use and distance traveled by individuals to obtain hospital care. We assessed how use and travel distance varied for uninsured and Medicaid patients who lived near versus farther away from an SNH in the base year.
We then examined how these measures changed in communities where SNHs closed or converted ownership relative to a set of comparison sites where the hospital safety net was stable.
Population Studied: Patients treated in urban hospitals in the study states were classified based on race/ethnicity and payer status. We additionally classified patients based on whether they lived near, at a moderate distance, or at a substantial distance to an SNH in the base year given analysis of local travel patterns. Our longitudinal analysis also identified those patients near a closed SNH or one that converted ownership status.
Principal Findings: For uninsured and Medicaid patients, the results indicate that race and ethnicity rather than proximity to an SNH was the primary factor in observed access problems in the base year. Non-Hispanic blacks and Hispanics typically had to travel farther to get hospital care than comparable white individuals and were generally less impeded by distance when making their hospital choices. However, when SNHs closed, and sometimes when their ownership converted from public to non-profit or for-profit status, minority populations living near the affected institution and who were uninsured or covered by Medicaid experienced larger detrimental effects than did similar white individuals.
Conclusion: SNH presence in a community does not appear to eliminate or sharply curtail racial or ethnic disparities in access to care, even for people living nearest to the SNH who are most able to avail themselves to their services. However, racial and ethnic minorities are disproportionately harmed by the exit of these facilities.
Implications for Policy, Delivery or Practice:
There has been much discussion about approaches to reducing health disparities among racial and ethnic minority groups. Although SNHs are not a panacea to eliminating disparities, it appears that their exit from a community can create additional access problems for already underserved minority populations.
Theme: Disparities
● Predictors of Observed Patient-Physician
Communication about Medication Costs in
Rheumatology Medical Visits
Ashley Beard, Ph.D., M.P.H.; Betsy Sleath, Ph.D.;
Susan Blalock, Ph.D.; Mary Roth, Pharm.D.,
M.H.S.; Morris Weinberger, Ph.D.; Gail Tudor,
Ph.D.
Presented by: Ashley Beard, Ph.D., M.P.H.,
Pharmaceutical Policy Research Fellow,
Department of Ambulatory Care & Prevention,
Harvard Medical School & Harvard Pilgrim Health
Care, 133 Brookline Avenue, 6th Floor, Boston, MA
02215, Phone: (617) 509-9849; Email:
Ashley_Beard@harvardpilgrim.org
Research Objective: Although it is generally agreed that medication costs should be considered in prescribing decisions, little research has examined patient-physician communication about medication costs during medical visits. The objective of this study was to examine the predictors of actual communication about medication costs in medical visits between rheumatologists and patients with rheumatoid arthritis (RA). RA is a model chronic illness for this study because medications are the predominant therapy and there is large variation in the costs of medications to treat RA.
Study Design: Data, including audiotapes of medical visits, patient interviews, medical records, and questionnaires, were collected from March
2003 to December 2005. Using a detailed protocol,
2 trained coders reviewed the medical visit audiotapes and recorded whether the patient or physician discussed direct medication costs (e.g. out-of-pocket costs incurred by patients) or indirect medication costs (e.g. prescription drug assistance programs or prescription drug insurance coverage).
Intercoder reliability for 35 tapes was 0.94. We examined the role of the following predictors on discussions of medication costs: patient characteristics (gender, age, education, income, marital status, race/ethnicity, functional status, and prescription drug coverage), physician characteristics (gender, race, age), relationship characteristics (prior visits), and medication characteristics (cost of RA medications, number of
RA medications prescribed).
Population Studied: 8 rheumatologists and 200 of their adult patients with RA in North Carolina.
Rheumatologists were eligible if they saw patients age 45 years or older with RA. Patients were eligible if they 1) were being seen by a participating rheumatologist, 2) were 45 years of age or older, 3) had a diagnosis of RA, and 4) were able to speak and read English.
Principal Findings: Physicians’ mean age was
46.2 years, one was Black/African-American, two identified as Hispanic, and four were women. The mean patient age was 62.3, 74% were women,
22.5% were Black/African American, 1% were
Hispanic, 35.7% had yearly household of incomes <
$20,000, and 19.5% lacked prescription drug coverage. The mean number of RA medications per patient was 4.3 and total median per-month medication costs were $246.91. Medication regimen changes were made in 51% of visits.
Approximately 34% of recorded visits included discussions of medication-related cost issues,
47.7%% of which were initiated by the patient. In multivariable logistic regression analysis stratified by physician, communication about medication costs was significantly more likely to occur when patients were Caucasian (AOR 2.51, 95% CI 1.32 –
4.78) and had no prescription drug coverage (AOR
3.15, 95% CI 1.75 – 5.70).
Conclusion: Although medication changes were made in over half of the visits, medication costs were only discussed in one-third of visits.
Communication about medication costs was more common among patients who were Caucasian and lacked prescription drug coverage. Disparities in communication about medication costs have the potential to negatively impact prescribing and subsequent medication use. Further research should examine potential disparities in communication about medication costs.
Implications for Policy, Delivery or Practice: To foster communication about medication costs, previsit questionnaires could be used to ask patients about medication cost concerns and physicians should routinely ask about medication costs in medical visits.
Funding Source(s): Arthritis Foundation
Theme: Prevention and Treatment of Chronic
Illness
● Predictors & Costs of Preventable Nursing
Home Hospitalizations: The Role of Serious
Mental Illness & Dementia
Marion Becker, R.N., Ph.D.; Timothy Boaz, Ph.D.;
Ross Andel, Ph.D.; Amber Gum, Ph.D.; Airia
Papadopoulos, M.P.H.
Presented by: Marion Becker, R.N., Ph.D.,
Professor, Aging & Mental Health, University of
South Florida, 13301 Bruce B Downs Boulevard,
Tampa, FL 33612, Phone: (813) 974-7188; Email: becker@fmhi.usf.edu
Research Objective: Preventable nursing home
(NH) hospitalizations place an enormous economic burden on an already overtaxed American healthcare system. To better understand this issue, we examined hospitalization costs and risk factors for hospitalization of NH residents with ambulatory care-sensitive (ACS) conditions. These conditions are considered preventable as they can often be treated safely in a NH.
Study Design: In this retrospective study we used three years (FY 2003-2006) of Florida Medicaid
administrative data to identify resident characteristics such as age, race, and mental health and physical diagnosis. We obtained facility characteristics data such as ownership and occupancy status from the Online Survey
Certification and Reporting (OSCAR) system.
OSCAR data are collected annually through surveys of all certified nursing facilities in the United
States. Inpatient cost data included payments made by Medicaid, Medicare, other, third party payers, and recipients as indicated on the Medicaid claim records. In the main analyses, we estimated a relative risk of ACS hospitalization using a Cox proportional hazards regression in SAS procedure
PHREG which included all available individual and facility characteristics. Cox proportional hazards regression is one approach to survival analysis that is used to study the occurrence and timing of events by accounting for the proportion of time spent in the study. A 95% confidence interval was used, setting the significance level at a twotailed.05. Data were right censored for end of the
NH stay, end of Medicaid enrollment, death, or end of the study. We also examined overall expenditures associated with ACS hospitalizations.
Population Studied: The study sample included
72,251 Medicaid-enrolled nursing home residents in
647 nursing homes in Florida. We identified 10,091
ACS hospitalizations (18.5% of all hospitalizations).
Principal Findings: In the fully-adjusted proportional hazard regression, younger age, non-
White race, dementia and serious mental disorder were associated with a greater risk of ACS hospitalization. Among facility characteristics, participants from for-profit facilities, facilities that were not a member of a chain, facilities with more
Medicaid recipients and facilities of smaller size
(less than 120 beds) had increased risk of ACS hospitalizations. The total cost of ACS hospitalizations for all residents in our sample during the study period was almost $34 million.
Residents with a diagnosed mental disorder and no dementia incurred relatively high expenditures for
ACS hospitalizations.
Conclusion: Our findings underscore a need for further research on key related issues such as the effect of chronic medical conditions on the treatment of serious mental disorders and viceversa, and how these interactions impact early identification and treatment that can improve resident care within the NH setting and avoid unnecessary, traumatic, and costly transfers to other facilities.
Implications for Policy, Delivery or Practice:
Given the elevated risk of adverse health events associated with NH hospitalization, increased attention from policymakers, clinical practitioners and NH administrators to identified predictors of
ACS hospitalization is warranted. These issues promise to become only more urgent as the population ages and healthcare expenses continue to escalate.
Funding Source(s): Florida Agency for Health
Care Administration
● Female Racial Disparities in In-Hospital
Mortality Rates for On- & Off-Pump Coronary
Artery Bypass Surgery: 2000-2006
Edmund Becker, Ph.D.
Presented by: Edmund Becker, Ph.D., Professor,
Health Policy & Management, Emory School of
Public Health, 1518 Clifton Road, Northeast,
Atlanta, GA 30322-1369, Email: ebeck01@sph.emory.edu
Research Objective: With the advent and diffusion of off-pump coronary artery bypass surgery
(CABS), some studies have suggested that women undergoing on-pump CABS face twice the risk of early death as women who had off-pump CABS.
Few studies, however, have used national data or examined the extent of off-pump and on-pump inhospital mortality differences for women by their race/ethnicity. We seek to address these issues by analyzing the in-hospital mortality trends in on- and off-pump CABS for women by race/ethnicity with data from a national 7-year database.
Study Design: A retrospective 7-year national study of 20% of the nation's hospitals CABS discharges using descriptive and logistic regression techniques. We evaluated female CABS in-hospital mortality rates by race and on-pump versus offpump procedure controlling for 60 covariates. With in-hospital mortality as the dependent variable, major variables controlled in the logistic regression model were: six age categories; patient's race and
CABS technique - Caucasian, African American,
Hispanic, and Asian and whether they had their
CABS procedure off-pump or on-pump; insurance coverage - Medicare, Medicaid, private, self-pay, no charge, and other; risk factors and co-morbid conditions (smoker, history of tobacco use, COPD,
S/P CABG, S/P PTCA, insulin-dependent diabetes, non-insulin dependent diabetes, conduction disorders, intracranial hemorrhage, chronic renal failure, unspecified renal failure, cardiogenic shock, hypertension, acute MI, old MI, cardiomyopathy, congestive heart failure, peripheral vascular disease, unstable angina, acute liver necrosis, endocarditis, mitral valve disease, aortic valve disease, mitral and aortic valve disease, and chronic hepatitis); procedure characteristics
(number of vessels bypassed, diagnostic cath, intra-aortic balloon pump, and hypothermia use); medications (GP2B3A and thrombolytics); year dummies; and hospital dummies (dummy variables for each of the hospitals performing CABS surgery with one hospital excluded). All procedures, complications, and co-morbid conditions were identified using ICD-9 codes.
Population Studied: 580,942 female inpatients, whose primary procedure was a CABS graft, aggregated from the Healthcare Cost and Utilization
Project (HCUP) national database for the years
2000-2006.
Principal Findings: 28% of all women over this 7year period underwent off-pump CABS versus 72% who received on-pump CABS. For the 7-year period, for all female CABS patients, unadjusted inhospital mortality rates averaged 3.3% but have fallen 33% - from 3.97% in 2000 to 2.66% in 2006.
In-hospital mortality rates have fallen for both groups of CABS women over the seven-year period. For off-pump CABS women, in-hospital mortality dropped from 4.14% in 2000 to 2.89% in
2006 – a 30.2% decline. For on-pump CABS women, in-hospital mortality declined from 3.68% in
2000 to 2.55% in 2006 – a 30.7% decline. Over the
2000-2006 period, mean female off-pump CABS inhospital mortality rates were 7.7% higher than onpump CABS rates (3.5% vs. 3.25%, respectively;
P<0.001). In 2006, mean in-hospital mortality rates for women undergoing off-pump CABS were 13.3% higher than off-pump women (2.89% vs. 2.55%: p=0.035). Across racial/ethnic groups these unadjusted findings varied. Over the 2000-2006 period, Asian women undergoing off-pump surgery had the highest in-hospital mortality of any of the racial/ethnic groups with 4.97% while on-pump
Asian women experienced a 3.04% rate of inhospital mortality (p<0.001). Hispanic women undergoing off-pump CABS mortality had lower inhospital mortality rates than Hispanic women undergoing on-pump CABS although the results were not statistically significant (2.93% vs. 3.25%; p=0.141). For black women, in-hospital mortality for on-pump CABS vs. off-pump CABS was 3.83% vs.
3.22%, respectively (p=0.004) while for white women the off- vs. on-pump in-hospital mortality differential was 3.49% vs. 3.26% (p=0.001).
Multivariate logistic results indicate that, when compared to white women undergoing on-pump
CABS, black and Asian women undergoing offpump CABS have significantly higher in-hospital mortality rates. Compared to white women undergoing on-pump CABS, black women undergoing off-pump CABS experienced 18%
(p=0.007) higher in-hospital mortality rates and
Asian women undergoing off-pump CABS experienced 35% (p=0.012) higher in-hospital mortality rates.
Conclusion: Peer-reviewed studies have suggested that off-pump CABS techniques appear to decrease early deaths for females. Our investigation of in-hospital mortality rates over the past seven years using a 20% sample of the nation’s discharges does not find support for these assertions. Looking at national results for the past seven years, the potential benefit of on-pump CABS surgery for women appears to be better for black and Asian women and no worse than off-pump surgery for white and Hispanic women. The analysis provides evidence that U.S. gender and race disparities in CABS in-hospital patient inhospital mortality appear to be declining and some of the underlying gender and race disparities are ameliorated when other patient comorbidities and risk factors are included. However, significant differences in in-hospital CABS mortality for women still remain.
Implications for Policy, Delivery or Practice: The potential advantages of both off-pump CABS and on-pump CABS procedures are widely marketed with conflicting claims. The public would be wellserved by more systematic and objective analyses.
Theme: Disparities
● Behavioral Economics Model for Patient
Safety
Raj Behal, M.D., M.P.H.
Presented by: Raj Behal, M.D., M.P.H., Senior
Patient Safety Officer, Rush University Medical
Center, 600 South Paulina Street 544AF, Chicago,
IL 60612, Phone: (312) 942-6706; Email: raj_behal@rush.edu
Research Objective: Healthcare providers’ behaviors that affect patient safety are poorly understood. The current thinking about human errors treats providers as passive actors that are prone to err. The objective of this research study is to develop a behavioral economics model to characterize why and how workers’ own utility, values and risk preferences emerge as behaviors we might later label as errors.
Study Design: This qualitative study is based on observations of harm events and associated phenomena at several hospitals. An approach known as cycles of theory building was applied to classify and organize such observations into a model. Provider behaviors were examined using concepts from behavioral economics. Specific constructs from Game Theory and Prospect Theory were applied to the emergent model. The model was re-evaluated and modified over time with further observations.
Population Studied: Healthcare providers involved in events that resulted in harm or that were near misses.
Principal Findings: Healthcare provider behaviors that result in harm or in harm avoidance can be adequately characterized by constructs of bounded rationality, bounded self-interest, and bounded willpower. Behaviors potentially resulting in harm are further shaped by providers’ own utility and a threshold or tolerance for risk. Nash equilibriums denoting conflicts among healthcare providers were identified.
Conclusion: A model based on behavioral economics can provide additional insights into how behaviors affecting patient safety emerge among healthcare providers.
Implications for Policy, Delivery or Practice:
Further research building on this model can be useful in devising specific management
interventions to reshape provider behaviors and to improve patient safety.
Theme: Organizational Performance and
Management
● Estimating the Need for Pediatric
Subspecialists: The Case of Pediatric
Rheumatology
Heather Beil, M.P.H.; Michelle Mayer, Ph.D.,
M.P.H.; Christy Sandborg, M.D.
Presented by: Heather Beil, M.P.H., Ph.D.
Student, Health Policy & Management, University of
North Carolina at Chapel Hill, CB #7411, Chapel
Hill, NC 27599, Phone: (919) 966-7769; Email: hbeil@email.unc.edu
Research Objective: In contrast to most physician specialties, pediatric subspecialties can be characterized as having relatively small numbers of providers who are concentrated in urban areas and academic medical centers. Despite these unique characteristics, few physician workforce studies have examined pediatric subspecialties. This study aimed to illustrate an approach to assessing the supply of pediatric medical subspecialists and estimate the number needed nationwide, using pediatric rheumatology as an example.
Study Design: Physician data from the American
Board of Pediatrics (ABP) and population data from
Claritas were used to estimate the number of hospital referral regions (HRR) without pediatric rheumatologists and provider to pediatric population ratios for each HRR. Additionally, we surveyed all pediatric rheumatologists listed in the American
College of Rheumatology Membership directory
(n=236). Survey responses were used to estimate average patient volume and inform our need-based estimates. Estimates of the number of pediatric rheumatologists needed in each HRR are calculated under three scenarios: 1) based on the volume of patients per provider reported in the survey and national prevalence estimates; 2) based on previous estimates of a needed population size to support one provider and pediatric population only; and 3) based on the needed pediatric population size and the number of general pediatric residencies in each HRR. We also performed sensitivity analyses at the state level.
Population Studied: This study provided national estimates on pediatric rheumatologists and the patients they serve.
Principal Findings: Approximately 200 pediatric rheumatologists are located in 25% of hospital referral regions in the United States. Our estimates suggest that, on average, there are 2500 children with rheumatic diseases per pediatric rheumatologist per HRR. Respondents of the survey reported caring for an average of 464 patients per provider. The current distribution of pediatric rheumatologists creates a situation in which a substantial portion of the under-18 population in the U.S. lives more than 40 miles, i.e., an approximately one-hour drive, from a provider.
Of the providers we surveyed, 80% reported that the farthest distance 2 or more patients travel to see them is greater than 80 miles. Using the reported patient volume of 470 patients per provider and prevalence estimates of 390 per 100,000 children, 535 pediatric rheumatologists are needed nationally. Using HRR level population data, we estimate that at least 257 pediatric rheumatologists are needed nationally based on pediatric population only. When medical education is incorporated into need estimates, the number of pediatric rheumatologists needed nationwide is 331.
Conclusion: The current number of pediatric rheumatologists in the United States is inadequate, and the patient volume of current pediatric rheumatologists may be too low. The problem is particularly acute in a number of heavily populated hospital referral regions that have either no pediatric rheumatologists or a small number relative to their pediatric population.
Implications for Policy, Delivery or Practice:
Efforts to increase the availability of pediatric rheumatology expertise are needed to meet demand for patient care and medical education.
Moreover, the model developed here could be applied to assess the supply of other academic pediatric subspecialties.
Funding Source(s): Bureau of Health Professions
Theme: Child Health
● Determining Drivers of Excess Mortality in an
Improvement Collaborative of Acute Care
Hospitals
Kathy Belk, B.A.; Richard Bankowitz, M.D., M.B.A.,
F.A.C.P.; Eugene Kroch, Ph.D.
Presented by: Kathy Belk, B.A., Senior Research
Analyst, Healthcare Informatics, Premier Inc., 2320
Cascade Pointe Boulevard, Charlotte, NC 28208,
Phone: (704) 733-5638; Email: kathy_belk@premierinc.com
Research Objective: To determine which disease groups (defined by APR-DRG) from a population of
144 acute care facilities represent the populations with the largest number of “excess deaths” (deaths in excess of predicted by a normative riskadjustment regression model).
Study Design: The Premier QUEST collaborative is a dynamic program designed to assist hospitals and healthcare systems achieve top performance across five dimensions: reduction of mortality, prevention of harm, optimization of cost, deliver of evidence based care and patient satisfaction. In order to direct the development of targeted mortality improvement collaboratives, we examined the number of “excess deaths” by APR-DRG for each hospital in the cohort using a regression model developed by CareScience designed to adjust for age, sex, severity of illness, comorbidities present
on admission, socioeconomic and other factors.
Excess mortalities defined as (observed – expected) * population N, were determined for every hospital by every APR-DRG. At the hospital level excess deaths for an APR-DRG can be negative (when observed mortality is less than expected) Because we were interested in finding the total opportunities for improvement across the entire QUEST cohort, the total excess mortality for any APR-DRG was calculated using only from those hospitals with positive excess mortality in that
APR-DRG. The results were sorted by APR-DRG in a Pareto fashion resulting in a prioritized list of
“actionable” clusters for improvement for the cohort.
Population Studied: The population included more than 2 million acute inpatients for QUEST participating hospitals during the baseline timeframe, which spans the period July 1, 2006 through June 30, 2007, that had APR-DRGs
(N=144). Patients were excluded if they were transferred to another acute-care facility or if they were missing a value that is critical to the risk adjustment methodology. Sub-populations of patients are defined using APR-DRGs. Participants represent teaching and non-teaching hospitals that range in bed size from 24 to 986 (median=283) and are distributed across 30 states.
Principal Findings: The top 10 APR-DRGs (out of
286 APR-DRGs with opportunity) in the resulting sorted list represented 27.4% of all excess deaths.
The sub-groups that included Septicemia, respiratory diagnoses, AMI, heart failure, and
CVA/Precerebral Occlusions with Infarct were identified as the areas that had the most opportunity for mortality reduction.
Conclusion: A very small number of disease groupings (10) accounted for almost 30% of all deaths in excess of predicted by a regression model specifically developed to adjust for patient risk. Sepsis accounted for the single largest gross number of excess deaths (593 excess deaths).
Implications for Policy, Delivery or Practice: In a large, diverse population of hospitals reasonably representative of all acute care facilities in the US, a small number of “usual suspect” disease groupings were found to account for the majority of “excess” deaths. Such a finding suggests that targeting specific populations with collaborative interventions designed to eliminate or reduce preventable deaths in these populations would be effective. In every case, it is possible to identify “top performers” whose observed mortality was much less than predicted by the model. Such institutions represent opportunities for discovery and dissemination of best practices to others in the cohort.
● Access to Care, Discrimination & Distrust on
Cervical Cancer Prevention & Control in a
Population-Based Survey of Women in South
Carolina
Jessica Bellinger, M.P.H., Ph.D.; Janice Probst,
Ph.D.; Heather Brandt, Ph.D.; Saundra Glover,
Ph.D.; Amy Brock Martin, Dr.P.H.; James Hardin,
Ph.D.
Presented by: Jessica Bellinger, M.P.H., Ph.D.,
Postdoctoral Fellow, University of South Carolina,
Health Services Policy & Management, South
Carolina Rural Health Research Center, 220
Stoneridge Drive, Suite 204, Columbia, SC 29210,
Phone: (803) 251-6317; Email: bellingj@mailbox.sc.edu
Research Objective: Pap tests and other effective screening methods have contributed to reduced cervical cancer mortality. However, the positive effects of appropriate screening have not been distributed equally across groups, contributing to excess mortality. Perceived discrimination and distrust in the health care system have been posited as potential contributors to health disparities. The objective of this study was to examine the effects of access to care, experiences of discrimination, and health system distrust on cervical cancer prevention and control in South
Carolina.
Study Design: Random-digit dial telephone interviews were used to collect data in a populationbased survey of women in South Carolina with an oversample of African Americans. This research was conducted as a substudy of a parent study designed to measure women’s knowledge, behavior, and attitudes related to cervical cancer and HPV. The survey was administered using random-digit dialing computer assisted telephone interviewing (CATI) from a database of all operating telephone numbers in the state. Self-reported measures of access included health insurance, geographical and travel impediments, and a regular source of care. The prevalence of perceived discrimination and health system distrust was measured using Krieger’s Experiences of
Discrimination Scale and Rose’s Health System
Distrust Scales. Appropriate cervical cancer screening was measured by a Pap test within three and five years of interview. Logistic regression was used to determine outcome predictors using appropriate survey design procedures.
Population Studied: Potential subjects were randomly selected. Inclusion criteria were adult women living in South Carolina (18-70), working landline telephones and the ability to hear, speak, and understand English. The interview response rates for the primary sample and the African
American oversample were 36.7% and 34.0% with
1,029 respondents. Incomplete interviews and
“other” minority groups (Hispanics and Asians) were excluded from analysis (n=986).
Principal Findings: Respondents reported experiences of discrimination in distinct situations, including health care. Over half (57.3%) of the sample reported never experiencing any perceived discrimination in any situation during their lifetime.
More African Americans reported high experiences
of perceived discrimination (34.2%) than whites
(4.0%) (p<.0001). Access was a greater predictor of appropriate cervical cancer screening than perceived discrimination or health system distrust.
Women with a regular provider (OR=2.819 CI
1.449, 5.483; OR=2.356 CI 1.128, 4.920) and private health insurance (OR=2.970 CI 1.364,
6.469; OR=2.769 CI 1.146, 6.694) were more likely to report 3- and 5-year Pap tests. African American women with low perceived discrimination were less likely to report Pap tests within five years than those with no perceived discrimination (OR=0.239 CI
0.060, 0.949).
Conclusion: The prevalence of perceived discrimination and distrust could contribute to health disparities. Both racial groups reported at least some lifetime experience with perceived discrimination, while African Americans had markedly higher odds of reporting such experiences. All respondents reported distrust; however African Americans had higher levels of distrust. Access was the strongest predictor of adherence to Pap test recommendations.
Implications for Policy, Delivery or Practice:
Improved access, especially to safety net health centers, would increase cervical cancer screening and ultimately lead to reduced cervical cancer mortality. Educating providers and at-risk women about revised recommendations would ensure eligible women receive appropriate Pap tests.
Funding Source(s): NCI
Theme: Disparities
● CDC’s Cervical Cancer Study (Cx3): An
Intervention Pilot Study of HPV in Illinois
NBCCEDP
Vicki Benard, Ph.D.
Presented by: Vicki Benard, Ph.D., Epidemiologist,
Division of Cancer Prevention, Centers for Disease
Control & Prevention, 4770 Buford Highway, Mail
Stop K-55, Atlanta, GA 30341, Phone: (770) 488-
1092; Email: vdb9@cdc.gov
Research Objective: The National Breast and
Cervical Cancer Early Detection Program
(NBCCEDP) is the only organized national screening program in the U.S. that offers breast and cervical cancer screening to underserved women.
HPV test is not currently a reimbursable expense under NBCCEDP guidelines. Adopting HPV testing along with Pap testing in women over 30 could help the program better utilize resources by extending the screening interval of women who have a normal
Pap test and negative HPV test, estimated to be 80-
90% of women. The proposed study will examine whether select NBCCEDP providers adhere to the recommended 3-year cervical cancer screening interval for women in the target age range with a normal Pap test and negative HPV test.
Study Design: The proposed pilot intervention study will be conducted in 6 health systems (with 14
NBCCEDP clinics) in Illinois in the summer of 2009.
The 6 health systems will be assigned to one of two study arms. Clinics in the intervention group will receive HPV tests to administer to eligible patients presenting for a routine Pap test PLUS a multicomponent educational intervention involving both providers and patients. Clinics in the comparison group will receive the HPV tests but will not receive the educational intervention. Patients of providers in both arms will receive an HPV test in addition to the
Pap test at the time of routine cervical cancer screening. Both arms will be matched on geographical location (urban/rural), racial/ethnic diversity, HPV policy, hospital versus non-hospital, provider specialty mix, and patient volume.
Population Studied: Women 35 to 60 years of age who are visiting one of 14 participating clinics for routine cervical cancer screening.
Principal Findings: Clinic Coordinators at each of the 14 participating clinics will be surveyed monthly for the first year of the study to obtain information regarding resources associated with participating in the study. Approximately 8,000 women will be tested with the HPV test and their follow-up will be reviewed via medical records at the end of the study. All providers in both study arms (n= 70) who routinely perform Pap tests at the participating clinics will be surveyed at four points in time to assess knowledge, attitudes, and beliefs regarding cervical cancer screening practices. A sample of
2,600 patients (1,300 in each arm) will be asked to complete a baseline survey and two follow-up surveys to assess knowledge, attitudes, beliefs, and behavior regarding cervical cancer screening.
Conclusion: The results of this study will provide information regarding the extent to which providers are willing to extend the cervical cancer screening interval to 3 years for women in the target age range with a normal Pap test and a negative HPV test.
Implications for Policy, Delivery or Practice: The findings from this study will help inform policy development regarding the HPV test on a national level for cervical cancer screening in the
NBCCEDP.
Funding Source(s): CDC
● The San Mateo County Medical Center
Redesign: An Evaluation of Efforts to Improve
Efficiency & Optimize Paitent Care
Sarah Benatar, Ph.D.; Embry Howell, Ph.D.; Dana
Hughes, Dr.P.H.
Presented by: Sarah Benatar, Ph.D., Research
Associate, Health Policy Center, The Urban
Institute, 2100 M. Street, NW, Washington, DC
20037, Phone: (202) 261-5259; Email: sbenatar@urban.org
Research Objective: The Urban Institute is conducting a two-year evaluation of the San Mateo
Medical Center’s (SMMC) effort to redesign its
service delivery system to achieve greater efficiency and optimal patient care. SMMC is an integrated inpatient and ambulatory care network in
San Mateo County, California, that includes a small public hospital and 11 community clinics, six of which are dedicated to adult medicine. Care provided by SMMC is widely perceived to be of generally high quality but it has struggled to meet the growing needs of the community. Patients experience long wait times to appointments, a scarcity of specialty services, and others barriers to having their health needs met. This work evaluates the county’s efforts to improve adult access to care, integrate innovative interventions to facilitate medical information exchange, and implement patient-centered system-wide efficiencies.
Study Design: This evaluation utilizes qualitative and quantitative analyses. Findings presented are based on the results of the year-one case study, which comprised an intensive one-week site visit, over thirty key informant interviews, and clinic waiting room observations. Furthermore, we have collected initial clinic-level data on patient satisfaction, cycle time (a measure of how long a patient waits to be seen), and wait-time to next appointment, as well as initial utilization data.
Population Studied: Adults served by SMMC, with particular focus on those with chronic conditions and those enrolled in the county’s Access to Care for Everyone (ACE) program, which extends coverage to uninsured adults up to 200 percent of the federal poverty level.
Principal Findings: Initial findings illustrate that access to specialty care in the county system is particularly challenging. Baseline cycle time data indicate that while most clinics have achieved the county’s goal of cycling patients through in less than one hour, the surgical specialty clinic and the medical specialty clinic still regularly report cycle times exceeding 70 minutes. Descriptive findings from the year-one case study indicate that there is a severe shortage of specialists working with the county, and that they are particularly resistant to using the electronic medical record system that is being rolled out. Moreover, only one-quarter of county physicians provide specialty care resulting in very long wait times to appointments and substantial unmet need. Since the introduction of
ACE, SMMC has enrolled nearly 5000 uninsured adults. Nearly 80 percent of those enrolled have utilized medical services and 44 percent have utilized pharmacy benefits. On average, patients have been enrolled for 6.1 months.
Conclusion: Each of the seven adult county clinics is in a different stage of implementing disease management registries, team-based care models, and meeting goals of improving visit cycle times and wait times for appointments. Access to specialty care for chronically ill adults is a particular challenge for the county.
Implications for Policy, Delivery or Practice:
SMMC’s shortage of specialty physicians poses a challenge for chronically ill adults who are reliant on the county health system. Moreover, these access barriers impact organizational costs and efficiencies and should be a focus of the county’s efforts to improve patient-centered care and disease management in this population.
Funding Source(s): San Mateo County, CA
Theme: Organizational Performance and
Management
● Caveat Medicus: The Evolving Landscape of
Federal & State Investigations of Health Care
Fraud
Charles Bennett, M.D., Ph.D., M.P.P.; June McKoy,
M.D., J.D., M.P.H.; Joseph White, J.D.; Thanh Ha
Luu, B.A.; Elizabeth Richey, B.A.; Oliver Sartor,
M.D.
Presented by: Charles Bennett, M.D., Ph.D.,
M.P.P., A.C. Buehler Professor,
Hematology/Oncology, VA Center for Management of Complex Chronic Care, Northwestern University,
710 North Fairbanks Court 8250, Chicago, IL
60612, Phone: (312) 503-0804; Email: cbenne@northwestern.edu
Research Objective: Federal and state health care programs account for 46% of all health care expenditures in the United States. An estimated
10% of these expenditures are attributed to health care fraud. We reviewed all Federal and state health care fraud investigations concluded since
1996.
Study Design: Data sources included the
Department of Justice’s website (www.usdoj.gov) ; state attorneys general press release archives; the
Taxpayers Against Fraud (TAF) website
(www.taf.org); and LexusNexus searches (key words: the qui tam relator’s name, “false claims act,” “qui tam,” “fraud,” or “Whistleblower”).
Searches were from 1996 through 2008. Median and total financial recoveries are reported for
Federal and state investigations and for payments to qui tam relators.
Principal Findings: Resolution of 939 Federal health care fraud cases and 23-state-led Medicaid fraud cases resulted in fines/recoveries of $16.9 billion and $1.1 billion, respectively. Mean annual recoveries for Federal and state-led investigations were $500 million and $1.7 million (1996 to 2000),
$1.5 billion and $8.2 million (2001 to 2005), and
$2.3 billion and $105.5 million (2005- 2008), respectively. Among Federal cases, $9.3 billion in recoveries were paid to resolve allegations that pharmaceutical/ device manufacturers provided kickbacks, promoted off-label marketing, or reported inflated average wholesale prices Two hospital chains paid $2.6 billion to resolve allegations of paying kickbacks or filing fraudulent cost reports. Among 14 Medicaid fraud investigations concluded prior to 2006, 13 involved only one state. Since 2006, eight Medicaid fraud
investigations concluded; seven involved several states. The two largest Medicaid cases involved allegations that Merck pharmaceuticals had submitted fraudulent cost information (settled by 34 states for $400 million) and allegations of not providing prenatal care to women enrolled in a health maintenance organization (settled for $225 million by Illinois).
Conclusion: Fraud against Federal and state health care programs is extensive and involves a wide range of health care providers.
Implications for Policy, Delivery or Practice:
With ever-increasing scrutiny of health care costs, the scope of Federal and state health care fraud investigations will continue to expand.
Theme: Quality and Efficiency: Policies and
Incentives
● Accelerated Approval of Cancer Drugs:
Improved Access to Therapeutic Breakthroughs or Early Release of Unsafe & Ineffective Drugs?
Charles Bennett, M.D., Ph.D., M.P.P.; Elizabeth,
Richey, B.A.; Allison Lyons, B.A.; Jonathan
Nebecker, M.S. M.D.; Steve Trifilio, R.Ph.; Steven
Hirschfeld, M.D., Ph.D.; Antonio Grillo-Lopez, M.D.
Presented by: Charles Bennett, M.D., Ph.D.,
M.P.P., A.C. Buehler Professor,
Hematology/Oncology, VA Center for Management of Complex Chronic Care, 710 North Fairbanks
Court 8250, Chicago, IL 60612, Phone: (312) 503-
0804; Email: cbenne@northwestern.edu
Research Objective: Accelerated approval (AA) of oncology drugs used to treat serious illnesses by the Food and Drug Administration (FDA) was initiated in 1995. This regulatory review is designed to shorten development times for drugs that treat serious illnesses based on identification of improvements in surrogate clinical outcomes.
Pharmaceutical sponsors are required to confirm efficacy in post-approval clinical trials. Recently, policy makers have raised concern that AA results in approval of ineffective or unsafe drugs and that sponsors fail to complete the required confirmatory clinical trials.
Study Design: We reviewed details of AA and regular approvals by the FDA of new molecular entities (NMEs) for therapeutic oncology indications from 1995 to August 2008. Data were reviewed on orphan drug status, clinical development time, black box warnings, and status of confirmatory efficacy clinical trials.
Principal Findings: AA accounted for 78% of FDA approvals for oncology NMEs between 2000 and
2003 versus 38% more recently. Median development times for AA NMEs (n= 19 drugs) and regular-approved oncology NMEs (n= 32 NMEs) were 7.3 and 7.8 years, respectively. Efficacy was confirmed for 42% of orphan drug oncology NMEs that received AA and 86% of non-orphan drug oncology NMEs that received AA. In survival analysis, confirmatory clinical trials for orphan oncology NMEs that received AA were 10 times less likely to be successfully completed than for non-orphan oncology NMEs that received AA. Black box warnings were added to package inserts of two regular approved oncology NMEs and four AA oncology NMEs.
Conclusion: AA oncology NMEs are safe and effective, although development times are not accelerated.
Implications for Policy, Delivery or Practice:
New strategies to re-invigorate the AA process for oncology NMEs, particularly those designed to treat orphan drug cancer indications, are needed.
Punitive actions included in recently proposed legislation such as assessing fines for failing to complete post-approval commitment studies should be undertaken with caution for sponsors of drugs that receive AA for oncology indications. Such an approach would represent a particular disservice to patients who have cancers associated with orphan drug indications.
Funding Source(s): NCI
Theme: Quality and Efficiency: Policies and
Incentives
● Rural Minorities & Health Status: The Dually
Disadvantaged
Kevin Bennett, Ph.D., M.S.
Presented by: Kevin Bennett, Ph.D., M.S.,
Assistant Professor, Family & Preventive Medicine,
University of South Carolina, 3209 Colonial Drive,
Columbia, SC 29203, Phone: (803) 434-3611;
Email: kevin.bennett@sc.edu
Research Objective: To explore the health status, chronic disease prevalence, and disability disparities experienced by rural minorities.
Study Design: Cross-sectional analysis using the
2005-2006 Behavioral Risk Factor Surveillance
Survey merged with the 2005 Area Resource File.
Bivariate analysis will identify the proportion, by demographic characteristics and residence, of those with the outcomes of interest. Multivariate logistic regression will identify the significant contributors to each outcome. Independent variables such as health indicators, insurance status, and environmental factors will be utilized as well. Rural residence was defined using urban influence codes, with four categories; urban, micropolitan, small adjacent rural, and remote rural.
All analyses were weighted to produce nationally representative estimates.
Population Studied: Adults, age 18 and over, included in the 2005 or 2006 Behavioral Risk Factor
Surveillance Surveys.
Principal Findings: This analysis utilized 289,877
(179,988,705 weighted) adults over the age of 18, with 89.8% living in urban areas, 8.0% in micropolitan areas, 1.5% in small adjacent rural areas, and 0.7% in remote rural areas. The overall
population was 67.9% white,10.2% black,6.8
Hispanic, and 15.1 other race.
In general, as the rurality of a location increased, the health of the population decreased. Selfreported health status was lower among rural residents (18.5% reported fair or poor health) compared to urban residents (15.5%). Chronic disease prevalence was significantly higher among small remote residents than urban residents ; the diabetes prevalence ranged from 7.9% in urban areas to 10.5% in small remote rural areas.
Physical limitations were also higher among rural residents, with 28.5% of small remote rural residents reporting a limitation (compared to only
19.5% among urban residents). Overweight prevalence, however, was lower among rural residents (48.9%) than urban residents (51.8%).
Minorities living in rural areas had worse health indicators than others. Nearly two-thirds of blacks living in remote rural areas (33.0%) reported fair or poor health, compared to only 12.5% of whites living in urban areas. Blacks in remote rural areas also had more than twice the rate of diagnosed diabetes than urban whites. In multivariate analyses, rural residence remained associated with health status. Remote rural residents were less likely than urban residents to report good to excellent health (OR = 0.73 (0.62, 0.85)). Race was also a significant factor, with blacks (OR = 0.66
(0.58, 0.75)) and Hispanics (OR = 0.65 (0.55, 0.77)) less likely to report good to excellent health. Even controlling for morbidity and those with insurance were more likely to report good to excellent health.
Similar trends are found for limitations in activities of daily living.
Conclusion: These findings indicate that rural minorities, controlling for other factors, report being in poorer health than other rural residents as well as whites. The lower reported health status, combined with higher disease prevalence and limitations, indicate a greater need for population-specific health care services.
Implications for Policy, Delivery or Practice:
Policy makers need to be aware of differences in need across communities. These findings indicate that certain groups, such as rural minorities, are in greater need of chronic disease care coordination as well resources to aide this with physical limitations. Given the difficulties in obtaining necessary resources experienced by many rural communities, novel approaches need to be identified to provide such services.
Funding Source(s): HRSA
Theme: Disparities
● Preventive Service Delivery among Rural
Adults
Kevin Bennett, Ph.D., M.S.
Presented by: Kevin Bennett, Ph.D., M.S.,
Assistant Professor, Family & Preventive Medicine,
University of South Carolina, 3209 Colonial Drive,
Columbia, SC 29203, Phone: (803) 434-3611;
Email: kevin.bennett@sc.edu
Research Objective: To estimate the compliance rate among rural minorities for several preventive services, such as mammography and colorectal cancer screening, and to identify the relevant factors that contribute to the receipt of these services.
Study Design: Cross-sectional analysis using the
2005-2006 Behavioral Risk Factor Surveillance
Survey merged with the 2005 Area Resource File.
Bivariate analysis will identify the proportion, by demographic characteristics and residence, of those with the outcomes of interest. Multivariate logistic regression will identify the significant contributors to each outcome. Dependent variables include having a physical exam, having a dental exam, and obtaining a recommended mammogram,
Pap smear, colorectal cancer screen, influenza vaccine or pneumococcal vaccine. Independent variables such as health indicators, insurance status, and environmental factors will be utilized as well. Rural residence was defined using urban influence codes, with four categories; urban, micropolitan, small adjacent rural, and remote rural.
All analyses were weighted to produce nationally representative estimates.
Population Studied: Adults, age 18 and over, included in the 2005 or 2006 Behavioral Risk Factor
Surveillance Surveys. Certain measures further delimit the population; mammography is limited to women age 40 and over, colorectal cancer screens are limited to adults age 50 and over, and the influenza and pneumococcal vaccines are limited to adults 65 and over.
Principal Findings: This analysis utilized 289,877
(179,988,705 weighted) adults over the age of 18, with 89.8% living in urban areas, 8.0% in micropolitan areas, 1.5% in small adjacent rural areas, and 0.7% in remote rural areas. The overall population was 67.9% white,10.2% black,6.8
Hispanic, and 15.1 other race.
In general, as the level of rurality increased, the residents were less likely to receive services. The largest gaps in service delivery were in annual dental exams, mammography, colorectal cancer screens, and Pap smear. Influenza and pneumococcal vaccination rates were similar for among urban and rural residents, although small adjacent rural residents had significantly lower rates than other residents. Blacks were more likely than those of other races to have an annual physical exam or a Pap smear, but were less likely to have had a dental exam, or screen for colorectal cancer.
Holding all other factors equal in multivariate analysis, rural residents and blacks were less likely to report an annual physical exam. Having a usual source of care was the strongest predictor of an annual exam (OR = 3.76 (3.50, 4.05)). Similar results were found for annual dental exams, with having insurance being a strong factor as well (OR
= 2.44 (2.22, 2.67)). Despite the initial bivariate analyses, the multivariate analysis showed that blacks were more likely to obtain a screen for colorectal cancer (OR = 1.13 (1.01, 1.27)), while rural residents are still less likely to obtain such a screening; similar results were found for mammography, but to a lesser degree. Rural residence was less of a factor for Pap smear, influenza, and pneumococcal vaccination than for the other services studied.
Conclusion: These results indicate that rural minorities obtain some preventive services as recommended, even at higher rates than urban residents or whites, but continue to lag behind in others. The overall compliance rates for some services (e.g. 54.7% obtaining a colorectal cancer screen among small remote rural residents) indicate a large proportion of the population not getting recommended care.
Implications for Policy, Delivery or Practice:
Rural minorities still lag behind whites and urban residents in obtaining several recommended preventive services. The finding that some services are delivered to blacks at a higher rate
(e.g. mammography and colorectal screening) indicate that targeted interventions can improve these rates, but more work needs to be done target rural residents to meet their needs.
Funding Source(s): HRSA
Theme: Public Health
● Tool for Evaluating the Quality of
Observational Studies Included in Systematic
Reviews of the Literature
Nancy Berkman, Ph.D.; Meera Viswanathan, Ph.D.
Presented by: Nancy Berkman, Ph.D., Senior
Health Policy Research Analyst, Program for Health
Care Quality & Outcomes, RTI International, 3040
Cornwallis Road, Research Triangle Park, NC
27709, Phone: (919) 541-8773; Email: berkman@rti.org
Research Objective: Systematic reviews of the literature require the evaluation of the quality of each study included in the review. Reviews that include observational studies must critically appraise bias in selection, performance, detection, and attrition (Higgins 2006) in addition to domains typically evaluated for randomized studies. These reviews are handicapped by the lack of validated and adaptable instruments for evaluating the quality of these studies. The idiosyncrasies of topics often results in each new review developing its own quality rating tool, leading to inconsistent standards within and across reviews.
The objective of the project to create a practical and validated tool for evaluating the quality of observational studies of interventions or exposures that: (1) captures all of the domains critical for evaluating this type of research; (2) can be easily lifted "off the shelf" by different researchers; and (3) can be adapted to different topic areas and observational study types (e.g., cohort, case control, cross-sectional and case series studies).
Study Design: The study to date has two phases.
The first focused on development of the tool
(completed) and the current phase on validation, including cognitive and inter-rater reliability testing.
Phase 1 included the compilation of a large number of potential quality review items which, primarily, had been used previously in AHRQ-sponsored systematic evidence reports. To ensure that items encompassed important domains related to risk of bias and threats to validity, items were sorted into domains identified by Deeks et al. (2003). A prototype instrument was created containing a parsimonious list of items within all relevant domains. Additionally, instructions for use and directions for interpreting individual items for
Principal Investigators and abstractors were included. Phase 2 of the project includes: (1) testing the face validity of the instrument; (2) cognitively testing the instrument for readability; (3) further validating the instrument through intra- and interrater reliability testing; and (4) revising the instrument based on these reviews.
Population Studied: Published observational studies of interventions or exposures.
Principal Findings: A validated tool can be developed for evaluating the quality of observational studies of interventions or exposures.
Conclusion: A validated tool evaluating the quality of individual observational studies can be developed and used across types of studies and topic areas. This tool will also be relevant to evaluate the risk of bias across studies when grading the strength of the body of evidence.
Implications for Policy, Delivery or Practice:
Increasingly, government payers are requiring the evaluation of the effectiveness or efficiency of individual treatments or the comparative effectiveness of competing treatments. As the focus of these studies expands to include a more generalized population than can be observed through randomized controlled trials, long term outcomes and harms (side effects), evidence is being gathered from observational studies. The evaluation of these studies requires new validated tools, such as the one created through this project
Funding Source(s): AHRQ, RTI International
Theme: Quality and Efficiency: Measurement
● Wealth, Income & the Affordability of Health
Insurance
Didem Bernard, Ph.D; Jessica Banthin, Ph.D.;
William Encinosa, Ph.D.
Presented by: Didem Bernard, Ph.D, Senior
Economist, Committee For a Constructive
Tomorrow, Agency for Healthcare Research &
Quality, 540 Gaither Road, Rockville, MD 20850,
Phone: (301) 427-1682; Email: dbernard@ahrq.gov
Research Objective: The importance of wealth in determining consumption has long been recognized. However, the role of assets and wealth in health insurance purchase decisions is only recently being recognized. The recent fall in the financial markets and the decline in housing values make estimates of wealth effects on health insurance especially relevant.
Study Design: There have been debates over how many of the uninsured can afford insurance but refuse to purchase it. Affordability models based on income lead to a puzzle: some with low income buy insurance while some with high income do not. For example, in 2006, 20.6 percent of non-elderly adults
(3.8 million) with income below the federal poverty line purchased private insurance. In contrast, 20.7 percent of non-elderly adults (32.5 million) above the poverty line remained uninsured. Therefore, income alone does not fully explain insurance purchase decisions. Using the Medical Expenditure
Panel Survey, we examine the difference in asset holdings between the privately insured and the uninsured, controlling for income. Furthermore, we estimate the effect of wealth on insurance take up using an instrumental variable approach with panel data from 1996 to 2006. Market-level variation in housing values over time is used as a source of exogenous variation in asset values.
Population Studied: Nonelderly adult population aged 21 to 64.
Principal Findings: Our preliminary findings show that the difference in purchasing power is not fully revealed by income comparisons. For example, using pooled data from 2002 and 2003, we find that while median income among the privately insured is
2.9 times the median income of the uninsured, median wealth among those with private insurance is 23.2 times the median wealth of the uninsured.
Conclusion: Our result suggest that assets are an important determinant of effective affordability, and need to be taken into account when determining factors associated with being uninsured.
Implications for Policy, Delivery or Practice:
Estimates of the effect of wealth on health insurance take up can help policymakers predict the impact of economic recessions on insurance take up.
Funding Source(s): AHRQ
Theme: Coverage and Access
● Small-Area Variation (SAV) in the Extent of
Physician Financial Self-Interest (FSI) in
Imaging & Its Effect on Variations in Imaging &
Non-Imaging Costs
Mythreyi Bhargavan, Ph.D.; Jonathan Sunshine,
Ph.D.
Presented by: Mythreyi Bhargavan, Ph.D.,
Director, Research, American College of Radiology,
1891 Preston White Drive, Reston, VA 20191,
Phone: (703) 715-4394; Email: mbhargavan@acr.org
Research Objective: The literature on small-area variation (SAV) in healthcare has repeatedly demonstrated large variations in healthcare utilization and costs among small areas across the country, with potentially huge savings possible from eliminating the variations. However, this literature has not identified the causes of variation. The literature on financial self-interest (FSI) finds substantially higher utilization of services when the treating physician has an FSI in providing the service (such as imaging or services in a physicianowned specialty hospital). This study brings together these two literatures. The objectives of this study are (a) to examine whether SAV in the extent of physician FSI in one type of service, namely imaging, explains much of the SAV in per beneficiary costs of that service, and (b) to examine whether the extent of FSI in imaging can explain variation in per beneficiary healthcare costs for other services.
Study Design: Data are from Medicare’s 5%
Research Identifiable Files for 2005. The small areas are the 306 hospital referral regions (HRRs) from the Dartmouth Atlas of Healthcare. Because
FSI can be difficult to identify unambiguously, four alternative definitions are used to identify it. The extent of FSI in imaging in an HRR is measured in two ways: percent of treating physicians who have
FSI in imaging, and the share of imaging relative value units (RVUs) initiated by physicians with FSI in imaging. Total healthcare costs and imaging costs will be calculated using total Medicare allowed charges (including patient cost-sharing).
Multivariable regression will used to measure the contribution of FSI in imaging to SAV in per beneficiary costs of imaging and non-imaging care in the HRR. The analysis will control for a number of factors: demographics (percent elderly, beneficiary age distribution, gender composition), socioeconomics (poverty rate, education, etc.), physician supply (radiologists, other specialists, and primary care physicians per 100,000 population), disease profile of the population (presence of comorbidities, county risk profiles), and price index for healthcare in the area.
Population Studied: Medicare enrollees, 2005
Principal Findings: Preliminary results show substantial variation in the utilization of imaging, with this variation strongly related to variation in physician FSI. With no controls, variation in percent of physicians with FSI explains 9% to 17%
(depending on the definition of FSI) of the variation in number of imaging services per fee-for-service beneficiary. Similarly, variation in the share of imaging ordered by physicians with FSI explains
17% to 35% of the variation in number of imaging services per fee-for-service beneficiary.
Conclusion: Further analysis is underway.
Multivariable regression results will more definitively
determine how much of the variation in costs FSI explains.
Implications for Policy, Delivery or Practice: If
FSI explains a substantial portion of the variation in total healthcare costs (or explains variations in imaging costs without offsetting effect on other healthcare costs), we will be able to devise a potential strategy for cutting costs that is centered on limiting FSI in imaging. Such a strategy should achieve some of the savings that the SAV research has suggested are possible.
Theme: Health Care Markets and Competition
● Underuse of Breast Cancer Adjuvant
Treatment: Patients’ Knowledge, Beliefs &
Medical Mistrust
Nina Bickell, M.D., M.P.H.; Jessica Weidmann,
B.A.; Kezhen Fei, M.S.; Howard Leventhal, Ph.D.
Presented by: Nina Bickell, M.D., M.P.H.,
Associate Professor, Health Policy, Mount Sinai
School of Medicine, One Gustave L. Levy Place,
Box 1077, New York, NY 10029, Phone: (212) 659-
9565; Email: nina.bickell@mountsinai.org
Research Objective: Little is known about why women with breast cancer who access surgery do not receive proven effective post-surgical adjuvant treatments.
Study Design: We surveyed 258 women recently surgically treated at 6 NYC hospitals for early-stage breast cancer about their care, knowledge and beliefs about breast cancer and its treatment. As per national guidelines, all women should have gotten an adjuvant treatment. Adjuvant treatment data were obtained from in & out-patient charts.
Factor analysis was used to create scales scored to
100 of treatment beliefs and knowledge, medical mistrust and physician communication about treatment. Bivariate and multivariate analyses assessed differences between treated and untreated women.
Population Studied: Women with a new primary stage I or II breast cancer undergoing surgical treatment at 1 of 6 participating NYC hospitals between 2004- 2006.
Principal Findings: Compared to treated women, untreated women were less likely to know adjuvant therapies increase survival (66 vs 75 on 100-point scale; p<0.0001), had greater mistrust (64 vs 53; p=0.001) and less self efficacy (92 vs 97; p<0.05); there was no association between physician communication about treatment and patient knowledge of treatment benefits (r=0.8; p=.21).
Multivariate analysis found that untreated women were more likely to be >70 years (aRR=1.11;
95%CI: 1.00-1.13), have comorbidities
(aRR=1.10;1.04-1.12), express mistrust in the medical delivery system (aRR=1.003;1.00-1.007) and believe adjuvant treatments beneficial
(aRR=0.99; 0.98-0.99) (model c=0.84; p=<0.0001).
Conclusion: Patient knowledge and beliefs about treatment and medical mistrust are mutable factors associated with underuse of effective adjuvant therapies.
Implications for Policy, Delivery or Practice:
Physicians may improve cancer care by ensuring that discussions about adjuvant therapy include a clear presentation of the benefits not just the risks of treatment and by addressing patient’s trust in and concerns about the medical delivery system.
Theme: Quality and Efficiency: Policies and
Incentives
● Breast Cancer Patients' Perceived Quality of
Care: The Importance of Trust &
Communication
Nina Bickell, M.D., M.P.H.; Rebeca Franco, M.P.H.;
Kezhen Fei, M.S.; Kathie-Ann Joseph, M.D.
Presented by: Nina Bickell, M.D., M.P.H.,
Associate Professor, Health Policy, Mount Sinai
School of Medicine, One Gustave L. Levy Place,
Box 1077, New York, NY 10029, Phone: (212) 659-
9565; Email: nina.bickell@mountsinai.org
Research Objective: As insurers consider paying for performance and quality measures grow in importance, factors that affect patients´ perceived quality of cancer care matter. We undertook this study to assess key predictors of women´s ratings of the quality of their breast cancer care.
Study Design: Study enrolled patients were telephone-surveyed to assess care experiences, knowledge, attitudes and beliefs about breast cancer and its treatment. Trust is based on a validated instrument and calibrated to a 100 point scale (Cronbach a=.73). A scale of 5 items assessing physician communication was created and calibrated to100 points (Cronbach a=.83). This scale includes items asking about their physician´s discussions about treatment, its pros and cons, what to expect and patient treatment preferences.
Race is self-reported. We used bivariate analyses and logistic models to identify factors associated with patient ratings of excellent quality of care. Trust and physician communication scales were dichotomized at the median when entered in the logistic models. Interaction terms were not significant and dropped out of the final model.
Population Studied: 210 of 300 eligible women with a new primary stage I or II breast cancer undergoing surgical treatment at 1 of 8 participating
NYC hospitals were enrolled in a RCT of Patient
Assistance to Reduce Disparities in Care in 2006-
2008 and responded to our survey (70% response rate): 43 (20%) were African-American, 85 (40%) were white, and 63 (30%) were Hispanic and 19
(9%) were other races.
Principal Findings: Overall, only 55% of women rated their quality of cancer care as excellent.
Compared to women who did not rate their care as excellent, those who did had greater trust in their
physician (97 vs 91 on 100 point scale; p<0.0001), better communication with their physician (89 vs 78; p<0.0001), indicated that were treated very well by their physicians´ office staff (59% vs 41%; p=0.01), and knew which physician to ask when they had questions about their breast cancer (87% vs 64%; p=0.0001) ; age, education and income were not significantly related to patient report of excellent cancer care. Of all ethnic groups, African-American women were less likely to rate their care as excellent (35% vs white 62% vs Hispanic 65%; p=0.004). African American women also had lower levels of trust in their physician (AA=92; W= 95, H=
96; p=0.02); there were no racial differences in physician communication. Patients reporting greater levels of physician communication also had greater trust (r=0.38; p<0.0001). Multivariate models evaluating the role of patient race, education, income, knowing which physician to talk to and how well the staff treated the patient found that being African American (aRR=0.47; 95%CI:
0.21-0.88), having greater trust in physician
(aRR=1.72; 95%CI:1.49-1.85) and better physician communication (aRR=1.38; 95%CI: 1.03-1.65) were significantly associated with patient perception of excellent quality care (model c=.80; p<0.0001).
Conclusion: Greater levels of physician communication about treatment and patient trust of their physician affect women´s ratings of excellent cancer care quality.
Implications for Policy, Delivery or Practice:
Efforts should be made to improve physician communication about treatment, particularly among
African American women, to improve levels of trust and ratings of cancer care quality.
Funding Source(s): NCI
Theme: Disparities
● Family Health History-New Generation Tool
Mary Beth Bigley, Dr.P.H., M.S.N.; Gregory
Downing, D.O., Ph.D.; W. Gregory Feero, M.D.,
Ph.D.; Christopher Lamer, Pharm.D., M.H.S.,
B.C.P.S., C.D.E.
Presented by: Mary Beth Bigley, Dr.P.H., M.S.N.,
Acting Director of Science & Communication, Office of the Surgeon General, Department of Health &
Human Services, 4228 South 32nd Street,
Arlington, VA 22206, Phone: (703) 969-3300;
Email: marybeth.bigley@hhs.gov
Research Objective: Family health history is an underutilized tool in health care and could play a central role in enhancing the uptake and effectiveness of preventive services and early detection for a variety of disorders of major public health importance. In recognition of this, the U.S.
Surgeon General launched the Family History
Initiative in 2004, and has designated Thanksgiving as National Family History Day every year since the
Initiative’s inception. Obtaining a family health history is time consuming; numerous studies show that health care providers often fail to obtain a family health history due to time constraints. A potential solution would be the development and adoption of interoperable, patient completed, family health history tools in health information technology
(HIT) systems.
Study Design: In 2008 the U.S. Department of
Health and Human Services spearheaded, through the activities of the Personalized Health Care workgroup of the American Health Information
Community (AHIC), collaboration between multiple public and private partners to develop a new webbased tool for the collection of family history. This tool replaces the very well received U.S. Surgeon
General’s My Family Health Portrait (MFHP) tool.
Principal Findings: The ‘next generation’ version of MFHP offers the advantage of being based on accepted data standards and terminologies for family history, and was programmed using open source software. Importantly, the new family history tool is being made fully and freely available for adoption and modification by entities that provide health care services. Subsequent to the creation of the tool there have been a number of public and private projects launched that were designed to demonstrate the utility of the tool.
Conclusion: This presentation will outline the value of family history in health care, describe and demonstrate the newly developed MFHP tool, and present examples of ongoing demonstration projects.
Implications for Policy, Delivery or Practice: The goal will be to engage the audience in a discussion of both how the tool can be improved and how to best drive adoption of the tool among individuals as well as care providers of patients with hereditary conditions.
Funding Source(s): HHS
Theme: Health Information Technology
● Family Income at Birth & Childhood Obesity:
The Developmental Origins of an Obesity
Gradient
Lucy Bilaver, M.P.P., M.S.
Presented by: Lucy Bilaver, M.P.P., M.S., Ph.D.
Candidate, School of Social Service Administration,
University of Chicago, 969 East 60th Street,
Chicago, IL 60637, Phone: (773) 288-5863; Email: labilave@uchicago.edu
Research Objective: To investigate whether family income gradients in childhood overweight begin at birth. The majority of the work detailing inverse obesity gradients by parental income and socioeconomic status(SES) has relied on crosssectional data. Given the evidence that obesity is a condition with developmental origins, this work will explore the effect of family income and SES during pregnancy on the development of childhood overweight. The analysis will also also test whether
racial disparities exist after controlling for conditions during the prenatal period.
Study Design: Longitudinal data from the children of the National Longitudinal Study of Youth 1979
(NLSY 79) were used. Repeated measures of height and weight were available for children born to the women in the NLSY 79. The first phase of the study used the 11 waves of data to determine if racial disparities in the income-obesity gradient were ameliorated by controlling for prenatal economic conditions. Logistic regression incorporating the survey weights were estimated using SAS v9.
Population Studied: The children of the NLSY 79 include all children born to the representative sample of female respondents in the original survey. The population of children were interviewed every two years beginning in 1986. For this study sample, children born after 1978 with at least one observation after age 2 were included.
After excluding children with missing data in the key covariates, 6,499 children were included in the study sample.
Principal Findings: Total net family income and the family specific poverty threshold was used to create a poverty income ratio (PIR). The PIR was categorized into tertiles. In 2006, having family income in the low or middle PIR tertile was associated with 1.7 and 1.9 times the odds of overweight (BMI>95th percentile) compared with children in the upper tertile. There was no association between birth PIR tertile and childhood overweight status. In 1996, the opposite was true.
Current family PIR had no association with childhood overweight status while PIR during the prenatal period was associated with a significant negative gradient. In models stratified by survey year and race, there was evidence of some racial disparities in PIR, but the results were inconsistent.
During the mid-1990's, for example, Hispanic youth displayed a strong negative gradient in both current and prenatal PIR. There was no income gradient in overweight status among African American youth in this data.
Conclusion: The inconsistent income-obesity gradient in these data suggest that complex interactions may be confounding the simple relationship observed in cross-sectional data.
While birth PIR was not a consistent predictor of childhood overweight, it did display a strong negative gradient for some racial groups at some points in time. Future longitudinal analysis of these data will be better able to control the interaction of time with income effects over childhood.
Implications for Policy, Delivery or Practice: The income gradient in childhood obesity observed in most cross-sectional studies suggests that lowincome children are most at risk for the development of obesity. Policies that target low income children rely on the assumption that racial disparities in the income-obesity gradient do not exist and that the gradient is the result of current conditions. The recognition that the development of obesity may begin in-utero and that income gradients begin at birth for some children may lead to interventions during pregnancy.
Funding Source(s): AHRQ
● Surfactant Replacement Therapy May Improve
Survival for Infants Born with Congenital
Diaphragmatic Hernia: An Instrumental Variable
Analysis
Tommy Bird, M.S.; J. Mick Tilford, Ph.D.; Patrick
Casey, M.D.; James Robbins, Ph.D.; Robert Lyle,
M.D.
Presented by: Tommy Bird, M.S., Director of
Medical Informatics Research, Pediatrics, Center for Applied Research & Evaluation, University of
Arkansas for Medical Sciences, 11 Children's Way,
Little Rock, AR 72202, Phone: (501) 364-4936;
Email: birdtommym@uams.edu
Research Objective: Congenital diaphragmatic hernia (CDH) is a major structural birth defect with mortality rates ranging from 20% to 60%. Large observational cohort studies examining surfactant therapy in infants with CDH have shown either no effect on mortality or a negative effect.
Observational studies of CDH are complicated by many factors, including selection bias due to both physician preference and clinical severity. There have been no large randomized controlled trials of surfactant therapy in CDH. Instrumental variable analysis is one method of controlling for selection bias and approximating the results of a RCT with observational data. This study used instrumental variable analysis to control for selection bias while examining the relationship between surfactant therapy and mortality among infants with CDH.
Study Design: Data came from the Pediatric
Health Information System (PHIS) over years 1999-
2008. PHIS is a detailed hospital discharge database of 40 stand alone children’s hospitals.
PHIS contains demographic, diagnosis, and procedure data, as well as pharmacy and clinical service data. A multivariable bivariate probit model with instrumental variables was used to calculate the predicted probability of mortality for CDH infants who did and did not receive surfactant therapy. The model also controlled for demographics, birth weight, and clinical factors such as use of ECMO, comorbid birth defects, surgical repair of intestines, and surgical repair of liver.
Population Studied: Neonates admitted to one of
40 stand alone children's hospitals within the first 10 days of life with a diagnosis of congenital diaphragmatic hernia.
Principal Findings: A total of 2,306 infants with
CDH were included in the study, only 114 of whom received surfactant therapy. Naïve multivariable probit models, which do not accont for selection bias, indicate a non-significant increased probability of mortality for infants receiving surfactant therapy
(21.8% (13.5%-30.9%) compared to 19.2% (17.1%-
21.5%), p=0.521). Bivariate probit models with instrumental variables, which can account for selection bias, indicate a significant reduction in the probability of mortality for infants receiving surfactant therapy (8.0% (3.5%-16.0%) compared to 20.3% (18.0%-22.7%), p=0.009).
Conclusion: Surfactant replacement therapy, almost abandoned for the treatment of infants with
CDH on the basis of naïve observational studies, could greatly reduce mortality rates in these very fragile neonates. These findings should be confirmed in a multi-center randomized controlled trial.
Implications for Policy, Delivery or Practice:
Surfactant replacement therapy should be reevaluated as a front line treatment option for newborns with CDH.
Funding Source(s): Arkansas Biosciences Institute
Theme: Child Health
● Patient Involvement in Patient Safety
Yvonne Birks, D.P., B.S.; Dorothy, McCaughan,
M.S.; Jill Hall, M.S.; Maggie Peat, Ph.D.; Vikki
Entwhistle, Ph.D.; Brian Williams, Ph.D.
Presented by: Yvonne Birks, D.P., B.S., Research
Fellow, Health Sciences, University of York,
Heslington, York, YO10 5DD, UK, Phone:
+44(0)1904321328; Email: yfb1@york.ac.uk
Research Objective: Patient involvement in patient safety is at the centre of growing international interest and debate, yet little is known about the factors that affect patients being involved in their own safety, or about patients’ views of taking on this role.We have conducted an empirical study in order to: investigate patients’ awareness of, and concerns about, health care safety; identify situations and ways in which patients (or their representatives) have acted, or wanted to act, to secure their own safety and to identify the kind of actions they took, or considered taking; explore patients’ (and their representatives’) perceptions of the various roles they might play in ensuring their own safety, and factors that appeared to facilitate or impede them; explore patients’ views about contributing to national and local safety reporting systems
Study Design: Qualitative methods (depth interviews) have been used to elicit perceptions from a wide range of patient populations. Data have been coded using the ATLAS.ti computer software package, and analysed in accordance with the Framework approach advocated by Ritchie and
Lewis.
Population Studied: Patients (and their family members or representatives where appropriate) from six different groups were strategically selected in order to explore the views of patients with diverse clinical, demographic and social characteristics, and to consider health care contexts that raise a range of issues in terms of both patient involvement and patient safety. Groups comprise: adults with type 2 diabetes; women treated for breast cancer; parents of children who have been hospitalised with asthma; patients scheduled for joint replacement surgery; people with severe and enduring mental health problems; people who have raised a safety related issue or made a complaint about their health care.
Principal Findings: Most respondents initially professed limited awareness of patient safety issues, but all identified instances in which they had felt concerned about mistakes or shortfalls in the care they received. Patients’ decisions about whether or not to act in these situations were apparently influenced by: personal characteristics e.g. self-identification as assertive or timid; their knowledge and experience of their condition and/or treatment which could affect their confidence about whether or not something was a threat to their safety; their capacity to act which might depend on their health condition; and the response they anticipated from healthcare staff. Previous experience led many to be sceptical that staff would respond helpfully.
Conclusion: Patients engaged with the topic and showed a broad understanding of issues.They identified a number of safety roles, some more challenging than others. The potential for carrying out a role could be affected by a range of factors including: characteristics of the patient and their disease/condition;characteristics of the HCP;factors relating to the healthcare system.
Implications for Policy, Delivery or Practice:
Although policy makers emphasise the potential for patients to contribute to the safe delivery of healthcare, our findings suggest that current healthcare practices and cultures often impede patients’ active participation in the safe delivery of healthcare. Efforts to promote patient involvement in safety-enhancing activities need to reflect patients’ perspectives on safety in healthcare and address systemic as well as individual barriers to action.
Funding Source(s): Department of Health UK,
National Patient Safety Agency, UK
Theme: Consumer Choices in Health Care
● Emotional Intelligence in Healthcare
Yvonne Birks, D.P., B.S.; Ian Watt, B.S., M.B.,
Ch.B., M.P.H.
Presented by: Yvonne Birks, D.P., B.S., Research
Fellow, Health Sciences, University of York,
Heslington, York, YO10 5DD, UK, Phone:
+44(0)1904321328; Email: yfb1@york.ac.uk
Research Objective: The principles of patientcentred care are increasingly stressed as part of healthcare policy and practice. Explanations for why some practitioners seem more successful in achieving patient-centred care vary but a possible
role for individual differences in personality has been postulated. One of these, emotional intelligence, is increasingly referred to in healthcare literature. This paper reviews the literature on emotional intelligence in healthcare and poses a series of questions about the links between emotional intelligence and patient-centred outcomes.
Study Design: A range of databases in several subject areas were searched to identify documents discussing EI. The current study draws on published literature to inform a critical discussion of the area.
Population Studied: Initial searches employed several databases, including Medline, Cinahl and
Psych Info, using the term ‘emotional intelligence’.
Reference lists from identified papers were also hand-searched to identify any further literature which had not been identified in the initial searches.
Principal Findings: Empirical studies of EI in health care settings were few, with the majority of papers being editorials and opinion pieces. Six empirical studies were identified. The study discusses the findings in relation to several applications for EI in healthcare. EI training in the business community is a lucrative business, and testing using current instruments is expensive and complex. Without the empirical evidence to support the idea that many health care outcomes can be improved by increasing EI in health care professionals, widespread adoption of programmes to increase EI should be questioned.
Conclusion: The questions posed in this review call for a systematic examination of the role of EI in health care rather than the uncoordinated scattered approach which is currently evident.
Implications for Policy, Delivery or Practice:
While EI is an appealing prospect to some, its benefits to clinical practice, education and selection in any health care discipline have yet to be adequately explored. We have only recently begun to explore the possibility that EI may be of benefit to either the professional or the patient.
Funding Source(s): Department of health UK,
National Institute of Health Research
Theme: Organizational Performance and
Management
● Exploring the Relationship Between
Emotional Intelligence & Patient Centred Care
Yvonne Birks, D.P., B.S.; Ian Watt, B.S., M.B.,
Ch.B., M.P.H.
Presented by: Yvonne Birks, D.P., B.S., Research
Fellow, Health Sciences, University of York,
Heslington, York, YO10 5DD, UK, Phone:
+44(0)1904321328; Email: yfb1@york.ac.uk
Research Objective: Increasingly, NHS policy emphasises that the patient experience should be at the heart of any healthcare consultation.
However, healthcare professionals vary in their ability to achieve an understanding of the patient perspective and provide patient centred care. This project will employ a psychological approach to determine whether the emotional intelligence of health care professionals is a determining factor in these differences. The study aims to explore the relationship between emotional intelligence (EI) in health care professionals and the quality of patient centred care.
Study Design: Cross-sectional study of Primary
Care practitioners in the UK.
Population Studied: 93 general practitioners, primary care nurses and complementary practitioners were recruited and completed EI measures. Each practitioner also collected data from their patients on empathy and enablement.
The EI and patient data will be analysed to determine whether differences in EI exist between groups and how and whether EI relates to the patient-centredness of their care.
Principal Findings: Analysis of this study is ongoing but will report the findings in relation to relationships between emotional intelligence in individual practitioners and patient reports of enablement and empathy as a reflection of patientcentredness.
Conclusion: Conclusions will be reported as appropriate.
Implications for Policy, Delivery or Practice: The implications for any relationships between EI and patient-centred care for policy, healthcare delivery and practice will be discussed.
Funding Source(s): Department of Health UK,
National Institute of Health Research
Theme: Organizational Performance and
Management
● Scoping Review of Interventions Intended to
Involve Patients in Patient Safety
Maggie Peat, Ph.D.; Yvonne Birks, D.Phil.; Su
Golder, M.S.; Ian Watt, B.S., M.B., Ch.B., M.P.H.;
Vikki Entwistle, Ph.D.
Presented by: Yvonne Birks, Health Sciences,
University of York, Heslington, York, Y010 5DD,
UK, Phone: +44(0)1904321329; Email: yfb1@york.ac.uk
Research Objective: Estimates from several countries, including the UK suggest that around 1 in
10 patients admitted to hospitals will experience some sort of unintended harm. Strategies to reduce adverse events have so far focused mainly on changing systems of care and professional behaviour. However, more recently, there has been a growing interest, in the UK and other countries, in involving patients in safety initiatives. These include the use of interventions to promote the roles of patients and their family members/representatives in securing their own safety in health care. At present, the evidence has been relatively
unexplored and insufficiently critiqued. The aim of this study was to investigate, as one component of a wider project, how patients and their family members and other representatives might appropriately be involved in their healthcare to effectively promote their own safety, and to explore how this might vary by context, place or demography. In addition we aimed to develop categorisations of safety interventions which involve patients according to how they are intended to improve safety in health.
Study Design: We carried out extensive searches for any citations reporting (1) comment, opinion or research about the ways in which patients’ actions might (intentionally or inadvertently) enhance their own or others’ safety; (2) descriptions or evaluations of interventions to encourage or enable patients to adopt actions that might enhance their own or others’ safety; and/or (3) issues relating to patient’s willingness to adopt actions that might enhance their own or others’ safety. Relevant information from the publications was summarised into tables. Commentary and opinion pieces were summarised and used to construct a ‘patient safety timeline’. Categorisations were developed from the interventions described and populated with examples from the literature reviewed.
Population Studied: Cochrane Database of
Systematic Reviews, Database of Abstracts of
Reviews of Effects, CENTRAL, CINAHL, EMBASE,
HMIC, MEDLINE, MEDLINE in-process, PsycINFO and ASSIA to August 2008. We also searched databases of reports, conference proceedings, grey literature, ongoing research, relevant patient safety organisations and hand searched two journals.
Principal Findings: 1933 reports were identified as potentially relevant and of the 745 of these that were included in the review, 437 were descriptions of interventions, 299 comments or opinion pieces and 42 explored or described patients’ willingness and ability to adopt safety promoting behaviours.
We identified three broad routes by which patients’ actions might contribute to their safety.
Interventions were categorised according to how they are intended to improve safety. Patients might help to make sure that: their treatment is appropriate for them; that treatment is given as planned and according to appropriate protocols; that health systems are safe.
Conclusion: We have reviewed a wide range of literature detailing actual and purported patient involvement in promoting their own safety while using health services. Through focusing on the intended aim and mechanism of action of interventions we have identified three broad routes by which patients’ actions might contribute to their safety. This has been combined with what we know about patients’ willingness and ability to adopt these actions. In this way we have begun to clarify how patients can contribute to their safety and what actions might influence their contribution.
Implications for Policy, Delivery or Practice:
There appears to be potential in patients using their expert knowledge of themselves to ensure that health professionals have appropriate information to make treatment decisions. Monitoring and delivering their own care safely also appears to be within the capabilities of most patients. Some patients are also able to draw on their experience to contribute to making healthcare systems safer.
However, a key feature of all these broad routes is facilitation of patient action by health professionals or the healthcare system. This can range from provision of tools to manage anticoagulation treatment to asking a patient about his/her current medication. Patients who are discouraged from adopting safety promoting behaviours because they are concerned about a negative response from a health professional could, presumably, be encouraged to act if assured of a positive response.
Patients who are capable of safely delivering and monitoring their own treatment are, potentially, capable of monitoring the treatment given to them by others, given the appropriate information, tools and environment.
Funding Source(s): Department of Health UK,
National Patient Safety Agency, UK
Theme: Quality and Efficiency: Policies and
Incentives
● Effectiveness of Interventions Designed to
Promote Patient Involvement to Enhance
Safety: A Systematic Review
Maggie Peat, Ph.D.; Yvonne Birks, D.Phil.; Su
Golder, M.S.; Ian Watt, B.S., M.B., Ch.B., M.P.H.
Presented by: Yvonne Birks, Health Sciences,
University of York, Heslington, York, YO10 5DD,
UK, Phone: +44(0)1904321329; Email: yfb1@york.ac.uk
Research Objective: Estimates suggest that between 300,000 and 1,400,000 adverse events occur each year in the UK NHS alone. Current strategies to reduce adverse events are mainly focused on changing systems of care and professional behaviour. However, there is increasing international interest, in involving patients in safety initiatives. The evidence for the effectiveness of interventions to promote patient involvement in patient safety has not been systematically reviewed. The aim of this review was to identify evaluations of strategies or interventions that have been used with the intention of promoting patients’, and/or their family members’ involvement in their care, with a view to enhancing their own, or others’ safety in a healthcare context.
Study Design: Systematic review.
Population Studied: Experimental and quasiexperimental studies interventions which promoted or supported patients in activities with the explicit intention of enhancing patient safety were included.
Outcomes of interest included implementation
aspects, the impact on patient safety and intermediate outcomes.
Principal Findings: We identified 14 individual experimental and quasi-experimental studies plus 1 systematic review. The majority of studies fell into the monitoring and ensuring safe delivery of treatment by self category and were all related to enhancing medication safety. Authors’ reported improved patient safety incident outcomes for the intervention groups compared to controls where the interventions aimed to encourage patient involvement in: Monitoring and ensuring safe delivery of treatment by self (self-management of anticoagulation, ‘easy’ read information leaflet, nurse-led education to promote self-medication in hospital, patient package insert using lay terminology); Informing the management plan/ monitoring and ensuring safe delivery of treatment by self (individualised teaching plan by nurse, pharmacist counselling). It was not possible to draw any clear conclusions as to the effectiveness of the interventions (with the exception of one specific aspect of self-medication i.e. self-management of anticoagulation) due to concerns about the methodological quality of the studies.
Conclusion: There is limited evidence for the effectiveness of interventions designed to promote patient involvement on patient safety incidents and in general is poor quality. Existing evidence is confined to the promotion of safe self-management of medication, most notably relating to the selfmanagement of oral anticoagulants.
Implications for Policy, Delivery or Practice:
There is a major international movement to increase patient involvement with a view to enhance patient safety. However, there is scarce evidence of benefit. Future research should focus on areas other than medication safety. In particular, interventions in most urgent need of evaluation are those that are currently widely used but unevaluated, for example 20 tips to help prevent medical errors and Please Ask – Be Informed. In addition, evaluations should include the assessment of outcomes related to patient safety incidents.
Funding Source(s): Department of Health UK,
National Patient Safety Agency, UK
Theme: Quality and Efficiency: Policies and
Incentives
● Sociodemographic, Market & Political Factors that Predict Registered Nurses who Work
Outside of Nursing
Lisa Black, Ph.D., R.N.; Joanne Spetz, Ph.D.;
Charlene Harrington, Ph.D., R.N., F.A.A.N.
Presented by: Lisa Black, Ph.D., R.N., Assistant
Professor, Orvis School of Nursing, University of
Nevada, Reno, 1664 North Virginia Street, Mail
Stop 0134, Reno, NV 89557, Phone: (775) 682-
7155; Email: lblack@unr.edu
Research Objective: Little empirical research examines registered nurses (RNs) who participate in the labor market, but choose to work outside of nursing. The purpose of this study was to describe sociodemographic, market, and political factors that predict non-nursing work
Study Design: Data from the 2004 National
Sample Survey of Registered Nurses (NSSRN) was used to describe nurses who work outside of nursing. The NSSRN is the Nation’s most comprehensive data source on all individuals with active licenses to practice nursing in the United
States, whether or not they are employed in nursing. Market data from the 2004 HRSA Area
Resource File and political liberalism data from the
National Journal were used to identify market and political factors that predict non-nursing work. A market wage was created using OLS to control for endogeneity of the nursing wage. Married and unmarried nurses were empirically different, and were analyzed separately. A logistic regression model was estimated using two-stage residual inclusion.
Population Studied: NSSRN data (N = 35,635) were weighed using HRSA sampling weights to represent a National estimate of 2,909,357 registered nurses in the United States.
Principal Findings: Registered nurses who worked outside of nursing earned approximately $1 per hour more than those who worked in nursing, and were, on average, 5.5 years older, had been out of school six years longer, were more likely to be white and educated in the United States, be male, and have young children in the home. Logistic results showed that the non-nursing wage was a significant factor in unmarried nurses only.
Unmarried RNs working in the Southern and
Western US Census Regions were less likely to work outside of nursing than nurses living in the
Northeast, where the NSSRN (2004) reports greater numbers of RNs per capita. Nurses who worked as licensed practical/vocational nurses or certified nursing assistants prior to RN licensure were less likely to work outside of nursing. Nurses living in more politically liberal counties were less likely to work outside of nursing than nurses in politically conservative counties
Conclusion: Registered nurses who work outside of nursing were different from nurses who work in nursing on a number of study factors. Different strategies are needed to retain nurses who are on the margin of leaving nursing, and to incentivize those who have left to return to the profession.
Implications for Policy, Delivery or Practice:
Workplace remedies are needed that address the specific needs of the nursing sub-populations who are most likely to leave the profession, and to incentivize those who have left to return. Ergonomic and scheduling changes may address the needs of older nurses and nurses with children in the home.
Though they represent a small percentage of the overall nursing labor market, male nurses require
different workplace retention strategies than female nurses. Changes in the nursing workplace are likely to be more successful than further enhancements to the nursing wage, as the wage was significant for unmarried nurses only in this study.
Theme: Health Care Workforce
● Hospital Location & Hospital Performance:
Equity Implications for P4P
Jan Blustein, M.D., Ph.D.; Melissa Valentine,
M.P.A.
Presented by: Jan Blustein, M.D., Ph.D., Professor of Health Policy & Medicine, Health, New York
University/Wagner, 295 Lafayette Street, New York,
NY 10012, Phone: (212) 998-7427; Email: jan.blustein@wagner.nyu.edu
Research Objective: Pay-for-performance is predicated on the assumption that all hospitals have the capacity to perform well, and that underperforming institutions can improve over a relatively short time frame. Yet hospital capacity to perform (and improve performance) depend on local economic and human resources. In this study, we quantified the association between local resources and hospital performance. We also explored the extent to which local resources influence hospital capacity to improve performance over time.
Study Design: Panel data study of the association between hospital locational resources and hospital performance, during the years 2004-2007. We characterized hospitals as being relatively locationally “advantaged” or “disadvantaged” along
11 dimensions of local workforce and economy.
For example, “advantage” in the local workforce was reflected in an adequate availability of health professionals and a relatively higher educational level among working-age adults. Advantage in the local economy was reflected in relatively low levels of poverty in preceding 4 census periods, and relatively high local employment levels. Hospital clinical performance was assessed using Medicare
Hospital Quality Initiative [HQI] measures of the process of care for three clinical conditions (AMI, heart failure, and pneumonia). Various individual characteristics of hospitals as well as hospital case mix were included as control variables in multivariate analyses.
Population Studied: U.S. hospitals responding to the Medicare Hospital Quality Initiative from 2004 through 2007 (n = 3084).
Principal Findings: In 2004, hospitals that were relatively advantaged in terms of local workforce and economic conditions performed substantially better than their relatively disadvantaged counterparts. For example on process measures of
Acute Myocardial Infarction [AMI] care, hospitals in non-Health Professions Shortage Areas [HPSAs] had a mean composite score of 86.4, compared with a mean of 74.3 for hospitals that were located in HPSAs (p < .001) Hospitals located in counties at the highest quartile of college graduates in the population had a mean composite score of 91.2 compared with a mean of 78.7 for hospitals located in counties at the lowest quartile of college graduates (p <.001; test of trend for all four quartiles p < .001). Hospitals that were located in counties without a history of chronic poverty earned mean scores of 89.2, compared with a mean of
77.5 for hospitals in counties that had a history of chronic poverty (p <.001). These and other local workforce and economic conditions continued to have substantial and independent effects on performance in multivariate analyses that controlled for individual hospital characteristics. Analyses of heart failure performance measures showed similar patterns; examination of pneumonia performance measures yielded less consistent findings. The relationship between local resources and clinical performance persisted through 2007, with hospitals located in resource-advantaged locations performing substantially and significantly better than their less locationally advantaged peer institutions.
Moreover, compared with hospitals in resourcedisadvantaged locations, those in advantaged locations were more able to improve their performance over time.
Conclusion: Hospital performance on clinical process measures is associated with the quantity and quality of local human and economic resources. In addition,ability to improve performance is associated with local human and economic resource levels.
Implications for Policy, Delivery or Practice:
Conclusions: Hospital P4P is likely to reward institutions in relatively resource-rich areas, raising important questions of allocational equity.
Theme: Quality and Efficiency: Policies and
Incentives
● Measuring Post-Hospital Mortality after
Intensive Care
Megan Bohensky, M.P.H.; Vijaya Sundararajan,
M.D.; Damien Jolley, M.S.; Caroline Brand,
M.B.B.S.; Sue Evans, Ph.D.; David Pilcher,
M.B.B.S.
Presented by: Megan Bohensky, M.P.H., Ph.D.
Student, Epidemiology & Preventive Medicine,
Monash University, Burnet Building, Alfred Hospital-
Level 3, Prahran, VIC, 3182, AU, Phone:
+61399030053; Email: megan.bohensky@med.monash.edu.au
Research Objective: In intensive care medicine, hospital mortality is a commonly used quality measure, as it is an unambiguous outcome based on routinely collected data. Moreover, there are validated scoring systems (e.g. APACHE and
SAPS ) that enable reliable risk-adjustment for comparisons. Criticisms of hospital mortality as a
flawed measure of the quality of patient care have prompted the search for alternative measures. With an ageing population, post-hospital survival of patients admitted to the intensive care unit (ICU) is becoming an important factor to consider, as many patients may receive costly treatments only to fare poorly after hospital discharge.
Study Design: We sought to measure post-hospital survival of patients receiving intensive care at three hospitals in the state of Victoria, Australia by linking three existing data sources: 1) The Australian and
New Zealand Intensive Care Adult Patient
Database (ANZICS APD) is a clinical registry that contains data from patients admitted to nonpediatric intensive care units in Australia and New
Zealand. This data-set contains the elements necessary to risk-adjust patient outcomes using
APACHE or SAPS models. 2) The Victorian
Admissions Episodes Data is collected for all hospital episodes across the state of Victoria for funding, health policy and evaluation and population surveillance. Routine, independent audit of coding quality have established it as a reliable source of hospital diagnostic and procedural data. 3) The
Victorian Registry of Births Deaths and Marriages registers all deaths that occur in Victoria.
Population Studied: 18,603 patients aged 16 years and older with an index intensive care admission at one of three teaching hospitals between 1 January 2001 and 31 December 2005.
Survival status was followed until 30 June 2008.
Principal Findings: 2,381 (12.8%) patients died in hospital. Within the first year after discharge, a further 1,240 (7.7%) patients had died. Factors relating to worsening post-hospital survival included older age (OR = 4.3 for 90 plus age group), widowed/divorced/separated marital status
(OR=1.4-1.5), higher GCS within the first 24 hours of ICU admission(OR=1.1), higher Charlson comorbidity score (OR=1.2), higher APACHE3 and
SASPS2 scores within the first 24 hours of ICU admission (OR=1.1) and ventilation (OR=2.1).
Conclusion: It is possible to link existing data sources to study post-hospital survival of patients receiving intensive care in Victoria, which enables a more comprehensive assessment on the long-term patient outcomes after ICU care. Further work will be required to enhance the case ascertainment and quality of existing data-sets and to determine factors predisposing patients to post-hospital mortality.
Implications for Policy, Delivery or Practice:
Studying post-hospital mortality rates of patients will provide evidence to inform better allocation of ICU resources and improve the ability to assess patient outcomes during and after discharge from hospital.
It will also be important in informing how patients are best managed and followed up post discharge.
Funding Source(s): Monash PhD scholarship
Theme: Quality and Efficiency: Measurement
● Changes in Older Adults’ Health Care Costs
Before & After an Incident Fall
Alex Bohl, B.S.; Marcia Ciol, Ph.D.; Barb Williams,
Ph.D.; James LoGerfo, M.D., M.P.H.; Paul
Fishman, Ph.D.; Elizabeth Phelan, M.D., M.P.H.
Presented by: Alex Bohl, B.S., Research Assistant,
Health Services, Health Promotion Research
Center--University of Washington, 1107 Northeast
45th Street, Suite 200, Seattle, WA 98105, Phone:
(206) 543-2891; Email: aabohl@u.washington.edu
Research Objective: Falls are a leading cause of fatal and non-fatal injury among the elderly. Direct medical care costs resulting from falls have been estimated on an individual and population level; however, a paucity of research has examined fallers’ total health care costs over time. By explicating the temporal occurrence of cost after a fall, policy makers and administrators can begin to build a business case for fall prevention programs.
Thus, we aimed to compare total and component
(e.g. inpatient) direct health care costs for those who fall (“fallers”) and non-faller controls two years before and up to one year after a fall that required medical attention.
Study Design: Fallers were identified by E-Codes describing mechanism of the fall (880-1, 884-5, and
888) and injury-related ICD-9 codes (800-48, 850-4, and 920-4) recorded through medical records and claims between 1/1/04 and 12/31/06. Non-fallers were frequency matched to fallers on age group and gender, using random selection within agegender groupings. Outcomes of interest were total, outpatient, inpatient, emergency, long-term care, radiology, and pharmacy costs, inflated to 2007
United States dollars. Costs were classified into quarter-year time periods relative to an index date, which represented the day of the first fall for which medical attention was sought (for fallers) or a corresponding date (for non-fallers). Quarterly costs were modeled using generalized estimating equations with log link and gamma family.
Covariates included age, gender, comorbidities, and number of medications.
Population Studied: 8,958 fallers and 8,969 nonfallers were selected from all persons aged 67 and over enrolled in an integrated health care delivery system serving Western Washington populations.
The cohort was 62% female, with a mean age of 78 years (s.d. = 7.0). Nineteen percent had diabetes,
18% had cancer, and 38% had heart disease. A greater percentage of fallers had diabetes and heart disease, and fallers had significantly more outpatient visits in the year prior to the index date
(13 versus 9 visits, p<0.01).
Principal Findings: Unadjusted average quarterly costs for fallers were higher than non-fallers in all quarters. Adjusting for covariates, total, pharmacy, outpatient, emergency, long-term care, and radiology costs were higher for all four quarters following the index date for the faller cohort
compared to non-fallers (from Wald tests), while faller inpatient costs were significantly higher for only the first two quarters post-index. Average predicted total costs for the faller (non-faller) cohort were $7,886 ($1,929) in the first quarter after index,
$3,848 ($2,086) for the second, $3,326 ($2,113) for the third, and $4,665 ($2,862). Inpatient and longterm care costs were the main drivers of total cost differences.
Conclusion: Falls for which medical evaluation is sought increase costs of care substantially, and this increase is primarily driven by inpatient and longterm care costs.
Implications for Policy, Delivery or Practice:
Evidence-based fall prevention programs are available and relatively inexpensive compared to hospitalization or a skilled nursing facility stay. A concerted effort should be made to target persons at highest risk for serious fall-related injury leading to hospitalization.
Funding Source(s): CDC
Theme: Medicare
● Public Opinion on Health & Health Disparities
Bridget Booske, Ph.D., M.H.S.A.; Stephanie Robert,
Ph.D.
Presented by: Bridget Booske, Ph.D., M.H.S.A.,
Senior Scientist, Population Health Institute,
University of Wisconsin-Madison, 610 Walnut
Street, Suite 760, Madison, WI 53726, Email: bbooske@wisc.edu
Research Objective: Despite growing research evidence about the social and economic determinants of health, and of persisting health disparities, there is little public and policy attention to these issues in the U.S. This lack of attention may result, in part, from the public’s lack of knowledge about social and economic determinants of health and health disparities, or from low concern about these issues. Yet there is little evidence regarding what the public actually knows and thinks about health and health disparities. The primary goal of our project is to understand U.S. adults’ knowledge and attitudes about health and health disparities, and to determine which policy domains they view as relevant for improving health and reducing health disparities.
Study Design: The study involves a national random-digit-dial survey of 2800 adults, including a randomized experiment with approximately 400 people randomly assigned into one of seven health disparity modules.
Population Studied: Approximately 2800 US adults, with oversamples of racial/minority and low income respondents.
Principal Findings: The telephone interview includes questions intended (1) To describe the public’s views about the determinants of health; (2)
To describe the public’s views about the existence of and causes of health disparities (by race, income, and education); (3) To describe attitudes about the policy domains viewed as relevant to improving health and reducing health disparities;
(4) To describe people’s priorities for remedies to improve health and reduce health disparities; (5)
To examine how people’s demographic, social, ideological, and political, characteristics contribute to their knowledge and attitudes about health and health disparities; and (6) To understand how people’s views about health disparities and policy priorities differ by the type of health disparity is being discussed.
Conclusion: Conclusions are not yet available because data collection is still underway (data collection will be completed by the end of February
2009).
Implications for Policy, Delivery or Practice: The primary research goal of this project is to understand U.S. adults’ knowledge, opinions, and attitudes about health and health disparities, and to determine which policy domains they view as relevant for improving health and reducing health disparities. These results can be used to target information and policy campaigns to the general public and to track changes in public opinions over time.
Funding Source(s): RWJF
Theme: Disparities
● The Influence of Health Confidence on Racial
& Ethnic Disparities in Consumers' Ratings &
Reports of Health Care: Findings from the
MEPS
Ty Borders, Ph.D.; Shelley Lensing, M.S.
Presented by: Ty Borders, Ph.D., Associate
Professor, Health Policy & Management, University of Arkansas for Medical Sciences, 4301 West
Markham Street, Little Rock, AR 72205, Phone:
(501) 526-6641; Email: tfborders@uams.edu
Research Objective: Consumer assessment measures are useful tools for identifying racial and ethnic disparities in more subjective aspects of the accessibility and quality of health services. Health status has been shown to be correlated with consumer assessments, with persons in poorer health tending to have worse evaluations of their care. In addition, other individual differences, including attitudes about health care, may influence consumer assessments. The objective of this study was to examine whether health-related attitudes influence consumers’ ratings and reports as assessed by the CAHPS® and whether adjusting for attitudes attenuates ethnic or racial disparities in consumer assessments.
Study Design: To address these issues, we conducted cross-sectional statistical analyses of
CAHPS® questions included in a nationally representative survey of adults. The dependent variables were: 1) a single item “overall health care” rating, 2) a multi-item “getting needed health care”
composite 3) a multi-item “getting care quickly” composite, and 4) a multi-item “how well doctors communicate” composite. The primary independent variable of interest was health confidence, which was measured by a single item measure of one’s self confidence in overcoming illnesses without medical help. Health-related attitudes were assessed by a single item measure of health confidence and health status was measured using the physical and mental components of the SF-12 (PCS-12 and MCS-12).
Other covariates were age, sex, race, ethnicity, language spoken most at home, marital status, rural/urban residence, education, family income, and insurance coverage.
Population Studied: 20,679 adults in the U.S. who participated in the 2005 Medical Expenditure Panel
Survey (MEPS), which was conducted by the U.S.
Agency for Healthcare Research and Quality
(AHRQ).
Principal Findings: Persons with greater health confidence had lower adjusted odds of rating their overall care very highly (OR=0.76, 95% C.I.=0.68-
0.85), worse reports of getting care when needed
(OR=0.81, 95% CI=0.69-0.96), and worse reports of doctor communication (OR=0.79, 95% C.I.=0.70-
0.89). Adjusting for health confidence attenuated many ethnic and racial disparities. When adjusting for health confidence and other potential confounders, there were no Hispanic/non-Hispanic disparities in ratings or reports. African Americans had lower overall care ratings and reports of getting needed care than whites. Asians had lower overall care ratings and reports of getting needed care, getting care quickly, and doctor communication.
Conclusion: Consumer assessment measures are helpful tools for identifying dimensions of the accessibility and quality of health care services in need of improvement. However, the findings of this study indicate that health confidence influences consumers’ assessments of health care and may partially explain some racial and ethnic disparities in consumer assessments.
Implications for Policy, Delivery or Practice: To accurately compare ratings and reports across health plans, hospitals, and ambulatory care organizations, health policy makers and researchers should consider adjusting for measures of health confidence or other health-related attitudes. However, even when adjusting for health confidence, some racial disparities in ratings and reports of health care persist among African
Americans and Asians, suggesting that strategies should focus on improving the organization and delivery of care for these particular population subgroups.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Measurement
● Inpatient Quality Indicator Rates in Select
Veteran Populations
Ann Borzecki, M.D., M.P.H.; Priscilla Chew, M.P.H.;
Susan Loveland, M.A.T.; Patricia Loya, M.S.; Amy
Rosen, Ph.D.
Presented by: Ann Borzecki, M.D., M.P.H.,
Clinician-Researcher, Center for Health Quality,
Outcomes & Economic Research, 200 Springs
Road (152), Bedford, MA 01730, Phone: (781) 687-
2870; Email: amb@bu.edu
Research Objective: Select patient populations with chronic conditions, such as serious mental illness (SMI) or spinal cord injuries (SCI), may be at high risk of adverse healthcare outcomes. Because of this concern, the Veterans Health Administration
(VA) has implemented several initiatives to improve care of these patients. These include: establishing registries to identify and track special patient groups, establishment of condition-specific research centers under the Quality Enhancement and
Research Initiative (QUERI), and separate tracking of a limited number of national performance measures in chronic disease populations.
However, additional quality measures such as
AHRQ’s Inpatient Quality Indicators (IQIs), which are easily generated from administrative data, may be useful to provide a more comprehensive picture of care received by such patients and identify populations at particularly high risk of poor outcomes. The IQIs screen for potential inpatient quality problems by measuring: a)mortality rates from specific procedures and conditions where high mortality may be associated with poorer care, and b)utilization rates of procedures where concerns exist about over-/under-, or misuse. As part of a larger IQI study, we identified high-priority special populations and determined their IQI rates.
Study Design: We convened a panel of key VA stakeholders who helped prioritize both IQIs and special populations for further study. We then derived risk-adjusted rates (adjusted for age, gender, comorbidities and illness severity using
APR-DRGs) for selected populations: diabetes,
SMI, spinal cord injured (SCI), HIV, and stroke patients, using FY2004-2007 VA inpatient files and the AHRQ IQI software (v3.1). Rates were compared to the remaining VA population.
Population Studied: All FY04-07 admissions of veterans with diagnoses of diabetes, SMI, SCI, HIV, or stroke.
Principal Findings: Of 2.3 million admissions, the subpopulation breakdown was as follows: diabetes
34%, SMI 13%, SCI 2%, HIV 1%, and stroke 6%.
Diabetes and stroke patients were older than the remaining VA population (65.5+/-11.6 and 69.0+/-
11.5 vs 61.7+/-14.3 years; p<0.05); SMI, SCI and
HIV subjects were younger (respective mean ages
57.5+/-13.8, 59.3+/-12.9, 51.4+/-8.8; p<0.05). SMI patients had higher mortality rates from several medical conditions, including acute myocardial infarction (AMI), heart failure, stroke, and pneumonia than the non-SMI population (e.g., for
AMI, risk-adjusted rates were 9.7+/-2.4 vs 7.4+/-2.8, per 100 AMI admissions; p<0.05). There was a trend toward higher mortality for SCI subjects across the same conditions but differences weren’t significant presumably due to low numbers. No consistent trends were seen among the other cohorts, or with respect to procedure-related mortality or utilization indicators.
Conclusion: SMI patients are at higher mortality risk from various medical conditions compared to the general VA population. This increased vulnerability may be due to numerous factors, such as delayed admission or inattention to medical problems.
Implications for Policy, Delivery or Practice:
Further research is needed to understand reasons for the higher mortality risk in SMI patients, determine whether similar trends exist outside the
VA, and develop quality improvement initiatives that will decrease the mortality risk of these patients.
The IQIs appear useful in the VA for tracking and monitoring quality of care problems that occur in the inpatient setting, and for highlighting subpopulations that may be at increased risk of adverse outcomes.
Funding Source(s): VA
Theme: Quality and Efficiency: Measurement
● Deriving Health Status Information for
Children from Pharmacy Codes in 5 Spanish
Primary Care Clinics: Implications for Risk
Adjustment
Paulo Boto, M.D., M.S.; Jonathan Weiner, Dr.P.H.
Presented by: Paulo Boto, M.D., M.S., Ph.D.
Candidate, Health Policy & Management, Johns
Hopkins Bloomberg School of Public Health, Av.
Carolina Michaelis, 22, 4º Esq, Linda-A-Velha,
2795-049, Portugal, Phone: +351914595634;
Email: pboto@jhsph.edu
Research Objective: Risk adjustment tools have application for payment, case-management and research. Originally these tools were based on diagnostic codes and they are now progressively deriving information from drug prescription to gain further insights into health status. Little research has been done on the application of drug based risk adjustment tools in children. The objectives of this paper are: 1) to compare and contrast health status/risk information derived from the drug prescribing of primary care physicians in a clinic system in Spain; and 2) to assess the extent to which a pharmacy based risk adjustment tool - the
Johns Hopkins ACG "Rx-MG" - performs in paediatric populations, namely its ability to explain variation in total primary care health care resource use measures, given these differences.
Study Design: Observational, retrospective, crosssectional. The study uses a dataset with demographic, diagnostic, pharmacy and cost information on the study population (see below) for two consecutive years, 2006 and 2007. The Rx-MG method (originally based on the US FDA's NDC codes) has been calibrated to the WHO's "ATC" system which is in use in Spain.
Population Studied: The study population includes approx. 67,000 universally insured persons, attending a set of 5 primary care clinics in
Badalona, Spain; of these, approximately 11,000 are children aged 0 to 14.
Principal Findings: Prescription of drugs varies widely with age; adults have higher rates in most
Rx-MG categories, and these are particularly larger in all categories in the cardiovascular group, most of the endocrine, and also in more specific areas like contraception - naturally -, but also glaucoma, gout, genito-urinary acute minor, most of the psychosocial conditions with the exception of the
ADHD, and respiratory chronic medical. Adults are also much more likely to be medicated for general signs and symptoms, like pain. Children, on the other hand, have higher rates in only a handful of conditions, mostly the respiratory - acute minor, airway hyperreactivity and cystic fibrosis -, but with much lower ratios. Probably due to these wide age differences, the tool is able to explain variation in primary care costs much better in adults than in the paediatric population - r2=0,55 compared to r2=0,13.
Conclusion: Prescription patterns in children are quite different from those seen in adult populations.
Partly due to the these variations the Rx-MG system performed better in explaining variation in terms of resource use measures between adult patients than in the paediatric population.
Implications for Policy, Delivery or Practice:
Researchers, managers and policymakers acknowledge the need for risk adjustment for several purposes. Results in this paper suggest that children have different patterns of both morbidity and drug prescription which might affect the ability of pharmacy based risk adjustment tools to be effectively used in paediatric populations, and this population should thus warrant special attention.
While this and other similar pharmacy based risk adjustment tools have many useful applications for child health services research, finance and management, there are also limitations that need to be acknowledged. This is particularly true if risk adjustment methods developed for adults are then applied to paediatric populations.
Funding Source(s): Fundação Gulbenkian,
Portugal; FCT, Portugal
Theme: Child Health
● Analysing Heterogeneity in a Diabetic
Population
Paulo Boto, M.D., M.S.
Presented by: Paulo Boto, M.D., M.S., Ph.D.
Candidate, Health Policy & Management, Johns
Hopkins Bloomberg School of Public Health, Av.
Carolina Michaelis, 22, 4º Esq, Linda-A-Velha,
2795-049, Portugal, Phone: +351914595634;
Email: pboto@jhsph.edu
Research Objective: Patient classification systems are helpful in managing health services by grouping people with similar conditions and/or resource use.
However, they inevitably group different patients.
The main objective of this study is to illustrate just such variability, namely in terms of resource use, in a diabetic population, using for this the Adjusted
Clinical Groups (ACGs) software.
Study Design: Observational, retrospective, crosssectional.
Population Studied: A population of about 56,000 users of primary care services over 14 years of age, in 5 clinics in Badalona, Spain, during the years of
2006 and 2007.
Principal Findings: Diabetes has in this population an overall prevalence of 10,22%, ranging from 8,9 to 11,8 across the 5 clinics (p<0,05). The number of diabetic patients seen by each doctor is also significantly different, ranging from 3,4 to 15% of each panel of patients (p<0,05). Diabetic patients in the highest morbidity resource utilization band
(RUB) tend to visit their primary care physician 10 times more often than the healthiest diabetic patients (p<0,05). The same pattern emerges when we look at total primary care costs, with the most expensive group spending approximately 5 times more than the healthier group of diabetics (p<0,05), and particularly 29 times more on pharmacy prescriptions (expected, since pharmacy represents approximately 10% of costs in the least expensive group, but 53% in the most expensive one).
Variation in the number of visits is explained
(adjusted r2=0,30) by age, sex, the number of chronic conditions patients have, the presence of major Aggregate Diagnosis Groups (ADGs), the presence of a frailty marker, and the presence of a hospital dominant condition, as well as the presence of 7 other chronic conditions, increasing most notably with diagnoses of asthma, CHF and depression (although diagnoses of COPD, hyperlipidemia and ischemic heart disease seem to have the opposite effect). Variation in costs across resource utilization band is on the other hand partially explained (adjusted r2=0,35) by age (but not sex), the number of chronic conditions, the presence of major ADGs, and of a frailty marker, and again the presence of 7 other chronic conditions, increasing in this case most notably with diagnoses of asthma and depression. The hospital dominant condition marker seems to have in this case a saving effect.
Conclusion: If we look at patients from a single disease perspective (e.g. a single disease like diabetes), we will be overlooking significant differences, namely in terms of their use of resources. Patients usually have more than one condition and that determines at least part of those differences.
Implications for Policy, Delivery or Practice:
Patients are becoming more complex, both in terms of their characteristics and their needs; health services management tools need to accommodate this fact to avoid dangerous simplifications.
Funding Source(s): Fundação Gulbenkian,
Portugal; FCT, Portugal
Theme: Quality and Efficiency: Measurement
● Accreditation, Benchmarking & Mortality
Paulo Boto, M.D., M.S.; Carlos Costa, Ph.D.; Sílvia
Lopes
Presented by: Paulo Boto, M.D., M.S., Lecturer,
Health Services Management, National School of
Public Health/Escola Nacional de Saúde Pública,
Av. Carolina Michaelis, 22, 4º Esq, Linda-A-Velha,
2795-049, Portugal, Phone: +351914595634;
Email: paulo.boto@ensp.unl.pt
Research Objective: Given the growing interest in benchmarking and accreditation projects in
Portugal, this study assesses the extent to which participation in these impacts the quality of health care, as measured by risk adjusted in-hospital mortality.
Study Design: Observational,cross-sectional, retrospective. In-hospital risk adjusted mortality rates of public sector hospitals involved in the HQS and/or JCI accreditation programs, as well as in the
IQIP benchmarking scheme, were analysed and coded as a binary variable (observed mortality above or below the expected). Chi-square statistic was used to compare results in 2x2 tables
(involvement vs mortality).
Population Studied: All public sector, acute care hospitals in Portugal (n=76).Hospital discharge databases from 2001 to 2006 were used to calculate in-hospital mortality. Results were adjusted for patient severity, using the Disease
Staging software.
Principal Findings: No statistically significant differences in terms of risk adjusted in-hospital mortality were found between participating and nonparticipating hospitals (p>0.5). The length of involvement with any of these projects or the participation in more than one had no effects either
(Fisher's Exact Test=0,61 and 0,54, respectively).
The direction of O/E mortality ratios (above or below 1) did not change for 80% of the hospitals involved this far. Hospitals are not significantly different in terms of mortality ratios at the time of entering the projects. Mortality ratios are moderately correlated from year to year (with correlation coefficients ranging from 0,51 to 0,82).
The mortality ratio (O/E) in the previous year is the most important predictor of the same ratio in the next year, as assessed by logistic regression models.
Conclusion: This study looks at accreditation and benchmarking processes not yet analysed in this perspective and context. Although the authors
believe that accreditation and benchmarking programs can bring substantial benefits to the institutions involved in them, at a time when, nationally and internationally, evidence of their effectiveness and cost-effectiveness starts being called for, decision makers, be they politicians or managers, need to know if the investment in these programs is worth being generalized, and, given competing options, which looks more promising. The answer, at this moment and in this context, is that none seems associated with better performance.
Implications for Policy, Delivery or Practice:
This study should promote a reflection on the impact of these programs in Portuguese hospitals, to help define a strategy for the future that is more evidence based and universal.
Theme: Quality and Efficiency: Measurement
● Introducing Rx-MG, a New Pharmacy Based
Risk Adjustment Tool, In an International
Context
Paulo Boto, M.D., M.S.
Presented by: Paulo Boto, M.D., M.S., Ph.D.
Candidate, Health Policy & Management, Johns
Hopkins Bloomberg School of Public Health, Av.
Carolina Michaelis, 22, 4º Esq, Linda-A-Velha,
2795-049, Portugal, Phone: +351914595634;
Email: pboto@jhsph.edu
Research Objective: Risk adjustment tools have come to include information on drugs prescribed.
This paper: 1. describes Rx-MG, a new pharmacy based risk adjustment tool; 2. uses it to describe the drug prescription patterns found in a specific primary care setting in Spain, 3. compares these to those of a reference population from the US; and 4. assesses the ability of the classification system to explain variation in health care resource use measures.
Study Design: Observational, retrospective, crosssectional.
Population Studied: A population of about 67,000 users of primary care services in 5 clinics in
Badalona, Spain, during the years of 2006 and
2007.
Principal Findings: Prescription of drugs varies with age, with the elderly - those over 65 - consuming many more medications classified by the Rx-MG system under the categories neurologic
Alzheimer’s disease - ratio ~63:1 when compared to those under 65 -, cardiovascular congestive heart failure - ratio ~15:1 -, cardiovascular chronic medical and genito-urinary acute minor - both with ratios of ~9:1. Prescription rates are higher in the younger only in very specific conditions, like those related to the female reproductive system, attention deficit hyperactivity disorder and acne, for instance.
Prescription rates vary too across the two settings compared, with the largest differences in the Rx-
MG categories female reproductive, allergy/immunology chronic inflammatory - ratios
~7:1 to 9:1 -, and musculoskeletal inflammatory conditions - ratios ~5:1 to 6:1, with US > Spain -, and gastrointestinal hepatic chronic stable, gastrointestinal hepatic inflammatory bowel disease, neurologic chronic medical and respiratory cystic fibrosis - with Spain > US. In spite of these cross-country differences, the tool is able to explain variation in primary care costs in this setting as well as diagnoses based systems used previously - r2=0,46 compared to r2=0,43 -. The results are comparable to those attained by the same tool in the US.
Conclusion: The Rx-MG system performed well in a setting very different from that where it was developed, with very different prescribing patterns and using the WHO’s ATC coding system, instead of the USA’s NDC. There was only a small percentage - ~5% -, of codes that could not be assigned to a category, which is a good measure of how well it adapts to a different context and coding system. The system highlighted differences in prescription rates between age groups that are consistent with the common clinical sense; it also highlighted differences between countries that can either be due to real differences in the prevalence of conditions in the two settings, differences in prescribing patterns or simply differences in the way delivery systems are organized (which might be the case for the differences for drugs related to the female reproductive system). The tool proved to be as useful in this setting as elsewhere in terms of explaining variation between patients in terms of resource use measures.
Implications for Policy, Delivery or Practice: In a time when risk adjustment is more and more required and accepted as necessary, this study validates the use of a new tool in a specific setting.
Results confirm the widely held belief that crosscultural comparisons should however be done very carefully, since a number of different factors can explain the differences found.
Funding Source(s): Fundação Gulbenkian,
Portugal; FCT, Portugal
Theme: Global Health
● Prevalence, Correlates & Impact of Substance
Use Disorders in PTSD Disability Seekers at the
Department of Veterans Affairs
Michel Boudreaux, B.A.; Maureen Murdoch, M.D.,
M.P.H.
Presented by: Michel Boudreaux, B.A., Research
Assistant, Health Policy & Managment, School of
Public Health, University of Minnesota, 420
Delaware Street, MMC 729, Minneapolis, MN
55455, Phone: (651) 503-4079; Email: boudr019@umn.edu
Research Objective: Some policy makers have called to reduce disability awards to veterans with secondary substance use disorders (SUDs),
despite limited knowledge about the prevalence of
SUDs and their relative impact on veterans seeking disability support. Secondary SUDs arise as the result of a primary disabling condition (e.g., when substances are used to self-medicate) and they may contribute to the level of disability. The objectives of this study were to examine the prevalence of SUDs in a sample of PTSD disability applicants, to evaluate the characteristics of these substance abusers, and to examine the association of SUDs to PTSD severity and functional impairment. Because women are the second fastest growing demographic of veterans and may express a clinical profile different from men, we examined these relationships within gender.
Study Design: Data is from the first wave of a nationally representative, longitudinal panel of veterans who have been followed for over 10 years.
Survey responses were supplemented with administrative records.
Population Studied: U.S. veterans who applied for
PTSD disability benefits at the Department of
Veterans’ Affairs (VA) between 1994 and 1998.
Women were oversampled. 5,395 veterans were sampled and 3,337 returned usable surveys.
Principal Findings: Over one in five respondents
(21.1%) were SUD positive. Men were more likely than women to have a SUD (25.1% versus 17.3%; p<.001). Those with a SUD diagnosis were disproportionately black, unmarried, less educated, and low-income. They were more likely to be exposed to combat traumas and reported more current life stressors. Overall, SUD respondents had worse social and physical functioning scores; however SUD positive and negative men had equivalent physical functioning. Risk factors for
SUD differ by gender. For women, substance users are more likely to have experienced sexual assault
(69.2%) than non-users (58.7%) and compared to similar men are more likely to be unmarried (82.3% versus 55.3%). Unlike men, race is not associated with SUD status for women. SUD respondents had higher PTSD severity (50.3 versus 43.5; p<.001) than veterans without SUDs. This pattern held for both men and women. Multivariate models showed that after controlling for demographics, service characteristics, trauma, and other health indicators, the mean difference in PTSD severity was less pronounced (3.8), but still significant. The interaction between SUD and gender was not significant.
Conclusion: These data suggest that SUDs are a common problem for men and women applying for
PTSD awards. SUDS are more pronounced in disadvantaged social groups, but the risk profile differs by gender. Most importantly, women with
SUDs are more likely to have a history of sexual assault. SUDs are associated with more PTSD severity and functional impairment.
Implications for Policy, Delivery or Practice:
Practitioners should be aware that gender, race, marriage status, and trauma history might put veterans with PTSD at risk for SUDs and consequently greater disability. Our study suggests that decision makers should proceed with caution before reducing benefits to veterans with PSTD and comorbid substance use problems as they may have greater financial need and a greater degree of disability.
Funding Source(s): Department of Veterans’
Affairs, Health Services Research and
Development Service
Theme: Military and Veterans Health Care
● Incremental Value of Using Medicaid Claim
Files to Study Comorbid Conditions &
Treatments in Dually Eligible Beneficiaries
Cathy Bradley, Ph.D.; Bassam Dahman; Paul
Bataki; Siran Koroukian
Presented by: Cathy Bradley, Ph.D., Professor,
Health Administration, Virginia Commonwealth
University, 1008 East Clay Street, Richmond, VA
23298, Phone: (804) 828-5217; Email: cjbradley@vcu.edu
Research Objective: Although investigations using
Medicare claims files are ubiquitous in the health services research literature, Medicaid claims files are used less frequently. Nonetheless, Medicaid is the major payer for healthcare among low-income persons. However, Medicaid agencies have limited resources to devote to extracting and linking claim files with other data sources and they are often wary of sharing their data with researchers.
Nevertheless, Medicaid is the most comprehensive source for studying health care utilization among the very poor. Using Medicaid and Medicare claim files from two Midwestern states, we assessed the added value of Medicaid claim files for identifying comorbid conditions and cancer treatments in a dually eligible sample. This assessment informs researchers weighing the potential gain from
Medicaid claim files in studying dually eligible beneficiaries diagnosed with cancer.
Study Design: Data were obtained from linked statewide tumor registries from two contiguous
Midwestern states (Michigan and Ohio), Medicare and Medicaid enrollment files, and Medicare and
Medicaid claims files. We estimated the prevalence of Charlson Comorbidity Index conditions by counting the number of patients with these conditions in the Medicaid claim files and adding them to the number identified from Medicare claims files alone. We then estimate the expected percent increase in the prevalence of comorbid conditions
(along with the 95% confidence interval) that could be obtained by using both Medicare and Medicaid claim files. We followed a similar procedure to identify treatments provided to dually eligible patients.
Population Studied: We selected all dually eligible patients diagnosed with female breast, colorectal, or prostate cancer between January 1, 1997 and
December 31, 2000. These cancer sites and years of data were common to both states’ datasets.
Patients were continuously enrolled in Medicaid for at least 12 months prior to the month of diagnosis.
We excluded patients that were not enrolled in
Medicare Parts A and B or were enrolled in a
Health Maintenance Organization (HMO) because we were concerned that claims data for these patients were not complete. The sample size of eligible patients from Michigan was 1107 and from
Ohio the sample size was 2463 patients.
Principal Findings: Medicaid claims added very few comorbid conditions over the conditions identified with Medicare claim files. The increase in the prevalence of comorbid conditions was between
5% and 6%. Likewise, Medicaid claims identified few cancer treatments that were not already present in Medicare claim files.
Conclusion: This study suggests that the incremental value of Medicaid inpatient, outpatient, and carrier claims is relatively small over what can be obtained from Medicare claims and strongly cautions researchers from using Medicaid files alone when studying dually eligible beneficiaries.
Implications for Policy, Delivery or Practice:
Within the elderly dually eligible samples, Medicaid claim files added little information, in terms of comorbid conditions and treatments, over Medicare claim files. Therefore, researchers may want to carefully consider whether Medicaid claims are worth the investment in time, dollars, and effort to supplement inpatient, outpatient, and carrier claims in the Medicare files.
Funding Source(s): NCI
● Poor Glycemic Control Increases Likelihood of Job Loss among Employed Patients with
Diabetes
Cathy Bradley, Ph.D.; Jennifer Elston-Lafata, Ph.D.;
Sandeep Vijan, M.D.; Elizabeth Dobie, M.S.;
George Divine, Ph.D.
Presented by: Cathy Bradley, Ph.D., Professor,
Health Administration, Virginia Commonwealth
University, 1008 East Clay Street, Richmond, VA
23298, Phone: (804) 828-5217; Email: cjbradley@vcu.edu
Research Objective: As the prevalence of diabetes in the U.S. has increased, so too have associated economic burdens. The increased prevalence among younger individuals suggests that diabetes will become more common in the working-age population. Consequently, employment of individuals with diabetes and its control are important issues for patients, clinicians, and employers. We examine the effect of glycemic control on patients’ probability of employment.
Study Design: We conducted a longitudinal study of patients receiving care from an integrated delivery system in southeast Michigan. Insured patients with diabetes aged 21 to 64 years were identified at the time of an HbA1c test. Primary data were collected using a mixed
(mailed/telephone) mode questionnaire and covered topics including socio-demographic characteristics and family background, labor market outcomes, and health status. HbA1c test results were extracted from the delivery system’s automated laboratory system at baseline and within
90 days of follow-up survey, which occurred 1 year after baseline. Multivariable logistic regression was used to predict the likelihood that patients who were employed at baseline were non-employed 1 year later as a function of their baseline and follow-up
HbA1c. Patient characteristics along with job characteristics were controlled in the analyses.
Population Studied: The response rate to the mailed survey was 69% (N=1066). Patients’ average age was 54 years (SD=7.15); 76% were married, 55% were female, and a substantial portion of the patients were African American
(41%). The majority of patients had some college or a college degree (72%).
Principal Findings: In the follow-up survey, 11% of diabetic patients were no longer employed. The likelihood of continuing to be employed at follow up decreased with increasing baseline HbA1c levels
(OR=0.88, 95% C.I.=0.78-0.99). The mean baseline HbA1c for patients who continued to be employed was 7.9% whereas patients that left the workforce had a mean HbA1c of 8.1% at baseline.
Poor HbA1c control at the time of follow up was not associated with non-employment (OR=0.96, 95%
C.I.=0.85-1.09).
Conclusion: To our knowledge, this is first study to use primary data to demonstrate that poor HbA1c control in a base year predicts transitions from employed to unemployed 1 year later among patients with diabetes. Although the effect of poor
HbA1c control may not be immediately apparent, either in terms of health or economic outcomes, increasing HbA1c levels are predictive of economic consequences within a year. More research is needed to examine whether diabetes complications or other factors are the direct cause of lower employment in those with poor glycemic control.
Implications for Policy, Delivery or Practice: The long-term economic impact of retaining diabetic patients in the workforce could increase patients’ and their families’ financial stability through maintenance of earnings and health insurance.
Retaining employed patients in the workforce may also reduce costs to employers and result in substantial savings in productivity, which greatly benefits society. Interventions that encourage diabetic workers to adhere to their medications and other self-management behaviors known to help sustain HbA1c control are likely to be highly costeffective in terms of both health benefits and productivity savings.
Funding Source(s): NIDDK
Theme: Prevention and Treatment of Chronic
Illness
● Potential Saving from Colorectal Cancer
Prevention & Control Strategies
Cathy Bradley, Ph.D.; Iris Lansdorp-Vogelaar; K.
Robin Yabroff; Bassam Dahman; Eric Feuer; Martin
Brown
Presented by: Cathy Bradley, Ph.D., Professor,
Department of Health Administration, Virginia
Commonweath University, 1008 East Clay Street,
Richmond, VA 23298-0203, Phone: (804) 828-
5217; Email: cjbradley@vcu.edu
Research Objective: Colorectal cancer (CRC) is the third most common malignancy and second most common cause of cancer death among
Americans. We developed a model that projects
CRC medical, patient time, and lost productivity cost savings that may result from implementing strategies involving reduction of risk factor prevalence, increasing CRC screening, and/or improvements in treatment.
Study Design: The model estimated the potential savings from five CRC prevention and control strategies from 2005 to 2020 with expected population trends: (1) risk factors, screening and chemotherapy use stay at 2005 levels; (2) risk factors, screening and chemotherapy use develop more favorably than current trends suggest; (3) risk factors are reduced in the population, but screening and treatment trends remain at the 2005 level; (4) screening rates increase, but risk factors and chemotherapy stay at the 2005 level; and (5) chemotherapy use improves, but risk factors and screening stay at the 2005 level.
Population Studied: Incidence and mortality projections and associated life-years with CRC and life-years lost to CRC were based on the National
Cancer Institute’s (NCI) Colorectal Cancer Mortality
Projections simulation of a large population of individuals from birth to death. SEER-Medicare claims data were used to estimate CRC medical care costs. Patient time cost estimates were derived from service frequency differences between colorectal cancer cases and controls from SEER-
Medicare. The human capital method was used to estimate the expected earnings.
Principal Findings: By 2010, a strategy that involves simultaneous interventions results in nearly
5100 fewer deaths. The greatest long-term gains in life years and deaths avoided for a single strategy is risk factor reduction, followed by improved screening, and improved chemotherapy use.
Current trends in CRC incidence and mortality are expected to result in medical and patient time costs of approximately $38 billion in 2010 and $43.1 billion in 2020. Simultaneous interventions can reduce these costs by $7.2 billion in 2020, whereas improvements in chemotherapy will result in a slight increase in costs. Productivity costs using 2005 trends are expected to approach $32 billion in 2020.
These costs can be reduced by approximately $3 billion in 2020 by improving treatment or screening, or by $6 billion with simultaneous interventions.
Compared to current trends, reductions in total costs are greatest with simultaneous interventions
(18.0%), followed by increased screening (9.7%), reduction in risk factor prevalence (8.0%), and improved treatment (1.0%).
Conclusion: Current trends in CRC incidence and mortality will result in $75 billion in medical, patient time and lost productivity costs in 2020. Strategies that reduce risk factor prevalence and improve screening led to the greatest long-term savings relative to the other strategies.
Implications for Policy, Delivery or Practice: A policy maker faced with having to choose only one strategy to reduce CRC incidence and mortality would likely choose to improve screening rates— assuming that costs to implement the programs are roughly equivalent. A strategy that attacks CRC on several fronts—screening, risk factors, and treatment—results in saving exceeding $13 billion in 2020, suggesting that CRC interventions are likely to be cost-effective.
Funding Source(s): NCI
Theme: Prevention and Treatment of Chronic
Illness
● Electronic Health Records: The Gap Between
Desirability & Availability of Functions
James Bramble, Ph.D.; Mark Siracuse, Pharm.D.,
Ph.D.; Kimberly Galt, Pharm.D., Ph.D.; Karen
Paschal, P.T., D.P.T., M.S.; Andjela Drincic, M.D.;
Amy Abbott, Ph.D., R.N.
Presented by: James Bramble, Ph.D., Associate
Professor, School of Pharmacy & Health
Professions, Creighton Health Services Research
Program, 2500 California Plaza, Boyne, 140,
Omaha, NE 68178, Phone: (402) 280-4129; Email: jbramble@creighton.edu
Research Objective: Electronic health records
(EHRs) can improve efficiency and patient safety in our health care system. EHRs have many functions, including the ability to record patient and clinical data, provide clinical decision support, and assist with medication prescriptions. EHRs offer physicians the ability to electronically send prescriptions, order laboratory tests and radiological images, and view the results. This study explores
EHR functions physicians desire and those that are available.
Study Design: A cross-sectional survey was administered using a modified Dillman technique between July 2007 and November 2007. Potential
EHR functions were identified from an extensive literature review of existing instruments. Physicians who identified themselves as using an EHR selected functions currently available on their systems, while physicians planning to use an EHR selected those functions they desired.
Population Studied: All office-based physicians in
Nebraska and South Dakota were asked to complete a survey about EHRs.
Principal Findings: The survey was completed by
955 physicians. Half of these physicians (50.8%) identified themselves in the planning stages of EHR adoption with 29.1% reported using them, referred to as planners and users, respectively. For all functions, the percentage of physicians who desired a specific function was always greater than the percentage of physicians reporting the function was available. Functions desired by planners and those available to users included: recording patient demographics (99.6% to 94.6%), patient’s presenting complaint (91.3% to 82%), medical history (95.9% to 83.3%), encounter notes (93% to
92.8%), and current medication history (93.8% to
88.1%). A larger gap existed among other EHR functions between planners’ desires and those available to users. Examples include the ability to view laboratory results (95.3% to 69.1%), order prescriptions (82.5% to 66.9%), send prescriptions electronically (83.9% to 51.4%), exchange patient data with hospitals (76.7% to 30.9%), exchange patient data with other physicians (74.4% to
25.5%), and reference drug information (75.9% to
51.8%).
Conclusion: The results indicate that EHR functions fall into two distinct groups. One group included functions with relatively little gap between what is desired and what is available. The use of these functions is self-contained to the EHR system within the physician’s practice. There is no need for the system to communicate with other systems.
The second group had functions where there was a much greater gap between physicians’ desire for the function and function availability. The common characteristic of these functions was the dependency of the EHR system communicating with other systems or the interoperability of the
EHR system. For example, the exchange of patient data with the hospital requires interaction with a system outside the physician’s office. The lack of interoperability becomes a barrier for physicians who desire to use this EHR function.
Implications for Policy, Delivery or Practice:
This study’s findings suggest that both policy makers and private stakeholders must facilitate successful interoperability between relevant parties to realize the full potential of EHRs. Factors that must be addressed include: 1) infrastructure readiness at the local, regional, and national level;
2) economic, user, and consumer/patient acceptance; and 3) legal and privacy concerns.
Funding Source(s): AHRQ, Nebraska Medical
Association
Theme: Health Information Technology
● The Effect of Intergenerational Relationships on Financial Preparations for Long-term Care
Shuli Brammli-Greenberg; Revital Gross, Ph.D.
Presented by: Shuli Brammli-Greenberg, Senior
Researcher, Health Economist, Smokler Center for
Health Research, Myers-JDC-Brookdale Institute,
JDC Hill P.O. Box 3886, Jerusalem, 91037, IL,
Phone: (972) 655-7481; Email: shuli@jdc.org.il
Research Objective: To examine the effect of various aspects of parents' relations with their children on their financial arrangements for longterm care (LTC).
Study Design: Telephone interviews were conducted in late 2006 with a representative national sample of residents of Israel aged 22+ using a structured questionnaire in Hebrew, translated into Arabic and Russian. Some 1,700 individuals were interviewed. The response rate was 80%. The data were analyzed using advanced econometric methods to neutralize offsetting and other effects of the individual's characteristics on their relations with their children and their financial arrangements for LTC. In the first stage of the twostage analysis, we examined the effect of the individuals' characteristics on aspects of their relationship with their children; in the second, we examined the effect of their characteristics and aspects of their relationship with their children on their financial arrangements for LTC in old age.
Population Studied: Individuals with children in
Israel.
Principal Findings: Factor analysis identified four distinct aspects of intergenerational relationships.
Culture had the strongest effect on the relationship with children along with other background variables
(gender, age, income, family status). Each aspect was found to have a different, independent effect on individuals' financial preparedness for long-term care. Agreement with the statement "Children have to help their parents when they're old" was found to have a strong negative independent effect on such arrangements while agreement that there is "a need for parents' economic independence" had a strong positive effect. Having "a strong emotional bond with the children" and "trusting that the children will provide long-term care" have a positive independent effect only on ownership of LTC insurance.
Conclusion: Caring for the elderly and people with disabilities is a key challenge for health and welfare systems worldwide. Given the limited public funding, it is becoming increasingly important for individuals to make their own financial preparations for long-term care. The study contributes to an indepth understanding of individuals' relationship with their children, which is one of the main factors affecting their financial decisions regarding old age.
The study also contributes to the professional literature exploring the effect of intergenerational relationships on individuals' financial decisionmaking.
Implications for Policy, Delivery or Practice: The study findings show that policymakers can affect individuals' financial arrangements for LTC through
education and by changing individuals' perceptions about their relationships with their children.
However, it is necessary to relate in a different way to different population groups (distinguished by culture, age, income, etc.)
Funding Source(s): T he Israel National Institute for Health Policy and Health Services Research &
Brookdale Institute
Theme: Long Term Care
● Unit-Level Staffing, Workload & Adverse
Events in Army Acute Care Hospitals: 2003-2006
Sara Breckenridge-Sproat, M.S.N., R.N., Ph.D.;
Meg Johantgen, Ph.D.; Pat Patrician, Ph.D.
Presented by: Sara Breckenridge-Sproat, M.S.N.,
R.N., Ph.D., Lieutenant Colonel, U.S. Army Nurse
Corps, Nursing Research, Walter Reed Army
Medical Center, 6900 Georgia Avenue, NW,
Washington, DC, 20307, Phone: (703) 528-1186;
Email: breck.sproat@comcast.net
Research Objective: 1. To describe staffing levels, skill mix, and patient workload changes over time in an inpatient acute care setting; 2. to examine the relationship of staffing, skill mix and workload to medication errors and falls; and 3. to determine if the nursing practice environment moderates this relationship.
Study Design: A descriptive correlational longitudinal design was used to conduct a secondary data analysis of unit level data from the
Military Nursing Outcomes Database (MilNOD) from
2003-2006 to examine the relationships between nurse staffing and quality of care. A cross sectional design using only 2006 data was used to examine the moderating relationship of the practice environment. Linear mixed modeling was used to accommodate the nested data.
Population Studied: The MilNOD data sample used in this study represents 22 inpatient nursing care units in four Army hospitals, over four years.
The units are comprised of a mix of medical surgical, step down and intensive care. Survey data from 506 nurses from the study units reflected measures of the practice environment in 2006.
Principal Findings: Despite volatility in the organizations, preliminary analyses indicate little variation in staffing levels and adverse events.
However, there were changes in staff mix, including increased use of reserve and civilian nurses. Linear mixed modeling showed limited effects of total staffing and staff mix on adverse events. The practice environment was a significant moderator of the relationship between staffing and outcomes but explains only a small amount of variance in the overall model.
Conclusion: Findings of this study are consistent with other unit level studies that found a relationship between nurse staffing, practice environment and patient outcomes. However, the variance explained in patient outcomes remains limited.
Implications for Policy, Delivery or Practice:
This unique longitudinal, unit-level analysis contributes to understanding the dynamic nature of nurse staffing at the micro level. It also supports the growing literature on the variation in nursing staffing and the influence on patient outcomes. As policy-makers struggle to identify optimum staffing levels, understanding the day-to-day and shift-to shift variation is important.
Funding Source(s): Tri-Service Nursing Research
(TSNRP)
Theme: Organizational Performance and
Management
● Predicting Wound Deterioration in Home
Health Care Patients Using Competing Risks
Survival Analysis
Carlin Brickner, M.S.; Iordan, Slavov, Ph.D.; Robert
Rosati, Ph.D.
Presented by: Carlin Brickner, M.S., Statistical
Analyst, Center for Home Care Policy & Research,
The Visiting Nurse Service of New York, 1250
Broadway - 20th Floor, New York, NY 10001,
Phone: (212) 609-5793; Email: carlin.brickner@vnsny.org
Research Objective: Medicare spending on hospitalizations for wound related problems doubled from 2000 to 2005, and the health care industry sees home health care an as an important component in a patient’s wound treatment. In June
2008 The Center for Medicare & Medicaid Services
(CMS) released risk adjusted models for predicting the improvement in the status of surgical wounds and the number of surgical wounds. The model measuring the improvement of a surgical wound requires both a baseline measurement at start of care (SOC), and an end point measurement at discharge. Patients in home health care who are hospitalized before a follow-up assessment will not have their wound status evaluated and thus the
CMS predictive model will underestimate the risk.
The purpose of this study is to correctly identify and estimate the parameters of factors that are significantly correlated with the deterioration of a pressure wound or stasis ulcer using a different approach.
Study Design: In this cohort study a competing risk survival analysis was used to measure the number of days until an event or censoring. The two competing risks in the proportional hazards regression model were: a) hospitalization for a wound complication; b) any other hospitalization.
Patients who were discharged from care without a hospitalization were treated as censored events. A wide variety of baseline data sources were used to select the best model predicting the competing risks such as CMS mandated Outcome and Assessment
Information Set (OASIS), medication counts, payer eligibility, and the clinician’s plan of care for the patient.
Population Studied: All patients admitted to
Visiting Nurse Service of New York (VNSNY) in
2007 with a pressure wound or stasis ulcer indicated on the OASIS at SOC (N= 9799).
Principal Findings: The overall hospitalization rate was 45.14% with a median time to hospitalization of
23 days, and 8.04% of all patients admitted with a stasis ulcer, and 4.64% with a pressure ulcer were hospitalized because of a wound complication.
After controlling for baseline measurements such as the quantity of wounds, and stage and status of the most severe wounds, several factors emerged as significant predictors for hospitalizations because of a wound complication. Patients with diabetes (OR
= 1.48, 95%CI [1.25, 1.76]) and ischemic heart disease (OR = 1.50, 95%CI [1.13, 2.00]) and patients with a pressure ulcer that could not be observed (OR = 1.77, 95%CI [1.32, 2.37]) increased the hazard of hospitalization with a wound complication.
Conclusion: Using a competing risks model in a home health care setting isolates the factors that are critical in identifying patients at most risk of wound deterioration at SOC.
Implications for Policy, Delivery or Practice:
Identifying patients at admission who are at an elevated risk of wound related hospitalization can allow home health care agencies to quickly assign a wound care specialist and advanced wound treatment technology to the patient.
Theme: Quality and Efficiency: Measurement
● Characterizing Polypharmacy in Long-Term
Care Homes: Any Association with Quality?
Susan Bronskill, Ph.D.; Sudeep Gill, M.D., M.S.;
Michael Paterson, M.S.; Geoffrey Anderson, M.D.,
Ph.D.; Paula Rochon, M.D., M.P.H.
Presented by: Susan Bronskill, Ph.D., Scientist,
Institute for Clinical Evaluative Sciences, 2075
Bayview Avenue, Toronto, M4N3M5, CA, Phone:
(416) 480-4055 ext. 3873; Email: susan.bronskill@ices.on.ca
Research Objective: Old and frail residents of long-term care (LTC) homes are often prescribed several drug therapies at the same time. While the treatment of many chronic diseases requires the use of multiple medications, this practice also increases the risk of adverse drug events and drugdrug interactions. Concurrent use of nine or more medications has previously been used as an indicator of polypharmacy and applied to assess quality in LTC homes. In this project, we examine the clinical and behavioral characteristics of LTC home residents and characterize patterns of use in the drug therapies they receive across LTC homes in order to explore the usefulness of the concurrent dispensing of nine or more medications
(polypharmacy) as a facility-level marker of quality of care.
Study Design: In this population-based study, we examined a prevalent cohort of LTC home residents. Variation in the concurrent dispensing of nine or more medications was assessed across residents and facilities with multivariate, logistic regression models. Based on control limits derived from a funnel plot, facilities were identified as having high, low or expected rates of polypharmacy among residents. Rates of polypharmacy across facilities were compared with rates of use of Beers criteria drugs and antipsychotic medication.
Population Studied: A prevalent cohort of 64,394 residents 66 years of age and older residing in 589
Ontario LTC homes in 2005.
Principal Findings: Nine or more medications were dispensed concurrently to 10,007 (15.5%)
Ontario LTC home residents. Rates of polypharmacy varied 3-fold across prescribing categories: 26.2% of residents in high-rate facilities received nine or more medications versus 7.9% of residents in low-rate facilities. Compared with those dispensed fewer medications, residents with polypharmacy were more likely to have increased clinical comorbidity indicated by Charlson scores of
2 or greater (odds ratio (OR) 4.14, 95% confidence interval (CI) 3.81-4.49) but were less likely to have memory problems (i.e. resident cannot locate own room) (OR 0.78, 95% CI 0.73-0.84). Diuretics
(68.2%), proton pump inhibitors (54.8%) and angiotensin-converting-enzyme (ACE) inhibitors
(51.7%) were the most commonly prescribed medications. Facility-level rates of polypharmacy were modestly correlated with rates of use of Beers criteria drugs (r=0.27, p <0.001) and antipsychotic medications (r=0.16, p <0.001).
Conclusion: Residents in Ontario LTC homes commonly received nine or more medications concurrently.
Implications for Policy, Delivery or Practice:
While at the resident-level the practice of polypharmacy is likely appropriately concentrated in individuals with multiple chronic conditions, the high rate of polypharmacy at certain facilities and its correlation with other measures of inappropriate prescribing suggest some value for this measure as a mechanism for guiding medication review at the facility level.
Funding Source(s): Canadian Institutes of Health
Research
Theme: Long Term Care
● Current Thinking on Ethics & Public Health
Preparedness: Overview & Practitioner Reaction
Janet Bronstein, Ph.D.; Andrew Rucks, Ph.D.
Presented by: Janet Bronstein, Ph.D., Professor,
Health Care Organization & Policy, University of
Alabama at Birmingham School of Public Health,
1665 University Boulevard, Birmingham, AL 35294-
0022, Phone: (205) 975-8962; Email: jbronstein@uab.edu
Research Objective: Disaster situations challenge the standard codes of ethics that form the basis both for professional codes and for public expectations of legitimate behavior. For example, triage during disasters is oriented towards saving the most lives, not the most severely ill, while plans for scarce vaccine distribution during a pandemic incorporate measures of social worth along with vulnerability to disease. Dialogue on appropriate alternative ethical values is occurring in the bioethics literature, but is not generally accessible to public health practitioners who are currently planning responses to disaster situations.
Study Design: We conducted a bioethics literature review on six areas relevant to preparedness practice: resource distribution in disasters, public compliance with directives, professionals’ duty to serve in threatening situations, commandeering of private property, incorporation of voluntary offers to assist and risk communication. This overview will be presented to public health preparedness specialists at the Preparedness Summit in February
2009. These practitioners will evaluate the current state of the bioethics dialogue for its utility in helping them plan for and practice during disaster situations in the context of scenarios which will include ethical dilemmas in the six domains.
Principal Findings: Current thinking on resource distribution highlights the principle of equity but challenges the public to define the criteria to be used to formulate equivalent categories.
Procedural and deliberative justice principles are key for legitimate resource distribution. Current thinking on public compliance highlights the value of public health prevention but identifies approaches for limiting prevention activities in order to protect individual liberties, such as requiring use of least restrictive effective approaches and offering opportunities to comply voluntarily before mandatory requirements are invoked. Current thinking on professionals’ duty to serve identifies the limits of personal risk that professionals are expected to take and urges explicit agreements to serve that should be included in preparedness planning activities. Current thinking on use of private property distinguishes between property used for public benefit and property seized because it represents a hazard. Current thinking on involvement of volunteers expands on the value of social reciprocity, while thinking on risk communication highlights the danger that fear represents in disaster situations, while distinguishing between informational and reassurance messages.
Conclusion: There is much within the expanding bioethics literature that can be useful to preparedness professionals, but it is not known at this point how to transmit these ongoing dialogues in a way that is useful for these practitioners.
Implications for Policy, Delivery or Practice:
Because effective action in disaster situations may require alternatives to standard codes of ethics, there is a risk that these actions will be seen as illegitimate by the public and unethical by professionals. Both perceptions will hinder our ability to protect the public’s health. Thus it is essential to engage in a practitioner-oriented dialogue on the ethics that should appropriately guide preparedness activities.
Funding Source(s): CDC
Theme: Public Health
● Executive Training for Research Application
(EXTRA): Smarter Decisions… Stronger Health
System
Mireille Brosseau, M.A.
Presented by: Mireille Brosseau, M.A., Program
Officer, Executive Training, Knowledge Exchange,
Canadian Health Services Research Foundation,
1565 Carling Avenue, Ottawa, CA, Phone: (613)
728-2238; Email: mireille.brosseau@chsrf.ca
Research Objective: To improve the knowledge transfer and uptake of evidence-based decisionmaking by individuals and organizations.
Study Design: EXTRA is a two-year fellowship that uses a blended learning approach (residency sessions, e-learning platform, intervention projects, mentoring, and networking) designed to increase decision-maker and organizational capacity for research use. Using Kirkpatrick’s model for evaluating training effectiveness (Kirkpatrick 1994,
1998), data is gathered systematically through questionnaires, focus groups, interviews and document reviews to address the following questions: Do fellows acquire the necessary skills to use research-based evidence to bring about organizational change?; Do fellows apply the skills learned and use research-based evidence to bring about organizational change?; Are the skills needed for improved use of research in management spread beyond those formally enrolled as fellows in
EXTRA?; and, Do fellows improve their capacity to collaborate as evidence-based decision-makers between professional streams?
Population Studied: Mid-career nurse, physician and other healthcare executives occupying senior leadership positions in direct-care-delivery organizations.
Principal Findings: According to the 2007 program evaluation data: 81% of fellows felt that their organization engaged in evidence-informed change management strategies based on the fellow’s intervention project; 95% of fellows felt that colleagues involved in their intervention project increased their capacity to critically examine their actions related to problem-solving and decisionmaking; and 76% of fellows felt that their intervention project (and change management experiences derived from the project) were transferred beyond their organization. Over the duration of the fellowship the proportion of fellows who felt their: knowledge of research evidence was
‘excellent’ to ‘good’ increased from 37% to 90%; skill-set for assessing quality of evidence was
‘excellent’ to ‘good’ increased from 37% to 90%; organization used research evidence 'frequently',
'most', or 'all of the time' increased from 29% to
72%.
Conclusion: As demonstrated by the data, evaluation results indicate that research use skills and strategies are being transferred to fellows and their organizations. Also, the data indicate that changes are being effected within leadership and interprofessional ranks. Organizational case studies are underway (2008/09) to document the nature and impact of these changes in organizations highly involved in the EXTRA program.
Implications for Policy, Delivery or Practice: The
EXTRA experience is transferable to other contexts and jurisdictions. In 2008 the program was expanded to include interdisciplinary teams and policy makers from government ministries. In 2007 and 2008 the Centre for Health Systems Research at the National Institute of Public Health in Mexico adapted the EXTRA curriculum for use in its
Research Use and Application course designed for senior healthcare managers from Central America and Mexico. There has also been interest in piloting an EXTRA-like program in several African countries affiliated with the EVIPNet initiative.
Funding Source(s): Health Canada
● Orthopaedic Surgery Venous
Thromboembolism Prophylaxis: Pay-For-
Performance Guidelines Are Harming Our
Patients!
Gregory Brown, M.D., Ph.D.
Presented by: Gregory Brown, M.D., Ph.D.,
Associate Chief of Surgery for Outcomes,
Orthopaedic Surgery, Park Nicollet Health Services,
6490 Excelsior Boulevard, Suite E400, Saint Louis
Park, MN 55426, Phone: (952)993-5663; Email: brownga@parknicollet.com
Research Objective: (1) to conduct an evidencedbased medicine pooled analysis of randomized controlled trials (RCTs) cited by the American
College of Chest Physicians’ (ACCP) guidelines for venous thromboembolism (VTE) prophylaxis to evaluate differences in pharmacologic class rates for symptomatic deep vein thrombosis (DVT), pulmonary embolism (PE), fatal PE, and operative site bleeding; and (2) to compare the 30 day hospital readmissions, 30 day reoperations, and 90 day VTE events after total hip arthroplasty (THA) or total knee arthroplasty (TKA) using low molecular weight heparins (LMWH) and/or vitamin K antagonists (VKA) versus aspirin for venous thromboembolism prophylaxis.
Study Design: (1) an evidenced-based medicine pooled analysis of RCTs for pharmacologic VTE prophylaxis was conducted; and (2) a retrospective review of elective lower extremity joint replacement operations at a large-volume community hospital over 24 months was conducted evaluating 30 day readmissions, 30 day reoperations, and 90 day VTE events.
Population Studied: (1) secondary data collection included all RCTs evaluating pharmacologic VTE prophylaxis for THA, TKA, and hip fracture surgery cited by the ACCP guidelines; and (2) primary data collection included an integrated healthcare system and community hospital with high-volume total joint replacement program using an integrated inpatient and outpatient electronic medical record. All patients were included in the cohort (no exclusion criteria).
Principal Findings: (1) pooled analysis: symptomatic DVT rates: aspirin 84/8726 (0.96%),
VKA 67/3326 (2.01%), LMWH 95/7436 (1.28%), pentasaccharides 34/3616 (0.94%); PE rates: aspirin 54/8726 (0.62%), VKA 18/4518 (0.40%),
LMWH 42/9269 (0.42%), pentasaccharides 29/3616
(0.80%); fatal PE rates: aspirin 19/8726 (0.22%),
VKA 2/4518 (0.04%), LMWH 12/9269 (0.13%), pentasaccharides 11/3616 (0.30%); operative site bleeding/wound complication rates: aspirin 40/8726
(0.46%), VKA 60/2681 (2.24%), LMWH 217/7423
(2.92%), pentasaccharides 69/3616 (1.91%).
Aspirin had a significantly lower symptomatic DVT rate compared to VKA (p= 3.5x10-6); aspirin had significantly lower operative site bleeding rate compared to VKA (p=5.4x10-18), LMWH
(p=1.0v10-35), and pentasaccharides (p=4.7x10-
15). (2) Post-operative VTE prophylaxis measures were: LMWH and/or VKA (n=1545) and aspirin
(n=386). 30 day readmissions for patients receiving
VKA and/or LMWH were 61/1545 (3.95%) and for patients receiving aspirin were 5/386 (1.30%), p=0.010. 30 day reoperations for patients receiving
VKA and/or LMWH were 41/1545 (2.65%) and for patients receiving aspirin were 3/386 (0.78%), p=0.027. VTE event rates for patients receiving
VKA and/or LMWH were 12/1545 (0.78%) and for patients receiving aspirin were 6/386 (1.55%), p=0.16.
Conclusion: The evidenced-based medicine pooled analysis demonstrated that symptomatic
DVT, PE, and fatal PE rates were not significantly different for aspirin compared to VKA, LMWH, and pentasaccharides. The operative site bleeding relative risks of VKA, LMWH, and pentasaccharides compared to aspirin, are 4.9 (95% CI 3.3-7.3), 6.4
(95% CI 4.6-8.9), and 4.2 (95% CI 2.8-6.1), respectively. The strict exclusion criteria of RCTs negate the generalizability of their results to routine clinical practice. The retrospective review of a highvolume community hospital confirmed the increased operative site bleeding rate of patients receiving
LMWH and/or VKA. In the community hospital, there was no significant increase in VTE rates in patients receiving aspirin as compared to LMWH and/or VKA.
Implications for Policy, Delivery or Practice:
Practice guidelines must include risk stratification
for VTE and bleeding risks. The pooled analysis provides the evidence needed to justify the use of aspirin as a VTE agent after THA or TKA in patients without additional VTE risk factors. The ACCP guidelines have been incorporated into the Surgical
Care Improvement Project (SCIP) and Medicare’s pay-for-performance (P4P) project. When P4P incentives are linked to inadequate guidelines, P4P has the potential to harm the very patients whose healthcare is supposedly “improved.”
Theme: Quality and Efficiency: Policies and
Incentives
● Using Econometric Analysis when Long-Term
Randomized Trials are Unethical: The Case of
Dental Care & Cardiovascular Disease Events in
Women
Timothy Brown, Ph.D.; Erin dela Cruz, B.A.
Presented by: Timothy Brown, Ph.D., Assistant
Adjunct Professor, School of Public Health - Petris
Center, University of California at Berkeley, 2150
Shattuck Avenue, Suite 525, Berkeley, CA 94704,
Phone: (510) 643-4103; Email: tbpetris@berkeley.edu
Research Objective: To determine if dental care prevents cardiovascular disease events (angina, fatal and non-fatal heart attacks and strokes, and fatal and non-fatal occurrences of congestive heart failure) in women
Study Design: This study uses a quasiexperimental design to determine a causal relationship between dental care and subsequent cardiovascular disease events. Contemporaneous controls include age, race, education, marital status, and foreign birthplace. Controls for cardiovascular risk factors (health status, body mass index, alcohol use, smoking status, diabetes status, high blood pressure status, physical activity, and depression) were also included, lagged by one period to avoid endogeneity bias. To correct bias from omitted variables and measurement error, the instrumental variable method was applied to predict lagged dental care.
Population Studied: We use data from the Health and Retirement Study (HRS). The HRS is a biennial longitudinal survey of individuals aged 50 and over (and their spouses) funded by the National
Institutes of Aging (spouses may be younger than
50 years old).
Principal Findings: The probability of a cardiovascular disease event over the period examined was reduced from 1.2% to 0.2% for women receiving dental care who would not otherwise have received it.
Conclusion: Dental care is highly effective in the prevention of cardiovascular disease events among women.
Implications for Policy, Delivery or Practice:
These results suggest an enlarged role for public dental insurance programs, particularly those aimed at older adults. While a cost-effectiveness study has yet to be done, it may be the case that the cost of expanding dental care to middle-aged and older individuals who are not currently receiving regular dental care may be more than offset by savings from the subsequent reduced need to treat such individuals for cardiovascular disease events.
Theme: Prevention and Treatment of Chronic
Illness
● Facility Level Variation in Adherence to Oral
Diabetes Medications
Christopher Bryson, M.D., M.S.; Chuan-Fen Liu,
Ph.D.; Matthew Maciejewski, Ph.D.; Stephan Fihn,
M.D., M.P.H.; George Jackson, Ph.D.; John Piette,
Ph.D.
Presented by: Christopher Bryson, M.D., M.S.,
Investigator, Health Services Research &
Development, VA Puget Sound Health Care
System, 1100 Olive Way, Suite 1400, Seattle, WA
98101, Phone: (206) 277-1770; Email: cbryson@u.washington.edu
Research Objective: Daily, ongoing drug therapy is a principal treatment for patients with diabetes.
While prior research has identified patient-level factors associated with poor medication adherence, these factors are either largely immutable or require enormous resources to change. Potential organizational barriers to adherence receive far less attention. Little research has focused on variation in adherence to medications among practices in a large health care system such as the VA. This study assessed the magnitude of variation in medication adherence across VA facilities for patients with diabetes who were actively managed in primary care clinics.
Study Design: Patients with diabetes in the VA nationwide were identified from electronic data sources using a validated algorithm of 2 diagnoses of diabetes in FY2005-2006 and at least one primary care visit per year in FY2006 and FY2007.
Users of a specific oral hypoglycemic agents (OHA, including sulfonylureas, metformin, and glitazones) were identified by two or more non-partial fills of the drug in the baseline year (FY2006) recorded in central VA electronic pharmacy records. Subjects on insulin other than NPH were excluded to remove those who may have transitioned to insulin therapy.
Facilities with fewer than 100 OHA users were excluded in order to assure stable facility-level estimates. Medication adherence for each patient was estimated for the first 3 months of FY2007.
Adherence to each drug was assessed individually by a validated algorithm, and then the proportion of days adherent was averaged across drugs to produce an adherence score for the patient’s overall regimen. Patients were considered adherent if they filled at least 80% of their OHA prescriptions during this period. Facility-level adherence
measures were calculated based on the proportion of adherent patients.
Population Studied: All VA patients with diabetes.
Principal Findings: This analysis included 552,455 patients from 692 VA medical centers and community-based outpatient clinics, with 100 to
7110 patients in each qualifying facility. Adherence to sulfonylureas was better overall than adherence to metformin (average adherence was 74.7% for sulfonylurea, 66.7% for metformin, 72.6% for glitazone). There was substantial variation in facility-level adherence to OHAs, with the top ranking facility having 85.5% of patients classified as adherent, and the bottom facility having 41.3% of patients classified as adherent. The 10th 25th,
50th, 75th, and 90th percentiles were 65.6%,
69.5%, 72.2%, 75.3%, and 77.9%, respectively.
CBOCs had more variation than VA medical centers.
Conclusion: Overall adherence to OHAs is suboptimal. There is substantial variation among
VA facilities in the proportion of patients who are adherent to their medications for diabetes.
Implications for Policy, Delivery or Practice:
Examining organizational characteristics associated with this variation could identify modifiable factors to improve medication adherence.
Funding Source(s): VA
Theme: Organizational Performance and
Management
● Immigration, Acculturation & Chronic Back &
Neck Problems in Asian- & Latino-Americans
Quynh Bui, M.D.; Mark Doescher, M.D. M.S.P.H.;
Vicky Taylor, M.D.
Presented by: Quynh Bui, M.D., Senior
Fellow/Acting Instructor, Department of Family
Medicine, University of Washington, Box 356390,
Seattle, WA 98195, Phone: (206) 856-1504; Email: buiq@u.washington.edu
Research Objective: Because some immigrant groups have been shown to have high risk for musculoskeletal pain, this study investigates the association between acculturation and the report of chronic back and neck problems among Asians and
Latinos in the US. It aims to: (a) compare the prevalence of chronic back and neck problems among US-born and immigrant Asians and Latinos; and (b) examine whether level of acculturation is associated with the report of chronic back and neck problems.
Study Design: This study examines the association between the report of spine problems and four measures of acculturation (nativity, percentage of lifetime in the US, generation in the
US, and English proficiency) by secondary analysis of data from 2095 Asian and 2554 Latino adults from the 2002-2003 National Latino and Asian
American Survey (NLAAS), a nationally representative cross-sectional survey. Data were analyzed using logistic regression with Taylor series linearization for standard error estimation. Results are presented as adjusted proportions.
Population Studied: Non-institutionalized Asians and Latinos older than 18 years residing in the US.
Principal Findings: After adjusting for demographic, socio-economic, and physical and mental health indicators, we found that a significantly higher proportion of US-born Asians,
20.1%, (95% Confidence Interval [CI]: 16.6%,
24.2%) reported chronic back and neck problems than immigrant Asians, 12.0% (95% CI: 10.2%,
14.1%). Likewise, 18.7% (95% CI: 15.5%, 22.4%) of US-born Latinos reported chronic spine problems compared to 11.9% (95% CI: 10.2%, 13.9%) of immigrant Latinos. For both Asians and Latinos, higher generation in the US was significantly associated with a higher 12-month prevalence of chronic spine problems. However, among Asian and Latino immigrants, the association between back and neck problems and the percentage of lifetime in the US was not significant. For Latinos but not Asians, English proficiency was associated with higher report of spine problems independent of other measures of acculturation.
Conclusion: Our findings suggest that the report of chronic back and neck problems is higher among
Asians and Latinos who are more acculturated to
US society. These findings are independent of health status, obesity, and the presence of depressive disorders. Whether our findings reflect greater spine and neck pain with increasing acculturation, a higher likelihood of reporting this pain, or both cannot be determined from these data.
Implications for Policy, Delivery or Practice:
Clinicians who care for ethnic minority populations should be aware that acculturation may influence the experience or reporting of chronic back or neck pain. Future research should identify causal relationships between specific acculturative processes and chronic pain that are amenable to prevention.
Theme: Disparities
● Burden of Influenza-Related Hospitalizations among Children with Sickle Cell Anemia
David Bundy, M.D., M.P.H.; John Strouse, M.D.;
Marlene Miller, M.D., M.S.; James Casella, M.D.
Presented by: David Bundy, M.D., M.P.H.,
Assistant Professor of Pediatrics, Pediatrics, Johns
Hopkins University School of Medicine, 600 N
Wolfe Street; CMSC 2-121, Baltimore, MD 21287,
Phone: (410) 502-8623; Email: dbundy@jhmi.edu
Research Objective: Children with sickle cell anemia (SCA) are considered a high risk population for influenza infection and annual vaccination is recommended. The burden of influenza infection in this population, however, is largely unknown. Our objective was to compare the rates and
epidemiology of influenza-related hospitalizations
(IRHs) among children with and without SCA.
Study Design: We conducted a cross sectional analysis of the AHRQ-sponsored Healthcare Cost and Utilization Project State Inpatient Databases from New York, Florida, and Maryland for the 2003-
4 and 2004-5 influenza seasons. US Census data were used to estimate appropriate denominators for rate calculations.
Population Studied: We analyzed hospitalizations of children less than 18 years of age with an
International Classification of Diseases, 9th
Revision, Clinical Modification discharge diagnosis code for influenza, stratified by the presence or absence of a co-occurring discharge diagnosis code for SCA.
Principal Findings: There were 5,256 pediatric influenza-related hospitalizations (2003-4: 3,277;
2004-5: 1,879), 1 in 40 of which included a cooccurring diagnosis of SCA. Among children without SCA, there were 2.6 IRHs per 10,000 children per year (2003-4: 3.3; 2004-5: 1.9), compared with 201 IRHs per 10,000 children with
SCA per year (2003-4: 254; 2004-5: 148). Across the 2 influenza seasons, IRHs among children with
SCA occurred at 79 times the rate of IRHs among children without SCA. IRHs associated with SCA involved older children (mean age 5.5 years vs. 2.6 years, P<0.001) and were more likely to involve children from large metropolitan areas (87% vs.
70%, P<0.001), compared with IRHs not associated with SCA. IRHs with SCA were more likely to originate in the emergency department (78% vs.
67%, P=0.03) and be paid for by Medicaid (64% vs.
54%, P=0.04) than those without SCA. Mean length of stay was similar for IRHs with and without co-occurring SCA (3.3 vs. 3.8 days respectively,
P=0.41).
Conclusion: Children with sickle cell anemia are hospitalized for influenza at substantially higher rates than those without SCA.
Implications for Policy, Delivery or Practice:
Characterizing children with SCA as high risk for influenza is justified and supports the need for expanded efforts to measure and improve influenza vaccination rates in this population. Medicaidbased immunization outreach efforts, had they been in effect, could have reached two-thirds of children with SCA hospitalized with influenza in this cohort.
Theme: Child Health
● Pediatric Vaccination Errors: Application of the
David Bundy, M.D., M.P.H.; Andrew Shore, Ph.D.;
Laura Morlock, Ph.D.; Marlene Miller, M.D., M.S.
Presented by: David Bundy, M.D., M.P.H.,
Assistant Professor of Pediatrics, Pediatrics, Johns
Hopkins University School of Medicine, 600 N
Wolfe Street; CMSC 2-121, Baltimore, MD 21287,
Phone: (410) 502-8623; Email: dbundy@jhmi.edu
Research Objective: Vaccinations comprise a widely used and highly efficacious preventive service. Little is known, however, about the epidemiology and causes of pediatric vaccination errors. Our objective was to evaluate the epidemiology of pediatric vaccination errors using the "5 Rights" framework (right vaccination, right time, right dose, right route, right patient) and to determine whether vaccination error types are predictable based on vaccine and patient attributes.
Study Design: We conducted a cross sectional analysis of medication errors from a nationwide, voluntary medication error reporting system
(MEDMARX) using methods which account for possible clustering of error properties by facility.
Population Studied: We analyzed errors that involved children ages 0-17 years, involved vaccines, and occurred in outpatient settings from
2003 to 2006.
Principal Findings: 607 error reports listed 691 vaccines. Influenza was the most commonly reported vaccine (10% of all reports). Wrong vaccine errors were more common among lookalike / sound-alike groups (Td/Tdap/DTaP/DT
(36%), pneumococcal conjugate / pneumococcal polysaccharide (29%), haemophilus influenzae / influenza (23%)) than among vaccines with no lookalike / sound-alike group (13%, P=0.008).
Scheduled vaccines (e.g., pneumococcal conjugate) were more often involved in wrong time errors than seasonal (e.g., influenza) and intermittent (e.g., tetanus booster) vaccines (44%,
18%, and 29% respectively, P=0.006). Wrong dose errors were more common for vaccines whose dose is weight-based (57% of palivizumab errors) and age-based (38% of influenza errors) than for vaccines whose dose is uniform (8.2%, P<0.0001)).
Wrong route errors were rare and more common in subcutaneous (5.7%) compared with intramuscular
(2.0%) vaccines, although this difference was not statistically significant. Wrong patient errors were also rare (4.1% of reports); at least 44% of wrong patient errors involved sibling confusion.
Conclusion: In this largest-ever analysis of pediatric vaccination errors, error types were associated with vaccine and patient attributes.
Predictable human factors challenges, such as confusion between vaccines with similar names, were associated with which vaccines and patients were involved in which types of errors.
Implications for Policy, Delivery or Practice:
Efforts to reduce vaccination errors in children should focus on these human factors; such approaches may have medication safety benefits extending beyond vaccinations.
Funding Source(s): AHRQ
Theme: Child Health
● Defining & Measuring Integrated Patient Care:
Promoting the Next Frontier in Healthcare
Delivery
Jako Burgers, M.D., Ph.D.; Sara Singer, Ph.D,
M.B.A.; Meredith Rosenthal, Ph.D; Mark Friedberg,
M.D., M.P.P.; Lucian Leape, M.D.; Eric Schneider,
M.D., M.Sc.
Presented by: Jako Burgers, M.D., Ph.D.,
Harkness Fellow 2008/2009, Health Policy &
Management, Harvard School of Public Health, 677
Huntington Avenue, Kresge 429, Boston, MA
02115, Phone: (617) 432-6448; Email: jburgers@hsph.harvard.edu
Research Objective: The growing prevalence of co-morbidity and complex health problems increases the demand for specialized patient care, challenging the coordination and integration of healthcare services. Innovations, such as the
‘medical home’, may help, but may not be sufficient to assure integrated patient care. In this study we aim to assess existing patient-experience surveys and to develop a survey instrument that can be used to measure the degree of integration of healthcare services and determine the association between integration, better patient outcomes, and lower costs.
Study Design: First, we identified previous definitions and developed a theoretical framework delineating important aspects of integrated patient care based on literature review and interviews with leaders of organizations representing different forms of structural integration. Second, we identified potential measures of integration and related terms
(e.g. coordination, multidisciplinary and interprofessional care) from literature selected by experts and using a search of PubMed, generating a list of potential survey items related to integrated patient care, which we grouped into categories. Our team then generated new potential survey items to address aspects of the integrated patient care framework that were not well-captured in existing measures. From this comprehensive list, the authors developed, in consultation with several external experts, a shorter list of potential survey items.
Population Studied: Literature related to integrated, coordinated, patient-centered, interprofessional patient care and prior surveys regarding patient experience of integrated patient care.
Principal Findings: Our integrated patient care framework includes an organizational process of coordination which seeks to achieve care that is (1) seamless across professionals, medical facilities, and other social support systems, (2) continuous over time, (3) tailored to the patients’ needs and preferences, and (4) based on a comprehensive view of the patient. Different models of care delivery (e.g., multi-specialty group practice, medical home) may integrate patient care through a variety of structural mechanisms (e.g., electronic medical record, care teams). We identified seven prior patient-experience surveys that addressed integrated patient care, including more than 100 unique potential items. Most items in available measures address care coordination and patient centeredness by documenting satisfaction and perceived quality of care regarding individual facilities, clinicians, or office visits. However, few items assess directly the extent to which health and medical care services are delivered in an integrated way for patients across facilities, systems, and time and with comprehensive consideration of the patient. Additional items were formulated to address these gaps. Pilot testing and instrument validation is ongoing.
Conclusion: Initiatives to improve integration of care are needed to ensure quality of care for patients with multiple or complex chronic conditions.
We developed a survey instrument measuring most relevant aspects of integrated patient care based on a review of literature and expert opinion. Further field testing is needed to improve the measurement properties.
Implications for Policy, Delivery or Practice: In the absence of evidence that integrated patient care is associated with better clinical quality, it is difficult for policymakers and payers to recognize and provide incentives for organizations that provide such care. A measure of integrated patient care allows policymakers and payers to determine the extent to which various forms of organization deliver integrated patient care and the extent to which integration promotes quality.
Funding Source(s): CWF, Harvard School of
Public Health
Theme: Quality and Efficiency: Measurement
● Presence & Correlates of Racial Disparities in
Adherence to Colorectal Cancer Screening
Guidelines: Results of a National Survey of
Veterans
Diana Burgess, Ph.D. Michelle Van Ryn, Ph.D.,
M.P.H.; Joan M. Griffin, Ph.D.; Sally Vernon, Ph.D.;
Deborah Fisher, M.D., M.H.S.; Melissa Partin,
Ph.D.
Presented by: Diana Burgess, Ph.D., Assistant
Professor, Center for Chronic Disease Outcomes
Research, Minneapolis VA Medical Center, One
Veterans Drive 152/2E, Minneapolis, MN 55417,
Phone: (612) 467-1591; Email: diana.burgess@va.gov
Research Objective: We conducted a survey informed by Social Cognitive Theory to examine the presence and correlates of black/white racial disparities in adherence to guidelines for colorectal cancer screening (CRCS).
Study Design: Cross-sectional observational study using a mail survey with telephone follow-up.
Hierarchical logistic regression analysis were used to assess the contribution of background, cognitive, and environmental characteristics to racial
disparities in adherence to CRC screening guidelines
Population Studied: The study sample included
339 African American and 1749 white veterans who responded to a national survey of patients age 50-
75 who received care from the Veterans Health
Administration (VHA) and who had no history of bowel disease.
Principal Findings: In unadjusted analyses, blacks had lower rates of adherence to CRCS recommendations than whites (76% vs 82%, p <
.001). However, these differences became nonsignificant after controlling for demographic and health-related characteristics, several of which
(younger age, poorer health, and diagnoses of substance abuse) were associated with CRCS adherence and black race in bivariate analyses.
These racial differences in CRCS were driven by lower rates of colonoscopy among blacks compared with whites.
Conclusion: Although rates of CRCS for both black and white veterans were significantly higher than the national average, blacks were less likely to be adherent to CRCS using colonoscopy compared with whites.
Implications for Policy, Delivery or Practice:
High rates of CRCS underscore the important role of the VHA in providing healthcare to a diverse, low income population who may lack other sources of care. However, findings of lower rates of colonoscopy among black compared with white veterans in the VHA raise concerns given the recent joint guidelines by the American Cancer
Society, the US Multi-Society Task Force on
Colorectal Cancer, and the American College of
Radiology, which recommend CRCS tests like colonoscopy that can prevent as well as detect cancer.
Funding Source(s): VA
● Selecting Weights To Calculate A Composite
Measure of Quality: Two Approaches That
Allow Flexibility & Reward Results
James Burgess, Jr., Ph.D.; Michael, Shwartz,
Ph.D.; Dan Berlowitz, M.D.
Presented by: James Burgess, Jr., Ph.D., Senior
Investigator, Center for Organization, Leadership &
Management Research, VA Boston Healthcare
System (152M), 150 South Huntington Avenue,
Boston, MA 02130, Phone: (857) 364-5683; Email: james.burgess@va.gov
Research Objective: A composite quality score derived from individual quality indicators (QIs) is useful to policy makers, providers, and consumers.
The standard approach for creating a composite score is to weight individual QIs equally or to use opportunity-based weights that vary facility weights based on the proportion of patients qualifying for the QI. However, in addition to differences in the proportion of patients qualifying for each QI, differences in facility performance on specific indicators may reflect legitimate local factors related to patients, staff, facility and the external environment, as well as differences in strategic priorities, factors not taken into account under the standard approaches. There may be some value in recognizing these facility differences in the weights used to calculate the composite score. We illustrate two approaches for adjusting baseline weights (which may be determined by equal weighting or opportunity-based weighting) to reward facilities for particularly strong performance on individual indicators. We call these “flexibilitybased” approaches.
Study Design: We compare several standard approaches to two flexibility-based approaches that allow limited modification of the standard weights on QIs. The first approach constrains the sum of the weights to equal one. The second constrains the composite score to be between 0 and 1. Both approaches use linear programming to find optimal weights subject to policy-imposed constraints on the extent to which baseline weights can be modified.
Population Studied: We apply the standard and flexibility-based approaches to data on five riskadjusted QIs from a convenience sample of 32 VA nursing homes.
Principal Findings: The flexibility-based approaches do not have a significantly greater impact on the weights than occurs from switching between existing standard weighting approaches.
To illustrate, when opportunity-based weights are based on the overall prevalence of each QI across all facilities, the weights vary from .044 (for pressure ulcers) to .439 (for mortality). When facility-specific opportunity-based weights are used, the weight assigned to pressure ulcers varies across facilities from.013-.091; the weight assigned to mortality varies from .195-.747. In contrast, when flexibility is allowed to adjust weights around those derived from overall prevalence, the weight assigned to pressure ulcers varies from .011-.078; the weight assigned to mortality varies from .110-
.768. Though composite scores are improved numerically if flexibility is allowed, relative performance is not necessarily better. Facilities that benefit the most from the flexibility-based approaches are those that have achieved relatively strong performance in particular areas.
Conclusion: Flexibility-based approaches, because they allow facilities to reflect local preferences and conditions, should be attractive to facilities and increase their acceptance of composite scores. Widely-accepted composite scores will increase capability for learning from quality improvement and organizational transformation efforts, as well as facilitate pay-forperformance programs.
Implications for Policy, Delivery or Practice:
Policy makers can impose weights on QIs to form composite measures of quality, but ordinarily
choose uninformative equal weighting or patientdriven opportunity-based weights. The flexibilitybased approaches we discuss here have the potential to balance social policy incentives with local context to improve acceptance by all parties of composite measures of quality.
Funding Source(s): VA
Theme: Quality and Efficiency: Measurement
● How States are Financing Health Coverage
Initiatives
Courtney Burke, M.S.
Presented by: Courtney Burke, M.S., Director,
Health Policy Research Center, Health Policy
Research Center, Rockefeller Institute of
Government, 411 State Street, Albany, NY 12203,
Phone: (518) 443-5243; Email: burkec@rockinst.org
Research Objective: Learn how states are financing health coverage initiatives, the advantages and disadvantages of different financing mechanisms, and the sustainability of financing. Use the information to inform policymakers about the tradeoffs and implementation challenges of various financing options for health insurance coverage initiatives.
Study Design: This study began with a literature review regarding how states have financed health insurance coverage expansions. Interviews were conducted with a select number of state officials to learn about the challenges of implementing financing mechanisms, as well as the perceived advantages and disadvantages of the approaches.
A comparative table of financing approaches was developed. A specific analysis was done regarding
New York’s proposed financing for coverage expansions. This data served as the basis for a policy paper, which will be published in March 2009.
Population Studied: This paper focuses on the population of uninsured that are targeted by state health insurance coverage expansions. In some states, the target population is all of the uninsured; in other states it is a select group of individuals such as those that work for small businesses.
Principal Findings: State methods for financing health insurance coverage expansions vary, but fit in nine general categories: 1. Leveraging federal dollars; 2. Sin taxes (e.g., on non-diet soft drinks, cigarettes, alcohol); 3. Enrollee cost sharing; 4.
Surcharges on providers or third party payers; 5.
Redirecting financing from other initiatives; 6.
Shared responsibility with insurers; 7. Cost containment; 8. One-time, non-recurring sources
(e.g., tobacco settlement funds); and 9. Health plan conversion funds. The methods that states use to finance coverage initiatives are similar, regardless of whether the initiative is large or incremental or if the state is populous or not. Some methods of financing are more widespread than others, and there are advantages and disadvantages for every type of financing mechanism. Leveraging of federal funds is the most common form of financing and generally more sustainable. Cost efficiency and repurposing of existing sources is more sustainable, but harder to implement.
Conclusion: The research concludes that some financing methods are easier to enact (e.g., leveraging federal financing), while others are easier to implement (e.g., one-time, non-recurring revenues). Some financing mechanisms receive more political support (e.g., tobacco tax) than others (e.g., enrollee cost sharing). The research makes recommendations about which financing mechanisms might be better depending on the scope and scale of a state’s planned coverage initiative, and reaches conclusions about what types of financing might work better under which circumstances. It also specifically concludes what types of financing mechanism might work better in
New York State.
Implications for Policy, Delivery or Practice: The research has implications for policymaking, and in particular, for budgeting for policy initiatives, because it is the first large scale study to examine financing for health coverage. Because financing is a necessary part of almost any coverage expansion, the research has wide-ranging implications for policies and policymakers that seek to expand insurance coverage.
Funding Source(s): NYS Health Foundation
Theme: Coverage and Access
● Managing Risk in the Small Group Insurance
Market Through Merged Markets, Reinsurance,
Mandates & Pooling Arrangements: Does One
Strategy Work Better than Another & What Are the Tradeoffs?
Courtney Burke, M.S.
Presented by: Courtney Burke, M.S., Director,
Health Policy Research Center, Health Policy
Research Center, Rockefeller Institute of
Government, Health Policy Research Center, 411
State Street, Albany, NY 12203, Phone: (518) 443-
5243; Email: burkec@rockinst.org
Research Objective: Determine the tradeoffs of different risk management strategies (e.g., merging the small group and individual insurance markets, creating high-risk pools, and establishing individual mandates for insurance coverage, and reinsurance) that increase insurance coverage. Review the advantages and disadvantages of each strategy to enable policymakers to make more informed decisions about implementing initiatives designed to improve private health insurance coverage.
Study Design: The study reviews the literature and then examines initiatives in a small number of states in more depth. The state-level information is gleaned through interviews with state officials.
National expertise is used to help review the various strategies for risk management. A comparative table briefly reviews the advantages
and disadvantages of each strategy. The final paper discusses what policymakers could consider in weighing different interventions and briefly discusses specific options as they apply to New
York State.
Population Studied: The population studied for this research is people who have difficultly affording private insurance and those who are uninsured and not likely to qualify for public insurance. In some states, the target population may be a large percent of the population (e.g., all those who are uninsured and not eligible for public insurance) or it may be a particular population for which private insurance is less affordable (e.g., people in the individual insurance market).
Principal Findings: The research finds that each of the methods to manage risk in the insurance market for the purpose of increasing insurance coverage has advantages and disadvantages. For instance, although high-risk pools are advantageous because they allow otherwise uninsurable people to obtain coverage, a disadvantage is that those in the pool pay much higher rates for insurance coverage. Another risk management strategy — reinsurance — helps reduce adverse selection in the insurance market, but it also requires a large public subsidy in order to be effective. Some methods, especially pooling, are more widespread than others, such as reinsurance or merging the individual and small group markets.
Conclusion: The exact degree to which different risk management strategies reduce risk or increase coverage is not well evaluated but this research provides helpful information for policymakers about the advantages and disadvantages of various policy interventions. A risk management strategy that is preferable in one state may not be preferable in another. The issues that governments should consider in making their choices of a risk management strategy include: the primary goal of the risk intervention (e.g., to reduce the number of uninsured; to make insurance more affordable, to whom, and to how many people); available revenue; administrative mechanisms needed; and the implementation techniques of other states.
Implications for Policy, Delivery or Practice:
This research has potential implications for future state or national efforts to bolster the private insurance market because it allows state and federal policymakers to better evaluate the tradeoffs of different risk management strategies that can potentially increase private insurance coverage and it provides recommendations regarding implementation.
Funding Source(s): NYS Health Foundation
Theme: Coverage and Access
● Children with Special Health Care Needs:
Evaluating Care Coordination Services for
Children Diagnosed with an Autism Spectrum
Disorder
Christine Burns, Ed.M., M.B.A.; Mark Orlando,
Ph.D.; Stephen Sulkes, M.D.; Daniel Mruzek, Ph.D.;
Elizabeth Hebert, M.S.; Jacalyn Yingling, M.S.
Presented by: Christine Burns, Ed.M., M.B.A., Co-
Director Leadership Education in
Neurodevelopmental Disabilities, Pediatrics,
University of Rochester, 601 Elmwood Avenue,
Box 671, Rochester, NY 14642, Phone: (585) 275-
6681; Email: Christine_Burns@urmc.rochester.edu
Research Objective: The purposes of this investigation were to determine the usage of electronic medical records or electronic information systems for providing care coordination services, to evaluate professional perception of effort in coordinating services for children who have a diagnosis of an ASD, and to evaluate professionals’ satisfaction of care coordination services they provide to families of children with an autism diagnosis.
Study Design: The survey was designed after thorough literature review and qualitative data gathered through focus groups and then stringently tested and reviewed for its face and content validity by 10 subject matter experts in the field of ASD.
The survey is composed of the following sections:
Demographics, Background, Current Disclosure
Practices, Barriers to Effective Disclosure,
Recommendations, Electronic Medical Record Use,
Care Coordination with EMR/EIS, Care
Coordination Practices, and Transitions. Subjects were contacted via-email from their respective networks and directed to Survey Monkey website.
Data were submitted on-line through the website and stored for subsequent analyses by the researchers. Data were analyzed using Statistical
Analysis Software, or Statistical Package for Social
Sciences – PC, or EPI Info 6.04B Version 2000
(Centers for Disease Control, 1997). Participants’ responses to the survey were analyzed through quantitative statistical methods (i.e., descriptive statistics, including percent, mean response levels, range of responses), as well as qualitative methods
(i.e., review and analyses of written responses). A p-value of 0.05 was considered the level of statistical significance throughout the analysis.
Population Studied: Eight hundred participants from four groups: a) the Autism Treatment Network
(ATN); b) the Autism Special Interest Group of the
Association for University Centers on Disabilities
(AUCD); c) the Leadership Education in
Neurodevelopmental and related Disabilities
(LEND) network; d) the University Center for
Excellence in Developmental Disabilities (UCEDD) network received an on-line survey.
Principal Findings: Preliminary data analysis revealed that considerable variability in care coordination practices continues to exist and significant barriers in communication among providers in health care, financial and school settings and in accessing information about
community resources prevail. Satisfaction with care coordination for children with ASD was associated with the scope and extent of electronic information system use. EMR use facilitated decision-making, receiving immediate responses to problems, and assistance with treatment plan implementation and transition planning.
Conclusion: Use of Electronic Information Systems and Electronic Information Records must be increased to assure improved care coordination enhancing health and functional outcomes of children with ASD and to enhance family-centered care.
Implications for Policy, Delivery or Practice:
Integrated Electronic Information Systems using
Electronic Information Records are needed to promote successful care coordination and provide additional supports that benefit children with a diagnosis of an ASD, their families, and professionals.
Funding Source(s): HRSA
Theme: Health Information Technology
● Spending at the Thin Margin of Health: Out-of-
Pocket Health Care Expenditures & Medicaid
Beneficiaries with Disabilities
Marguerite Burns, Ph.D.; Nilay Shah, Ph.D.;
Maureen Smith, M.D., Ph.D.
Presented by: Marguerite Burns, Ph.D.,
Postdoctoral Trainee, Population Health Sciences,
University of Wisconsin, 610 North Walnut Street,
Room 634, Madison, WI 53705, Phone: (608) 265-
3296; Email: meburns@wisc.edu
Research Objective: The budgetary impact of adult Medicaid beneficiaries with disabilities places them at the center of ongoing health care redesign efforts across the country. However, the evaluation efforts that accompany these reforms frequently lack an examination of cost-related barriers to health care despite the sensitivity of adults with disabilities (AWDs) to the accessibility of their health care system. In this study, we examine one potential measure of cost-related barriers to care, out of pocket (OOP) health care spending.
Study Design: We estimate total OOP health care spending among adult Medicaid beneficiaries with disabilities, the services and products which command the largest share of OOP expenditures, and characteristics of beneficiaries who incur them.
Using the Medical Expenditure Panel Survey (1996-
2004), we construct a dataset in which each observation is a person-year. The analyses describe the distribution of OOP health care spending in the aggregate, per beneficiary, and by selected beneficiary characteristics in the study population overall and among beneficiaries who incur the highest level of Medicaid expenditures.
Estimates are weighted according to the survey design of the MEPS and dollars are converted to constant 2004 dollars.
Population Studied: Our nationally representative sample includes non-institutionalized Supplemental
Security Income (SSI) beneficiaries aged 18-64 who are enrolled in Medicaid, and not in Medicare
(N=2,438 unweighted person-years). Maximum income eligibility for SSI, the primary pathway to
Medicaid for AWDs, is 75% of the federal poverty level, $6983 in 2004.
Principal Findings: Between 1996-2004, the proportion of beneficiaries with any OOP health care expenditures in each year remained stable at
83%. Aggregate annual OOP health care spending varied from $1.2 billion to a low of $590 million.
Median annual OOP expenditures per beneficiary ranged from $115 to $288 across the years. More than half of OOP spending was allocated to prescription medications each year. Central nervous system, cardiovascular, and psychotherapeutic medications consistently ranked among the top three drug classes in terms of OOP spending. Higher OOP spending on prescription medications was observed among women, the more educated, those in poorer physical or mental health, rural beneficiaries, and HMO enrollees.
Among high cost beneficiaries, those at the 90th percentile of Medicaid spending on adults with disabilities, OOP costs exceeded $1,100 annually— with a majority of OOP spending on prescription drugs.
Conclusion: Adults with disabilities have been described as having a thin margin of health. Their disabling physical or mental health impairment places them at increased risk for secondary conditions and co-morbidities. The health of
Medicaid beneficiaries with disabilities is possibly more precarious as they possess few financial resources to apply to health maintenance. Our findings suggest that for most AWDs, the Medicaid program effectively limits their personal spending on health care-- as intended. However, for the most expensive beneficiaries, OOP spending is substantial in light of the programmatic income eligibility thresholds.
Implications for Policy, Delivery or Practice:
Policy-makers should monitor OOP spending and its health and welfare consequences as they redesign care systems to minimize Medicaid expenditures.
Funding Source(s): AHRQ
Theme: Medicaid, SCHIP and State Health Reform
● Immunization Trends in Nursing Homes
Before & After Public Reporting: Experience from New York
Shubing Cai, M.S.; Helena Temkin-Greener, Ph.D.
Presented by: Shubing Cai, M.S., Ph.D Candidate,
Department of Community & Preventive Medicine,
University of Rochester, 601 Elmwood Avenue, Box
644, Rochester, NY 14642, Phone: (585) 275-
0369; Email: shubing_cai@urmc.rochester.edu
Research Objective: In October 2005, Centers for
Medicare and Medicaid Services (CMS) started to require nursing homes (NH) to provide residents with flu vaccines. A year later, CMS began to publish NH immunization rates on Nursing Home
Compare (NHC) website. This study: (1) describes
NH influenza immunization trend starting in late
2005, and (2) examines the association between influenza immunizations and related health outcomes in New York State (NYS) nursing homes.
Study Design: NHC data were linked with
Minimum Data Set (MDS) for NYS. We selected the first quarter of 2007 and 2008 NHC data, reflecting
NH immunization rates before (10/2005~03/2006) and after (10/2006~03/2007) public reporting. MDS records covering the corresponding flu seasons
(10/2005~05/2006 and 10/2006~05/2007) were identified. Facility was the unit of analysis.
Population Studied: Our sample included 212,370 residents during 10/2005~05/2006, and 251,642 residents during 10/2006~05/2007 flue seasons, in
648 NHs. Immunization rates before and after public reporting were compared. A first-difference model was fit to examine the association between immunization rates and health outcomes (influenzarelated infections and hospitalizations). These two outcomes were measured as log-transformed ratio of number of affected residents to total patient days during flu seasons. Main variables of interest were
NH influenza immunization rates for long-term (LT) and for short-term (ST) residents. We also controlled for facility-level case-mix. All NYS facilities were included in the analysis. Thus, the interpretation was based on actual effect size and no statistical inferences were made.
Principal Findings: Flu immunization rates in NY were higher than the national average:
79.96%(SD=15.16) for ST and 91.20%(SD=7.66) for LT residents during 2005~2006 flu season; and
81.57%(SD =14.44) for ST and 90.77%(SD=9.15) for LT residents during 2006~2007 season. For facilities with a relatively high baseline immunization rates (above the 25% quantile), a 10% increase in immunization rates, for LT residents, was associated with 3.4% decrease in baseline infection rates. A 10% increase in immunization rates, for ST residents, was associated with a 3% decrease in baseline hospitalization events. For facilities with a relatively low baseline immunization rates (below
25% quantile), a 10% increase in immunization rate for LT care residents was correlated with about 7% decline in baseline hospitalization rates.
Conclusion: Our analysis was based on a state with relatively high baseline immunization rates, leaving relatively little room for improvement. State average influenza rates for ST residents improved slightly over the two seasons. However, since influenza immunization rates among communitydwelling elderly have also improved slightly in this time period (64.7% in 2006 to 70.5% in 2007), we could not ascribe the observed NH trend to the effect of public reporting alone. Furthermore, some benefits of immunizations obtained in NHs may be difficult to measure, particularly for ST residents who are likely to return to the community quickly.
Nevertheless, improvements in immunization rates were associated with improvements in some healthrelated outcomes.
Implications for Policy, Delivery or Practice:
Even at relatively high rates of immunizations, improvements in flu vaccination rates and their associated outcomes can be realized, suggesting potential benefits to public reporting of this measure.
Funding Source(s): NIA
● Hospitalization Risk in Nursing Homes: Does
Payer Source Matter? Evidence from New York
State
Shubing Cai, M.S.; Dana Mukamel, Ph.D.; Peter
Veazie, Ph.D.; Paul Katz, M.D.; Helena Temkin-
Greener, Ph.D.
Presented by: Shubing Cai, M.S., Ph.D.
Candidate, Department of Community & Preventive
Medicine, University of Rochester, 601 Elmwood
Avenue, Box 644, Rochester, NY 14642, Phone:
(585) 275-0369; Email: shubing_cai@urmc.rochester.edu
Research Objective: High rates of hospitalizations among nursing home (NH) residents may indicate poor quality of care and account for significant healthcare cost to society. This study examines whether NH residents’ payer type (Medicaid vs. private-pay) and NH bed-hold status impact hospitalization.
Study Design: The 2003 Minimum Data Set
(MDS), Medicaid Analytic Extract (MAX) personal summary and long-term-care files, Medicare
Provider and Analysis Review File (MedPAR), and
Medicare denominator file are linked for all NH residents. Payer status (Medicaid vs. private-pay) is jointly determined by the MDS, MedPAR and MAX files. NH-originating hospitalizations are identified using MDS and MedPAR. Facility ownership and occupancy rates are determined from Medicaid cost reports.
Population Studied: The analytical sample includes 69,307 unique residents (18 % private-pay
& 82% Medicaid) in 564 NHs in NYS. The analysis is based on individual-level data. All observations are defined to start from a non-Medicare reimbursement period. Hospitalization is the dependent, dichotomous variable. Payer status is the main independent variable of interest. Other control variables include residents’ health status/conditions, socio-demographics, lifesustaining treatment preferences, identified at the beginning of each observational period, and facility characteristics. Facilities are stratified into 4 groups, based on ownership and occupancy rate (95% occupancy rate is required for bed-hold payment in
NYS), and each of them are analyzed separately.
Base logit and conditional fixed-effects logit models are fit for these subgroups. The impact of the
Medicaid payer status on hospitalization risk across these subgroups is examined.
Principal Findings: About 28% of residents experience at least one hospitalization during non-
Medicare periods. Overall Medicaid residents are more likely to be hospitalized than private-pay residents in all four subgroups. The differences in hospitalization risk between Medicaid and privatepay residents decline after accounting for facility fixed-effects, suggesting that the overall difference can be explained by both within- and across-facility variations. Among facilities that are eligible for bedhold payment, Medicaid residents in for-profit facilities with an average payer-mix level
(%Medicaid reimbursed days) are 34% more likely to be hospitalized than their private-pay counterparts (P<0.001). This risk increases with facility payer-mix level. In non-profit facilities,
Medicaid residents are 17% more likely to be hospitalized than their private-pay counterparts
(P<0.001). The effect of payer status is statistically significantly different between for-profit and nonprofit facilities. Among facilities that are not eligible for bed-hold payment, there is no statistically discernable difference in hospitalization risk between Medicaid and private-pay residents in for-profit facilities (OR=1.02, P=0.764). However,
Medicaid residents in nonprofit facilities have a higher hospitalization risk than private-payers
(OR=1.25, P<0.01).
Conclusion: Financial incentives may motivate
NHs to consider payer source in deciding whether to hospitalize residents; the incentive seems particularly strong among for-profit NHs that are eligible for bed-hold payments. Results also suggest that across-facility variation in hospitalization risk for all residents is due to the overall proportion of Medicaid residents in the NH.
Future studies to identify facility characteristics contributing to hospitalization risk are warranted.
Implications for Policy, Delivery or Practice:
Policymakers should focus on reducing hospitalization of Medicaid residents in an effort to improving both quality and costs.
Funding Source(s): NIA
Theme: Quality and Efficiency: Policies and
Incentives
● Racial Disparity in Pressure Ulcer Prevalence among NYS Nursing Home Residents
Shubing Cai, M.S.; Dana Mukamel, Ph.D.; Helena
Temkin-Greener, Ph.D., M.P.H.
Presented by: Shubing Cai, M.S., Ph.D.
Candidate, Department of Community & Preventive
Medicine, University of Rochester, 601 Elmwood
Avenue, Box 644, Rochester, NY 14642, Phone:
(585) 275-0369; Email: shubing_cai@urmc.rochester.edu
Research Objective: The occurrence of pressure ulcers (PUs) in nursing homes (NHs) is considered to be a marker for poor quality of care. Studies have shown that Blacks have higher rates of NHacquired PUs than Whites. This study examines whether racial differences in PU rates among nursing home residents are due to within or acrossfacility disparities.
Study Design: 2006-2007 Minimum Data Set records for NYS nursing facilities were linked with the On-Line Survey Certification and Reporting
System (OSCAR). Long-term care (LTC) residents with high risk for PUs were identified using definitions provided by the Center for Medicare and
Medicaid Services (CMS). For residents with multiple quarterly/annual assessments one assessment was randomly selected.
Population Studied: The analytical sample includes 61,153 unique LTC high-risk residents
(18% black & 82% white) in 648 NHs in NYS. The analysis is based on individual level data. The dependent variable is dichotomous indicating presence/absence of PU on any given quarterly assessment. Race (black/white) is the main variable of interest. Other control variables include individual characteristics (e.g. age, gender, and health conditions), and facility characteristics (e.g. profit status, %blacks in the facility, and staffing).
Data for all LTC high-risk residents were randomly divided into two parts. A training sample was used to fit the risk-adjustment model, which was subsequently tested on the validation sample using
Hosmer-Lemeshow test and C statistics. We then fit three sets of models using selected risk adjustors. First, a simple logistic regression was estimated to examine the overall correlation between race and PU occurrence. Then a conditional fixed-effects logistic model was employed to test for the within-facility racial disparity in PU occurrence. Finally, a random effect model with bootstrapping SE was compared with the fixed-effects model.
Principal Findings: The overall unadjusted prevalence of PUs is 14.49%, 18.03% for Blacks and 13.73% for Whites. Overall, Blacks are 21% more likely to have PUs than Whites, controlling for all risk factors (OR=1.21, P<0.001). However, we find no such effect within facilities, after controlling for facility fixed effect (OR=0.98, P=0.66). The two estimates are significantly different across these two models (P<0.001). The random effect model, compared with the fixed effects, shows similar results and demonstrates that higher PU presence among Blacks is associated with greater concentration of Black residents in a facility
(P=0.01).
Conclusion: Consistent with previous studies, we find that Blacks are more likely to have PUs than
Whites. However, this difference is likely caused by across-facility differences rather than within-facility disparities. Greater PU occurrence among Blacks does not appear to be the result of differential
within-facility treatment of Black versus White residents. Rather, Blacks are more likely to reside in facilities with poorer quality of care.
Implications for Policy, Delivery or Practice: To improve the PU quality of care for Blacks in NHs, efforts should be focused on improving the overall quality of care for black-centered facilities.
Funding Source(s): NIA
Theme: Disparities
● Assessing & Promoting Breast Cancer
Literacy & Preventative Behaviors Among
Immigrant Latinas in South Los Angeles
Jose Luis Calderon, M.D.; Richard Baker, M.D.;
Nisa Sangasubana, Ph.D.; Mohsen Bazargan,
Ph.D.; Patrick Hardigan, Ph.D.; Sylvia Rabionet,
Ed.D.
Presented by: Jose Luis Calderon, M.D., Associate
Professor, Division of Research, Charles Drew
University of Medicine & Science, 1731 East 120th
Street, Los Angeles, CA 90059, Phone: (760) 521-
5511; Email: drcalderon@sbcglobal.net
Research Objective: Untimely participation in breast cancer screening behaviors (self-breast exam-SBE, clinical breast exam and screening mammography) may mediate disparities in disease free survival and mortality experienced by immigrant Latinas at any age and stage of diagnosis. Health literacy is independently associated with health status and health outcomes.
Since immigrant Latinas face language, literacy and healthcare access barriers, enhancing breast cancer health literacy using accessible, non-written venues may help to mitigate breast cancer health disparities. We tested the efficacy of two interventions at enhancing breast cancer health literacy and promoting screening behaviors as measured by the Breast Cancer Health Literacy
Survey; 1) focused discussion groups (FDG); 2)
FDG plus viewing an animated novella about breast cancer screening and SBE, and SBE training using latex models (FDG-plus). FDG differ from focus groups in collecting qualitative data while promoting health literacy through moderator participation and participant sharing and interaction.
Study Design: Non-randomized, pretest-posttest research design. Data was collected qualitatively by audio-taping FDG and generating abridged transcripts and quantitatively by pre/post intervention surveys. We used Atlas.ti for computerized qualitative data analysis and descriptive, non-parametric and logistic regression as needed for survey data analysis. Outcome variables were breast cancer knowledge and screening behaviors.
Population Studied: Two hundred immigrant
Latinas aged 30 and older were recruited into each intervention group (n=400). A community-based research approach was employed by implementing the study in community stakeholder homes, and local schools and churches and coordination by a
Promotora/health navigator.
Principal Findings: There was no significant variance between groups in the mean value of demographic, knowledge and screening behavior variables at baseline. Qualitative results revealed that Latinas harbored a marked fear of diagnosis, and if diagnosed a fear of not fulfilling family role, perceived hopelessness and fatalism because of financial barriers to healthcare. There was significant within group increase in breast cancer knowledge scores and screening behaviors three months post intervention for both groups (p<.05).
There was a significantly greater increase in self breast exams in the FDG-plus group (p<.05).
Conclusion: Community-based FDG represent a viable and potentially cost effective alternative to written information for enhancing health literacy and promoting breast cancer screening among populations with limited literacy. The animated novella (trailer may be viewed at www.healthlampanimation.com) and SBE training promoted behavioral change where embarrassment is reported as a barrier to SBE. In this study seven women <age 40 were referred for mammography.
Three women in the study were diagnosed with breast cancer. By promoting health literacy FDG may be a cue to action for timely breast cancer screening that may result in greater disease free survival and a diminution of breast cancer health disparities.
Implications for Policy, Delivery or Practice:
Immigrant Latinas are in need of usable health information. Policy mandating the use of nonwritten venues and public health campaigns using multi-media to promote screening for breast cancer among populations with limited literacy skills are needed. Trained Promotora/health navigators are culturally appropriate for recruitment and retention of hard to reach populations. Latinas may benefit as a group and outcomes many improve if screening mammography is encouraged at age <40 years.
Funding Source(s): CMS
Theme: Disparities
● Surviellance Mammography for the Dually
Eligible
Patricia Carcaise-Edinboro, Ph.D.; Cathy Bradley,
Ph.D.; Bassam Dahman, M.S.
Presented by: Patricia Carcaise-Edinboro, Ph.D.,
Research Associate, Health Administration, Virginia
Commonwealth University, 1008 East Clay Street,
Richmond, VA 23298-0203, Phone: (804) 828-
4712; Email: pcedinboro@comcast.net
Research Objective: The objective of this study was to assess the receipt of surveillance mammography for Medicaid/Medicare breast cancer survivors and to identify variance in
surveillance by race, age, dual eligibility, or cancer treatment.
Study Design: We used logistic regression and proportional hazards analysis to assess the receipt of mammography breast surveillance for up to 3 years post treatment while controlling for patient’s age, race, comorbidity, treatment, stage at diagnosis, and dual eligibility.
Population Studied: The study population included
Michigan Medicare and Medicaid dually eligible patients 66 years of age or older who were diagnosed with early stage breast cancer between
1997 and 2000.
Principal Findings: In the year following treatment, patients who received BCS and radiation therapy
(OR= 1.71; 95%CI=1.48-1.98) were more likely to receive surveillance mammography than those receiving BCS alone, regardless of race or dual status. Proportional hazards analysis indicated patients who received BCS and radiation treatment
(hazard ratio (HR)=1.72; 95%CI=1.56-1.89) had a greater probability of receiving surveillance mammography sooner than those treated with BCS alone. However, the probability of receiving mammography sooner after treatment was reduced for older patients (80+ years) (HR=0.55;
95%CI=0.45-0.66) than for younger and for and those with greater comorbidity HR=0.59;
95%CI=0.43-0.81) than for healthier patients.
Conclusion: Regardless of race and dual eligibility, there is a greater likelihood of breast cancer patients who receive BCS and radiation treatment to not only receive surveillance mammography but to receive it sooner than patients who receive BCS without radiation. However because the dual eligible, black, and the elderly are less likely to receive BCS with radiation, disparities across the treatment and surveillance continuum need to be further investigated.
Implications for Policy, Delivery or Practice:
Elderly, dual eligible breast cancer patients who were diagnosed in an early stage would seem to have minimal access to care barriers. However, in this same population we see disparity in breast cancer surveillance. Elderly breast cancer patients need continued and appropriate surveillance, particularly those who have received minimal surgical treatment without the benefit of adjuvant radiation.
Funding Source(s): NCI
● Is there an Ideal GINI Index?
Marcelo Cardarelli, M.D., M.P.H.
Presented by: Marcelo Cardarelli, M.D., M.P.H.,
Assistant Professor, Surgery & Pediatrics,
University of Maryland, 22 South Greene Street,
Baltimore, MD 21201, Phone: (410) 328-5842;
Email: mcard001@umaryland.edu
Research Objective: The Gini index, a relative measure of income distribution is extensively used in the public health literature as a predictor of health outcomes. Developed societies have a tendency toward lower Gini indexes and better health outcomes. We attempt to investigate how low should the Gini index of a developed society be, in order to achieve health indicators better than the average for the group.
Study Design: Relationships between health indicators, Gini index and independent variables were investigated through a multiple regression model. Mean values for the health indicators and
Gini indexes of all 73 societies were calculated.
Population Studied: We collected data on income distribution and health status indicators (life expectancy at birth, infant and neonatal mortality rates), as well as demographic variables, from 23
European countries and individually for each of the
50 U.S. States (total n = 73). We controlled for 14 commonly used health determinants as independent variables(size and rates for urban, foreign, retired and unemployed populations, obesity rate, alcohol and tobacco consumption per capita, public and total healthcare expenses per capita Gross Domestic Product per capita, physicians, nurses and hospital bed ratio).
Principal Findings: The results of the multivariable analysis demonstrated a significant negative effect p=0.034 (Estimate -0.08, SE+/- 0.0037) of income distribution on life expectancy. Societies with a lower Gini coefficient had statistically significantly higher life expectancies. Income distribution also had a statistically significant direct effect on neonatal and infant mortality rates (p <0.0001 in both cases). The mean Gini index for the 73 studied societies was 39.6 (range 23 - 48). Societies with a
Gini index at or below this level had a Life
Expectancy above the mean value for the cohort
(77.5 years). Similarly, the neonatal and infant mortality rates for societies with a Gini =39.6 or lower, was better than the mean for the cohort (4.2 deaths/100.000 live births and 6.1/1000 live births respectively). Life expectancy was also influenced by the prevalence of obesity within each society, with a statistically significant negative effect p=0.01
(Estimate -0.055, SE +/- 0.02). Neonatal and infant mortality did not seem to be statistically associated to other variables.
Conclusion: This study confirms a relationship between these health outcome measures and income distribution. A Gini index of 39.6 seems to represent a threshold below which health outcomes surpass the mean health indicator values for developed societies.
Implications for Policy, Delivery or Practice:
While lowering the level of disparities within a society is likely to improve its health outcomes, having a specific target in mind would greatly facilitate planning for such effort.
Theme: Disparities
● Consumer Behavior & Market Share in the
First Two Years of Medicare Part D
Colleen Carey, M.A., Ph.D.
Presented by: Colleen Carey, M.A., Ph.D.,
Doctoral Candidate in Economics, Department of
Economics, Johns Hopkins University, 440
Mergenthaler, 3400 North Charles Street,
Baltimore, MD 21211, Email: colleen.carey@aya.yale.edu
Research Objective: In 2006, Medicare beneficiaries began enrolling in subsidized prescription drug insurance through the new Part D program. Little is known about what was important to seniors deciding among the dozens of available
Part D plans, or about how the decision-making process differed between the first and subsequent years.
Study Design: I use enrollment decisions in 2006 and 2007 in a discrete choice demand estimation that determines seniors' willingness-to-pay for plan characteristics such as out-of-pocket costs
(measured on 12 dimensions for 151 sampled drugs), lower deductible, coverage in the doughnut hole, and six enrollee evaluations of plan quality.
Population Studied: The study analyses the market share of 1116 stand-alone prescription drug plans in 2006 and 1462 plans in 2007, representing the decisions of 15.0 and 16.1 million beneficiaries, respectively. Plans with very low enrollment were excluded due to data availability.
Principal Findings: I report results from four models: a non-nested multinomial logit, a nested logit where plans are grouped by benefit type
(Defined Standard, Actuarially Equivalent, Basic
Alternative, or Enhanced Alternative), a nested logit where plans are grouped by parent organization, and a multinomial logit with plan fixed effects to correct for potential endogeneity. Testing for the
Independence of Irrelevant Alternatives suggests the non-nested model is appropriate. Under the non-nested specification, I find beneficiaries are willing to pay for a lower deductible, higher customer service ratings, and lower out-of-pocket costs. I then use the multinomial logit coefficients determined for 2006 and 2007 to decompose, using a Oaxaca decomposition, the change in market share between the two years into a portion due to changes in plan characteristics and a portion due to changes in consumer taste for these characteristics. I find that changes in market share are overwhelmingly due to changes in consumer taste between the two years, particularly reduced sensitivity to premium and mean out-of-pocket costs.
Conclusion: While a compositional change in Part
D beneficiaries or consumer learning about the Part
D environment could also produce changes in consumer tastes, other results suggest that the decision-making process changed between the two years under analysis. The model’s R-squared fell from 0.7 in 2006 to 0.3 in 2007. The intercept term, which in the multinomial logit denotes the mean utility across insurance plans that is not attributable to any of the measured characteristics, rose from
0.1 (SE 1.1) to 14.0 (SE 2.4) over the same time period. Then the same plan characteristics that effectively explained plan choice in 2006 did not do so in 2007. My weak ability to explain consumer choice in 2007 suggests that beneficiaries, facing costs associated with collecting information about numerous Part D plans, remained in their 2006 Part
D plan instead of reoptimizing in 2007. If so, Part D plans with large market share in 2006 hold market power.
Implications for Policy, Delivery or Practice: My results suggest that in 2006 Medicare beneficiaries reacted as expected to features of free-standing
PDPs. However, in 2007, implicit switching costs may have inhibited the competitiveness of the Part
D market.
Funding Source(s): CMS
Theme: Medicare
● Internal Psychometric Properties of the
Children with Special Health Care Needs
Screener: Findings from the National Survey of
Children with Special Health Care Needs
Adam Carle, M.A., Ph.D.; Stephen Blumberg,
Ph.D.; Charlie Poblenz
Presented by: Adam Carle, M.A., Ph.D., Assisstant
Professor, Psychology, University of North Florida,
1 UNF Drive, Jacksonville, FL 32224, Phone: (904)
620-3573; Email: adam.carle@unf.edu
Research Objective: Too little research has established the psychometric (measurement) properties of the Children with Special Health Care
Needs (CSHCN) Screener. This leaves unclear whether Screener-based estimates reliably identify
CSHCN. To address this, we sought to establish the CSHCN Screener’s psychometric properties.
Study Design: The CSHCN Screener asks parents a series of questions to determine whether children experience one of five health related consequences and whether the consequence(s) result from a medical, behavioral, or other health condition lasting at least 12 months. Any child who meets criteria for at least one of the five consequence receives classification as a CSHCN. We used classical and modern test theory, including confirmatory factor analysis for ordered-categorical measures (CFA-OCM) and item response theory
(IRT), to establish the CSHCN Screener’s psychometric properties in a large, nationally representative survey of US children.
Population Studied: Data came from the 2005-
2006 National Survey of Children with Special
Health Care Needs (NS-CSHCN). The National
Survey of CSHCN provides state- and national-level data on the prevalence of special health care needs and their impact on children and their families.
Children (n = 359,154) ranged in age from 0 to 18
years. Design weights make the data representative of children nationally and within states.
Principal Findings: Cronbach’s alpha equaled
0.76. Confirmatory factor analysis for orderedcategorical measures (CFA-OCM) indicated that a single underlying health-condition-complexity trait underlies Screener responses. Item response theory (IRT) showed that responses provide particularly precise measurement among children experiencing elevated health-condition-complexity levels.
Conclusion: Findings demonstrate that responses to the CSHCN Screener as used in the NS-CSHCN have good internal psychometric properties and include minimal random measurement error. CFA-
OCM results support the suggestion that a healthcondition-complexity continuum describes
CSHCN’s health conditions. IRT analyses show that
Screener responses provide excellent discrimination among children experiencing above average health-condition-complexity levels in concordance with the CSHCN definition.
Implications for Policy, Delivery or Practice:
Health service providers, epidemiologists, clinicians and others can rely on CSHCN Screener responses to reliably identify CSHCN experiencing the consequences included on the CSHCN Screener.
Additionally, results show that children who need specialized physical, occupational, or speech therapies likely experience the greatest levels of health-condition-complexity. Programs designed to improve systems of care for CSHCN with the greatest health-condition-complexity levels may find that this question proves particularly useful for identifying CSHCN with the highest healthcondition-complexity levels.
Theme: Child Health
● An Examination of Insurance Status Across
Race, Ethnicity & Hispanic Nativity in Predicting
Emergency Department Utilization
Erin Carlson, M.P.H.; Fernando Wilson, Ph.D.
Presented by: Erin Carlson, M.P.H., Graduate
Research Assistant, Health Management & Policy,
University of North Texas Health Science Center,
3500 Camp Bowie Boulevard, Fort Worth, TX
76107, Phone: (402) 305-3144; Email: ecarlson@hsc.unt.edu
Research Objective: This study examines the relationship of insurance status, race/ethnicity and
Hispanic nativity status with emergency department
(ED) utilization.
Study Design: Data is obtained from the 2005
National Health Interview Survey (NHIS).
Descriptive analysis is provided using bivariate analyses. Multivariate logistic regression is used to model race/ethnicity and nativity as predictors of ED utilization, while adjusting for effects of insurance status, poverty, age, marriage, education, gender, having a usual source of care, and health status.
Post-estimation analyses stratify race, ethnicity, and nativity by insurance status. Odds of ED utilization within the past year are calculated comparing the uninsured to the insured and the publicly insured to the privately insured within race, ethnicity, and
Hispanic nativity.
Population Studied: Only respondents who report race/ethnicity as non-Hispanic White, non-Hispanic
Black, or Hispanic, and are aged 18 to 64 years are included in the study. The resulting study sample consists of 54,776 respondents, including 8,573
Hispanics of foreign nativity, 5,126 Hispanics of
U.S. nativity, 33,964 Whites of U.S. nativity, and
7,113 non-Hispanic Blacks of U.S nativity.
Principal Findings: Hispanics are significantly associated with lower ED use compared to non-
Hispanics [OR=0.76 (95%CI 0.67,0.87)] and Blacks are significantly more likely to have used the ED in the past year than Whites [OR=1.31 (95%CI
1.12,1.50)]. U.S. nativity is significantly associated with higher ED utilization compared to foreign nativity [OR=1.71 (95%CI 1.44, 2.04)]. After controlling for nativity, Hispanic ethnicity is no longer significant [OR=0.99 (95%CI 0.86,1.16)]. In post-estimation analyses, utilization is similar by race, ethnicity, and nativity when stratifying by insurance status. The uninsured are significantly less likely to have used the ED than the insured within White [OR=0.92 (95%CI 0.89, 0.95)] and
U.S. born [OR=0.94 (95%CI 0.1, 0.96)] groups.
Within each racial/ethnic group, the publicly insured have nearly the same odds of ED utilization and are significantly more likely to use the ED than the privately insured [White OR=1.61 (95% CI
1.53,1.70); Black OR=1.62 (95% CI 1.31,1.93);
Hispanic OR=1.62 (95% CI 1.57, 1.68); U.S. Born
OR=1.62 (95% CI 1.43, 1.80); Foreign Born
OR=1.62 (95% 1.38, 1.86)].
Conclusion: Differences in ED utilization by race, ethnicity and Hispanic nativity are not significant after stratifying by insurance. Differences in utilization may be explained by other factors such as being publicly insured.
Implications for Policy, Delivery or Practice:
Policies seeking to reduce ED overcrowding should look beyond health insurance expansion and beyond cultural factors of Hispanic immigrants.
Rather, policies should target increasing access to primary care for the publicly insured by addressing barriers to care such as difficulty finding providers who accept public insurance, wait lists to be seen, and inability to take time off work to be seen during business hours. Policy options include increasing reimbursement or other incentives to providers treating publicly insured patients and expanding number and/or capacity of walk-in primary care clinics with extended hours that accept public insurance.
Funding Source(s): RWJF, University of Kentucky
PHSSR Dissertation Grant
Theme: Coverage and Access
● Does Lean/Toyota Production System (TPS)
Redesign Work in Health Care? Findings &
Implications from a Systematic Literature
Review
Kristin Carman, Ph.D.; Kelly Devers, Ph.D.; Lauren
Smeeding; Karen Frazier, M.A.; Steven Garfinkel,
Ph.D.; Michael Harrison, Ph.D.
Presented by: Kristin Carman, Ph.D., Co-Director,
Health Policy & Research, Health Program,
American Institutes for Research, 1000 Thomas
Jefferson Street, NW, Wasington, DC 20007,
Phone: (202) 403-5090; Email: kcarman@air.org
Research Objective: To conduct a systematic review of empirical (quantitative and qualitative) studies, peer reviewed literature and select grey literature on the implementation and impact of
Lean/Toyota Production Systems (TPS), an organizational redesign approach that is increasingly used in health care. Lean/TPS focuses on elimination of waste, defined as any activity that consumes resources without enhancing value to those being served by the process. Using the broader literature on organizational learning, innovation, and change and quality improvement, we developed a conceptual framework to help identify gaps in the current literature and to guide future empirical studies of Lean/TPS, including the case studies that we will conduct in the second stage of our project.
Study Design: Several major electronic databases
(e.g., PubMed, EBSCOBusinessSource) were searched using the following key or MeSH terms:
Lean, healthcare, hospitals, Toyota production system, and quality improvement. After reviewing the articles using pre-established criteria (i.e., empirical studies), 23 articles remained. We also searched the Web and contacted Lean experts to identify grey literature that went beyond simple description of Lean/TPS, its principles, and its potential application in healthcare. At least two people reviewed and abstracted the documents.
The findings were assessed in multiple domains
(e.g., organizational type and characteristics, organizational environment, Lean/TPS implementation, impact, and sustainability).
Population Studied: The total sample included
131 documents: 23 peer-reviewed articles and 108 documents from the grey literature. Numerous other documents were eliminated based on exclusion criteria.
Principal Findings: We reviewed the literature on
Lean/TPS using an organizing frame derived from broader literature on organizational learning, innovation, and change and quality improvement.
Most studies of Lean/TPS lack data on key domains that are important for informing potential adopters, namely factors that may affect their success with implementation. Moreover, there is little rigorous reporting on implementation, impact, and sustainability. Although useful, most studies were conducted in a single organization (e.g., academic medical center or hospital), reported on a single
Lean/TPS project, and did not employ rigorous quasi-experimental or case study designs. When reporting on the key implementation and impact domains occurred, their findings were valuable and consistent with existing literature.
Conclusion: The conceptual frame presented here, along with the literature review, should help foster and guide more rigorous studies of the implementation, impact, and sustainability of
Lean/TPS and other promising interventions in health care settings.
Implications for Policy, Delivery or Practice: As pressure grows for providers to redesign care for greater efficiency, more research is needed on the effects of promising redesign strategies, such as
Lean/TPS, and on the best ways to implement and sustain these strategies. To develop evidence for providers and policymakers about the potential of
Lean/TPS, rigorous, practice-based research needs to examine the whole range of domains identified in our conceptual framework. Our own research project will respond to this need by examining
Lean/TPS implementation in different hospital systems and different units within these systems.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Organized
Processes
● Quality of Care for Medicaid & Privately
Insured Children in a Commercial Insurance
Plan
Henry Carretta, Ph.D., M.P.H.
Presented by: Henry Carretta, Ph.D., M.P.H.,
Adjunct Professor, School of Dentistry/Pediatric
Dentistry, Virginia Commonwealth University, P.O.
Box 980566, Richmond, VA 23298, Phone: (804)
828-0172; Email: hjcarret@vcu.edu
Research Objective: The purpose of this study is to examine whether exposure to a managed care environment can eliminate or reduce disparities between children insured by publicly funded programs, such as Medicaid or the State Children’s
Health Insurance Program, and a privately funded health insurance plan for a common chronic disease such as pediatric asthma.
Study Design: Asthma quality-of-care measures were chosen using evidence-based guidelines and served as the outcomes of interest for the 14 hypotheses. The Andersen Behavioral Model was used to develop a conceptual framework and guide selection of independent variables representing predisposing characteristics of the patient, health beliefs, community or health plan enabling resources, and medical need defined by the ACG® severity score.
Population Studied: Study data were extracted from three years of enrollment and administrative claims records of a managed care organization with
both publicly and privately insured members.
Asthma cases were defined using medical and drug claims occurring in both a six month base period and a 6 to 24 month follow-up period. Hypotheses were tested using logistic and negative binomial regression methods and adjusted for potential endogeneity using propensity scores or instrumental variables.
Principal Findings: Hypotheses predicting that public members would have higher asthma severity and higher probability of emergency department visits and lower probability of spirometry procedures, optimal drug ratio, and follow-up with a physician after an emergency department visit during the base period were supported.
Hypotheses predicting that longer enrollment would provide disproportional benefit to public members and reduce disparities were generally not supported. One model suggested that increased enrollment for pubic members increased the likelihood of follow-up after and inpatient hospital discharge. Continuity of care was, an enabling factor in the Andersen Model, was associated with improved outcomes in both populations.
Conclusion: Disparities in selected asthma qualityof-care measures, between publicly and privately insured members of a single managed care organization, persisted during 12 to 36 months of observation. These disparities persisted over time and may be related to unmeasured family or provider characteristics or health plan factors
Implications for Policy, Delivery or Practice: The ability of MCOs to influence persistant disparities may be hampered by their limited ability to influence provider and patient behaviors. The assignment of a PCP to each member is probably a necessary first step but it is not sufficient to reduce disparities for common outcomes without additional interventions.
Funding Source(s): CDC
Theme: Medicaid, SCHIP, and State Health Reform
● Fragmentation & Continuity of Care Among
Diabetic Medicaid Beneficiaries Seeking Care at
Safety-Net Hospitals & Clinics
Emily Carrier, M.D.; Marc Gourevitch, M.D., M.P.H.;
Maria Raven, M.D., M.P.H., M.S.; Shane Tay, M.A.;
Iryna Lobach, Ph.D.; Nirav Shah, M.D., M.P.H.
Presented by: Emily Carrier, M.D., Fellow,
Medicine & Public Health Research, Emergency
Medicine, New York University, 423 East 23rd
Street 15N-028BN, New York, NY 10010, Phone:
(917) 848-9216; Email: emily.carrier@nyumc.org
Research Objective: The objective of this project is to describe utilization patterns and measure continuity of care among diabetic Medicaid beneficiaries seeking care in a large safety-net health system. We compared groups with high and low continuity of care using multiple measures.
Study Design: Retrospective database review.
Population Studied: 16,985 diabetic Medicaid beneficiaries with at least two primary care visits to one site in a large safety-net health system between January 1 2004 and December 31 2007 and at least one Hemoglobin A1c (A1c) level greater than 7%.
Principal Findings: We followed subjects’ visits across fifteen safety-net hospitals and ambulatory care centers. During the three-year study period, the typical (median) subject was seen at one facility and had 9 primary care visits during which they saw
3 primary care providers. 16,136 of subjects were seen by specialty as well as primary care providers
(2,549 were seen by endocrinologists). Sixty-four percent were seen in an emergency department
(ED); among subjects with ED visits, the typical subject had 3.67 visits. Forty percent had at least one inpatient admission, and the typical admitted subject was admitted twice. When Continuity of
Care indices were calculated, the average CoC was
.44. The typical subject had a mean A1c level of
8.6%. When subjects were divided by initial A1c measurement, those groups with poorer glycemic control (initial A1c values over 9%) showed a significant improvement over time, although these improvements would not be sufficient to bring values to <7% for most. Among subjects with more than five A1c measurements during the study period, sixteen percent showed a decreasing slope on linear regression while sixteen percent tended to rise and the remainder had no significant change.
There was no connection between decreases in
A1c during the study period and continuity of care as measured by (1) the number of facilities visited;
(2) visiting multiple facilities that share information
(vs. multiple facilities that do not); (2) concentration of care in primary care clinic (vs. specialty clinics) and (3) Continuity of Care index.
Conclusion: We followed diabetic Medicaid beneficiaries seeking care in a large safety-net health system for three years and found that most subjects sought care in only one facility. However, they had high rates of specialist and ED utilization as well as high rates of inpatient admission during our study period. For most subjects, A1c did not change significantly over time. Among those subjects for whom A1c tended to rise significantly or fall significantly during the study period, there was no association with any measure of continuity.
Implications for Policy, Delivery or Practice:
Continuity of care is an important feature of many proposed innovations in care delivery, particularly the medical home. Our study found that in a chronically-ill Medicaid population, continuity of care with an individual provider was low (particularly compared with published data from private health plans) and specialty and ED utilization was high.
Among this population, continuity was not associated with diabetes control. Safety-net providers may face a greater struggle to ensure continuity for their patients.
Funding Source(s): CDC
Theme: Coverage and Access
● Factors Associated with Repeat Use of Crisis
Substance-Use Detoxification Services
Emily Carrier, M.D.; Maria Raven, M.D., M.P.H.,
M.S.; Jennifer McNeely, M.D.; Shane Tay, M.A.;
Iryna Lobach, Ph.D.; Marc Gourevitch, M.D., M.P.H.
Presented by: Emily Carrier, M.D., Fellow,
Medicine & Public Health Research, Emergency
Medicine, New York University, 423 East 23rd
Street, 15N-028BN, New York, NY 10010, Phone:
(917) 848-9216; Email: emily.carrier@nyumc.org
Research Objective: In New York State, crisis detoxification services cost $319,883,637 in 2005 in
Medicaid reimbursement alone. Increasingly, this cost has been concentrated among subjects who have multiple crisis admissions but do not participate in noncrisis care (for example, inpatient or outpatient rehabilitation). New York State
Medicaid administrators plan to decrease reimbursement for crisis detoxification, giving providers and policymakers an incentive to identify subjects at risk for readmission so that they can be targeted for appropriate interventions. Proposed interventions include intensive case-management and a ‘warm handoff’ where subjects are escorted from crisis detoxification discharge directly to rehabilitation admission. Our study sought to use information currently being collected for crisis admissions to predict whether subjects were at risk for readmission.
Study Design: Retrospective database review.
Population Studied: 19,104 subjects with at least one admission for crisis chemical dependence detoxification to a certified New York State facility between January 1 and June 30, 2007. Crisis and noncrisis utilization was measured for all subjects in the six months following each initial detoxification admission.
Principal Findings: Thirty percent of subjects were admitted to crisis detoxification services more than once during the study period. Seven percent were admitted to crisis services more than three times.
The most prevalent substances used were alcohol
(59%) and heroin (30%). The study population was predominantly male (77%), racially diverse (38% white, 34% African American and 26% Hispanic) and urban (71% were from New York City and
Westchester County). 52% of the population was insured by Medicaid, of whom the majority (44%) were insured by fee-for-service Medicaid.Twentyseven percent were homeless. Psychiatric comorbities were not measured, but only 2% of subjects were referred to psychiatric services upon discharge. Neither information collected at admission nor initiating non-crisis services within 30 days of discharge predicted repeat utilization of crisis services within 6 months. Initiating noncrisis services during the six-month period was associated with increased crisis utilization.
Conclusion: A significant proportion of subjects had more than one crisis admission during a sixmonth period of observation. Demographic information, patterns of substance use, insurance status and utilization of noncrisis services did not strongly predict readmission.
Implications for Policy, Delivery or Practice: We were not able to generate a predictive model using available information, suggesting that intake data collection procedures should be expanded to include other information including psychiatric comorbidities which are not currently assessed.
Given the complexity of substance use and relapse, strategies to identify repeat users that do not require predictive modeling (such as a web-based intake system that can capture utilization in real time) may be appropriate targets for further study.
We found that timely access to noncrisis services after crisis discharge was not associated with a decreased likelihood of readmission, suggesting that proposed ‘warm-handoff’ interventions may require further study. Finally, we found a high rate of participation in fee-for-service Medicaid, suggesting that case management and care coordination interventions should be targeted at this population.
Funding Source(s): CDC
Theme: Behavioral Health
● Gender Differences in Utilization of Hospice
Services: Who Cares For The Latino Terminally
Ill?
Iraida Carrion, Ph.D., L.C.S.W.
Presented by: Iraida Carrion, Ph.D., L.C.S.W.,
Assistant Professor, School of Social Work,
University of South Florida, 4202 East Fowler
Avenue, MGY 132, Tampa, FL 33620, Phone:
(813) 974-7353; Email: icarrion@cas.usf.edu
Research Objective: This study examined varied caregiving arrangements for hospice and nonhospice terminally ill Latino individuals. Data was collected from 20 in-depth, semi-structured interviews with hospice caregivers of individuals utilizing hospice services and non-hospice caregivers of terminally ill individuals who were not utilizing hospice services, coupled with a statistical analysis of an existing archival dataset of 3,123
Latino patients who utilized hospice services from
August 2002 to June 2006. Interviews were translated and analyzed seeking to substantiate salience of themes across the sample related to the types of caregiving. These methods enabled the researcher to identify key themes related to culture, caregiving, and terminal illness with Latino families.
Study Design: Ten Latino hospice caregivers interviewed were caring for individuals utilizing hospice services. The ten non-hospice caregivers interviewed were caring for terminally ill individuals not utilizing hospice services. The Latino hospice caregivers and non-hospice caregivers consisted of
a combination of paid and unpaid family caregivers.
Two essential criteria were required for the comparative groups. First, it was necessary for both the hospice caregiver and the non-hospice caregiver to self-identify themselves and the terminally ill individuals as Latino. The second criterion was for the Latino individual to have a terminal illness. The qualitative and quantitative design methods facilitated understanding of the population along with the social and cultural contexts within which they live.
Population Studied: Latino hospice and nonhospice caregivers were studied.
Principal Findings: The findings indicate that factors related to language, immigration, income, education, and gender impacted care-giving and contributed to healthcare decisions and treatment choices at end of life. Themes emerged which distinguished hospice caregivers from the non- hospice caregivers. These pertained to age, gender, length of time in the US, income, education, language preference, unpaid/paid caregivers, documentation status and health insurance. All of the hospice caregivers were females and nine of non-hospice caregivers were females. The hospice caregiver group’s median length of time in the US is
35 years and 14.6 years for the non-hospice care group. Spanish was the language of preference for eight of the hospice caregivers and for all of the non-hospice caregivers.
Conclusion: The female Latino caregivers for hospice patients were unpaid family members and paid caregivers for non-hospice patients. The archival data analysis using SPSS 15.0 indicated hospice caregivers were 32.3 percent daughters,
29.0 percent spouses and 11.7 percent sons.
Implications for Policy, Delivery or Practice:
Clinicians and health care providers should be cognizant of the stress factors which Latina caregivers experience when caring for a terminally individual. As a result of their caregiving responsibilities they are at risk of experiencing additional stress which could lead to health and mental health problems, due to limited time to address their own health care needs. Furthermore,
Latina caregivers are coping with language barriers while managing employment demands and family responsibilities. All of these factors are critical for policymakers to consider as they develop eligibility criteria for state and federally funded caregiver programs.
Theme: Gender and Health
● Factors Associated with Turnover Among
Long-Term Care Administrators & Director of
Care in Ontario
Natalie Ceccato, M.S.; Whitney Berta, Ph.D.; Larry
Chambers, Ph.D.; Katherine McGilton, Ph.D., R.N.;
Ann Tourangeau, Ph.D., R.N.; Walter Wodchis,
Ph.D.
Presented by: Natalie Ceccato, M.S., Research
Associate, Department of Health Policy,
Management & Evaluation, University of Toronto,
Health Sciences Building, 155 College Street, 4th
Floor, Toronto, L1N9P4, CA, Phone: (416) 946-
3862; Email: n.perkins.ceccato@utoronto.ca
Research Objective: To determine the level of turnover among senior leadership (Administrators and Directors of Care (DoC)) in Ontario Long-Term
Care (LTC) Homes and to examine whether LTC home characteristics, human resource practices and quality improvement activities are associated with turnover.
Study Design: Data for this study came from: 1) a survey of Administrators and Directors of Care
(DoC) in 620 LTC homes, and 2) the Ontario
Ministry of Health and Long Term Care (MOHLTC) staffing data. The survey included information on home and management characteristics, quality improvement activities, human resource practices, and staffing. Survey and MOHLTC staffing data were linked for homes providing approval for the release of their staffing data from the MOHLTC.
Turnover was measured as the number of
Administrators and Directors of Care who had left the home (voluntarily or involuntarily) in the previous 3 years. Logistic analyses were used to determine factors associated with high turnover (2 or more departures) for each position (Administrator and DoC) separately. The following predictor variables were derived from survey responses and examined: facility, Administrator and DoC characteristics, retention strategies, training, staffing levels and other staff (e.g. personal support worker) turnover.
Population Studied: Administrators and Directors of Care at every LTC Home in Ontario were contacted via e-mail and/or telephone and asked to participate in an on-line survey between July and
October 2008. A total of 225 homes provided complete data for all measures and were included in this analysis.
Principal Findings: Nearly 14% of homes had two or more Administrators leave within the previous three years and 25% had lost two or more DoCs.
Administrator and DoC turnover was significantly correlated (85% of homes with high Administrator turnover also had high DoC turnover). Factors associated with high Administrator turnover included: presence of a management firm, not-forprofit ownership, and DoC turnover. Management training opportunity and quality improvement activities were not associated with Administrator turnover. A significant positive association between high DoC turnover and administrator turnover was identified. High DoC turnover was negatively associated with larger home size, government and not-for-profit ownership, and the presence of a dedicated staff member for quality improvement
(p<0.05).
Conclusion: LTC Homes require consistent leadership typically assumed within the
Administrator and DoC positions. Although
Administrator and DoC turnover are strongly related, factors significantly related to turnover were different for each group. Additional clinical supports within the home were important for DoC turnover while organizational structures were more important for Administrator turnover.
Implications for Policy, Delivery or Practice:
This study begins to address the gap in understanding senior leadership turnover in Ontario
LTC Homes. Identifying key factors related to turnover is an important step in exploring ways to reduce turnover, encourage talent to seek jobs in
LTC, and to improve resident care.
Funding Source(s): Canadian Institutes of Health
Research
Theme: Long Term Care
● Factors Associated with Personal Support
Worker Turnover in Ontario Long-Term Care
Homes
Natalie Ceccato, M.S.; Walter, Wodchis, Ph.D.;
Whitney Berta, Ph.D.; Larry Chambers, Ph.D.;
Katherine McGilton, Ph.D., R.N.; Ann Tourangeau,
Ph.D., R.N.
Presented by: Natalie Ceccato, M.S., Research
Associate, Health Policy, Management &
Evaluation, University of Toronto, Health Sciences
Building, 155 College Street, 4th Floor, Toronto,
L1N9P4, CA, Phone: (416) 946-3862; Email: n.perkins.ceccato@utoronto.ca
Research Objective: Personal support workers
(Nursing Assistants in the U.S.) provide the vast majority of direct care to residents in Long-Term
Care (LTC) homes. Consistency in these staff are considered essential in ensuring quality of care in
LTC. The objective of this research was to determine the level of turnover among full and parttime personal support worker (PSW) staff in Ontario
LTC Homes and to examine whether LTC home characteristics, retention and training opportunities are associated with turnover.
Study Design: Data for this study came from: 1) a survey of Administrators and Directors of Care
(DoC) in all Ontario LTC homes, and 2) staffing data collected by the Ontario Ministry of Health and
Long Term Care (MOHLTC). Administrators were contacted via e-mail and/or telephone and asked to complete an on-line survey between July and
October 2008. The survey included information on home management characteristics, quality improvement activities, human resource practices, and staffing. Survey and MOHLTC staffing data were linked for homes providing approval for the release of their staffing data from the MOHLTC.
Turnover was defined as percent of PSW staff who left voluntarily or involuntarily in the previous 12 months. This was calculated separately for full and part-time staff. For analyses we dichotomized turnover rates to identify facilities with turnover in the highest quartile. Regression models included the following predictor variables: facility,
Administrator and DoC characteristics, training and related human resources practices as well as turnover among other staff in the home.
Population Studied: The population of interest for this study was PSWs in Ontario LTC homes. A total of 163 homes provided complete data for all measures relating to PSWs and were included in this analysis.
Principal Findings: Average turnover was 7% among full-time and 18% among part-time PSW staff with the top quartile defined as above 8% and
30% respectively. Forty-nine percent of PSW care hours were provided by part-time staff. Larger homes were less likely to have high turnover among either full or part-time PSW staff. PSW participation in training and the home’s use of explicit written skills and competencies in regular evaluations reduced the likelihood of having high levels of turnover among full-time staff. On the other hand, facilities with high registered (nursing) staff turnover and higher proportion of full-time staff were more likely to have high levels of PSW turnover.
Conclusion: Part-time PSWs play a substantial role in LTC Home operations. This study identifies the important role human resource programs may have in reducing PSW turnover. Identifying factors associated with reducing overall turnover among professional and non-professional staff is an important extension to this work.
Implications for Policy, Delivery or Practice:
PSWs are critical resources for providing quality care in LTC Homes. Understanding factors related to turnover (including registered nurse staff) provides potential ways to reduce PSW turnover and improve resident quality of care in LTC homes.
Based on this study, increased participation in training and use of skills and competencies in staff evaluations may reduce the likelihood of turnover among full-time PSW staff.
Funding Source(s): Canadian Institutes of Health
Research
Theme: Long Term Care
● Impact of Consumer Directed Health Plan
Membership on Costs & Utilization
William Cecil, M.B.A.; Teresa Waters, Ph.D.; David
Mirvis, M.D.; Panagiotis Kasteridis, Ph.D.; Cyril
Chang, Ph.D.
Presented by: William Cecil, M.B.A., Director,
Health Policy Research, BlueCross BlueShield of
Tennessee, 801 Pine Street - 1E, Chattanooga, TN
37402, Phone: (423) 535-3372; Email: bill_cecil@bcbst.com
Research Objective: This study examines the effects of consumer-directed health plan (CDHP) on the costs and utilization of prescription drugs and
other health care services, using a unique dataset drawn from health plan benefit and service utilization data from Blue Cross Blue Shield of
Tennessee (BCBST), Tennessee’s largest health insurance carrier.
Study Design: Individuals who were enrolled in a
PPO in 2005 and continuously enrolled for the study time period (2005 – 2007) were identified.
Individuals who switched from a PPO plan to a
CDHP plan in 2006 or 2007 (n=1,376) were compared to individuals who remained in a PPO using a two-step process. First, propensity-score matching was used to select a comparison group of enrollees who remained in PPO plans during the study period. Second, difference-in-differences analyses (DID) was used to compare costs and utilization across the groups. A series of Hurdle
Bayesian Poisson panel data models were estimated to predict the probability of any use and the level of utilization for each area of healthcare utilization. Two-part log-normal panel data models were used to estimate the probability of any expenditure and level of expenditures. All models were Bayesian and parametric.
Population Studied: The study population included individuals continuously enrolled in BCBST commercial health plans (PPO or CDHP) over the study period.
Principal Findings: While analyses are still ongoing, our initial results suggest that membership in a CDHP plan had a modest negative impact on overall drug utilization, as measured by total number of prescriptions (DID coefficient = - 0.076; p
= 0.06), but little if any impact on prescription expenditures. These results may be driven by increases in the probability of filling ‘essential’ (DID coefficient = 0.589; p <= 0.01) or ‘non-preferred’ medications (DID coefficient= 0.275; p <=0.05).
Our results also suggest that CDHP membership may have a modest negative impact on PCP visits
(DID coefficient -0.090; p <= 0.05), but no significant impact on other areas of utilization such as emergency room or specialist visits.
Conclusion: Our study differs from most previous studies of the impact of CDHP plans on utilization because we use data from multiple employers across a southern state, most of whom are relatively small. Our findings suggest that membership in CDHP plans is associated with some differences in health care utilization, with modest changes to prescription drug utilization and
PCP visits. We also find some switching within medication types, suggesting that consumers may further “optimize” their utilization when subject to different constraints under CDHP plans relative to those found under a PPO (e.g. using more ‘nonpreferred drugs’).
Implications for Policy, Delivery or Practice: We do not find significant reductions in health care utilization or use of specific types of medications associated with CDHP membership. We are currently investigating whether specific subgroups of enrollees (e.g. chronically ill, high users) may have greater responses to the changes in deductibles and copayments associated with switching to a CDHP plan.
Funding Source(s): Pharmaceutical Economics
Policy Council of Pfizer, Inc.
Theme: Consumer Choices in Health Care
● An Open Source Architecture to Collect &
Analyze Performance Measurement Data of
LHAs' in the Tuscany Region
Domenico Cerasuolo, C.S.; Sabina Nuti
Presented by: Domenico Cerasuolo, C.S.,
Management & Health Laboratory, Scuola
Superiore Sant'Anna, Via San Francesco, 18, Pisa,
56127, IT, Phone: +39 050883881; Email: cerasuolo@sssup.it
Research Objective: The aim of this article is to present an hardware and software architecture that combines different open source technologies to realize an efficient collecting and analysing system using data from a web survey (interview). The framework presented has been developed within a
Performance Evaluation System, to measures the
Tuscany LHAs’ employee’ satisfaction level, compared to some specific variables of internal organizational climate.
Study Design: The study is based on the use of open source technologies in particular:apache as web server, PHP as script language and a relational database management system extremely suitable for web applications as mysql. The current employee's satisfaction survey was activated from december 1 to January 8 2008; dataset and reports have been constantly monitored and final results were available immediately on January 8 th .
Population Studied: The system has been applied to a web survey involving all the 42,000 employee of the 15 Local Health Authorities and the 4 teaching hospitals of the regional health care system. The response rate was about 48.32%
(20,294 cases). The items submitted were 55.
Principal Findings: The advantages of the presented solution are the full automatization of the collecting and analysing processes, the web as a paradigm of deployment, the extreme flexibility of the system that allows to produce a complete and timely/quick reporting output immediately available to the management, adoption of open source solutions, deployed and updated from a community of experts and professionals.
The system allows to produce/generate complex reports using a specific syntax which enables flexible selections of variables and/or values within the data set.
Conclusion: The combination between open source technologies and on-line direct data-entry allowed a strict control of input quality (no missing values, coherency check) of data entry behaviours , and full automation of final reporting
Implications for Policy, Delivery or Practice: The use of open source architecture proved to be flexible and suitable for the employee's satisfaction survey; however Tuscan Regional health care system is already considering to extend this model also to other application areas such as performance indicators of helth services, patient satisfaction surveys and obiuvsly other web surveys. Finally this system provides the possibility to get results, immediately at the end of the survey. As a consequence, it allows managers to take strategic decision faster.
Funding Source(s): Tuscany Region and LHAs
Theme: Health Information Technology
● Improving Pharmaceutical Coverage Decision
Making Through Deliberative Processes
Roger Chafe, Ph.D.; Mark Dobrow, Ph.D.
Presented by: Roger Chafe, Ph.D., Post-Doctoral
Fellow, Cancer Services & Policy Research Unit,
Cancer Care Ontario, 620 University Avenue,
Toronto, ON M5G2L7, CA, Phone: (416) 971-9800
Ex. 3232; Email: roger.chafe@cancercare.on.ca
Research Objective: To make recommendations on how to improve processes for making coverage decisions for new pharmaceuticals for both public and large private insurance plans.
Study Design: Key informant interiews with decision makers in the US, Canada and the UK; document analysis; direct observation.
Population Studied: Processes for making coverage decisions for pharmaceuticals for both public and private insurance programs.
Principal Findings: There is a move in a number of countries to use more structured, deliberative approaches for evaluating whether to grant coverage to new phramaceuticals. These process, like those employed by NICE (UK), the Common
Drug Review (CAN) or Blue Cross (US), hold the promise to improve decision making for drug coverage decisions by making these decisons more transparent, systematic, and evidence informed.
Although based on systematic reviews of the scientific evidence, these processes need to be in part deliberative in order to combine the heterogenous types of evidence which are relevant to the decison whether to cover a specific drug.
Conclusion: With the increased focus on phramaceuticals and the cost of new phramaceuticals, it is likely that we will continue to see the increased use of structured, deliberative processes, for making these decisions.
Implications for Policy, Delivery or Practice: The presentation will discuss the elements needed for good decision making relating to the coverage of new pharmaceuticals and present some of the best practices identified for conducting deliberative processes for making coverage decisions for new pharmaceuticals.
Funding Source(s): Canadian Health Services
Research Foundation
Theme: Global Health
● Is the More the Better? Primary Care
Physician Supply & Ambulatory Care Sensitive
Conditions Hospitalizations in Taiwan’s
Universal Health Insurance Setting
Hsiao-Feng Chang, M.B.A.; Yia-Wun, Liang, Ph.D.;
Jwo-Leun Lee, Ph.D.; Wen-Yi Chen, Ph.D.; Yu-Hsiu
Lin, M.B.A.; Chia-Ling Chuang, M.B.A.
Presented by: Hsiao-Feng Chang, M.B.A.,
Student, Institute of Health Care Management,
Central Taiwan University of Science & Technology,
No.17-4, Lane 33, Hesiang Street, Beitun District,
Taichung City, 406, TW, Phone: +011886-9-
30070987; Email: tercelchf@yahoo.com.tw
Research Objective: The first purpose of this study is to describe patterns in ACSHs at the county level based on county demographics and other characteristics. These patterns include trends overtime, persistence within county over time, and variation between and within socioeconomic strata.
Secondly, we will calculate the ACSH rate for 64 counties respectively, using AHRQ definition. Last, this study intends to explore the impact of primary care physician supply on ACSH rate.
Study Design: Descriptive analysis will be analyzed cross-sectional and overtime. Ordinary least square regression and quantile regression will be used to determine the impacts of primary care physician supply on average and extreme values of the distribution of ACSHs rate while controlling for other variables may also affect ACSHs.
Population Studied: Data for this study comes from 2000-2006 NHI Research Database, a nationally representative dataset, and other government databases. We will examine sixteen principal or first-listed diagnoses identified by
AHRQ as ACSHs. The independent variable of interest, the primary care physician supply, is measured as a continuous variable (MDs/100,000).
Principal Findings: For the present time, we don’t have any results yet, because we are at dealing with the data cleaning and data editing. As a result, the empirical statistical analysis can not be performed.
Conclusion: Though, we do not have the empirical results to show the relationship between primary care physician supply and ACSHs yet. We speculate that there might exist a relationship between primary care physician supply and ACSHs, but the association may vary by some health system factors and population health factors.
Implications for Policy, Delivery or Practice:
Based on previous conclusion assumption, though we still do not know whether the association between primary care physicians supply and
ACSHs in Taiwan is positive or negative, policies promoting practitioner placement in underserved
areas may need further consideration to provide a heath care safety net.
● Evaluation of Smoking Effect on Incidence &
Progression of Chronic Obstruction Pulmonary
Disease & Associated Healthcare Services Cost
Yaojen Chang, Dr.P.H., M.S., M.P.H.; Todd Lee,
Ph.D.
Presented by: Yaojen Chang, Dr.P.H., M.S.,
M.P.H., Post-Doc Fellow, Institute for Healthcare
Studies, Northwestern University, 750 North
Lakeshore Drive, 10th Floor, Chicago, IL 60611,
Phone: (312) 503-5568; Email: ychang@northwestern.edu
Research Objective: Chronic obstructive pulmonary disease (COPD) is one of the leading causes of death and disability in the United States.
Smoking is the leading cause of COPD; however, the impact of the amount of smoking as not been evaluated on the progression of COPD and the costs associated with COPD. Therefore, the goals of this study is: (1) to evaluate the impact of smoking dose on incidence and progression of
COPD; and (2) to estimate cost of healthcare services used to treat COPD.
Study Design: The Markov model was constructed to simulate incidence and progression of COPD and to estimate cost of COPD-related healthcare services consumed by the COPD patients over a
25-year time period. In the model subjects were categorized into four groups, including nonsmokers, light smokers, moderate smokers, and heavy smokers. COPD severity was classified to three stages, based on GOLD criteria (I, II, III/IV).
This analysis was restricted to males aged 50-54 years. Characteristics of the cohort were based on characteristics in the NHANES 2005-06 dataset.
Survival period and cost values were discounted to present values at annual rate of 3%.
Principal Findings: The heavy smoking group had the highest average probability of each COPD stage (I, II, and III/IV) from the model outcomes at each Markov cycle, followed by moderate smokers, light smokers and non-smokers. At the end of model simulation, the probability of heavy smokers in the COPD stage I, II, and III/IV was 1.02, 4.35, and 6.36 times greater than that of non-smokers, respectively. Except the heavy smoking group, the rest of the smoking groups or non-smoking group had an ascending trend in annual COPD prevalence rate on the scatter plot, as age increased. Influence of smoking on incidence and progression of COPD in the heavy smoking group was more significant than in the rest of the groups.
In comparison with the other three smoking groups, the non-smoking group had the lowest annual cost per person, $821 and the longest survival per person, 17.23 years. As the non-smoking group serving as the baseline group, the incremental cost of the light smoking group, the moderate smoking group , and the heavy smoking group was $546,
$681, and $2127 per person, but incremental survival of each group was -0.01, -0.05, and -0.16 year, respectively. The three smoking groups were dominated by the non-smoking group. Under the assumption that 20% and 80% of COPD patients were non-smokers and smokers, the average annual COPD-related healthcare services cost consumed by the male non-smokers aged 50-54 years was $146,121,634. The male light smokers, moderate smokers, and heavy smokers in the 50-
54 years age group respectively consumed
$376,555,801, $413,543,396, and $1,471,262,617 of COPD-related healthcare services per year.
Conclusion: There was an accelerating doseresponse relationship between number of cigarettes smoked and COPD progression. The cost of healthcare services used by COPD patients was highly associated with the accelerating doseresponse relationship.
● Telephonic Nurse Triage & Emergency Room
Utilization
Griselda Chapa, M.S., M.P.H.; Robert Simes,
M.B.A.; Thomas Kotsos, B.A.; Kennith Hunter,
D.P.P.; Herman Jenich, M.P.P.; Kejian Niu, M.S.
Presented by: Griselda Chapa, M.S., M.P.H.,
Director Clinical Research, Analytics & Research,
HMC, 5555 Sheridan Road #1614, Chicago, IL
60640, Phone: (773) 944-9172; Email: gchapa@choosehmc.com
Research Objective: This study had two related objectives. The first was to examine a sample (n =
408) of randomly selected calls to a telephonic nurse triage system (Nurseline or NL) to develop baseline data on emergency room utilization comparing member self-report and validating this with claims data. The second objective was to apply the knowledge gained from the sample on emergency room utilization to a larger population (N
= 2,779,110) to evaluate whether a telephonic nurse triage system can result in cost savings based on appropriate emergency room utilization.
Study Design: This was a retrospective observational study, which examined medical claims and self-reported data.
Population Studied: The pilot had a random sample of 408 individuals enrolled in managed care who had access to a telephonic nurse triage system. The larger analysis evaluated claims data for 2,779,110 individuals enrolled in managed care and disease management. The intervention group had access to the telephonic nurse triage system the control group did not. Using predictive modeling, the intervention and control groups were matched on risk scores.
Principal Findings: The results of the pilot study indicated individuals with access to a telephonic nurse triage system were less likely to utilize emergency room services for less severe care as
indicated by claims procedure codes. This finding was confirmed by the larger population study.
Conclusion: Our findings suggest, given an alternative, privately insured health plan members prefer not to use the emergency room when their condition can be treated in a non-emergency room setting.
Implications for Policy, Delivery or Practice:
Given the costs associated with an emergency room visit, which involves both professional and facility fees, our findings suggest nurse triage systems do produce cost savings while giving health plan members the ability to better select their care options.
Theme: Quality and Efficiency: Measurement
● IVR Messaging & Maternity Management
Griselda Chapa, M.S., M.P.H.; Britt Parrack, M.Ed.;
Jennifer Beale, M.P.H.; Ellen Silver, M.D.; Chang
Su, M.S.
Presented by: Griselda Chapa, M.S., M.P.H.,
Analytics & Research, HMC, 5555 Sheridan Road #
1614, Chicago, IL 60640, Phone: (773) 944-9172;
Email: gchapa@choosehmc.com
Research Objective: To test the acceptance of an
IVR system for use with pregnant women.
Study Design: Randomized cross-sectional telephonic survey. Women were read IVR messages inviting them to enroll in a maternity management program.
Population Studied: 300 English-speaking women
18 years of age or older who: 1) were currently pregnant; 2) had been pregnant in the previous year; or 3) had a miscarriage at any past time.
Principal Findings: Women who had either only 1 child or no children were 2.28 times more likely to say they would enroll in the maternity management program. This was significant at p. <.05. These findings suggest IVR messaging is a modality that can be helpful in enrolling women in a maternity management program.
Conclusion: Interactive Voice response messaging campaigns for health education and disease management may be more successful when these are more target appropriate. In this study, women who had more than one child were less interested in enrolling in a maternity management program.
Implications for Policy, Delivery or Practice:
Interactive Voice Response (IVR) messaging has been used extensively in health services but less so in maternity care. IVR messaging has a wide range of advantages including: economy, autonomy, confidentiality, access to certain population groups, improved data quality, standardized interviewing, and multi-lingual interfaces. However, our results suggest the IVR message is better received if the message is meaningful to the target audience.
● Using Geo-Mapping to Demonstrate the Need for Small Area Demographic, Health, & Health
Systems Data for Program Planning & Disease
Prevention
Theresa Chapple-McGruder, M.P.H.; Jonathan
Currie; Rebecca Lipton, Ph.D.; Diane Lauderdale,
Ph.D.; Stacy Tessler Lindau, M.D., M.A.P.P.
Presented by: Theresa Chapple-McGruder,
M.P.H., Epidemiologist, Obstetrics & Gynecology,
University of Chicago, 5841 South Maryland
Avenue, Chicago, IL 60637, Phone: (773) 834-
9877; Email: tchapple@babies.bsd.uchicago.edu
Research Objective: To evaluate patterns in demographic, health and healthcare variables within and between 32 community areas comprising the Southside of Chicago and comparing the
Southside to 15 community areas comprising the
Northside.
Study Design: We mapped the geographic dispersion of race, income, income inequality, transit lines, preterm births, diabetes deaths, and smoking during pregnancy using 2008 demographic estimates from Easy Analytic Software Inc (EASI) and 2004 health and health systems data obtained from the CDPH website. EASI estimates are derived from 2000 Census data, county-level and national Census Bureau benchmarks. We used
ArcGIS for mapping variables at the community area or, where available, census tract. Low income households were defined as those with annual income less than $25,000. Impoverished areas were communities with more than 50% low income households. Income inequality was calculated using
Gini coefficients, which ranges from 0-1; 1 denotes perfect income inequality.
Population Studied: Community areas on the
Southside of Chicago.
Principal Findings: The Southside is predominantly African-American (72%); the
Northside is predominantly White (58%). The
Southside includes 88% of the city’s impoverished areas, 75% of which surround UCMC. Income inequality is present throughout Chicago, primarily in areas surrounding the city’s center. Census tract income inequality is found in high and low income communities on the Southside, but only in high income communities on the north. The south-east side and the southernmost community areas have fewer healthcare facilities and are further from train lines. Of the 16 communities with the highest rates of preterm birth (15-19 per 1000 live births), 13 are on the Southside. Only one Southside community has a preterm birth rate below the national average of 11%. Similar results were found for diabetes deaths, with high rates clustered in communities south of 63rd street. Of the 10 communities with the highest diabetes death rates, only one is on the
Southside. The prevalence of smoking during pregnancy is 5-20% on the Southside, as compared to 0-5% on the north.
Conclusion: Comparisons of demographic, health and health systems data between the South and
Northsides of Chicago reveal socioeconomic and health disparities. Homogeneous patterns of income inequality at the community area level are more complex when viewed at the Census tract level. For income inequality, smaller area data reveals significant economic heterogeneity within several Southside community areas.
Implications for Policy, Delivery or Practice:
Small area health and health systems data can be useful in understanding social determinants of health in the community areas surrounding UCMC and the CBOs serving people on the Southside.
Funding Source(s): University of Chicago Medical
Center's Urban Health Initiative
● Effects of Health Savings Accounts on
Utilization, Expenditures & Preventive Services
Mary Charlton, Ph.D.; Barcey Levy, M.D., Ph.D.;
John Schneider, Ph.D.
Presented by: Mary Charlton, Ph.D., Assistant
Professor, Health Services Research &
Administration, University of Nebraska Medical
Center College of Public Health, 984350 Nebraska
Medical Center, Omaha, NE 68198-4350, Phone:
(402) 559-8436; Email: mcharlton@unmc.edu
Research Objective: To assess the impact of a health savings account (HSA)-eligible plan on utilization and expenditure patterns of, and use of preventive services by, enrollees of an employersponsored Midwestern health plan that offered one traditional plan from 2003 through 2004 and then offered only an HSA-eligible plan in 2005 and 2006.
Study Design: Retrospective pre/post design with a control group.
Population Studied: The medical and pharmacy claims experience of plan members under age 65 who were continuously enrolled throughout the 4year study period were used to evaluate expenditures and utilization (medical services and use of preventive services) before and after the full replacement of a traditional plan with an HSAeligible plan. These measures were compared to those of a control group consisting of plan members covered by employers in similar industries and geographic locations during the same period. The models use a panel data design with fixed patientlevel effects models.
Principal Findings: The HSA-eligible plan was associated with fewer annual office visits (0.45) and prescription fills (1.16) than the traditional plans and significantly lower annual pharmacy expenditures
($167) and amount per prescription ($9.66). The
HSA-eligible plan was also associated with significantly lower odds of using mammograms
(0.57) and routine physical exams (0.73).
Conclusion: Employer-sponsored HSA-eligible plans may lower health care utilization and expenditures related to the office and pharmacy places of service. However, they may discourage preventive care and thus may eventually lead to increased medical costs.
Implications for Policy, Delivery or Practice:
Employers offering HSA-eligible plans should take care in designing the preventive services benefit and may want to consider offering additional incentives and educational resources for employees and their families to receive appropriate preventive care.
Theme: Consumer Choices in Health Care
● Review & Analysis of Community-Based
Organizations Impacting South Asian
Americans
Nitasha Chaudhary, M.P.H.
Presented by: Nitasha Chaudhary, M.P.H., 914
Willowleaf Way, Rockville, MD 20854, Phone:
(301) 706-5598; Email: nitasha.chaudhary@gmail.com
Research Objective: The overall goal of this research was to better understand how the needs of the South Asian populations are being met via community-based organizations and how these services are provided by these organizations.
Study Design: The preliminary research was conducted via web searches to identify the various
CBOs that target their services to the South Asian
American population. Additionally, CBOs were identified based on information on public health listserv.
Population Studied: South Asian American
Community Leaders
Principal Findings: 1. Use of media to raise awareness/visibility of South Asian Americans 2.
Broadening the scope of discussions among this specific minority group to other minority groups in the U.S. 3. Positive Advocacy movement.
Conclusion: At this time, there is very limited available data on South Asian Americans. The initial thought of this project was to determine health indicators for South Asian immigrants; however, with limited data and awareness for this population, it became critical to understand and analyze the work of community based organizations and the effectiveness of their programs within this ethnic group.
● The Health Characteristics of South Asians: A
Comparison of Estimates from Two Survey
Methodologies
Neetu Chawla, M.P.H.; Ninez Ponce, Ph.D.; Beth
Glenn, Ph.D.; Zul Surani, B.S.; Roshan Bastani,
Ph.D.
Presented by: Neetu Chawla, M.P.H., Researcher,
Jonsson Comprehensive Cancer Center, University of California, Los Angeles, 650 Charles Young
Drive South, Los Angeles, CA 90095, Phone: (310)
825-7430; Email: nchawla@ucla.edu
Research Objective: To assess the effects of sampling methodology on demographic, access, and health status characteristics between two data sources on South Asians, an under-researched, but rapidly growing Asian subpopulation in the United
States.
Study Design: Two independent data sources were utilized. The South Asian Needs Assessment
(SANA) recruited a convenience sample of 344
South Asians living in Southern California, and was conduced between September 2004 - March 2005 in community sites such as places of worship, community centers, and individual homes. SANA was conducted in English (50 %), Bengali, Gujarati,
Hindi, Punjabi, and Urdu through in-person interviews (69%), telephone interviews (18%), and self-administered questionnaires (13%). The 2005
California Health Interview Survey (CHIS) was a statewide, random-digit-dial telephone survey of about 40,000 respondents, conducted in English,
Spanish, Vietnamese, Korean, and Chinese. The
400 South Asians in CHIS, all interviewed in
English, were included in this study. Demographic, access and utilization, and health status measures were compared between the two data sources. The
CHIS sample was compared to the total SANA sample (n=344) as well as the sub-set of the SANA sample that was administered in English (n=173).
Population Studied: South Asians living in
California who were 18 years and older.
Principal Findings: Chi-square analysis revealed that South Asians captured in the SANA sample were significantly older, lower-income, less educated, and less English proficient, when compared to the CHIS sample. These differences remained when comparing only those taking the interview in English in the SANA sample. The
SANA sample was also more likely to be uninsured, lack a usual source of care, and report communication problems with their provider compared to the CHIS sample. Englishinterviewees were still less likely to be insured
(p<0.05) and report communication problems with their provider (p<0.001) compared to CHIS respondents. The SANA sample (full and English only) also fared worse than CHIS respondents in health characteristics (daily smoking, heart disease, diabetes, and high blood pressure). After controlling for age, gender, and being insured, significant predictive margins were found between the full SANA sample and the CHIS for lifetime receipt of mammograms (p<0.05), lifetime receipt of
Pap tests and most recent cholesterol test (p<0.05).
When comparing the SANA English interviewees to
CHIS respondents, significant predictive margins only remained for most recent cholesterol test receipt (p<0.05). After controlling for age and gender, significant predictive margins were found between the full SANA sample and CHIS in health status (p<0.05), daily smoking (p<0.05), and diabetes (p<0.05). Diabetes (p<0.05) remained significant when comparing the English interviewees to CHIS.
Conclusion: Considerable differences in the two data sources suggest that sole reliance on population-based data sources could have consequential impact on measuring and addressing healthcare and health status disparities for South
Asians in California.
Implications for Policy, Delivery or Practice:
Data collection efforts should utilize multiple methods to help identify more vulnerable population segments of the South Asian population to better serve their needs.
Theme: Disparities
● Examination of Chronically Uninsured with
Chronic Medical Conditions
Nancy Cheak-Zamora, Ph.D.; Timothy McBride,
Ph.D.
Presented by: Nancy Cheak-Zamora, Ph.D.,
Research Manager, Public Health, Saint Louis
University, 3952 Cleveland Avenue, Saint Louis,
MO 63110, Phone: (512) 964-4551; Email: cheaknc@slu.edu
Research Objective: The literature on access to health insurance is vast and indicates that having a chronic condition decreases the probability of having adequate insurance. Although we know that having a chronic condition affects the likelihood of being offered and the ability to pay for health insurance little is known about the eight million
Americans that are both chronically ill and uninsured. Furthermore, none of these studies utilized advanced health status measures nor focuses on the dynamics of uninsurance spells for the chronically ill. This study explores how the length of an uninsurance spell impacts the physical and mental health of the chronically ill and examines the demographic and socioeconomic differences between subgroups with and without chronic conditions and with various length of uninsurance.
Study Design: The sample was divided into four subgroups based on chronic condition status and uninsurance spell duration (chronically ill insured, chronically ill intermittently uninsured, chronically uninsured without chronic conditions, and chronically uninsured with chronic conditions (the
“chronically chronic”)). Chi-square and t-tests were then preformed to determine if the chronically chronic were statistically different then other subgroups. Binomial logit regressions were conducted to assess how risk factors and personal characteristics were associated with increased likelihood of being in the chronically chronic subgroup rather than then never or intermittently uninsured with chronic conditions subgroup or the chronically uninsured without chronic conditions subgroup.
Population Studied: This project uses data from the Medical Expenditure Panel Survey (MEPS) representing a sample from the general civilian noninstitutionalized public. Data collected for Panel 8 of the MEPS household component was used in this analysis, created by combining the 2003 and 2004 full year data files (n=9057).
Principal Findings: The chronically chronic were similar to the general population with regards to demographic characteristics but are more likely to face economic and employment disadvantages.
For instance, the chronically chronics had significantly lower levels of education, lower employment rates and were more likely to be living in poverty then any other subgroup. The chronically chronics had significantly poorer physical health and more limitations then the intermittently uninsured with chronic conditions and the uninsured without chronic conditions. They had poorer mental health but fewer chronic conditions and physical limitation then the insured chronic condition subgroup.
Conclusion: To our knowledge this was the first study to investigate the chronically chronic and compare them to other chronically ill with varying lengths of uninsurance. This study finds that in fact for the chronically ill uninsured the biggest barrier facing them may in fact be that they lack the financial means to access health insurance.
Implications for Policy, Delivery or Practice:
These findings identify an insurance and labor market problem, both of which are working to keep the chronically ill out of the health insurance pool.
Policy makers should use these findings to evaluate why these individuals are not or cannot access public or private health insurance. These findings should spur more research on how the chronically chronic manage their illness and cost benefit analysis on costs associated with postponing care.
Theme: Coverage and Access
● Understanding the Experiences of
International Medical Graduate Primary Care
Physicians in the United States
Peggy Chen, M.D., M.Sc.; Leslie Curry, Ph.D.;
Marcella Nunez-Smith, M.D., M.S.; Susannah
Bernheim, M.D.; Aysegul Gozu, M.D.; David Berg,
Ph.D.
Presented by: Peggy Chen, M.D., M.Sc.,
Pediatrician; Post-Doctoral Fellow, Robert Wood
Johnson Clinical Scholars Program, Yale University
School of Medicine, IE-61 SHM; P.O. Box 208088,
New Haven, CT 06520, Phone: (203) 785-2478;
Email: peggy.chen@yale.edu
Research Objective: International Medical
Graduates (IMGs), physicians who attended medical school outside the US or Canada, currently comprise >25% of the US physician workforce, with even greater representation in primary care fields
(general pediatrics, general internal medicine and family medicine). Despite the frequency of transcultural interactions in primary care and their potential impact on workplace experiences, there is a paucity of literature that aims to systematically understand IMGs experiences and how they may best be supported in the primary care workplace.
The goal of this project is to characterize the professional experiences of non-US-born international medical graduates (IMGs) currently practicing as primary care physicians in the United
States.
Study Design: Because we were interested in studying a complex and potentially sensitive topic involving cultural and social interactions that are difficult to measure quantitatively, we used qualitative methods, in particular, in-depth, inperson interviews. Coding and analysis are being conducted iteratively with data collection, by a multidisciplinary team and will utilize the constant comparison method. Unifying themes in IMGs experiences will be identified to inform an understanding of how IMGs can best be supported in the primary care workplace.
Population Studied: We recruited a purposeful sample of participants with attention to country of origin and specialty. We targeted IMG physicians from among the 130 countries the World Health
Organization has classified as having < 2 physicians per 1,000 individuals (the US has 2.56 physicians per 1,000 individuals) in order to focus on IMGs from non-Western developing countries for whom the trans-cultural differences were likely to be substantial. We limited our sample to IMG physicians in primary care specialties who had completed US residency training and are practicing physicians.
Principal Findings: At this time, our sample consists of 19 respondents. We have a 90% participation rate with 32% of respondents from
Pediatrics, 32% from Internal Medicine and 37% from Family Medicine. We are currently engaged in the iterative process of coding and analysis and will continue recruitment and data collection until we reach theoretical saturation (anticipated at 5-8 additional interviews). A number of recurrent and unifying concepts are emerging, including IMG physicians perceptions of bias, discrimination and intimidation in the workplace, the identification of structures and supports that would enhance IMGs ability to function effectively in the workplace and feeling the burden of representation.
Conclusion: IMGs have a substantial and growing role in primary care. Current workforce trends indicate continuing shortages in this area, particularly for vulnerable populations. This study is the first to characterize the factors that impact the professional lives of IMG physicians.
Implications for Policy, Delivery or Practice: A systematic understanding of IMGs' workplace experiences can inform the development of effective targeted interventions to optimally support the delivery of primary care by a diverse physician
workforce. It may also stimulate the development of policies or practices to encourage continued cooperation and links between IMGs and countries of origin.
Funding Source(s): RWJF
Theme: Health Care Workforce
● High Cost Inpatients: Who They Are & Where
They're Treated
Lena Chen, M.D., M.S.; Jie Zheng, Ph.D.; Ashish
Jha, M.D., M.P.H.
Presented by: Lena Chen, M.D., M.S., General
Internal Medicine Fellow, Medicine, VA Boston
Healthcare System, 150 South Huntington Avenue,
Boston, MA 02130, Phone: (857) 364-4201; Email: lmchen@partners.org
Research Objective: High and rising Medicare expenditures pose a serious threat to the federal budget. Although prior studies have shown that costs are concentrated among a small fraction of patients, we know little about who these patients are or where they receive their care. Therefore, we sought to determine: 1) the concentration of inpatient costs 2) how high cost patients generate their expenditures, 3) the characteristics of these these patients and, 4) the characteristics of hospitals that care for such patients.
Study Design: We categorized Medicare beneficiaries with inpatient charges into two groups: the 5% with the highest inpatient expenditures (high cost patients), and all others (low cost patients). We compared annual admissions, average lengths of stay, daily inpatient charges, and patient-level characteristics of high and low cost patients. We also examined the structural characteristics, the quality of care (using summary scores from the
Hospital Quality Alliance measures), and the efficiency (using previously developed risk-adjusted cost measures) of hospitals that care for more versus fewer of these high cost patients.
Population Studied: Fee-for-service Medicare beneficiaries discharged from U.S. hospitals in
2006.
Principal Findings: The 5% of Medicare patients with the greatest inpatient expenditures (high cost patients) made up 26% of all Medicare inpatient expenditures. These high cost patients had higher yearly admissions (3.7 vs. 1.5, p < 0.0001) and longer lengths of stay (40 days versus 8 days, p <
0.0001). High and low cost patients had comparable mean costs per inpatient day ($2,954 versus $2,483, p < 0.0001). High cost patients were less likely to be the oldest old (>/= 80 years of age,
31% vs. 41%), and more likely to be men (51% versus 41%), and black (12 % versus 8%). High cost patients were more likely to be treated in large
(37% vs. 29%), urban (90% vs. 84%), and teaching
(26% vs. 16%) hospitals (all p-values<0.0001). High cost patients were 1.5 times more likely to be cared for by high quality compared to low quality hospitals
(proportion in highest vs. lowest hospital quality quartile: 30% vs. 19%, p < 0.0001). They were also more than four times as likely to be cared for by high cost hospitals (proportion in highest vs. lowest hospital cost quartiles: 42% vs. 9%, p < 0.0001).
Conclusion: High cost Medicare inpatients generated their higher costs through more frequent and longer hospital admissions, and less so through more intense daily resource use. Certain types of hospitals, such as large, urban and teaching hospitals care for a disproportionate share of these patients.
Implications for Policy, Delivery or Practice:
Prior efforts to reduce Medicare spending have focused on care management for high cost patients, but evidence in support of the cost-effectiveness of such programs is limited and mixed. Our data suggest that a second approach -- which assists hospitals which are more likely to care for high cost patients -- may also be effective.
Theme: Medicare
● Solo Practices: Their Patients, Quality of Care
& Resource Use
Lena Chen; Ashish Jha, M.D., M.P.H.
Presented by: Lena Chen, General Internal
Medicine Fellow, Medicine, VA Boston Healthcare
System, 150 South Huntington Avenue, Boston, MA
02130, Phone: (857) 364-4201; Email: lmchen@partners.org
Research Objective: Given the imperative to improve the quality and costs of care for chronically ill patients, policymakers have focused on reforming primary care, especially in large group settings.
However, not all Americans receive their care in such settings. Understanding how care is delivered in small practices will be critical to any national effort to improve primary care. Therefore, we sought to determine: 1) What proportion of primary care visits occur in solo practices? 2) What types of patients visit solo practitioners? 3) And, do solo practitioners provide comparable quality at similar cost to non-solo-practitioners?
Study Design: We used the National Ambulatory
Medical Care Survey (NAMCS) to examine all visits by adults to private, office-based primary care physicians between 1999 and 2005. We compared quality and resource use between solo practices, and all other practices. We defined high quality care as meeting one of nine validated counseling (diet, exercise), screening (blood pressure), and medication (appropriate treatment for atrial fibrillation, congestive heart failure, hypertension, hyperlipidemia, coronary artery disease, the elderly) indicators. The measures of resource use included visit duration and the proportion of visits resulting in referral to another physician, at least one laboratory test, or at least one imaging test. To account for potential confounders, we used multivariate regression models, with performance on individual
quality or resource indicators as the primary outcome and practice size as the primary independent variable.
Population Studied: U.S. office visits by adult patients who saw a primary care physician.
Principal Findings: Between 1999 and 2005, 38% of adult primary care visits to physicians were made to solo practitioners. There was no trend in the proportion of visits to solo practices over time (pvalue for trend = 0.10). Compared to visits to nonsolo-practitioners, a greater proportion of visits to solo practitioners were made by patients who were elderly (34% versus 30%, p<0.001), non-white
(26% versus 20%, p=0.005), and without private insurance (51% versus 42%, p<0.001). Solo practitioners were less likely to provide appropriate care for seven of the nine indicators examined, although only one was statistically significant: diuretic or beta-blocker use for hypertension (OR
0.77, p=0.005). Solo practitioners spent more time with their patients in a typical visit (21.1 minutes versus 20.1 minutes, p=0.01) than non-solopractitioners. Solo practitioners referred patients less frequently to other physicians (OR = 0.74, p<
0.001) and ordered fewer laboratory tests (OR =
0.89, p = 0.03). They also ordered fewer imaging tests, but this was not significant (OR = 0.91, p =
0.12).
Conclusion: Nearly four in ten primary care visits in the United States are made to solo practices.
Solo practitioners were more likely to see elderly, minority patients and those without private insurance. Although solo practitioners spent more time with their patients, they made fewer referrals, ordered fewer tests, and generally provided lower quality of care.
Implications for Policy, Delivery or Practice:
Understanding how to improve the quality of care delivered by solo practices without raising costs of care, will be critical to improving primary care for
Americans.
Theme: Quality and Efficiency: Organized
Processes
● The Impact of Hospital Financial Condition on the Provision of Unprofitable Services
Hsueh-Fen Chen, Ph.D.; Gloria Bazzoli, Ph.D.; Hui-
Min Hsieh, M.S.
Presented by: Hsueh-Fen Chen, Ph.D., Assistant
Professor, Health Management & Policy, University of North Texas Health Science Center, 3500 Camp
Bowie Boulevard, Fort Wroth, TX 76107, Phone:
(817) 735-0311; Email: hschen@hsc.unt.edu
Research Objective: To examine the effects of hospital financial condition on the amount of unprofitable services that hospitals deliver.
Study Design: Economic theory and a longitudinal empirical study design are applied to address study objectives. The dependent variable is the logged total volume of unprofitable services for each study hospital. Unprofitable services were defined based on prior work of Darrell Gaskin and Jill Horowitz and included maternity care, newborns with complications, substance abuse, psychiatric care, trauma, and burn services. Discharges for these unprofitable services were extracted from the
Healthcare Cost Utilization Project State Inpatient
Data (HCUP-SID) and aggregated to the hospital level separately for insured and uninsured patients.
The primary independent variables of interest were financial performance measures, including the ratio of cashflow to total revenues and operating margin.
These were obtained from Medicare hospital cost reports. Control variables included health policy, hospital, and market factors from several data sources including the AHA annual survey, Area
Resource File, Interstudy, Census Bureau, and
Medicaid Statistical Information System. First difference models with instrumental variables were used to control for latent hospital-specific effects and endogeneity of hospital financial condition.
Models were estimated separately for not-for-profit
(NFP), public, and for-profit (FP) hospitals.
Population Studied: All nonfederal, general acute community hospitals included in the HCUP-SID from 1996-2003 for the states of Arizona, California,
Colorado, Florida, Iowa, New Jersey, New York,
Washington, and Wisconsin. The sample was further restricted to the hospitals with no ownership changes during the study period and with staffed beds more than 30. Looking across all the study years, our hospital sample included 5,032 NFP hospital observations, 1,360 public hospital observations, and 1,160 FP hospital observations.
Principal Findings: NFP hospitals with strong financial performance provided more unprofitable services for both insured and uninsured patients than did those NFP hospitals with worse financial condition. Financial performance did not influence public and FP hospitals provision of these services.
Conclusion: The results are consistent with the utility maximizing behavior of NFP hospitals in that they use their profits to expand services that may meet community need but are not adding to their profits. FP hospital’s behavior is consistent with profit maximization in that they provide these services as a business cost to attract and maintain consumers, especially insured patients.
Implications for Policy, Delivery or Practice:
NFP hospitals when confronted with financial distress may limit the amount of unprofitable services they provide, which means that they may be unable to meet community needs. Additionally, increases in demand for, and decreases in supply of, unprofitable services in a community may add to the burden of hospitals that currently provide these services. Policy reform to reduce the gap between payments and cost for certain services is advisable so that hospitals are better able to supply those services needed by the local community.
Funding Source(s): National Heart, Lung, and
Blood Institution
Theme: Quality and Efficiency: Policies and
Incentives
● Home Care Service Utilization & Congestive
Heart Failure Outcomes: Comparing Asian &
White Canadian Patients
Guanmin Chen, M.D., Ph.D.; Nadia Khan, M.D.,
M.S.; Kathryn King, Ph.D.; Brenda Hemmelgarn,
M.D., Ph.D.; Hude Quan, M.D., Ph.D.
Presented by: Guanmin Chen, M.D., Ph.D.,
Research Assistant Professor, Community Health
Sciences, University of Calgary, 3330 Hospital
Drive, Northwest, Calgary, T3K4H5, CA, Phone:
(403) 210-7696; Email: guchen@ucalgary.ca
Research Objective: Congestive heart failure
(CHF) is a major cause of death and hospitalization in the elderly and many of these patients will require formal home care services. Asian patients tend to rely on informal, family-based home care and may not utilize home care services. We investigated utilization of formal home care services among
Asian and White patients discharged from hospital with CHF and also explored whether there were ethnic differences in mortality and readmission to hospital for CHF among users and non-users of home care services.
Study Design: One-year mortality and hospitalization readmission for CHF after discharge were determined through linkage of vital and hospital discharge data. Asian ethnicity was identified using surnames and remaining patients were categorized as “White” after excluded aboriginal and Spanish ethnicity. Home care services utilization in the year following discharge was determined using home care services databases. Stratified by home care service utilization (yes/no), the mortality and hospital readmission rate were compared between the
Asian and White patients after controlling for the factors of age, sex, Charlson comorbidity score, and family neighborhood income.
Population Studied: CHF patients identified using hospital discharge abstract from March 31, 2000 to
April 1, 2004 in Calgary Hearth Region.
Principal Findings: There were 1182 Asian and
24989 White patients discharged from hospital with
CHF. The proportion of home care services utilization was similar between Asian and White
(58.0% for Asian versus 56.6% for White). But
Asian had a significantly lower proportion of home care service utilization than White [adjusted odds ratio (OR): 0.84, 95% confidence interval (95% CI):
0.74-0.89] after controlling for the potential cofounding factors of age, sex, Charlson comorbidity score, and family neighborhood income. Among those who did not utilize home care services, the mortality had no significant difference between them(6.3% for Asian versus 5.8% for
White); But Asian patients had a significantly lower mortality than White after controlling for the potential confounding factors(adjusted OR: 0.76,
95% CI: 0.60-0.96). Among those who utilized home care services, the mortality had no significant difference between Asian and White (7.1% Asian versus 6.7% White), and remained no significant difference between them after controlling for potential confounding factors (adjusted OR: 0.96,
95% CI: 0.81-1.13). There was no statistically significant difference in hospital readmission between Asian and White patients regardless of home care use (50.3% for Asian versus 49.7% for
White, adjusted OR: 1.02. 95% CI: 0.87-1.20 among those with home care and 29.2% for Asian versus 26.7% for White, adjusted OR: 1.02, 95%
CI: 0.83-1.26 among patients without home care).
Conclusion: Mortality was similar between Asian and White patients when home care services were utilized. However, Asian patients had a significantly lower mortality than White among those who did not utilize home care service.
Implications for Policy, Delivery or Practice: The causes to explain this mortality difference between
Asian and White CHF patients who do not use home care service, including differential utilization of informal family supports, warrant further study.
Theme: Disparities
● Adverse Cardiovascular Outcomes
Associated with Antecedent Hypertension after
Myocardial Infarction: Systematic Review &
Meta-Analysis
Guanmin Chen, M.D., Ph.D.; Brenda Hemmelgarn,
M.D., Ph.D.; Sami Alhaider, M.D.; Hude Quan,
M.D., Ph.D.; Norm Campbell, M.D.; Doreen Rabi,
M.D., M.S.
Presented by: Guanmin Chen, M.D., Ph.D.,
Research Assistant Professor, Community Health
Sciences, University of Calgary, 3330 Hospital
Drive, Calgary, T3K4H5, Phone: (403) 210-7696;
Email: guchen@ucalgary.ca
Research Objective: Many studies have examined the role of antecedent hypertension on adverse cardiovascular (CV) outcomes after myocardial infarction (MI), but this data has not been summarized and evaluated in a systematic manner.
We assessed the association of antecedent hypertension with adverse cardiovascular outcomes after MI using systematic review and meta-analysis method.
Study Design: A search of MEDLINE (1950 to April
2008) and an EMBASE (1980 to April 2008) database were supplemented by manual searches of bibliographies of key retrieved articles. Studies were included if they reported antecedent hypertension as risk for adverse outcomes (death, stroke, congestive heart failure (CHF), recurrent MI) among MI survivors. Both randomized controlled trials and cohort studies were included. The key exposure variable was antecedent hypertension.
The primary outcome was all-cause mortality. The
secondary outcomes included CVD mortality and adverse CV events (stroke, CHF or recurrent MI).
Two reviewers independently assessed studies for inclusion and quality, and extracted data from all eligible studies, following the recommendations of
QUOROM. We also extracted information on key indicators of study quality, using the standard of the
STROBE statement. Relative risks (RR) were pooled using a random-effects model and robustness of the pooled RR was evaluated in sensitivity analyses. Cumulative meta-analysis, by chronological year of study beginning, was also performed.
Population Studied: Study population was MI patients in the studies we selected for this systematic review and meta-analysis.
Principal Findings: The search yielded 17 studies
(N=56,748 participants) that reported antecedent hypertension with adverse outcomes for MI survivors: 13 for all cause mortality; 7 for cardiovascular disease (CVD) related mortality; 6 for both recurrent MI and CHF; and 4 for stroke.
RCTs (n=8) were pooled separately from cohort studies (n=9). For RCTs the pooled relative risk
(RR) was 1.19 (95% confidence interval [CI] 1.13-
1.26) for all cause mortality and 1.29 (95% CI1.09-
1.53) for CVD related mortality. For cohort studies, the pooled RR was 1.46 (95% CI 1.34-1.61) for all cause mortality and 1.54 (95% CI 1.22-1.93) for
CVD related mortality. Antecedent hypertension was also consistently associated with an increased risk of stroke, CHF and recurrent MI. Pooled estimates were robust in sensitivity analysis. The association antecedent hypertension with all cause mortality decreased over time in both cohort and
RCT studies.
Conclusion: Antecedent hypertension is associated with adverse outcomes for MI survivors.
However, the association of antecedent hypertension with all cause mortality has decreased over time.
Implications for Policy, Delivery or Practice: The trend of decreased association of antecedent hypertension with all cause mortality for MI survivors reflects improved treatment and management of hypertension in more recent years.
Theme: Prevention and Treatment of Chronic
Illness
● Modeling Clustered Binary Outcomes:
Hospital Performance Versus Hospital
Characteristics
Jacob Jen-Hao Cheng, Ph.D., M.S.; Samuel
Ogunbo, Ph.D., M.P.H.; Nikolas Matthes, M.D.,
Ph.D., M.P.H., M.S.; Carlos Alzola, M.S.
Presented by: Jacob Jen-Hao Cheng, Ph.D., M.S.,
Research Manager & Statistician, Quality Indicator
Project, Maryland Hospital Association, 6820
Deerpath Road, Elkridge, MD 21075, Phone: (410)
796-6250; Email: jcheng@mhaonline.org
Research Objective: Public reporting or pay-forperformance for hospitals is almost entirely based on proportion measures such as the National
Hospital Quality Measures, in which performance is defined as the proportion of patients who received appropriate care. Such proportional data, either rates for individual measures or composite scores across multiple measures, is often used by statistical models to understand the role of hospital characteristics in explaining performance variation.
Although linear model (LM) is the most commonly used modeling approach, it may violate normality and homoscedasticity assumptions. Several alternatives to deal with such clustered binary outcomes are (1) Arcsine transformation applied to
LM for proportional data, (2) generalized estimating equations (GEE) and (3) generalized linear mixed model (GLMM). In this study, we compared four models in model fit, inference, sample size effect and interpretation so as to propose appropriate suggestions for model selection based on their pros and cons under different situations.
Study Design: A retrospective cross-sectional study based on a patient-level performance data.
The quality of overall heart failure (HF) care for each hospital is determined by the pooled rate across four HF measures. The predictors include 6 common exogenous hospital characteristics; namely, region (NE, MW, S, W), bed size (0-99,
100-299, >=300), type (NFP, FP, GOV), urban
(Yes, No), teaching (Yes, No) and critical access
(Yes, No).
Population Studied: Patient-level performance data from 442 hospitals participating in NHQM through the Quality Indicator Project ® in any of four
HF measures during the period 2006-2007. The analysis was performed using two sample sizes: (1) one-year data (2007) with 216,412 opportunities and (2) two-year data (2006-2007) with 440,463 opportunities.rtunities and (2) two-year data (2006-
2007) with 440,463 opportunities
Principal Findings: Regression coefficients are not comparable across methods due to different scales and interpretations. Models were evaluated by aggregated R2, RMSE and Deviance (at hospital level) with resampling cross-validation for model fit and by generalized F-test for inference. Arcsine improved both normality and homoscedasticity for
LM, resulting in closer inference to GEE/GLMM but no better predictions. GEE had the best power for marginal predictions due to the robust estimations but GLMM provided additional hospital predictions due to the random effect. These differences among models reduced with the two-year data.
Conclusion: GEE is the most predictive model with reliable inference but needs patient-level data processing which is computationally intensive and may not be available. GLMM can handle complicated correlation structure and give hospitalspecific predictions for adjusted rankings, with the advantage to use the summary data. Arcsine transformation or larger sample size is effective to
improve LM to give consistent inference with
GEE/GLMM with the more flexibility to handle model diagnosis and nonlinear predictors.
Implications for Policy, Delivery or Practice:
Further exploration of the role of hospitals characteristics for NHQM such as for composite score or peer group design should carefully consider the limitations of linear models and evaluate other options in the context of the specific goals of the analysis.
● Variation in the Composition of Hospice
Interdisciplinary Staff
Emily Cherlin, M.S.W.; Melissa Carlson, Ph.D.
,M.B.A.; Dena Schulman-Green, Ph.D.; Ruth
McCorkle, Ph.D., F.A.A.N.; Rosemary Johnson-
Hurzeler, M.P.H., R.N.; Elizabeth Bradley, Ph.D.
Presented by: Emily Cherlin, M.S.W., Research,
School of Public Health, Yale University, 2 Church
Street South, Suite 409, New Haven, CT 06519,
Phone: (203) 764-9085; Email: emily.cherlin@yale.edu
Research Objective: Previous research in hospital and nursing homes found substantial variation in the proportion of professional staff, particularly registered nurses as a ratio of all nurses; however we know little about the variation in staffing patterns within hospice. As a method of profit maximization through cost minimization and liability avoidance, we anticipated that for-profit hospices would have a narrower range of staff types, fewer professional nurses and social workers as a percentage of all nurse and social work staff, fewer physicians as a percentage of all staff, and fewer volunteers as a percentage of all staff. Further, we anticipated that ownership differences in staffing patters would attenuate over time, with fewer differences by ownership type among more recent entrants into the industry. As the number of for-profit hospices have increased, understanding differences in the interdisciplinary staffing patterns is important for assessing the impact of such ownership changes in the provision of end of life care.
Study Design: We conducted a cross-sectional study of all Medicare certified hospices operating in
2006 in the U.S. We describe the range of staff employed, proportions of nurses who are registered nurses, psychosocial staff who are medical social workers, employees who are clinicians, and staff who are volunteers. We used bivariate and multivariable analyses to evaluate differences in staffing patterns by ownership type, and examine how these differences have changed using interactions with age of the hospice, adjusted for geographical location.
Population Studied: We used the 2006 Medicare
Provider of Services (POS) file, which provides a census of all Medicare certified hospices in the
United States.
Principal Findings: ere were a total of 3,927 hospices in the 2006 Medicare POS file. For-profit hospices were more likely to have a greater range of disciplines (OR = 1.80, p <0.05) We found forprofit hospices had fewer registered nurses as a proportion of all nurses (56.9% vs 63.6, p < 0.05), fewer medical social workers as a proportion of all psychosocial staff (57.9% vs 64.9%, p < 0.05), fewer clinicians as a proportion of all staff (18.5% vs
25.7%, p < 0.05), and fewer volunteers as a proportion of all staff (57.9% vs 64.9%, p < 0.05).
These differences remained significant in adjusted analyses, controlling for geographic location. In general, these differences by ownership were attenuated among hospices that had entered the market within the last 4 years as compared with older hospices (P-values on interactions < 0.05).
Conclusion: For-profit hospices on average have a wider range of disciplines among their staff but have a less professionalized staff, have fewer clinicians as a percentage of all staff, and have fewer volunteers as a percentage of all staff. These differences are most pronounced among older hospices.
Implications for Policy, Delivery or Practice: As the CMS considers quality measures and reporting requirements for hospices, attention to variation in staffing is warranted. Further research should evaluate the impact of these staffing differences on quality of care and satisfaction to understand the full impact of the changing ownership of hospices.
Funding Source(s): NCI
Theme: Long Term Care
● The Effects of Strategy Hospital Alliance on
Hospital Efficiency-Evidence from Public
Hospitals in Taiwan
Chia-Yu Chiang, M.A.; Hsuan-Lien, Chu, Ph.D.
Presented by: Chia-Yu Chiang, M.A., Accounting,
National Taipei University, 67, Sec. 3, Ming-shen E.
Road, Taipei, 104, TW, Email: s89632102@webmail.ntpu.edu.tw
Research Objective: In Taiwan, pressures from the National Health Insurance (NHI) Program and substantial decrease in government subsidies have been the major threats to the finance of public hospitals. Public hospitals search for effective operating strategies to improve their performance.
One of the most popular operating strategies adapted to improving public hospital performance is strategy hospital alliance (SHA). SHA arises when three or more hospitals with poor resources want to achieve a long-term purpose, which is not possible by one hospital. The purpose of SHA is not only to add value to each hospital but to decrease costs through centralized procurement mechanism. This study is to examine the effect of adapting strategy alliance on public hospital efficiency in Taiwan.
Study Design: First, we apply the Data
Envelopment Analysis model (DEA) to measure
public hospital efficiency. In addition, Tobit regression analysis is used to investigate the effects of adapting SHA on hospital efficiency after controlling for other factors (e.g., size, degree of competition, and the quality of patient care delivery) that may also affect hospital efficiency.
Population Studied: We select data from general hospitals in Taiwan including hospitals of department of health and municipal united hospitals during 2001-2003 periods.
Principal Findings: We find that public hospitals that adopted SHA are more efficient than hospitals that did not in Taiwan. In addition, our results also indicate that hospitals with smaller size, located in higher degree of competition, shorter length of stay, higher occupancy rate, and lower mortality are more efficient. We also conducted sensitivity analyses concerning the effect of adapting SHA on hospital financial performance (e.g., operating income rate and return of asset) instead of efficiency.
Conclusion: The results indicate that hospital financial performance improves after the implementation of SHA.
Implications for Policy, Delivery or Practice:
From Taiwan’s experience, we infer that SHA could have an impact on improving public hospital performance. Results obtained from this study should help health policy officials and other health care organizations consider implementing similar strategies for their own countries.
● Using DEA to Understand Cardiac Surgery
Performance over a Ten-Year Period
Jon Chilingerian, Ph.D.; Mitchell Glavin, Ph.D.
Presented by: Jon Chilingerian, Ph.D., Associate
Professor, Heller Graduate School, Brandeis
University, Heller 204, Waltham, MA 02454,
Phone: (781) 736-3828; Email: chilingerian@brandeis.edu
Research Objective: The primary goal was to conduct a 10-year study of performance of cardiac surgeons and to identify factors (patient, surgeon, and hospital characteristics) that explain variations in clinical efficiency for surgeons and hospitals involved in coronary artery bypass graft (CABG) procedures. The second goal was to identify surgeons and hospitals with both high efficiency and low mortality profiles that may serve as potential role models for clinical improvement.
Study Design: Data Envelopment Analysis (DEA) and a bootstrap for DEA estimators were both employed to evaluate the efficiency of surgeons and hospitals in producing CABG surgeries via a two-input, four-output production process. Utilizing a Malmquist approach, boundary shifts versus distance from the new (versus old) frontier were also explored. A multivariate Tobit analysis of surgeon and hospital DEA scores was employed to explore factors that explain differences in efficiency.
The specific factors examined included patient, surgeon, and hospital variables. The study used three different data sources. The chief data source was a database of all CABGs performed in
Pennsylvania hospitals in 1994, 1995, 2000, 2002,
2003, and 2004 from the Pennsylvania Health Care
Cost Containment Council (PHC4). The second data source employed in the study was the
American Medical Association’s Directory of
Physicians in the United States, 35th ed. The
American Hospital Association’s Annual Survey of
Hospitals Database represented the third data source for the study.
Population Studied: Surgeons and hospitals in
Pennsylvania performing CABG procedures over the 1994-2004 period. The main data base obtained from the Pennsylvania Health Care Cost
Containment Council contained over 90,000 discharges for DRG 107 and DRG 109 over the
1994-2004 period for 40 hospitals and 160 cardiac surgeons.
Principal Findings: DEA scores were consistent over consecutive years, and the paired differences were not significantly different from zero. This is evidence of the consistency of DEA scroes.
This research offers some evidence that clinical efficiency may be subject to investing in organizational capabilities that arise from operating strategies such as identifying best practices and offering open heart surgery as a specialty. Most of the best practicing surgeons were concentrated in a small number of hospitals. Although DEA estimates of efficiency were consistent over consecutive years, over the ten-year period, procedure volume was not associated with efficiency; however, the number of years in practice was associated with lower clinical inefficiency.
Conclusion: This longitudinal research offers some evidence that clinical efficiency may be subject to investing in organizational capabilities that arise from operational strategies such as developing open heart surgery as a special focus, offsetting increased years of experience with continuing medical training and, ultimately, developing a talent pool of “efficient” surgeons.
Implications for Policy, Delivery or Practice: A recent AHRQ report (April 2008) found that although work is needed to bring academic research tools such as DEA and SFA to operational application, they are reasonable and ‘easy to use’ products. The most obvious policy implication is to conduct this type of study with a large number of physicians, treating a variety of diagnoses and procedures over long time periods. Working more closely with clinical practitioners would enable health researchers to get a better understanding of the impact of physician clinical production on hospital costs and quality.
As indicators of health status and severity of illness improve, DEA models could become significant tools to reveal value-added activities in the care process. For example, over two decades ago Scott
(1979) proposed an multi-output-multi-input measure of clinical efficiency that divided the amount of improvement for each type of patient by the cost of each hospitalization (input). Measuring the change in severity of illness from admission to discharge would make a DEA efficiency measure clinically relevant by capturing a more complete picture of a physician's clinical performance relative to the resources used.
Funding Source(s): CWF
Theme: Quality and Efficiency: Measurement
● Diabetes Control & the Likelihood of High ED
Utilization
Shang-Jyh Chiou, Dr.P.H.
Presented by: Shang-Jyh Chiou, Dr.P.H., Health
Systems Management, Tulane University, 1440
Canal Street, Suite 1900, New Orleans, LA 70112,
Phone: (504) 988-5428; Email: chiouatg2@gmail.com
Research Objective: This study examines the diabetes control situation with the likelihood of diabetes-related emergency department (ED) visits among type 2 diabetic patients receiving care under a diabetes disease management (DM) program.
Study Design: Risk factors including: age, gender and race, comorbidity index, hospitalization, health plan, duration, and facility size were compared between these two groups using chi-square with
ANOVA and multivariate logistic regression (ED visits cut point in 4)
Population Studied: This study extracted 6,412 diabetic patients from the LSU Disease
Management Evaluation Database from 1999 to
2006. Only kept duration in the program over 4 years or more, type 2 diabetes patients consisted two groups: controlled (A1c < 7%, n=167) and uncontrolled (A1c > 9%, n=648) groups during the study period.
Principal Findings: Race, gender and first age of enrolling in this program were significant comparing two groups. Diabetes-related clinic visits were no significant differences while diabetes-related ED visits were significant between controlled and uncontrolled groups (2.93 v.s. 4.63). In the logistic model for the high ED visits over 4, after control other variables, blacks was more likely to use ED than white (OR= 1.837, 95% CI= 1.244 – 2.711, p=0.002). Individuals stayed longer would likely to use ED (OR=1.242,, 95% CI= 1.018 – 1.515, p=0.033). Patient with hospitalization experience increased the likelihood in high ED visits with an
OR of 1.518 (95% CI=1.070 – 2.153, p=0.019).
Finally, patients who did not keep their A1c in the normal level were 2 times (95% CI: 1.13 – 3.60,
P=0.018) more likely in high ED visits than those who keep well.
Conclusion: Racial disparities still is an issue in diabetes control. Disease managers should continually provide more clinic accesses to ease in high ED usage. Patients following the guideline to control their diabetes are still the best strategy to reduce the likelihood in high ED utilization.
Implications for Policy, Delivery or Practice: By reducing the likelihood of diabetes-related ED visits, unnecessary costs can be avoided and patients’ quality of care improved.
● The Role of Racial Composition on Nursing
Home Performance
Latarsha Chisholm, M.S.W.; Alex Laberge, P.T.,
M.B.A.; Robert Weech-Maldonado, M.B.A., Ph.D.;
Kathryn Hyer, M.P.P., Ph.D.
Presented by: Latarsha Chisholm, M.S.W.,
Graduate Research Assistant, Health Services
Research, University of Florida, 2915 Southwest
13th Street, Apartment 67, Gainesville, FL 32608,
Phone: (352) 745-6788; Email: lrchisho@ufl.edu
Research Objective: Nursing homes with a high proportion of Blacks have been associated with higher deficiencies, lower quality of care, and nurse understaffing. The purpose of this study was to examine the effects of residents’ racial composition on nursing home’s financial and quality performance. Unlike previous studies that have used payer mix as a proxy for financial performance, this study utilized actual measures of financial performance.
Study Design: This cross-sectional study used data from the Minimum Data Set Plus, the On-Line
Survey Certification of Automated Records, the
Medicare Cost Reports, and the Area Resource
File. Dependent variables for the study included: operating margin, total profit margin, expenses per patient day, revenue per patient day, percent of
Medicaid residents, RNs per 100 residents, pressure ulcer prevention processes, and quality of care deficiencies. The primary independent variable was racial composition of residents in nursing homes: no Blacks, medium proportion of
Blacks, and high proportion of Blacks. Categories were created based on an analysis of the distribution of proportion of Blacks, with the lower
23% of nursing homes representing nursing homes with no Blacks, the top 10% representing nursing homes with a high proportion of Blacks, and the medium group representing the nursing homes between these two categories. Control variables included facility (ownership status, chain affiliation, and case mix index) and market characteristics
(Herfindahl index, per capita income, and urban/rural). Ordinary Least Square (OLS) regressions were conducted with state and year fixed effects.
Population Studied: Medicare and Medicaid certified nursing homes across the United States that submitted Medicare cost reports between the years of 1999 and 2004 (11,472 average per year).
Nursing homes that were hospital-based,
government owned, or did not have any Medicare beds were excluded from the sample.
Principal Findings: Nursing homes with high proportion of Blacks had $3 lower patient expenses per day compared to nursing homes with no Blacks.
However, these facilities had a 0.06% lower total profit margin, 0.04% lower operating margin and
$11 lower revenue per patient per day. Nursing homes with a high proportion of Blacks had a higher proportion of Medicaid residents of 23% when compared to nursing homes with no Blacks.
Nursing homes with a high proportion of Blacks had a lower proportion of RN’s of one RN per 100 residents, lower pressure ulcer prevention processes of .10, and 0.4 more quality of care deficiencies compared to nursing homes with no
Blacks.
Conclusion: Despite having lower expenses than other nursing homes, nursing homes with a high proportion of Blacks have worse financial performance as a result of their high dependence on Medicaid and lower patient revenues. Similar to prior studies, we found that nursing homes with a high proportion of Blacks had lower quality of care in terms of RN staffing, pressure ulcer prevention, and quality of care deficiencies.
Implications for Policy, Delivery or Practice:
Changes in Medicaid policy may disproportionately affect nursing homes with a high proportion of minorities, due to their heavy reliance on Medicaid.
Policymakers should consider initiatives that may improve both the financial and quality performance of nursing homes serving predominantly Black residents.
Theme: Disparities
● Adolescent Health Literacy & Use of Online
Health Information
Deena Chisolm, Ph.D.; Lauren Johnson, B.S.
Presented by: Deena Chisolm, Ph.D., Assistant
Professor, Pediatrics, The Ohio State University,
700 Children's Drive, Room JW4982, Columbus,
OH 43205, Phone: (614) 722-6030; Email: deena.chisolm@nationwidechildrens.org
Research Objective: This study examines health literacy in adolescents with chronic disease and tests the relationship between health literacy and use of online health information.
Study Design: Youth age 13-18 were recruited from asthma and diabetes clinics at an academic health center and completed a battery of surveys including measures of health literacy, computer and internet access.
Population Studied: This preliminary analysis includes 63 youths, 38 diabetics and 25 asthmatics.
Principal Findings: One in five teens with chronic health conditions(19%) had low or marginal health literacy levels. Ninety percent of participants reported going online for e-mail or web surfing and
47% accessed online information on health, diet or fitness. Health literacy was significantly associated with white race and female gender but was not associated with age, condition (asthma vs. diabetes), or internet use. Low literacy youths were more likely to go online for health information on
“topics that are hard to talk about like drug use, alcohol or depression” (45% vs. 14%, p=0.02). Low literacy youths also trended toward higher internet use for finding general health information.
Conclusion: Half of teens with a chronic condition sought health information online and, although nearly 20% of participants demonstrated low or marginal health literacy, limited literacy was not a barrier to accessing online health information. Low literacy youths are actually more likely to use the internet for sensitive health topics.
Implications for Policy, Delivery or Practice:
Online health information is an effective tool for reaching teens with chronic conditions but web designers should be aware that many users will have limited literacy.
Funding Source(s): National Institute on Child
Health and Human Development
Theme: Child Health
● Revisit the Cost Factors of Dialysis Clinics in
Taiwan: Managerial Incentives Perspective
Chia-Ching Cho, Ph.D.; Shuen-Zen Liu, Ph.D.;
James Romeis, Ph.D.
Presented by: Chia-Ching Cho, Ph.D., Lecturer,
Department of Accounting, Yuan Ze University, No.
135, Far-East Rd., Chung-Li, Taoyuan, Taiwan
32003, Taoyuan, 32003, TW, Phone: +886 3-
4638800 ext.2194; Email: chiaching@saturn.yzu.edu.tw
Research Objective: This study provides a comprehensive analysis on factors influencing dialysis cost drivers using administrative data from a large renal clinic chain in Taiwan. The study differs from prior studies, by examining how cost drivers related to managerial incentives affect medical costs in dialysis, controlling for factors associated with medical treatments, patient characteristics and medical qualities.
Study Design: We use multiple linear regression analysis to examine factors that influence costs of dialysis in our study. In the analysis, we employ the total medical variable costs as dependent variable.
The total medical variable costs change with respect to a change in the volume of dialysis services. Our independent variables consist of five managerial factors (e.g. capacity utilization rate, shares holding rate, location of clinics, business model, and length of time managed by the case company) and other control variables.
Population Studied: The data used in the study is obtained from internal administrative sources of a large renal clinic chain. The company is the leading provider of dialysis and related services in Asia. It owns or operates about sixty dialysis facilities
across Asia, treating nearly four thousand patients.
In the regression analysis, there are 1,238 observations of dialysis treatments drawn from 16 dialysis facilities in Taiwan. These data are from
July, 2007; they are the most reliable data source available.
Principal Findings: In general, we find that total variable costs for dialysis treatments are significantly influenced by managerial factors such as capacity utilization rate, percentage of owners’ shares holding, and location of clinics. First, the capacity utilization rate is negatively associated with total medical variable costs. That is, when capacity utilization is low, physicians tend to incur higher costs in treatments. Second, the ownership structure matters. Specifically, the percentage of clinic ownership held by the company is positively associated with variable costs. That is, the less physicians own their clinics, they have less incentives to control costs. Finally, clinics locate at larger cities tend to incur higher total medical variable costs per patient than those locate in the rural areas. The result implies that the location of clinic is an important factor for operating.
Conclusion: Our results indicate that treatment costs of dialysis are influenced by managerial factors. Consistent with prior literature, cost drivers associated with medical treatments and patient characteristics also affect the medical variable costs. The findings help renal clinics better identify areas important to cost control and differences in profitability resulting from patient profiles. In addition, the dialysis procedure is complex and inappropriate to use a simple payment scheme for the healthcare policy officials.
Implications for Policy, Delivery or Practice: Our findings suggest that health regulators should be cautious that dialysis providers may reject costly patients and should establish appropriate monitoring mechanisms accordingly. Such caution deserves special attention because revenue pools are generally fixed by the Global Budget System implemented by Taiwan’s National Health
Insurance. As a result, dialysis providers might aggressively find ways to cut costs to increase their profitability. The same concerns are important for
US Medicare where costs of dialysis are substantial. Furthermore, the agency problem that commonly exists in profit-seeking setting also affects the operation of dialysis clinics. Thus, the clinic chain company may sell some portion of its shares to physicians to mitigate the problem. Or, if the case company wishes to maintain control power of clinics, it can offer performance incentives to motivate physicians to reduce costs.
● Designing a Wellness Portal for the Promotion of Patient-Centered Preventive Care: A Pilot
Study
Ann Chou, Ph.D., M.P.H., M.A.; Zsolt Nagykaldi,
Ph.D.; Cheryl Aspy, Ph.D.; James Mold, M.D.
Presented by: Ann Chou, Ph.D., M.P.H., M.A.,
Assistant Professor, Health Administration & Policy,
University of Oklahoma, 801 Northeast 13th Street,
CHB 355, Oklahoma City, OK 73104, Phone: (405)
271-2115 x4; Email: ann-chou@ouhsc.edu
Research Objective: Optimal delivery of preventive services requires appropriate information processing and patient involvement. However, information is limited in patient-centered approaches to deliver preventive services that integrate health information technologies (HIT). This study aimed to develop and pilot test an Internetbased patient Wellness Portal linked to a
Preventive Services Reminder System to facilitate patient-centered, preventive care in primary care practices.
Study Design: To guide Portal development, an advisory panel consisting of clinicians, staff, and patients was assembled to systematically identify
Portal elements/features. The Delphi Technique was employed to achieve consensus to finalize
Portal structure. To test of Portal, patients were randomly recruited from two practices in a Practicebased Research Network to participate in the pilot study, using chart audit. Patients completed a structured survey and semi-structured questionnaire regarding Portal use at the end of six months. A mixed method approach was used for evaluation where frequency statistics were compiled for all structured questions and content analyses were conducted to examine Portal utility.
Population Studied: The pilot included 67 patients who had been seen at the practice sites during the previous three months; 30 of whom were selected to complete the Portal evaluation. The age of the participants ranged from 23 to 83, with a mean of
41 years. About 78% were female and greater than
80% have at least some college education.
Ethnic/racial distribution of the sample included
77.8% White, 18.5% African Americans, and 3.7%
Native Americans.
Principal Findings: Based on consensus, the
Portal provides a personalized wellness plan to enhance patients’ ability to obtain preventive services based on demographic profile, risk factors, symptoms tracking, medication lists, visits/appointments, screening and laboratory tests, allergies, adverse reactions, personal preferences, as well as access to education materials and capability to communicate with providers. Patients participating in the pilot test rated the Portal in three areas: (1) ease of use; (2) importance; and (3) utility and value. Over 90% found the Portal easy to use in terms of navigation (93.3%), finding information
(90%), understanding information (93.3%) and instructions (93.3%). Eighty percent regarded the
Portal to be in important tools in managing their wellness and that it provided pertinent information
(86.7%). Sixty percent reported that the Portal is helpful in improving their health and patient-provider interactions, and 83.3% considered the Portal to be
a valuable resource. Contents analyses showed that patients found the Portal helpful as a tool, particularly its reminder and tracking functions.
Conclusion: Patients regarded the Portal as an easy to use, important, and valuable tool that helped them manage their health. One patient noted that, “Since [using] the Portal, my health has greatly improved. I've lost weight, my blood pressure has dropped, I've increased my activity level--and I feel and look so much better. Perhaps these positive attributes come simply from being more conscious of my health, but nevertheless, I'm thrilled.”
Implications for Policy, Delivery or Practice:
Patients with basic computer literacy may use a simple, consumer-oriented website to coordinate their own preventive care. The Portal is an example of how HIT may encourage active patient participation in their care and potentially improve outcomes.
Funding Source(s): AHRQ
Theme: Health Information Technology
● Surveying Organizational Characteristics in the Implementation of Infection Control in U.S.
Hospitals
Ann Chou, Ph.D., M.P.H., M.A.; Kimberly McCoy,
M.S.; Deanna Willis, M.D.; Bradley Doebbeling,
M.D., M.S.
Presented by: Ann Chou, Ph.D., M.P.H., M.A.,
Assistant Professor, Health Administration & Policy,
University of Oklahoma, 801 Northeast 13th Street,
CHB 355, Oklahoma City, OK 73104, Phone: (405)
271-2115 x4; Email: ann-chou@ouhsc.edu
Research Objective: The growing presence of antimicrobial resistance (AMR) has brought new challenges to the healthcare system, with serious implications on patient safety, costs, and quality improvement efforts that many hospitals have initiated. Although several AMR surveillance programs have demonstrated the extent of problems, there is little information on how they vary by organizational characteristics. To that end, this study’s objective is to describe current organizational practices, differentiated by AMR rates of Methicillin resistant S. Aureus (MRSA),
Vancomycin resistant enterococci (VRE),
Ceftazidime resistanct Klebsiella species (K-ESBL), and Quinolone resistant Escherichia coli (QREC).
Study Design: This study created an analytic dataset using survey responses of Infection Control
Professionals from the Epidemiology and Control of
AMR Study and hospital characteristics obtained from the American Hospital Association Annual
Survey. Point estimates for the AMR rates were calculated using the ridit method, and validated via antibiograms. AMR rates were compared across implementation of organizational processes for infection control, resources, leadership, and size using the Jonckheere-Terpstra test. Hospitals characteristics such as teaching, profit, and VA status were compared using the Cochran-Mantel-
Haenszel mean score statistic.
Population Studied: This study included 448
Infection Control Professionals, representing a 67% response rate of a nationally representative, stratified random sample of hospitals.
Principal Findings: Both MRSA and K-ESBL rates decreased significantly with increased hospital implementation of infection control measures, resources allocation, and leadership. VRE rates decreased with increased hospital implementation of infection control measures while QREC rates decreased with better resources. All AMR rates increased as the number of beds increased. Urban hospitals reported higher MRSA and VRE rates and teaching institutions had higher MRSA, VRE, and
QREC rates. VA hospitals reported higher MRSA and K-ESBL rates.
Conclusion: AMR rates decreased when hospitals implemented processes, dedicated resources, and demonstrated leadership for infection control.
Examining organizational characteristics in the context of AMR rates can provide insights into the challenging issue of preventing infections.
Implications for Policy, Delivery or Practice:
There is a need for better understanding of organizational factors in facilitating the implementation of infection control in US hospitals.
Identifying organizational characteristics associated with lower AMR rates may offer some concrete strategies in which hospitals can evaluate their current capabilities to implement effective processes or practices to foster and sustain a culture of patient safety.
Funding Source(s): VA
Theme: Organizational Performance and
Management
● Nursing Staffing & Patient Outcomes on
Taiwan's Acute Care Hospital Units
Chia-Ling Chuang, M.B.A.; Yia-Wun, Liang, Ph.D.;
Jwo-Leun Lee, Ph.D.; Wen-Yi Chen, Ph.D.; Chi-
Hsuan Asphodel Yang, Ph.D.; Yu-Hsiu Lin, M.B.A.;
Hsiao-Feng Chang, M.B.A.
Presented by: Chia-Ling Chuang, M.B.A., Student,
Institute of Health Care Management, Central
Taiwan University of Science & Technology, No.9,
Alley 26, Lane 492, Sec. 1, Yimin Rd., Dali City,
Taichung County, 41241, TW, Phone: +011886-9-
89770514; Email: apple0514@gmail.com
Research Objective: The purpose of this study is to analyze the associations between hospital nursing staffing (i.e., ratio of nurse to patients, skill mix) and outcomes potentially sensitive to nursing
(i.e., mortality, medication errors, patient falls, infection, pressure ulcers) with consideration of variables that could influence the primary association.
Study Design: Based on a comprehensive and systematic assessment of private and publicly available data, we can not identify any dataset as the best sources of data for constructing the outcomes potentially sensitive to nursing and hospital nursing staffing in Taiwan. Accordingly, we decided to collect the data by ourselves. Variables are measured retrospectively using consistently available data from surveyed hospitals each month for seven months. With stratified random sampling, we randomly select 69 hospitals across Taiwan, however only 33 hospitals were willing to complete the questionnaire, including two medical centers, eighteen regional hospitals, and thirteen district hospitals.
Population Studied: Based on a comprehensive and systematic assessment of private and publicly available data, we can not identify any dataset as the best sources of data for constructing the outcomes potentially sensitive to nursing and hospital nursing staffing. Finally, we decided to collect the data by ourselves. Variables are measured retrospectively using consistently available data from each month for seven months.
With stratified random sampling, we randomly select 69 hospitals across Taiwan, however only 33 hospitals were willing to complete the questionnaire, including two medical centers, eighteen regional hospitals, and thirteen district hospitals. A sample of 33 hospitals from Taiwan was used to analyzed the relationship between nurse staffing measured at the hospital level and 11 outcomes potentially sensitive to nursing. Data will be first analyzed by descriptive information for the
33 hospitals and 100 hospital units in this study, and for the nurses and patients that were sampled from those hospitals and units. Finally, linear regression models and hierarchical linear models
(HLM) will be used to estimate the effects of the nursing staffing on the different dependent variables, while controlling for hospital characteristics and patient acuity level.
Principal Findings: Results showed that patient falls, nosocomial infections, restraints, unplanned removal of the tubes, shock, mortality and failure to rescue are significantly different among hospital units. Nursing hours are significantly correlated with patient falls, restraints, unplanned removal of tubes, shock, mortality and failure to rescue.
Working years of unit nurses are significantly correlated with restraints, mortality. Patients have less restrains and unplanned removal of tubes when there are more N3s nurses in the staffing mix.
Conclusion: Obviously, the nursing staffing and the quality of nursing practice are associated.
However, owing to the ongoing collection of the survey data, therefore the empirical multivariate analysis needs to be examined in the future. We speculate the association between nursing staffing and the outcome potentially sensitive to nursing will also be found while controlling other variables.
Implications for Policy, Delivery or Practice: Our future analysis findings will provide evidence for improvements in nurse-patient ratio in hospitals have the potential to simultaneously increase outcomes potentially sensitive to nursing.
Funding Source(s): Department of Health,
Executive Yuan, R.O.C. (Taiwan)
● The Intertemporal Impact of Noncompliance on Separately Billable Medicare Allowable
Payments in Medicare ESRD Patients
Chien-Chia Chuang, M.S., M.A.E.
Presented by: Chien-Chia Chuang, M.S., M.A.E.,
Ph.D. Candidate, Health Management & Policy,
University of Michigan, Ann Arbor, 109 Observatory
Street, Ann Arbor, MI 48109, Phone: (734) 678-
0379; Email: ccchuang@umich.edu
Research Objective: Previous literature demonstrates that noncompliance in hemodialysis
(HD) sessions significantly affects dialysis patients’ outcomes and quality of care. Few studies have explored the association between noncompliance and economic consequences. Of these, most focused on renal transplant patients instead of all
HD patients. Therefore, the objective of this study was to understand whether a causal relationship exists between noncompliance and economic costs in the concurrent period and whether this impact extends to following periods for all Medicare HD patients.
Study Design: Noncompliance was measured by the number of dialysis sessions a HD patient skipped in each month. Economic costs were measured by Medicare Allowable Payments (MAP) for separately billable (SB) services including EPO, iron, vitamin D, other injectables, and certain laboratory services. I used the OLS model to estimate the association between noncompliance and SB MAP, controlling for patient case-mix and certain facility characteristics. I then applied a 2SLS model using the distance from patient residence to dialysis facility as an instrument to tackle the potential endogeneity that lies between skipped sessions and SB MAP. Additionally, three lagged effect models were fitted to see whether the fluctuation in SB MAP due to noncompliance stabilized after one, two or three months.
Population Studied: The study population consisted of 7,960,864 patient-month-facility data observations, corresponding to 349,984 Medicare
HD patients, from 2003-2006. SB MAP were obtained for all HD patients with Medicare as the primary payer. I used CMS Form 2728 and/or
Medicare claims to identify patient characteristics and factors associated with interrupted dialysis months.
Principal Findings: On average, 4.1% of the HD patients (n=14,505) had skipped at least one dialysis session, with a mean skipped sessions of
2.3. The coefficient estimate from the OLS model
showed that noncompliance was associated with a
$52.9 (p<0.0001) decrease in SB MAP, while the
2SLS model showed a drop of $90.9 (p<0.0001).
Preliminary results from the one month lagged model showed a drop of $5.9 (p<0.0001). The coefficient estimate of noncompliance was no longer statistically significant for the two month lagged model. Results from the three month lagged model suggested that noncompliance was associated with an increase of $1.1 (p=0.0367).
Conclusion: The findings obtained through the
2SLS model, a more robust approach, suggested that dialysis facilities would lose $90.9 in payment if a patient skipped one treatment, which was higher than the average SB MAP of $82.4. This provides a financial incentive for dialysis facilities to monitor and improve patient compliance. Results from the lagged effect models did not support the postulation that noncompliant patients would make up for missed treatments by using more drug injectables to reach their clinical target level in subsequent periods.
Implications for Policy, Delivery or Practice:
Medical costs and patient outcomes are two components to evaluate the effect of noncompliance on treatment efficiency.
Noncompliance could potentially be cost saving for the Centers for Medicare and Medicaid Services as long as the saved cost of missed treatments offsets the cost of increased morbidity. Future research should focus on measuring the cost of increased morbidity due to noncompliance in order to fully examine the net cost effect.
Theme: Medicare
● Patients Without a Medical Home: A Study of
Overuse of Emergency Departments in Florida
Askar Chukmaitov, M.D., Ph.D.; Robert Brooks,
M.D., M.B.A., M.P.H.; Anqi Tang, B.S.; Nir
Menachemi, Ph.D., M.P.H.
Presented by: Askar Chukmaitov, M.D., Ph.D.,
Assistant Professor, Family Medicine & Rural
Health, Division of Health Affairs, Florida State
University- College of Medicine, 1115 West Call
Street, Suite 3200-C, Tallahassee, FL 32306-4300,
Phone: (850) 645-6897; Email: askar.chukmaitov@med.fsu.edu
Research Objective: We evaluate the use of emergency departments (ED) for patients with
Ambulatory Care Sensitive Conditions (ACSCs) and for those who are frequent users of the ED.
Study Design: We hypothesized that patients without a “medical home” are more likely to get treated at EDs, or hospitalized through EDs, for conditions that can be effectively managed at the primary care level. We utilize Ambulatory Care
Sensitive Conditions (ACSCs) as described in the literature to define these conditions. We also studied patients who were treated at EDs, or hospitalized through EDs, frequently (more than 4 times per year) for all conditions, and specifically for
Ambulatory Care Sensitive Conditions (ACSC). We used all-payer, patient-level emergency department and inpatient hospital discharge data for the year of
2005 in Florida. A cross-sectional, descriptive study design was used. The unit of analysis was a visit to an ED and an inpatient admission made through an ED. We compared socio-demographic, payer type, severity, total charges, and other characteristics, using independent sample t-tests or chi-square tests. We then overlapped ACSCs and frequent ED use to identify the number of ED visits and inpatient admissions that were made through
EDs and estimated charges for patients who likely had no “medical home”.
Population Studied: We obtained data from the
Florida Agency for Health Care Administration. We considered all ED visits (n=4,914,933) and all inpatient admissions that were made through ED
(n=1,244,392)
Principal Findings: Around 17% of all ED visits were made for an ACSC. Thirty percent of all inpatient admissions for patients who were seen in
EDs and later hospitalized were due to ACSCs. We also discovered that about 22% of all ED visits and
14% of all hospitalizations were made four or more times. We found that 203,354 of all ED visits (or
4.14%) were made more than four times for
ACSCs, which corresponded to $256,862,867 in total charges. A total of 58,307 patients (4.69% of the sample) were frequently hospitalized for
ACSCs, which corresponded to $1,702,865,362 in total charges in 2005 in Florida.
Conclusion: ED utilization and associated charges due to ACSCs and frequent ED use are staggering.
While not every patient in the studied categories could be prevented from seeking ED care, generally speaking, most patients with the studied conditions could be potentially served as outpatients in offices and clinics. Unnecessary ED utilization can be avoided with the appropriate use of a “medical home” for these patients.
Implications for Policy, Delivery or Practice:
Various strategies can be used in order to contain
ED utilization and costs: (1) an increase in the total number of primary care practitioners available for the patients (including Medicaid recipients); (2) redesigning hours of operation of primary care clinics to accommodate a demand in primary health care services during weekends and at night hours;
(3) providing more support to local/county health departments and community health centers in provision of primary care to Medicaid, uninsured, self-pay, and charity population in the state.
Funding Source(s): Florida Agency for Healthcare
Administration
Theme: Quality and Efficiency: Organized
Processes
● The Carbon Footprint of Healthcare in the U.S.
Jeanette Chung, Ph.D.; David Meltzer, M.D., Ph.D.
Presented by: Jeanette Chung, Ph.D., Research
Associate Assistant Professor, Section of Hospital
Medicine, University of Chicago, 5841 South
Maryland Avenue, MC 2007, Chicago, IL 60637,
Phone: (773) 834-8237; Email: jchung1@bsd.uchicago.edu
Research Objective: Healthcare buildings are among the most energy-intensive commercial buildings in the U.S. The aim of this study was to estimate annual carbon dioxide (CO2) emissions from energy use -- i.e., the "carbon footprint" of the
U.S. healthcare service sector.
Study Design: We used data from the U.S. Energy
Information Administration 2003 Commercial
Buildings Energy Consumption Survey to estimate total and mean annual electricity and natural gas consumption in inpatient, outpatient, and nursing facilities for the nation as a whole and by region.
We converted electricity consumption to carbon dioxide (CO2) emissions using year- and regionspecific carbon emissions factors calculated from historical data from the U.S. Department of Energy.
Natural gas consumption was converted to CO2 using a factor of 120.59 lbs of CO2 per 1,000 ft3.
Emission rates per inpatient discharge were calculated using data on hospital utilization from the
2003 National Hospital Discharge Survey.
Population Studied: The population of buildings studied included all non-residential, nonagricultural, non-industrial commercial buildings in the U.S. with an area of 1,000 sq. ft. or more.
Within this population, we compared inpatient, outpatient, and residential nursing facilities to all other non-healthcare commercial facilities.
Principal Findings: The annual energy demand of the average inpatient facility in the U.S. was 30 times higher than that of the average nonhealthcare commercial building. In 2003, inpatient facilities generated over 43 million metric tons of
CO2 at a rate of 1.12 metric tons for every inpatient discharge. Outpatient and nursing facilities each contributed roughly 14 million and 13 million metric tons of CO2, respectively. Together, healthcare facilities comprised roughly 3% of the total commercial building stock in the U.S., but were responsible for 9% of total CO2 emissions from the commercial sector. Healthcare energy demand was highest in the Northeast and the Midwest, but due to regional differences in energy sources used for electricity generation, CO2 emissions were highest in the Midwest and the South, which averaged 1.36 and 1.44 metric tons of CO2 per discharge.
Conclusion: The healthcare service sector is highly energy-intensive. Annual CO2 emissions from electricity and natural gas consumption among inpatient, outpatient and nursing facilities are roughly equivalent to the annual emissions of 15 coal-fired power plants.
Implications for Policy, Delivery or Practice: The
U.S. healthcare sector has made strides in costcontainment, quality improvement, and patient safety over the last few years. We argue that sustainability and environmental impact are also important "bottom line" considerations that warrant additional research and attention. Ecological footprinting may be a useful accounting tool for measuring and tracking the environmental performance of healthcare providers.
Theme: Quality and Efficiency: Measurement
● Finding the High Performing Docs: Physician
Characteristics Associated with Quality of Care
& Care Improvement
Sukyung Chung, Ph.D.; Latha Palaniappan, M.D.,
M.S.; Haya Rubin, M.D., Ph.D.; Harold Luft, Ph.D.
Presented by: Sukyung Chung, Ph.D., Fellow,
Health Policy Research, Palo Alto Medical
Foundation Research Institute, 795 El Camino
Real, Palo Alto, CA 94301, Phone: (650) 853-4763;
Email: chungs@pamfri.org
Research Objective: To understand physician characteristics explaining variation in (1) quality of care measures and (2) changes in quality measure scores in response to a physician-specific pay-forperformance (P4P) program.
Study Design: The Palo Alto Division of Palo Alto
Medical Foundation (PAMF/PAD) has been assessing and periodically reporting physicians’ performance on established quality measures since
2004, using the electronic health records. In 2007,
PAMF/PAD implemented a physician-specific P4P incentive program linking the amount of the incentives (maximum of $5000/year per physician) to the physician’s performance on a subset of these quality measures, including diabetes outcomes
(HgbA1C, LDL, and blood pressure control), asthma controller prescription, preventive screening
(colon cancer, cervical cancer, Chlamydia), tobacco history documentation, and measurement of weight and height. We examined physician characteristics
(age, sex, years of practice, department, location of medical education, and practice site) associated with quality measures scores and improvement in scores during the physician-specific P4P program.
Dependent variables were percent scores (0-100: patients meeting the target/eligible patients)*100) of each physician for each measure. The data were analyzed with multivariate linear and quantile regressions.
Population Studied: The PAMF/PAD is a multispecialty group practice in Northern California. All the primary care physicians in three departments – family medicine, general internal medicine, and pediatrics – in five locations were included (n=167).
All active patients meeting the measure-specific criteria from 2006 to 2007, regardless of insurance type, of each physician were included in the quality measures.
Principal Findings: There was a wide variation in the performance on quality measures across
physicians (e.g. 8%-100% for HgbA1C control,
11%-73% for colon cancer screening). However, there was no significant correlation across measures within a physician (e.g. HgbA1C control score vs. colon cancer screening score of physician
AAA). In multivariate analysis, being a woman
(+3%) and years of practice (+0.2%/year) were significant predictors of quality scores (both p<0.01). However, these two predictors were no longer significant after controlling for previous year’s score, which was the strongest predictor
(p<0.01). Individual physicians’ scores were fairly consistent over time, with little difference in the improvement with the implementation of the P4P program between physicians whose scores were in the upper versus in the lower quantile for each measure.
Conclusion: Some physician characteristics
(female sex and practice years) are associated with better scores in quality -measures. Physicians’ scores for established quality measures were consistent over time and changed little in response to small financial incentives. No correlation across quality measures suggests it may be difficult to identify uniformly high performing physicians.
Implications for Policy, Delivery or Practice: In primary care practice, quality of care improves slightly with experience. However, physician practice style is fairly consistent over time, and relatively small financial incentives to individual physicians seem to have limited incremental effects on performance. In the primary care setting, a physician’s performance on one quality measure does not reflect performance on other quality measures. A wide variety of quality measures may be needed to comprehensively assess physician performance in future studies.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Policies and
Incentives
● Inter-State Patient Migration Patterns: A Case
Study of Illinois
Kyusuk Chung, Ph.D.; Duckhye Yang, Ph.D.; Jun
Hyup Lee, Ph.D.
Presented by: Kyusuk Chung, Ph.D., Associate
Professor, Department of Health Administration,
Governors State University, 1235 West Cottage
Place, Chicago, IL 60607, Phone: (708) 534-4047;
Email: k-chung@govst.edu
Research Objective: Determining the projected need for hospital services based on experienced demand for services in an area is fundamental for developing state health plans serving as a statewide policy blueprint for future resource allocation decisions. A recent review of projection methodologies used by state planning agencies revealed that no state methodology takes into account inter-state migration patterns. The objective of the study was to examine the extent to which inter-state migration occurs and to test if an adjustment for inter-state migration is needed in determining the projected need for hospital medical/surgical beds.
Study Design: Based on the year 2002 discharge data from hospitals in Illinois and its 6 neighboring states (Indiana, Kentucky, Missouri, Iowa, Michigan and Wisconsin), a matrix of 41 planning areas of residence by 41 areas of hospitalization was constructed to systematically calculate in-migration number and out-migration number. The in-migration number is defined as the number of patients from outside the planning area receiving services at area hospitals; out-migration is the number of area residents receiving services at hospitals outside the planning area.
Population Studied: Illinois resident discharges in
2002 from medical/surgical (MS) and Obstetrics
(OB) beds of hospitals in six neighboring states:
Indiana, Kentucky, Missouri, Iowa, Michigan and
Wisconsin, in addition to discharges (regardless of origin of residence) from Illinois hospitals.
Principal Findings: Illinois is primarily a net outmigration state, with 4 times greater (for MS) and
3.2 times greater (for OB) out-migration numbers than in-migration numbers. In region-wide estimates, downstate Illinois is a notable outmigration area. The magnitude of out-migration to neighboring states was largest among areas that share boundaries with a neighboring state. As many as eight areas had more than 20% of their resident
M-S discharges from the hospitals in the neighboring six states. Six areas had more than
15% of resident OB discharges from hospitals in the neighboring six states.
Conclusion: The present methodology that Illinois uses to project acute care hospital beds does not take into consideration the fact that a substantial portion of Illinois residents receive care out of the
State. Using extensive data from neighboring states, we verify many Illinois planning areas are
“out migration” areas, and the days of care received at out-of-state hospitals are not re-allocated back to the planning area of origin in Illinois.
Implications for Policy, Delivery or Practice:
“Recapturing” a portion of this extensive out-of-state migration into the planning areas where patients reside may be a reasonable consideration.
However, there is little agreement on why migration occurs. Some argue that residents in out-migration areas have to migrate due to lack of hospital beds.
Others argue that migration is a choice. Without data across areas differentiating between voluntary/rational migration and involuntary/forced migration, it is unclear at this time as to what an appropriate adjustment factor should be for interstate migration patterns.
Theme: Public Health
● Barriers to Diabetes Self-Management among
Rural & Urban Patients
Elizabeth Ciemins, Ph.D., M.A., M.P.H.; Betty
Mullette, M.S.N.; Patrician Coon, M.D.
Presented by: Elizabeth Ciemins, Ph.D., M.A.,
M.P.H., Research Director, Center for Clinical
Translational Research, Billings Clinic, 2800 Tenth
Avenue, P.O. Box 37000, Billings, MT 59107,
Phone: (406) 238-5724; Email: eciemins@billingsclinic.org
Research Objective: To identify and quantify healthcare provider-perceived barriers to selfmanagement for adults with uncontrolled type 2 diabetes vascular risk factors, and enrolled in a nurse practitioner-led diabetes self-management team intervention.
Study Design: A pre-post-intervention nested cohort design identified barriers to diabetes selfmanagement and tracked their progress to resolution for up to two years post-intervention. This study was nested in a larger study that evaluated the effectiveness of a nurse practitioner-led team approach to managing diabetes in urban and rural adult patients. Rural patients were managed using telemedicine. Specific barriers were classified into the following broad categories: Internal Physical,
External Physical, Educational, Psychological, and
Psychosocial. Provider-perceived barriers were identified through paper chart and electronic medical record review. The nesting of this study within a larger randomized trial enabled casecontrol comparisons of barrier category and proportion of barriers resolved.
Population Studied: Nine hundred and ninety-two
(992) barriers were identified and tracked over a two-year period in 61 urban and 60 rural intervention patients, and 46 control patients residing in Eastern Montana and Northern
Wyoming. All patients had at least one uncontrolled diabetes vascular risk factor, i.e., HbA1C > 7%, blood pressure > 130/80, or LDL cholesterol > 130, at time of study enrollment.
Principal Findings: Sixty-two percent of all identified barriers were educational (e.g., low diabetes knowledge) or psychological (e.g., depression, lack of motivation, emotional issues).
Low diabetes knowledge, limited range of available services and acute medical conditions were the most likely barriers to be resolved. Least resolvable barriers included obesity, lack of family support, and depression. Group differences in barrier identification and resolution were identified. The average number of barriers identified (resolved) per patient was 7(5.4), 6(3.3) and 4(2.5) for the urban intervention, rural intervention, and control groups, respectively (p<01). Significant group differences in time to resolve a barrier were identified; within one year of barrier identification, 62%, 55%, and 44% of barriers were resolved in the urban intervention, rural intervention group and control group, respectively (p=0.006).
Conclusion: The implementation of a nurse practitioner-led intensive team approach to diabetes management had a positive effect in identifying and addressing barriers to patient self-management of diabetes. This team model of care was most successful in resolving educational and motivational barriers, and less successful in resolving depression, emotional issues, work demands, and general health beliefs. Further studies are necessary to validate these findings and determine the effect of barrier resolution on diabetes vascular risk factor control. Additional studies are warranted to identify strategies to address the most difficult to resolve barriers.
Implications for Policy, Delivery or Practice:
This study demonstrates how an intensive diabetes management team approach may positively intervene to address barriers that prevent patients from successfully self-managing their diabetes.
Difficult barriers to overcome are described suggesting the need for alternative models of diabetes care to address such areas as behavioral health and psychosocial support.
Funding Source(s): NIDDK
● Clinical Inertia in the Management of Diabetes in the Rural Setting
Elizabeth Ciemins, Ph.D., M.A., M.P.H.; Patricia Jay
Coon, M.D.; Hillary Harris, M.S., M.P.H.
Presented by: Elizabeth Ciemins, Ph.D., M.A.,
M.P.H., Research Director, Center for Clinical
Translational Research, Billings Clinic, 2800 Tenth
Avenue, P.O. Box 37000, Billings, MT 59107,
Phone: (406) 238-5724; Email: eciemins@billingsclinic.org
Research Objective: To characterize office practice differences in the management of diabetes among primary care clinics in rural Montana and
Wyoming, focusing on provider behavior toward patients with borderline uncontrolled vascular risk factors.
Study Design: A cross-sectional study examined differences in control of risk factors in patients with type 2 diabetes, i.e., HbA1C, blood pressure, and
LDL cholesterol, as well as physician behaviors , e.g., the establishment of an active or passive plan, when a specific risk factor was considered at a borderline level of control. Comparisons were made for borderline levels of risk factor control, i.e., ‘grey zone’ levels defined as 7% < HbA1C < 8%, 130/80
< blood pressure<140/90, and 100 < LDL < 130.
Population Studied: Eight rural clinics from
Eastern Montana and Northern Wyoming with 908 patients were included in the study. All patients had a diagnosis of type 2 diabetes mellitus (ICD-
9=250.x) and were seen by a primary care provider in a rural clinic between 7/2004 and 6/2005. Most recent laboratory values for HbA1C and LDL and the most recent blood pressure reading were examined.
Principal Findings: Patients with borderline levels of risk factor control were significantly less likely to have had a provider acknowledge that risk factor and develop an active treatment plan (p<.01). This
‘grey zone’ effect was greater for blood pressure and lipids than for HbA1C, for which more providers implemented an active plan, regardless of how close the laboratory value was to the ADA established treatment guidelines. Clinical inertia, or the lack of acknowledgement and treatment of patients whose values fell in the ‘grey zone,’ was present at all sites with one exception. The differentiating factor of the one positive deviant clinic site was primary care provider turnover, i.e., higher staff turnover was associated with increased identification and treatment of patient risk factors, regardless of proportion of patient laboratory values/clinical readings falling in the ‘grey zone’
(p<.05).
Conclusion: Clinical inertia, or lack of proactive treatment for diabetes patients whose risk factor values were considered borderline, exists in rural primary care clinics, as has been previously demonstrated in the urban setting. However, factors that necessitate another provider reviewing a patient’s case, e.g., staff turnover, may break the cycle of inertia. In other words, a patient examination with a fresh set of eyes may result in the necessary action to improve control of vascular risk factors in patients with diabetes.
Implications for Policy, Delivery or Practice:
While provider continuity has long been touted as necessary for good patient care, this study suggests that a break in continuity may be necessary to interrupt inert cycles of care characterized by provider inaction. This may require re-thinking current care models to include an increase in shared decision-making in the care of patients with chronic conditions and supports the creation of a medical home for patients, in which a team of professionals becomes responsible for an individual patient’s care.
Funding Source(s): AHRQ, NIDDK
Theme: Prevention and Treatment of Chronic
Illness
● Market Impact of Specialty Hospitals: A Study of the Profitability of General Short-Term Acute
Care Hospitals Post Market Entry of Specialty
Hospitals
Robert James Cimasi, M.H.A., A.S.A., C.B.A.,
A.V.A., C.M., A.A.
Presented by: Robert James Cimasi, M.H.A.,
A.S.A., C.B.A., A.V.A., C.M., A.A., President, Health
Capital Consultants, 9666 Olive Boulevard, Suite
375, Saint Louis, MO 63132, Phone: (314) 994-
7641; Email: info@healthcapital.com
Research Objective: The purpose of this study was to determine whether the entrance of a specialty hospital(s) into a given core-based statistical area (CBSA) negatively impacted the financial viability of existing general short-term acute care hospitals (general hospitals) in the same
CBSA, and whether a change in profitability was comparable to that a nationally representative group of general hospitals. The association between bed utilization for specialty versus general hospitals and general hospital profitability was also examined.
Study Design: Financial indicators of profitability were utilized from a custom data set built by the
American Hospital Directory (AHD) for the time period spanning 1/1/1997 through 12/31/2005.
Profitability was measured via four representative variables: (1) operating income to beds; (2) operating income to discharges; (3) net income to beds; and, (4) net income to discharges, representative of the operating and overall profitability of facilities as a function of facility capacity and patient volume. A one-tailed paired ttest was conducted to determine whether there was a significant decrease in general hospital profitability from the period prior and subsequent to the eligible specialty hospital(s) opening in a given
CBSA. To determine differences between profitability of general hospital(s) in sample CBSAs and profitability trends for the national market of general hospitals (the “comparison group”) during the study period, the general hospitals within sample CBSAs were grouped according to “time periods,” determined by the opening year(s) for specialty hospital(s) located within each sample
CBSA (n = 18). Financial data was averaged across all general hospitals for the years prior and subsequent to the opening of specialty hospital(s) within a given time period for both the sample group and comparison group of general hospitals, respectively, and the difference in profitability was examined using a two-tailed one-sample paired ttest.
Population Studied: The unit of analysis for this study is a general hospital(s) [profitability] within a given CBSA. Eligible general hospitals were defined as hospitals operating within a recognized
CBSA containing one or more eligible specialty hospital(s) and excluding government hospitals, non-surgical specialty hospitals, and those general hospitals lacking at least one year of operating data available prior and subsequent to the opening of a specialty hospital(s) located within a given CBSA (n
= 405). All otherwise eligible general hospital(s) located in a CBSA not containing an eligible specialty hospital(s) were included in the comparison group (n = 2,120). Eligible specialty hospitals were identified from the Centers for
Medicare and Medicaid Services, AHD, and the
Physician Hospitals of America and were required to be located in a recognized CBSA with an opening date between 1/1/1998 and 12/31/2004, and have available financial data from AHD (n =
74).
Principal Findings: This analysis indicated that only 14 of 41 CBSAs (34.2%) indicated a statistically significant decrease in profitability; of those, 9 CBSAs (22.0%) showed significant results for only one or two of the four profitability indicators utilized in this analysis. In a comparison of relative capacity, for general hospitals indicating a significant decrease in profitability, 0.054-8.63% of inpatient beds were controlled by specialty hospital(s) within those CBSAs; this percentage ranged from 0.042-20.36% for general hospitals in
CBSAs not indicating a significant decrease in profitability for this analysis. Consequently, there appears to be no explicit trend in the data to support the hypothesis that a greater specialty hospital(s) capacity would be associated with decreased general hospital(s) profitability in a given
CBSA. In a subsequent analysis, a total of 22 sample CBSAs (53.7%) indicated that profitability significantly differed from that of the comparison group of general hospitals. Additionally, of the 14
CBSAs indicating a significant decrease in profitability in the previous analysis, all but one
(92.9%) indicated a significant difference in one or more measures of profitability when compared to profitability of the comparison group, such that no definitive conclusions may be offered regarding specialty hospital(s) impact on general hospital profitability, as external confounding factors (i.e., those other than the potential impact of eligible specialty hospital(s)) may have contributed to either inflated or reduced levels of profitability as seen in this analysis.
Conclusion: Pursuant to results indicating that there is inconclusive evidence that the entrance of a specialty hospital(s) has a significant negative effect on the financial viability of a general hospital(s) within a CBSA, further study is needed to provide more definitive conclusions regarding: (1) the longterm impact or effect of a specialty hospital(s) on a general hospital(s) profitability; (2) how to delineate particular market characteristics relevant to specialty hospital(s) impact on the susceptibility of a general hospital(s) financial viability; and, (3) the relative cost-benefit relationship regarding quality, efficiency, and beneficial outcomes of specialty hospitals in comparison to general hospitals to the overall continuance of care.
Implications for Policy, Delivery or Practice: In light of the ongoing legislative agenda related to restricting physician ownership of hospitals, the benefits of this research and the resulting informed discourse will be critical for legislators and health policy decision-makers in resolving policy issues related to physician ownership of specialty hospitals, and in responding to claims of advocacy groups pursuing market protection agendas.
Theme: Health Care Markets and Competition
● Risk of Hospitalization Among Patients
Receiving Specialized Outpatient HIV Services:
A New Risk Stratification Tool
Christopher Clark, M.P.A.; Ruben Amarasingham;
Tim Swanson; Kristin Snackey; Ying Ma; Gary
Sinclair, M.D.
Presented by: Christopher Clark, M.P.A., Analyst,
Center for Innovation & Knowledge Translation,
Parkland Health & Hospital System, Support
Building B, Room: G108, 5201 Harry Hines
Boulevard, Dallas, TX 75235, Phone: (214) 590-
8069; Email: c1clar@parknet.pmh.org
Research Objective: In order to allocate resources to HIV patients with highest need and reduce wait times to HIV specialty clinics, urban safety net health systems are attempting to carefully transition patients with HIV from specialty to primary care sites. This task has been hampered by a lack of sufficiently accurate risk stratification tools for community dwelling patients with HIV. We sought to develop an electronic model that would enable a public health system to inexpensively identify HIV patients at lowest risk for hospitalization over a 6 month period.
Study Design: A model for 6 month hospitalization risk was derived retrospectively using electronic data. Patients were randomized and split equally into a derivation and validation set, n=2,181 in each. We assessed severity of illness using the following electronic laboratory data to assign clinical risk: CD4 count, HIV viral load, creatinine, ALT,
AST. Race, sex, payer, no. of emergency contacts, history of illicit drug use or mental illness, no. of home address changes, emergency department-
ED visits, hospital admissions, and proximity to HIV specialty and primary care clinics were assessed through multivariable logistic regression. 6 month risk was calculated from time of first contact in
2007. We internally validated model coefficients by applying bootstrap sampling to candidate variables; variable selection was performed using step-wise regression and the Bayes Information Criterion. We assessed model fit through calibration, discrimination, and re-classification and assessed overall model improvement using the Area Under
Curve- AUC, for the Receiver Operating
Characteristic- ROC, Hosmer-Lemeshow test, and integrated discrimination improvement index. The final model was externally validated using 2,181 patients in the validation set.
Population Studied: 4,362 adult HIV patients who were seen at least once in a HIV specialty clinic within an urban academic safety net health system between January 2007 and October 2008.
Principal Findings: The unadjusted 6 month hospitalization rate was 11%. Significant predictors of stability, p<.05, after multi-variable adjustment included CD4 > 250- Odds Ratio: 0.63, RNA viral load <400- 0.47, no prior screening for cocaine use-
0.68, no prior hospitalizations- 0.28, and no prior
ED visits- 0.62 over the last year. The AUC for the derivation and validation model was 0.75- 95% CI:
0.71, 0.78 and 0.76- 95% CI: 0.72, 0.79
respectively. The following hospitalization rates reflect the ranked distribution of model prediction scores: 1st quartile: 2.2%, 2nd quartile: 6.79%, 3rd quartile: 9.72%, and 4th quartile: 23.99%. These groups were significantly different from each other, with a chi-squared value=164.03, p < 0.0001.
Conclusion: A set of automated clinical and nonclinical indicators can predict 6 month risk for hospitalization among community dwelling HIV patients in a large academic safety net hospital.
Implications for Policy, Delivery or Practice:
Electronically derived risk models may allow public health systems to identify HIV patients that could be successfully migrated to primary care sites without requiring expensive chart review.
Theme: Prevention and Treatment of Chronic
Illness
● The Effect of Relative Weight Status On
Education Accumulation
Timothy Classen, Ph.D.
Presented by: Timothy Classen, Ph.D., Assistant
Professor, Department of Economics, Loyola
University Chicago, 1 East Pearson Street,
Chicago, IL 60611, Phone: (312) 915-6184; Email: tclass1@luc.edu
Research Objective: This project estimates the role of relative weight status during high school in educational attainment. The potential underlying mechanisms for the previously identified asymmetric effect of obesity on educational attainment across genders are studied. Using a longitudinal, school-based survey of a representative population of students in 80 high schools, a student’s relative weight status within the distribution of Body Mass Index (BMI) in their high school is measured. This allows for addressing the question of whether having weight levels at higher points in one’s local distribution of weight status is associated with lower levels of education accumulation.
Study Design: National z-scores for a student’s
BMI relative to nationally representative measures of BMI growth by age and gender are first calculated to provide uniform measures of comparable body weights within a high school.
Then a local z-score is calculated based on the student’s BMI z-score relative to the population of students within their high school. Limiteddependent variable models for whether a student attended college are then estimated via probit regressions with local z-score of BMI relative to the in-school distribution of body weights and measures of ability, maternal education, family resources and demographic variables serving as control variables.
Population Studied: Data on 4,947 female and
4,881 male respondents from multiple waves of the
National Longitudinal Survey of Adolescent Health
(Add Health) are used to measure within-school relative weight status during high school (Wave I) and education accumulation seven years after the initial in-school measures (Wave III). Participants from 80 high schools were initially interviewed during high school in 1994-05 and follow-up interviews were conducted in 2001-02.
Principal Findings: Females with relatively higher weight levels (compared to other females within their high school) are less likely to attend college than their peers, with the largest effect size for obese females. Much weaker relationships between weight status and education accumulation is found among males, with no significant differences in the likelihood of college attendance for males at higher points in the local distribution of
BMI. However, contrary to previous findings, there exists some evidence that obese males with relatively higher BMI levels may be less likely to attend college.
Conclusion: Asymmetric consequences between genders of elevated weight levels for education accumulation are found with females at higher points in the local distribution of BMI less likely to attend college. The largest effect size is found for obese females. Weaker evidence is found for reduced likelihood of college attendance of obese males, but no significant differences are found for local measures of relative BMI among males.
Implications for Policy, Delivery or Practice:
These results for investments in education are consistent with the evidence that obese women face a wage penalty relative to their non-obese peers, while obese males do not (since such disparities may provide reduced incentives for human capital investments among females with relatively high weight levels). Programs that deal with issues of self-esteem or academic performance among obese students in high schools should be targeted toward females given the substantially larger disparity in long-run educational attainment.
Funding Source(s): Loyola University Chicago
Internal Grant
Theme: Obesity Prevention and Treatment
● Which Children with Asthma are at Greatest
Risk for Being Overweight?
Janet Coffman, M.A., M.P.P., Ph.D.
Presented by: Janet Coffman, M.A., M.P.P., Ph.D.,
Assistant Adjunct Professor, Philip R. Lee Institute for Health Policy Studies, University of California,
San Francisco, 3333 California Street, Suite 265,
San Francisco, CA 94118, Phone: (415) 476-2435;
Email: Janet.Coffman@ucsf.edu
Research Objective: Persons with asthma are more likely to be obese. Furthermore, several studies suggest that children who have asthma and who are also overweight are more likely to have poor health outcomes. To date, little research has been conducted to identify the children at greatest risk for co-morbid asthma and overweight. This
study uses data from a survey conducted in four states to assess whether the demographic characteristics of children with asthma are associated with being overweight.
Study Design: Data were obtained from the
National Asthma Survey. This cross-sectional random-digit-dial telephone survey was conducted in 2003 by the National Center for Health Statistics
(NCHS) in four states – Alabama, California, Illinois, and Texas. For children, data were obtained from parents or other adults knowledgeable about the child’s health. Demographic variables included age, sex, race/ethnicity, and annual household income.
All analyses were performed in STATA Version 9 using sampling weights provided by NCHS.
Population Studied: The survey response rate was 48.5%. Complete data on all variables analyzed were available for 1,562 subjects.
Children were classified as having asthma if, within the past three years, they had asthma symptoms, had taken asthma medication, or had visited a physician or other health professional regarding their asthma. A standard criterion (BMI > 95th percentile) was used to classify children as overweight or not overweight. BMI was calculated from data on children’s height and weight reported by the adult completing the survey on the child’s behalf that were not independently verified.
Principal Findings: Twenty-eight percent of sampled children with asthma had a BMI above the
95th percentile. Adolescents with asthma were less likely to be overweight than children under age 12
(OR 0.38, 95% CI 0.27-0.52) and girls were less likely to be overweight than boys (OR 0.67, 95% CI
0.48-0.93). Non-Hispanic white children were less likely to be overweight than children who were
Hispanic, non-Hispanic black, or from other/multiple racial/ethnic groups (OR 0.40, 95% CI 0.29-0.55).
Children from high-income households (> $75,000) were less likely to be overweight than children from low- and middle-income households (OR 0.52, 95%
CI 0.36-0.75). In a multivariate regression that incorporated all four demographic characteristics, the associations between the risk of being overweight and age, race/ethnicity, and household income remained statistically significant. Sex was not statistically significant in the multivariate analysis.
Conclusion: Findings from this study suggest that
28% of children with asthma are overweight, which is greater than the percentage of all children in the
United States who are overweight (17% per
NHANES 2003-2004). Racial/ethnic minority children and children from low-income families who have asthma are at increased risk of also being overweight.
Implications for Policy, Delivery or Practice:
These findings suggest that health professionals who care for children with asthma from racial/ethnic minority and low-income population should monitor their BMI and counsel parents regarding prevention and treatment of obesity. In addition, these findings suggest a need for holistic policies and interventions that address the multiple health care needs of vulnerable populations.
Theme: Obesity Prevention and Treatment
● Access to Health Care in a Consumer Driven
Market for Privately Insured Adults Aged 18-64
Years: Estimates from the 2007 National Health
Interview Survey
Robin Cohen, M.S., Ph.D.
Presented by: Robin Cohen, M.S., Ph.D.,
Statistician, Division of Health Interview Statistics,
National Center for Health Statistics, 3311 Toledo
Road, Hyattsville, MD 20782, Phone: (301) 458-
4152; Email: rcohen@cdc.gov
Research Objective: National attention to consumer-directed health care has increased following the enactment of the Medicare
Prescription Drug Improvement and Modernization
Act of 2003, which established tax-advantaged health savings accounts. Consumer-directed health care enables individuals to have more control over when and how they access care, what types of care they use, and how much they spend on health care services. This study investigates selected health care access measures among adults under 65 years of age with private health insurance. Access measures include flu/pneumoccal vaccinations, visits to doctors’ offices, dental visits, use of medical specialists, unmet healthcare needs due to cost, and barriers to receiving care.
Study Design: The source of data is the 2007
National Health Interview Survey (NHIS), a household survey nationally representative of the civilian non-institutionalized population. Data are derived from the Family Core component, which collects information on all family members in each household, and the Sample Adult component of the
NHIS, where one adult is randomly selected per family and asked additional questions.
Population Studied: Non-institutionalized privately insured adults aged 18-64 years.
Principal Findings: In 2007, 16.9% of adults18-64 years of age with private health insurance, were enrolled in a high deductible health plan (HDHP),
4.3% were enrolled in a consumer-directed health plan (CDHP), and 14.4% were in a family with a flexible spending account for medical expenses
(FSA). Among adults 18-64 years of age, more than 40% with directly purchased private coverage were enrolled in a HDHP compared with approximately 15% with private coverage obtained through the workplace or union. Working-age adults with a HDHP were less likely to have a usual source of medical care than those with private health insurance who were not enrolled in a HDHP.
More than 19% of working- age adults with a HDHP had an unmet medical need compared with almost
13% of those without a HDHP. Working-age adults with a CDHP were more likely to have a non-
medical barrier to care than those who were not enrolled in a CDHP. Working-age adults with private health insurance living in a family with a
FSA were more likely to have a usual source of medical care, and in the past year, visited a doctor, dentist, eye doctor, or medical specialist than those who were not living in a family with a FSA. More than 11% of privately insured working-age adults living in a family with a FSA had an unmet medical need compared with more than 14% of privately insured working-age adults who did not have a
FSA.
Conclusion: Working-age adults enrolled in a
HDHP were more likely to experience barriers to access than those who were in a private plan that was not a HDHP. The availability of a FSA increased health care utilization.
Implications for Policy, Delivery or Practice:
Although HDHPs, CDHPs, and FSAs are still a small share of the private health insurance market, their uptake has been increasing. The NHIS can play an important role in monitoring the uptake of these plans and the impact of them on access to care and utilization behavior.
Theme: Consumer Choices in Health Care
● Translating Behavioral Interventions into
Primary Care Practice
Deborah Cohen, Ph.D.; Bijal Balasubramanian,
M.B.B.S., Ph.D.; Nicole Isaacson, Ph.D., M.F.S.;
Rebecca Etz, Ph.D.; Elizabeth Clark, M.D., M.P.H.
Presented by: Deborah Cohen, Ph.D., Assistant
Professor, Family Medicine, University of Medicine
& Dentistry, New Jersey-Robert Wood Johnson
Medical School, 1 World's Fair Drive, Somerset, NJ
08893, Phone: (732) 743-3239; Email: cohendj@umdnj.edu
Research Objective: Health risk behaviors contribute significantly to morbidity and mortality in the US. Primary care is an important venue for addressing these risks due to the high prevalence of patients seen who need behavior change. There a is paucity of information on how best to integrate behavioral counseling into the primary care setting.
The purpose of this research is to describe and evaluate different approaches for integrating behavior counseling into primary care.
Study Design: A cross-project process evaluation of the Prescription for Health – P4H - program was conducted. The evaluation used extensive qualitative data to describe each project’s approach for integrating health behavior counseling into primary care practice. Using both qualitative and quantitative data, a cross-project comparative analysis was conducted to evaluate each intervention approach on two dimensions: success at implementing the intervention, and improvements in patients’ behaviors. Implementation success was assessed using qualitative data and a mixed method survey administered to project PIs.
Improvements in patients’ behaviors were assessed from each project’s published or unpublished findings.
Population Studied: 10 projects were funded through round 2 of the P4H program, 9 of which are included in this analysis. Each project was conducted in a practice-based research network and implemented a different intervention to improve diet, physical activity, risky drinking and smoking counseling.
Principal Findings: All of the interventions attempted to link primary care practices with resources to provide intensive behavioral counseling to patients. Four different approaches for linking practices with counseling resources were identified, and varied on the following dimensions:
1. delivery of brief counseling in the practice; 2. location and type of intensive counseling; 3. type of linkage between practice and counseling resource; and 4. how information was exchanged among practice, resource and patient. Approaches 1 and 2 included brief clinician counseling and referral to a behavioral counseling resource and were successfully implemented in practices. Of these,
Approach 1 fostered a collaborative relationship between practice and external counseling resources and included a way to easily exchange patient information. Two of the three projects in
Approach 1 reported significant improvements among patients in all four risky behaviors. Projects in Approach 2, which did not foster the same level of collaboration and information exchange, reported non-significant trends toward improvements in patients’ health behaviors. Approaches 3 and 4 attempted to by-pass clinician brief counseling and/or by-pass the practice entirely. These approaches encountered implementation barriers.
Improvements in patient behavior were not assessed due to small sample sizes.
Conclusion: Linking approaches that involve brief clinician counseling, establish strong collaboration and information exchange among practice, counseling resources and patients may be most effective in improving patients’ behaviors in primary care.
Implications for Policy, Delivery or Practice:
Primary care practices should not be bypassed when helping patients with behavior change.
Reimbursement reform is needed to pay clinicians for counseling patients, and infrastructure support is needed to develop efficient ways of connecting and fostering communication among primary care practices, patients and behavior counseling resources.
Funding Source(s): RWJF, AHRQ
Theme: Organizational Performance and
Management
● A Viable Alternative to Private Medicaid
Managed Care Models: Performance of the
Oklahoma Primary Care Case Management
Program in Controlling Preventable
Hospitalizations
Margaret Colby, M.P.P.; Samuel Simon, Ph.D.;
James Verdier, J.D.
Presented by: Margaret Colby, M.P.P., Health
Research Analyst, Health Research, Mathematica
Policy Research, Inc., 600 Maryland Avenue, SW,
Suite 550, Washington, DC 20024, Phone: (202)
264-3481; Email: mcolby@mathematica-mpr.com
Research Objective: To compare the care provided in two different Medicaid managed care models, we evaluated the impact of transitioning non-elderly Medicaid enrollees in Oklahoma from private managed care organizations (MCOs) to a state-run primary care case management (PCCM) program on the rate of preventable hospitalizations using a difference-in-difference approach.
Study Design: We linked Medicaid eligibility data from the state Medicaid agency and inpatient discharge records from the state Department of
Health for 2003 and 2006 to identify hospitalizations among enrollees in Oklahoma’s Medicaid managed care waiver program. Software from the Agency for
Healthcare Research and Quality was used to calculate standardized rates of preventable hospitalizations per 100,000 beneficiaries enrolled in MCOs (operating in urban areas during 1995-
2003), and the PCCM program (operating in rural areas since 1996 and statewide since 2004). We applied difference-in-difference methodology to estimate the effect on preventable hospitalizations of transitioning urban enrollees from MCOs to the
PCCM program in 2004. Rural enrollees, who were continuously covered by the PCCM program, served as the comparison group. Weighted logistic regression models for adults (four categories of preventable hospitalizations: diabetes, respiratory, cardiovascular, and acute conditions) and children
(four specific types of preventable hospitalizations: asthma, gastroenteritis, diabetes, and urinary tract infections) were run separately. Regression controls included age, sex, and county-level estimates of asthma and diabetes prevalence, ethnicity, and physician availability.
Population Studied: Enrollees up to age 64 in
Oklahoma’s Medicaid managed care waiver program in 2003 (n=442,413) and 2006
(n=477,359).
Principal Findings: Oklahoma’s managed care waiver covers roughly four times as many children as adults, so children accounted for 42 percent of all preventable hospitalizations in the covered population. Two-thirds of all preventable hospitalizations were admitted via the emergency room, nearly twice the rate observed for other hospitalizations. Among adults in both the transition and comparison groups, the rate of preventable hospitalizations in all categories except diabetesrelated conditions decreased between 2003 and
2006. The largest reductions were observed for admissions related to respiratory conditions
(decrease of 39% and 23% for the transition and comparison groups, respectively). Increases in preventable hospitalizations occurred for three outcomes among children who transitioned from
MCO to PCCM, but only the rate for gastroenteritis was significant (p=0.03). Transition from MCOs to the PCCM program was not associated with changes in the rate of preventable hospitalizations for respiratory, cardiovascular, or acute conditions for adults, or for most conditions for children.
However, the transition was associated with higher rates of preventable hospitalization for diabetesrelated conditions among adults (odds ratio=1.60, p=0.04) and for asthma among children (odds ratio=1.39, p=0.02) than those expected if the management model had not changed.
Conclusion: The state-run PCCM program in
Oklahoma generally performed as well as private
MCOs in controlling preventable hospitalizations among Medicaid beneficiaries. However, findings for preventable hospitalizations related to diabetes and asthma suggest the need for continued PCCM program improvements.
Implications for Policy, Delivery or Practice:
Evidence from Oklahoma demonstrates that PCCM models can be a viable tool for states to manage their Medicaid population without increasing preventable hospitalizations relative to the care provided by Medicaid MCOs.
Theme: Medicaid, SCHIP and State Health Reform
● Trends in Hospital Compare Reporting for
Acute Myocardial Infarction Patients in U.S.
Rural Hospitals, 2004-2007
Margaret Comins, M.P.A.; Elizabeth Pathak, Ph.D.;
Barbara Langland Orban, Ph.D.
Presented by: Margaret Comins, M.P.A.,
Student/Research Assistant/Teaching Assistant,
Health Policy & Management, University of South
Florida, 13201 Bruce B Downs Boulevard, MDC56,
Tampa, FL 33612, Phone: (813) 313-7732; Email: mcomins@health.usf.edu
Research Objective: Hospital Compare is a voluntary public reporting program, created with the intent of improving hospital quality of care.
Hospitals report on five ‘starter’ process of care scores for myocardial infarctions (AMI). Hospitals that are low performing, or do not report scores, are penalized financially. Rural populations are generally less healthy, less educated, more likely to be uninsured, and more expensive to treat than other populations; consequently rural hospitals are the most likely to have low or negative profit margins. Rural hospitals (RHs) may not have the funds to initiate or to continue their participation in public reporting quality improvement programs.
Study Design: The 5 starter set process of care scores for AMI for all U.S. hospitals were obtained from CMS Hospital Compare for the years 2004-
2007. Data were merged with urban-rural continuum codes by county and collapsed into three categories: metropolitan, non-metropolitan, and completely rural. Generalized linear models were used to compare the average score of each process of care measure by rurality, and to compare the average of the 5 AMI process of care measures.
Population Studied: AMI patients over the age of
18 treated at hospitals reporting to the Hospital
Compare program.
Principal Findings: The number of reporting hospitals increased 9% over the study period, with those in completely rural counties increasing 31%.
Higher scores for each of the measures were found in all 3 locations over time. Three of the 5 starter set scores in 2004, and all 5 in 2007, showed a pattern of high scores for metropolitan, lower scores for non-metropolitan, and lowest scores for completely rural. The beta blocker at arrival average score in
2004 was .88 for metropolitan, .79 for non-metro, and .71 for RHs, (F = 140.76, p<.0001), and in 2007 was .94 for metropolitan, .87 for non-metro, and .84 for RHs in 2007, F = 114.88, p<.0001) in 2007.
Rural hospitals scores for aspirin at discharge and ace inhibitor for LVSD fell from 2004-2007; beta blocker at arrival scores had the most dramatic improvement for completely rural hospitals, from .71 in 2004 to .82 in 2007. Metropolitan hospitals scored .88 in 2004, and .92 in 2007; 10 percentage points higher than rural hospitals.
Conclusion: Rural hospitals are participating in the program at an increasing rate, and show overall improvement in their process of care scores, although their scores started at lower values and even with improvements the scores are still lower than metropolitan hospitals.
Implications for Policy, Delivery or Practice: The
Hospital Compare Program appears to work as an incentive for public reporting, but it may not help hospitals that are performing lower from the beginning of the program. These hospitals may benefit from an alternate program where improvement is rewarded rather than rewards for the highest performers.
● Promising Therapies, Prohibitive Costs: A
Qualitative Assessment of the Effects of the
Medicare Part D Doughnut Hole on Access to
Costly Cancer Medications
Leslie Conwell, Ph.D.; Dominick Esposito, Ph.D.;
Daniel Ball, Dr.P.H.; Eric Meadows, Ph.D.; Martin
Marciniak, Ph.D.
Presented by: Leslie Conwell, Ph.D., Researcher,
Mathematica Policy Research, 600 Maryland
Avenue SW, Suite 550, Washington, DC 20024,
Phone: (202) 264-3489; Email:
LConwell@mathematica-mpr.com
Research Objective: This study aims to understand the impact of the Medicare Part D doughnut hole on patients’ financial access to oral targeted cancer therapies and patients’ strategies for handling out-of-pocket costs. Medicare Part B typically covers oral anticancer drugs that have injectable equivalents. Oral targeted therapies are usually covered under Medicare Part D but are costly, sometimes exceeding $100,000 annually.
Study Design: We conducted semi-structured, 30to-90 minute telephone interviews with oncology social workers and nurse practitioners who assist patients in overcoming financial barriers to targeted therapies. We analyzed interview notes to identify major themes across respondents.
Population Studied: Eleven social workers and two nurse practitioners from six United States cities.
Respondents were affiliated with large cancer centers, except one employed by a large private oncology practice.
Principal Findings: Respondents reported targeted therapies’ high costs, plus other cancerrelated expenses, pose financial barriers for patients. Respondents frequently encounter patients who question the affordability of targeted therapy copayments well before encountering the
Part D doughnut hole, of which many are unaware.
Patients who are aware of the doughnut hole are typically surprised that a 30-day supply of some targeted therapies places them in the doughnut hole. Some patients cease therapy upon reaching the doughnut hole. Although cost-sharing decreases when patients reach catastrophic coverage, respondents reported that these copayments are still burdensome for many. Social workers noted increasingly spending time helping patients obtain financial assistance for targeted therapies rather than providing psychosocial support. Respondents encourage patients to apply for the Part D Low Income Subsidy (LIS), which has facilitated access to targeted therapies among the poorest beneficiaries. Otherwise, respondents assist low-income patients in applying to copayment foundations or manufacturer patient assistance programs (PAPs), though securing assistance sometimes depends on a patient’s diagnosis and medication. Additionally, respondents may refer patients to community resources to assist with other living expenses. Respondents report that many patients are reluctant to ask family for financial assistance. Respondents reported that while some patients forego treatment, many patients eventually initiate targeted therapy treatment if financial assistance is available. Patients may delay or suspend treatment until this assistance is secured, but thereafter respondents thought patients generally were adherent, unless side effects were too burdensome. Self-pay patients draw upon retirement savings, resenting that Medicare does not fully cover their costs and worrying about exhausting their savings and impoverishing their spouse.
Conclusion: Although the LIS appears to have removed medication financial barriers for the lowest
income Medicare beneficiaries with cancer, barriers remain for others. Private foundations provide costsharing assistance, but many patients do not qualify. Respondents reported that patients who secure financial assistance more often adhere to their treatment regimen than those who face financial barriers.
Implications for Policy, Delivery or Practice: The market for targeted therapies is growing rapidly, but financial barriers associated with the Part D coverage gap and limited options for obtaining additional financial assistance may inhibit access to and reduce patient adherence to targeted agents.
Funding Source(s): Eli Lilly and Company
Theme: Medicare
● Adherence to Evidence-Based Guidelines &
Medication Compliance for Multiple Chronic
Diseases in a Medicare Part D Population
Christopher Cook, R.Ph., Pharm.D., Ph.D.; Julie
Priest, M.S.P.H.; Nicole Engle-Nitz, Ph.D.; Ami
Sklar, M.P.H.; Steven Burch, R.Ph., Ph.D.; C. Ron
Cantrell, Ph.D.
Presented by: Christopher Cook, R.Ph., Pharm.D.,
Ph.D., Sr. Manager, Health Outcomes Researcher,
Health Management Innovations, GlaxoSmithKline,
5 Moore Drive, Mail stop: B-4134, Research
Triangle Park, NC 27709, Phone: (919) 483-2834;
Email: christopher.l.cook@gsk.com
Research Objective: This study focused on establishing the level of adherence to nationally recognized treatment guidelines and medication compliance for patients enrolled in a Medicare
Advantage Prescription Drug (MAPD) plan with one of eight common chronic conditions. This is the fourth study in a series to examine the state of chronic disease care in the US for asthma, chronic obstructive pulmonary disease (COPD), diabetes, heart failure (HF), coronary artery disease (CAD), hyperlipidemia, hypertension (HTN) and depression. The program objective is to create a consistent analytic platform across disease states for differing populations including Commercial,
Medicaid, Veterans Health Administration and
Medicare. The intent is to establish comparable benchmarks to assist health plan administrators and payers in identifying areas for improving disease prevention, intervention, and innovation.
Study Design: This retrospective database analysis uses enrollment eligibility along with medical and pharmacy claims from a large US health plan offering Medicare Advantage health care and prescription drug benefits. Patients were identified with at least one condition in 2006 and all analysis was performed for January 1 – December
31, 2007. Disease identification criteria included standard definitions for ICD-9 codes following
National Guidelines and/or HEDIS definitions.
Adherence to treatment guidelines were assessed by disease specific HEDIS quality measures and appropriate medication usage. All data were analyzed overall and by phase of Part D coverage where applicable. Phases defined by Medicare based upon patients’ out-of-pocket expenditures for prescription medication include coinsurance phase, coverage gap phase, and catastrophic phase.
Medication compliance (medication possession ratio) and medication persistence (proportion of days covered) were calculated and utilized an 80% cutoff. All-cause and disease specific utilization and costs were assessed for each condition.
Population Studied: Patients aged 65 or older enrolled in a MAPD plan with evidence of target disease in 2006 and were continuously enrolled during calendar year 2007 were eligible for the study.
Principal Findings: Results to be presented at the conference.
Conclusion: Adherence to treatment guidelines and medication adherence potentially have serious implications to the health of an aging American population.
Implications for Policy, Delivery or Practice:
Identifying opportunities to reduce avoidable costs associated with chronic disease is central to promoting a better quality of life for patients and affordable health care for all Americans. Altering our current behaviors from managing complications of chronic disease to focusing on better disease prevention, early and consistent intervention, and continued innovation across the continuum of care can provide a much better framework for the
National health care system.
Funding Source(s): GlaxoSmithKline
● Back to the Future: Reinstatement of a Mailed
SCHIP Application
Julia Costich, J.D., Ph.D.
Presented by: Julia Costich, J.D., Ph.D., Associate
Professor & Chair, Health Services Management,
University of Kentucky College of Public Health,
121 Washington Avenue, Lexington, KY 40536,
Phone: (859) 257-6712; Email: julia.costich@uky.edu
Research Objective: The reinstatement of a mailed application and outreach for SCHIP and children’s Medicaid after a 7-year hiatus presented a natural state-level experiment that is assessed with regard to enrollment and retention impact as well as unintended consequences.
Study Design: We analyzed data from the first 6 months of mailed application reinstatement using descriptive statistics to identify trends and issues.
We solicited qualitative feedback from parents and community-based providers of application assistance to assess attitudes and identify concerns.
Population Studied: Uninsured children in lowincome families, their parents, and staff involved
with mailed application implementation and processing.
Principal Findings: Alternatives to the traditional face-to-face eligibility determination interview have long been identified as eliminating barriers to enrollment. This finding was reinforced by our state’s response to the reinstated mail option for
SCHIP and children’s Medicaid. Some 4000 mailed applications were received in the first six months, but over half were missing critical elements, including mandatory supporting documentation. A broad range of individuals and groups worked to assist families with the application process, most notably in school-based organizations. Telephone systems experienced overload in more populous counties, frustrating parents who were attempting to connect with caseworkers. Processing delays were reduced through a careful process of problem identification and support. Despite implementation concerns, parents received the policy change allowing mailed applications positively.
Conclusion: Eliminating the face-to-face interview requirement for SCHIP and children’s Medicaid increased the application rate, but a large proportion of parents continued to need assistance to complete the enrollment process successfully for their children. Communications with parents were most successful when trusted community- and school-based staff were involved.
Implications for Policy, Delivery or Practice:
Enrollment simplification requires both process improvements and careful attention to the needs of applicant families in order to achieve optimal impact on child health coverage.
Theme: Child Health
● Cost Analysis of the Geriatric Resources for
Assessment & Care of Elders (GRACE) Care
Management Intervention
Steven Counsell, M.D.; Christopher Callahan, M.D.;
Wanzhu Tu, Ph.D.; Tim Stump, M.S.; Greg Arling,
Ph.D.
Presented by: Steven Counsell, M.D., Mary
Elizabeth Mitchell Professor, Indiana University
Center for Aging Research, 1001 West 10th Street,
OPW M200, Indianapolis, IN 46202, Phone: (317)
630-7007; Email: scounsel@iupui.edu
Research Objective: The Geriatric Resources for
Assessment and Care of Elders, or GRACE, model of geriatric care management has been shown to improve quality and outcomes of care in low-income seniors, and reduce acute care utilization in those at high-risk of hospitalization. The purpose of this study is to provide a cost analysis of GRACE from the healthcare system perspective.
Study Design: Controlled clinical trial of 951 adults aged 65 or older who received care at one of 7 community-based health centers of an urban public healthcare system. Subjects were randomized to receive the intervention (n=474) or usual care
(n=477). The GRACE intervention includes a nurse practitioner and social worker who provide in-home assessment and care management over two years in collaboration with the primary care physician and a geriatrics interdisciplinary team, and guided by care protocols for common geriatric conditions. The
GRACE model uses an integrated EMR and Webbased care management tracking tool, and provides integration with affiliated pharmacy, mental health, home health, community-based, and hospital services. Main outcome measures were chronic and preventive care costs (including intervention costs), acute care (emergency department and hospital) costs, and total costs in the full sample
(n=951) and predefined groups at low-risk (n=725) and high-risk (n=226) of hospitalization. Cost data were obtained from a comprehensive database that contains the actual charges for all inpatient and outpatient facility fees, physician encounters, procedures and diagnostic testing, and rehabilitation and mental health services. Charge data were converted to costs using cost-to-charge ratios for the relevant year and cost center.
Population Studied: Baseline characteristics were similar between groups with mean age 72 years,
76% women, 60% black, 52% perceived health as fair/poor, and 15% needed help in basic ADLs; mean chronic disease count was 2.7 and hospital admission rate in prior year 0.23.
Principal Findings: In the full sample, mean twoyear total costs for intervention patients was higher but not significantly different than costs in usual care patients ($14,348 vs. $11,834; P=.20). Mean two-year total costs were significantly higher for intervention patients in the group at low risk of hospitalization ($13,307 vs. $9,654; P=.01).
Intervention patients in the high-risk group had lower but not significantly different mean two-year total costs compared to usual care ($17,713 vs.
$18,776; P=.38). The high-risk group had greater two-year costs for chronic and preventive care
($9,724 vs. 6,210; P<.001), including intervention costs ($2,691), but this was offset by reduced hospital costs ($7,343 vs. $11,731; P<.001). Cost analysis in the dually eligible for Medicare and
Medicaid subgroup (n=309) also revealed that intervention patients had higher chronic and preventive care costs ($8,168 vs. $5,769; P<.001) and lower hospital costs ($5,858 vs. $8,277;
P<.001) than usual care patients, resulting in similar two-year total costs ($14,612 vs. $14,677; P=.87).
Conclusion: The GRACE model of geriatric care management serves as an example of an organized system of care that provides improved quality at similar cost when delivered to a vulnerable population at high risk of hospitalization.
Implications for Policy, Delivery or Practice:
Payment reform such as the advanced medical home model will be required, however, for GRACE implementation beyond managed care Medicare systems since most services provided by GRACE
are not reimbursed under current fee-for-service
Medicare.
Funding Source(s): NIA, Nina Mason Pulliam
Charitable Trust and Wishard Health Services,
Indianapolis, Indiana
Theme: Medicare
● Outpatient Hysterectomy - A Study of Eight
States
Shanna Cox, M.S.P.H.; Samuel Posner, Ph.D.;
Susan Meikle, M.D., M.S.P.H.; Maura Whiteman,
Ph.D.; Polly Marchbanks, Ph.D.; Denise Jamieson,
M.D., M.P.H.
Presented by: Shanna Cox, M.S.P.H., Associate
Research Fellow, Division of Repoductive Health,
Centers for Disease Control & Prevention, 4770
Buford Highway Mail Stop K 20, Atlanta, GA 30341,
Phone: (770) 488-6477; Email: cio8@cdc.gov
Research Objective: The objective of this analysis is to examine outpatient hysterectomy in eight states in the US.
Study Design: Healthcare Cost and Utilization
Project data representing discharges from community hospitals and ambulatory surgery centers between 2000 and 2006 were used. Data was used from 2006 for California; 2000, 2002 and
2004 for Florida; and 2000, 2002, 2004, 2006 from
Colorado, Maryland, New Jersey, New York, North
Carolina and Wisconsin. Hysterectomies were identified using ICD-9-CM and CPT procedure coding. Hysterectomy rates per 1000 women aged
18 and older were calculated. Trends in hysterectomy rates were tested using least squares regression. Trends in the proportion of hysterectomies performed in the outpatient setting and trends in surgical approach were assessed using the Cochran-Armitage test for trend.
Differences in discharge characteristics and indications for hysterectomy by setting were examined using chi-square statistics.
Population Studied: Hysterectomy procedures among adult women in California, Colorado,
Florida, Maryland, New Jersey, New York, North
Carolina, and Wisconsin.
Principal Findings: The combined (inpatient and outpatient) hysterectomy rate did not change in any of the states examined and was highest in North
Carolina (6.5 per 1000 women) and lowest in New
Jersey (3.7 per 1000 women). The proportion of hysterectomies done in the outpatient setting increased for every state (p < .0001) and ranged from 0.4 % in New Jersey to 19.6% in North
Carolina. A higher proportion of hysterectomies done in the outpatient setting were performed by vaginal (85.2 %) or laparoscopic methods (29.2 %) as compared to the inpatient setting (55.4 % and
11.1 % respectively). The proportion of hysterectomies by vaginal route in the outpatient setting increased over the study period in Maryland, while decreasing in Colorado, Florida, New York,
North Carolina and Wisconsin (p < 0.05). The proportion of hysterectomies by vaginal route in the inpatient setting increased over the study period in
Wisconsin, while decreasing in Colorado, Florida,
Maryland and New Jersey (p < 0.05). The proportion of hysterectomies with laparoscopy increased in both the inpatient and outpatient settings in each state with multiple year data with the exception of New York which saw a decrease in vaginal hysterectomies with laparoscopy in the outpatient setting (p < 0.05). Outpatient hysterectomies were more likely to be younger than
55 years and have private insurance (p < .0001).
Outpatient hysterectomies were more likely to have an indication of endometriosis or other, while inpatient hysterectomies were more likely to have an indication of cancer, fibroids or uterine prolapse
(p < .0001).
Conclusion: Hysterectomies in the outpatient setting are becoming more common. While there is a general decline in the proportion of hysterectomies performed vaginally, laparoscopically-assisted hysterectomies are increasing in both the inpatient and outpatient setting. Age, insurance, indication and location influence the settings in which hysterectomies are performed.
Implications for Policy, Delivery or Practice:
Hysterectomy surveillance should include procedures done in both the inpatient and outpatient setting. Data that provides insights into regional differences in hysterectomy rates and proportion of hysterectomies done in the outpatient setting are needed.
● Validating Patient Harm Identified Using
Automated Data Streams
Chris Craver, M.A.; John Martin, M.P.H.; Richard
Bankowitz, M.D., M.B.A., F.A.C.P.; Eugene Nelson,
D.Sc., M.P.H.
Presented by: Chris Craver, M.A., Senior
Researcher, Research Services, Premier, Inc, 2320
Cascade Pointe Boulevard, Charlotte, NC 28028,
Phone: (704) 733-5241; Email: chris_craver@premierinc.com
Research Objective: In conjunction with the
QUEST initiative PREMIER INC and IHI have developed a set of 26 automated algorithms to identify patient harm. It is the goal of this study to validate these algorithms for widespread use to measure harm rates within and between hospitals.
These measures will be used to guide interventions that reduce harm by identifying priority areas contributing to avoidable illness and injury, and to monitor rates of harm in real time or as close to real time as possible.
Study Design: The validation study was conducted in four hospitals during the calendar year
2008/2009. Hospitals were selected from QUEST participants who were required to have all of the
appropriate data streams in place and were willing to commit to time and resource requirements necessary to complete the study. A random sample of approximately 150 patients was selected from each facility. This sample consisted of approximately 60% patients with harm identified through the algorithms and 40% that had no identified harm. An independent team of clinical experts where then deployed to each facility to perform a full chart review and IHI global trigger tool assessments on each sample patient discharge.
Approximately 20% of the charts were cross reviewed for purposes of inter-rater reliability.
Population Studied: Patients for the study were selected from inpatient discharges that occurred between October, 2007 and March, 2008 within each participating facility
Principal Findings: To date chart reviews have been completed at two participating facilities. This poster will present preliminary findings from the data analysis from these facilities. Initial results indicate that harm can be indentified and validated using a subset of the automated algorithms. A critical aspect that contributes to the success of the algorithms is the completeness of present on admission coding for ICD9 Diagnosis codes
Conclusion: Preliminary findings indicated that patient harm can be identified using select automated algorithms. Due to the infrequent nature of some types of harm events in hospitals, the use of this automated method to assess harm in the entire inpatient population -- compared to a chart review in a sub-sample of the population -- allowed facilities to more extensively assess patient safety with fewer burdens on resources.
Implications for Policy, Delivery or Practice: If the applications of the automated algorithms prove to be viable, hospitals within the QUEST program will have a lower cost, less resource intensive method for tracking rates of harm. However, data integrity and proper coding practices will be critical to the successful implementation of this automated system.
● Navigating Health Care Decision
Environments: The Effects of Age, Number of
Choices & Standardization
Janet Cummings, B.A.; Thomas Rice, Ph.D.; Yaniv
Hanoch, Ph.D.; Stacey Wood, Ph.D.
Presented by: Janet Cummings, B.A., Ph.D.
Candidate, Health Services, University of California,
Los Angeles School of Public Health, Campus Box
951772, Los Angeles, CA 90095-1772, Phone:
(310) 948-7154; Email: jrc12@ucla.edu
Research Objective: Traditional economic theory posits that more choice unambiguously benefits consumers. Recent research has questioned this notion, finding that information overload can reduce choice quality. Seniors who experience reduced cognitive abilities may be especially susceptible to information overload when facing complex decision environments as in Medicare Part D. One method of simplifying information used in the Medigap market is policy standardization. All participating companies must provide benefits that conform to one of several benefit packages. The idea is to facilitate apples-to-apples comparisons; once a benefit configuration is chosen, the consumer can simply compare premiums and company reputation.
We use an experimental design to examine whether younger individuals, those who face fewer choices, and those receiving standardized information are more likely to correctly answer factual questions about a set of long term care insurance plans.
Study Design: Data come from a randomized experiment conducted in 2007 in Claremont,
California. Participants were randomized into four groups to receive basic information about 5 or 15 hypothetical standardized or non-standardized long term care plans. Participants had to answer a simple factual question (identify the plan with the lowest premium) and a more complex factual question (identify the lowest premium for a plan with two specific characteristics). We estimated logistic regressions to examine whether age, number of plans, or standardization are associated with providing correct answers, controlling for race, education, and marital status.
Population Studied: 196 individuals, half age 18-
64 and half age 65+.
Principal Findings: Older adults were less likely than younger adults to answer each question correctly, with a marginal effect of -0.08 for the simple task (p=0.07) and -0.11 for the more complex task (p=0.04). Upon further investigation, numeracy (the ability to understand basic mathematical concepts) accounted for the age effect. When comparing performance for those assigned to more plans and those who received standardized plan information, there was no clear pattern of results for the simple task. For the more complex task, those assigned to non-standardized plans outperformed those assigned to standardized plans (marginal effect=0.27, p=0.002), in contrast to the hypothesized direction. One possible explanation is that participants in the nonstandardized group only had to use one table to answer this question, whereas those in the standardized group had to use two tables (a characteristic of standardized insurance).
Conclusion: Older individuals were less likely to answer factual questions correctly, which was mediated by numeracy. Those assigned to standardized plans were less likely to answer the complex question correctly, presumably because the information was necessarily located on two tables as opposed to just one.
Implications for Policy, Delivery or Practice:
Given the complex health care decision environment that the seniors face, our findings suggest that it is important to provide sufficient educational outreach efforts to Medicare
beneficiaries to help them navigate the expansive choice sets that they face in Part D. The findings also suggest that even if decision environments are standardized to aid plan comparison, it is critical to provide beneficiaries with education about how to correctly use such information.
Funding Source(s): RWJF
Theme: Consumer Choices in Health Care
● Globalization & Child Health Disparities
Jennifer Cushon, B.A., M.A., Ph.D.
Presented by: Jennifer Cushon, B.A., M.A., Ph.D.,
Doctoral Candidate, Community Health &
Epidemiology, University of Saskatchewan, 1017
10th Street East, Saskatoon, S7H0H8, CA, Phone:
(306) 966-7879; Email: jennifer.cushon@usask.ca
Research Objective: The objective of this research project was to investigate the political and economic pathways by which globalization impacts the determinants of health and health outcomes in a mid-sized North American city (Saskatoon,
Saskatchewan), with special reference to lowincome children.
Study Design: The analytical framework used to guide this study was developed by Labonte and
Torgerson (2005), and the theoretical framework was progressive political economy. The qualitative and quantitative research methods included: a detailed demographic profile; an environmental scan of national, provincial and municipal policy that has direct relevance for child health; process tracing; semi-structured interviews (n=26); Joinpoint regression; and Poisson regression.
Population Studied: The population of study was low-income children ages 0 to 5 in Saskatoon,
Saskatchewan, Canada.
Principal Findings: Low-income families in
Saskatoon, Saskatchewan, Canada have been negatively impacted by globalization due to declining family income; increasing income inequality; a rise in non-standard employment, usually in the service sector; declining housing affordability; and fewer and less generous social policies. Moreover, results indicate that globalization has contributed to increasing disparities between Saskatoon’s families in terms of income, education and occupation. These disparities in the determinants of child health have led to widening disparities in child health outcomes.
Conclusion: Globalization contributes to increasing disparities in the determinants of child health such as income level, the occupation of parents and the education level of parents. In addition, globalization contributes to increasing disparities in child health outcomes within a city.
Implications for Policy, Delivery or Practice:
Findings suggest a need for policy and program responses directed at addressing: income inequality
(e.g., decreasing the depth of poverty); improved working conditions for those in non-standard work arrangements (e.g., child care subsidies; sick leave benefits); housing that is affordable (e.g., decreasing the number of condo conversions; more social housing); and the retrenchment of the welfare state.
Funding Source(s): Canadian Institutes of Health
Research
Theme: Disparities
● Healthcare Experiences of Medicaid Enrollees in Florida
Rada Dagher, Ph.D., M.P.H.; Allyson Hall, Ph.D.,
M.P.H.
Presented by: Rada Dagher, Ph.D., M.P.H.,
Assistant Professor, Health Services Research,
Management & Policy, University of Florida, 101
South Newell Drive, Gainesville, FL 32611, Phone:
(612) 242-1638; Email: radadagher@yahoo.com
Research Objective: To examine the health care experiences of Medicaid enrollees under the Florida
Medicaid Reform. There is surprisingly little qualitative documentation of patients’ experiences with Medicaid in the research literature.
Study Design: This study employed purposive sampling of Medicaid enrollees throughout two waves of data collection, as part of an ongoing evaluation of Medicaid reform demonstrations in
Broward and Duval Counties. The first wave included 10 in-depth telephone interviews with
Medicaid enrollees and three focus groups (27 enrollees) between October 2006 and May 2007.
Interviewees constituted mainly of African
Americans (60%) and Caucasians (20%) and were primarily female (80%). Focus group participants were mostly African Americans (63%) and
Hispanics (26%). The second wave included 24 indepth telephone interviews with Medicaid enrollees and four focus groups (31 enrollees) between
October 2007 and June 2008. Interviewees were evenly divided between African Americans (37.5%) and Caucasians (37.5%) and were mostly males
(62.5%). Focus group participants were mostly
Caucasians (52%) and African Americans (39%) and constituted primarily of males (81%). The study team reviewed transcripts and developed a coding algorithm as well as verified overarching theme findings. In-depth interviews were coded using
Atlas.ti 5.0 software and focus group findings were coded using NVivo 2.0 software. Eight global themes were identified based on the focus group and in-depth interview instruments including
Medicaid Experience and Experience with Care in
General.
Population Studied: The study population consisted of all current Medicaid enrollees (adults and children) eligible to participate in Medicaid
Reform in Florida.
Principal Findings: Study participants reported a number of problems they experienced as Medicaid enrollees. These included difficulty with access and
referral to specialists, having to switch primary care providers, inadequate prescription drug coverage, lack of dental care, transportation barriers, lengthy wait times for services, and perceptions of discrimination under Medicaid. However, they also reported positive experiences such as being thankful for Medicaid and the services it provides them and being satisfied with their doctors. These experiences did not change over the two years of the study.
Conclusion: Pervasive problems with Medicaid continue to exist regardless of the changes implemented through Florida state reform efforts.
Key issues of access to care remain such as difficulty accessing specialists, problems maintaining continuity of care through primary care providers, and inadequate prescription drug coverage.
Implications for Policy, Delivery or Practice:
Study findings suggest a need for policymakers to address the underlying issues that prevent
Medicaid patients from having access to timely and quality health care.
Funding Source(s): Florida Agency for Health
Care Administration
Theme: Medicaid, SCHIP and State Health Reform
● Physical Abuse in U.S. Children: Risk Factors
& Disparities
Suzanne Dakil, M.D.; Hua Lin, Ph.D.; Jane
Brotanek, M.D., M.P.H.; Glenn Flores, M.D.
Presented by: Suzanne Dakil, M.D., Fellow of
Academic Pediatrics, Pediatrics, University of
Texas Southwestern, 5323 Harry Hines Boulevard,
Dallas, TX 75390, Phone: (214) 648-5031; Email: suzanne.roper@utsouthwestern.edu
Research Objective: Over 600,000 US children were reported for physical abuse in 2006, but little is known about risk factors for physical abuse or use of family services. Ou objective is to examine risk factors and racial/ethnic disparities for physical abuse in US children and services received.
Study Design: The 2006 NCANDS database was used. Risk factors for physical abuse were analyzed among reported (n=634,874) and substantiated physical abuse (n=161,630) and physical abuse deaths (n=158). Multivariate analyses adjusting for seven covariates examined reported and substantiated physical abuse, physical abuse deaths and support service use in whites, African-
Americans, Latinos, Asian/Pacific Islanders, and
Native Americans.
Population Studied: The database includes all children reported to child protective services in 49 states and Puerto Rico during 2006.
Principal Findings: Physically abusive homes have high proportions of domestic violence (20%), parental alcohol use (23%), single parents (47%) and public assistance receipt (31%), and physically abused children high proportions of behavior problems (10%) and prior abuse (30%). Domestic violence (OR 2.5; 95% CI 2.4-2.5), parental alcohol use (OR 2.2; 95% CI 2.1-2.2), prior abuse (OR 1.6;
95% CI 1.6-1.7), child behavior problems (OR 1.4;
95% CI 1.3-1.5), and public assistance receipt (OR
1.3; 95% CI 1.2-1.3) are associated with significantly higher odds of substantiated physical abuse. Public assistance receipt is associated with higher odds of physical abuse death (OR 2.5; 95%
CI 1.8-3.4). Compared to whites, African Americans have higher odds of reported (OR 1.1; 95% CI 1.1-
1.1) and substantiated (OR 1.3; 95% CI 1.2-1.3) physical abuse. Asian Pacific Islanders and Latinos have higher odds of reported PA, but similar odds
(vs. whites) of substantiated physical abuse. Native
Americans have lower odds of reported (OR 0.5;
95% CI 0.4-0.6) physical abuse and similar odds of substantiated physical abuse. Latinos have lower odds than whites of receiving any services (except mental health). Only 48% of parents with alcohol/drug use got substance abuse services.
Only 15% of families with public assistance received education/employment services.
Conclusion: Domestic violence, parental alcohol use, prior abuse, child behavior problems, and public assistance receipt are associated with higher adjusted odds of substantiated physical abuse in
US children, and public assistance receipt with greater odds of physical abuse deaths. Compared to whites, African Americans are at significantly higher risk, Latinos and Asian Pacific Islanders are overreported, and Native Americans are underreported for substantiated physical abuse.
Latino families have lower odds of receiving any services. Although parental alcohol use and public assistance receipt are major risk factors for substantiated physical abuse, less than half of parents with alcohol problems receive substance abuse services and 15% of families on public assistance receive employment/education services.
Implications for Policy, Delivery or Practice: By better understanding risk factors and disparities, the medical community can better identify at risk children and prevent abusive injury. Together with policy makers, we can better meet the needs of some families to not only keep children safe but also keep families together.
Theme: Child Health
● Early Results from Thailand’s 30 Baht
Universal Health Reform – Something to Smile
About
Kannika Damrongplasit, Ph.D.; Glenn Melnick,
Ph.D.
Presented by: Kannika Damrongplasit, Ph.D.,
Postdoctoral Research Fellow, Health Services,
University of California Los Angeles, Box 951772,
Los Angeles, CA 90095-1772, Phone: (310) 999-
8813; Email: kannikad@ucla.edu
Research Objective: In this paper, we use data from a large, nationally representative sample in
Thailand to explore whether Thailand’s 30-Baht program (i.e. universal health reform) succeeded in making coverage universal and, if so, whether it was able to avoid the potential problems of reducing access to previously insured or the implementation of informal payments by providers.
Study Design: Descriptive statistical methods.
Population Studied: We use data from Thailand’s
National Health and Welfare Survey (HWS) based on repeated cross-section of nationally representative households (2001, n=222,470 and
2005, n=67,815) collected by Thailand’s National
Statistical Office.
Principal Findings: We find that Thailand has added nearly 14 million people to the system and achieved almost universal coverage without compromising access for those with prior coverage and that, to date, no informal payment system has emerged.
Conclusion: Our data provide a system level assessment that suggests that programs overall objectives have been achieved – most of the previously uninsured have been added to the insurance system including 45 million covered by the 30 Baht program, contact rates for the previously insured have not gone down while contacts rates for the newly insured have gone up and it appears that health care providers are not collecting more from patients than the amounts prescribed by government policy.
Implications for Policy, Delivery or Practice: Our findings suggest that so far, Thailand has been able to succeed in its implementation of the universal coverage program. Other developing countries may use the Thai experience as an example when making their own health care reforms.
Funding Source(s): AHRQ
Theme: Coverage & Access
● The Effects of Prescription Drug Coverage on
Inappropriate Prescription Drug Use among
Community & Facility-Dwelling Aged Medicare
Beneficiaries
Laura D'Arcy, M.P.A.
Presented by: Laura D'Arcy, M.P.A., Ph.D.
Student, Health Policy & Management, University of
North Carolina at Chapel Hill, 508 Pittsboro Street,
Chapel Hill, NC 27516, Phone: (330) 606-0392;
Email: ldarcy@email.unc.edu
Research Objective: Evidence suggests that inappropriate prescription drug use (IRx) among the elderly is widespread. Studies indicate that upwards of 20% of community-dwelling elderly receive at least one inappropriate medication annually, and
IRx is associated with a substantial increase in the risk of nursing home admission. Approximately onethird of nursing home residents receive at least one inappropriate medication annually. Examining the extent of IRx is particularly salient in the era of
Medicare Part D. Recent studies suggest that the share of Medicare beneficiaries without prescription drug coverage has dropped substantially with the implementation of Medicare Part D, from rates of
25% to 38% before Part D to approximately 10% after implementation. Part D may also have important effects in the nursing home environment where residents eligible for both Medicare and
Medicaid were automatically enrolled in Part D.
This analysis provides a baseline assessment of the effect of prescription drug coverage on IRx and associated adverse drug events among Medicare beneficiaries prior to the implementation of Part D.
Study Design: I examine the association between prescription drug coverage and the risk of IRx in three groups of beneficiaries: those living in the community, those living in a nursing home, and those transitioning from the community to a nursing home. IRx is defined for a broad set of pharmaceuticals by the Beers criteria. Inappropriate use of antipsychotic drugs is measured using the
Beers criteria and criteria from the Centers for
Medicare & Medicaid Services’ State Operations
Manual.
Population Studied: The study sample comes from the 2004-05 Medicare Current Beneficiary
Surveys for approximately 8,000 community elderly and 1,200 beneficiaries who resided in a nursing home for at least three months during the study period.
Principal Findings: Existing data suggest that beneficiaries without drug coverage use fewer drugs and that Medicare Part D is associated with a slight decrease in cost-related medication nonadherence. I expect to find that prescription drug coverage is positively associated with drug use which is in turn linked to adverse drug events such as hospital use. I anticipate that coverage will be positively related to IRx.
Conclusion: IRx is an important and welldocumented problem among both community and facility aged Medicare beneficiaries. The availability of broader prescription drug coverage due to Part D may result in an increase in IRx.
Implications for Policy, Delivery or Practice: An increase in IRx associated with Part D would be harmful for beneficiaries and costly to the Medicare program. In future work, I intend to compare IRx among beneficiaries prior to and after the implementation of Part D as well as investigate important subgroups such as dual eligibles.
Funding Source(s): NIA
● Assisted Living in Pennsylvania - An Analysis of Potential Demand, 2009-2013:
Joan DaVanzo, Ph.D., M.S.W.; Allen Dobson,
Ph.D.; Steven Heath, M.P.A.; Patrick McMahon,
M.B.A.; Stuart Shapiro, M.D.; Anne Wantz
Presented by: Joan DaVanzo, Ph.D., M.S.W.,
C.E.O., Dobson DaVanzo & Associates, LLC, 9621
Meadowlark Road, Vienna, VA 22182, Phone:
(703) 281-3250; Email: joan.davanzo@dobsondavanzo.com
Research Objective: Advocates for publicly financed home and community-based long term care (LTC) have long maintained that transitioning residents out of nursing homes and into less intense assisted living (AL) settings saves state and federal Medicaid dollars. Under certain specific implementations of the policy, this assertion might be true. However, policymakers need to have a broad understanding of the entire LTC market in their state when implementing such a policy.
Recently enacted Act 2007-56 in Pennsylvania established a new category of LTC, which will permit payments for AL under a Medicaid waiver.
This study models future demand and the financial implications of various scenarios of usage if the
Pennsylvania Medicaid program were to begin paying for AL services.
Study Design: Three simulation models of how a potential Medicaid AL waiver might be implemented in Pennsylvania were created. The first model
(Nursing Home Only) allowed transitions only from nursing homes to AL facilities. The second model
(Budget Neutral) allowed transitions from all LTC settings, as long as the total fiscal impact was budget neutral. The third model (Unrestricted
Movement) allowed for transitions from all LTC settings and was limited only by the total number of
Medicaid AL “slots” available in any given year.
Principal Findings: We found that the acuity level of residents in Pennsylvania nursing homes was high, and that there were few residents who could safely transfer to a less intensive setting. All three of our simulation models assumed that a maximum of 1,000 Medicaid recipients would be able to transition from nursing homes to AL over five years.
Since Medicaid home and community-based LTC maintains a waiting list in Pennsylvania, we assumed that when a Medicaid recipient moved to
AL, their “slot” would be backfilled with a new recipient. We found that while the Nursing Home
Only model would save the state and federal governments $4.9 million over five years, approximately 4,000 Medicaid AL “slots” would go unfilled. The Budget Neutral model would not cost the state and federal governments money, but only
12 people could move to AL from home and community-based LTC, and 32 people new to
Medicaid could move to AL from the community.
Approximately 4,000 Medicaid AL “slots” would go unfilled. The Unrestricted Movement model would fill all 5,000 Medicaid AL “slots” from all LTC settings over five years, but would cost $477 million
($215 million is state and $262 is federal).
Implications for Policy, Delivery or Practice: In an effort to both expand consumer choice and save money, many states have sought to “rebalance”
Medicaid LTC spending by shifting resources from institutional LTC services to home and communitybased and AL. This study demonstrates that unless significant restrictions are placed on who may move into AL, making AL available under Medicaid may not meet the objective of cost savings.
Funding Source(s): Pennsylvania Health Care
Association
Theme: Long Term Care
● Increasing Burden of Methicillin-Resistant S.
Aureus (MRSA) Disease Complicating Influenza
& Pneumonia at U.S. Academic Medical Centers
(AMCs)
Michael David, M.D., Ph.D.; Sofia Medvedev,
Ph.D.; Sam Hohmann, Ph.D.; Bernard Ewigman,
M.D.; Robert Daum, M.D.
Presented by: Michael David, M.D., Ph.D., Fellow,
Medicine, The University of Chicago, 5841 South
Maryland Avenue, MC6054, Chicago, IL 60637,
Phone: (773) 702-3904; Email: mdavid@medicine.bsd.uchicago.edu
Research Objective: Community-associated (CA-) methicillin-resistant Staphylococcus aureus (MRSA) infections have been presumed to be increasing in incidence in the U.S. Because there is a long known association between influenza and severe S. aureus pneumonia, we hypothesized that the number of hospitalizations at U.S. academic medical centers (AMCs) for influenza complicated by MRSA pneumonia increased during the past decade and that the mortality rate was high.
Study Design: We analyzed data from the
University HealthSystems Consortium (UHC), a unique source of hospital discharge (HD) data for
>140 AMCs and affiliated hospitals. A MRSAassociated HD was defined as one with the V09.0
ICD-9 code. We determined the rate of HD with diagnosis codes for MRSA, influenza (I), pneumonia (P), and all combinations of these diagnoses (designated with a slash (/), e.g.,
I/MRSA) at U.S. AMCs for 2002-2008. The mortality for each diagnosis and combination of diagnoses was determined. All results were stratified by age group and U.S. region.
Population Studied: All patients discharged from
U.S. AMCs in the UHC database in 2002-2008.
Principal Findings: From 2002 to 2008, the rate of
P HDs increased slightly (2.31 to 2.94 per 1000). In contrast, I/P HDs increased nearly 7-fold (0.21% to
1.41% of P HDs). P/MRSA HDs doubled in 2002-
2008 (27.0% to 56.3% of P HDs). I/MRSA HDs increased from 0.65% to 2.14% of I HDs. In 2002-
2008, the I HD rate was highest for children 1-12 months (12.5 per 1000) and 1-17 years (5.1), and lowest (0.3) for children <1 month. I/MRSA HDs as a percent of I HDs were highest for patients 18-30
(1.95%) and 31-50 years old (1.93%), and lowest for children <1 year of age. I/P/MRSA HDs increased from 0.05% to 0.73% of P HDs in 2002-
2008. In 2002-2008, in-hospital mortality for I/MRSA
HDs was 8.2% (51/621) compared with 2.2%
(816/36,858) for all other I HDs.
Conclusion: In 2002-8, MRSA was an increasingly common concomitant diagnosis among patients hospitalized at AMCs for influenza or pneumonia.
Adults 18-50 years old had the highest percent of influenza diagnoses with concomitant MRSA diagnoses. Concomitant MRSA infection may increase mortality among hospitalized influenza patients.
Implications for Policy, Delivery or Practice:
Increasing rates of MRSA among patients with pneumonia and influenza hospitalized at U.S.
AMCs is a concerning trend. Empiric therapy in young adult patients with influenza and pneumonia should include antibiotics effective against MRSA.
This trend also raises concern for an epidemic of severe MRSA pneumonia cases in the event of an episode of pandemic influenza, given the likely synergy between influenza and CA-MRSA strains, now widely disseminated in the general population, and the high mortality in coinfected patients.
Theme: Public Health
● Factors Associated with Participation in
Medicare Part D in 2006 Among
Amy Davidoff, Ph.D.; Bruce Stuart, Ph.D.; Thomas
Shaffer, M.H.S.; J. Samantha Shoemaker; Melissa
Kim; Christopher Zacker
Presented by: Amy Davidoff, Ph.D., Assistant
Professor, Pharmaceutical Health Services
Research, University of Maryland School of
Pharmacy, 220 Arch Street, 01-214, Baltimore, MD
21201, Phone: (410) 706-5397; Email: adavidof@rx.umaryland.edu
Research Objective: A major policy goal of
Medicare Part D was to provide prescription drug coverage to beneficiaries unable to afford it, particularly those with low incomes and high drug costs associated with chronic health conditions. The objective of this study is to examine factors, particularly baseline health status, income, and access problems, associated with Part D participation among those without prior drug coverage.
Study Design: Bivariate and multivariate (logistic regression) methods were used to examine the relationship of beneficiary health status and other characteristics to Part D enrollment, MAPD enrollment among Part D participants, and enrollment in Low Income Subsidy (LIS) among low income beneficiaries. We compared characteristics of those without drug coverage in 2005 and 2006.
Beneficiary characteristics included health status, income relative to FPL, financial hardship obtaining drugs, cost-related non-adherence, demographics, and premium for the least costly PDP available in the state.
Population Studied: Community-dwelling
Medicare beneficiaries participating in both the
2005 and 2006 Medicare Beneficiary Survey
(MCBS) Access to Care surveys who had no prescription coverage in 2005 (N=3,038).
Principal Findings: 63% of beneficiaries with no drug coverage in 2005 enrolled in Part D, while another 9.6% gained some other type of coverage, and over one quarter (27.4%) remained without drug coverage. Among those who enrolled in Part
D, over four-fifths enrolled in a PDP. Adjusted results indicate that Part D enrollment was associated with age ((-) for non-elderly disabled and over age 75 relative to 65-69 years), male (-), poverty (+), financial hardship (+), premiums (-), supplemental medical insurance (+), number of chronic health conditions (+) all at p<.05. MAPD enrollment was associated with fewer chronic conditions, and with prior MA coverage. LIS participation was associated with age < 50 (+), black (+), supplemental coverage (public (+); private (-)), premiums (+), number of chronic conditions (+), and specific conditions: diabetes (+), depression (+), and COPD (+). Among beneficiaries who failed to enroll in 2006, 45% had three or more chronic conditions, and 37% had income below 150% FPL.
Conclusion: Patterns of participation are consistent with expectations that the most vulnerable beneficiaries, as evidenced by their burden of chronic illness, low income, and reported financial hardship, disproportionately enrolled in the program. MAPD plans enrolled healthier patients, consistent with prior evidence on selection into MA plans. Characteristics of beneficiaries remaining without coverage reflect a healthier subgroup, but there is a core with multiple chronic conditions and low income.
Implications for Policy, Delivery or Practice: Our findings suggest that the program was largely successful in providing coverage to vulnerable beneficiaries without other sources of drug coverage. The failure of some vulnerable beneficiaries to enroll indicates that barriers to participation, particularly among LIS eligible, must be addressed by policy makers. Longer term impacts on prescription drug use, spending, and health benefits will need to be monitored.
Funding Source(s): Novartis Pharmaceutical
Corporation
Theme: Medicare
● Consumer Perceptions of Seven Identified
Domains of Care Coordination
Ellen Dawson, Ph.D.; Jean Johnson, Ph.D;
Shoshanna Sofaer, Dr. P. H; Christine Pintz, Ph.D;
Andrea Brassard, D.N.Sc.
Presented by: Ellen Dawson, Ph.D., Chair,
Department of Nursing Education, Nursing
Education, George Washington University, 900
23rd Street, NW, Washington, DC 20037, Phone:
(202) 994-7961; Email: hspemd@gwumc.edu
Research Objective: NQF has named care coordination as a high priority area for measurement development, both in general and in terms of the impact of nurses on quality. Care coordination is a somewhat fuzzy concept that has been difficult to operationalize and measure. Yet effective care coordination is viewed by many as essential both to the overall experience of the patient and to critical patient outcomes and patient safety. Aim: (1) To identify key behaviors undertaken by nursing staff (especially the bedside nurse) in coordinating patient care over the course of a hospital stay; (2) to assess how patients respond to these potential measures; (3) to build toward a parsimonious set of items on the quality of care coordination by nurses, that could be incorporated into a patient experience survey such as Hospital CAHPS.
Study Design: Methods: To pursue Aim 1, we conducted interviews with hospital staff nurses to get their perspective on the important aspects of care coordination within hospitals. To elicit their views and experiences, patient care vignettes were used to illustrate situations where patients needed to receive care across hospital units and on discharge to other settings. Based on these interviews, we identified a set of seven domains of care coordination by nurses, with a limited number of example measures/behaviors in each domain.
To pursue Aim 2, we reality tested these domains and measures with consumers/patients in nine focus groups with recent hospital patients. We explored their views of the absolute and relative importance of the domains and measures, their appropriateness as an assessment of nurse performance, and the likelihood of finding variability in performance.
Population Studied: Hospital staff nurses and consumers who were patients within one year of the study.
Principal Findings: Findings: The seven domains identified were: Working with the patient;
Collaboration and cooperation; Keeping everyone informed and up to date: Mutual respect: Vigilance and follow-up; Working with the family; and
Ensuring a Safe Discharge. Recent patients found the first five domains very important based on their experiences, though their responses to specific examples/measures varied. Working with the family was viewed as less important; and were concerned that nurses not interfere with patient/family situations. Ensuring a safe discharge was uniformly viewed, in all groups, as least important, which is in stark contrast to the views of nurses themselves. An unexpected finding was that many participants felt that the primary responsibility for care coordination lay with the patients themselves and not nurses or other health care providers.
Conclusion: Conclusion: Using open-ended qualitative methods that focus on specific behaviors more than on general concepts, we have been able to identify a pool of nearly 25 potential items that reflect aspects of care coordination that the patient is likely to be able to observe.
Implications for Policy, Delivery or Practice: We plan to conduct iterative cognitive and psychometric testing of these items to identify a small set for incorporation into a survey such as Hospital
CAHPS.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Measurement
● Experience of a Large Urban Clinic
Implementing & Electronic Health Record with
Public Health Functionality on Provider
Productivity
Samantha De Leon, Ph.D.; Sarah Shih, M.P.H.;
Farzad Mostashari, M.P.H., M.D.
Presented by: Samantha De Leon, Ph.D.,
Research Scientist, Healthcare Access - Primary
Care Information Project, New York City
Department of Health & Mental Hygiene, 161
William Street, New York, NY 10038, Phone: (212)
788-5686; Email: sdeleon@health.nyc.gov
Research Objective: The objective of this study is to determine the economic impact, in terms of increased or decreased provider productivity, among providers at practices that implement an
Electronic Health Record (EHR) with public health functionality, by comparing levels of provider productivity in the pre-EHR period, to that in the post-EHR period.
Study Design: New York City has a citywide initiative (Primary Care Information Project) to implement EHRs with a health prevention focus to reduce health disparities and improve population health. Using historical billing data detailing the number and type of procedures performed each day for all patient encounters, productivity measures such as Relative Value Unit (RVU) work per day were tabulated for each provider during both the pre- and post-EHR periods, in order to capture changes in the volume and complexity of patient visits documented by providers over time.
Random coefficient models for multi-level longitudinal data were conducted, where each provider’s data is modeled as a separate growth curve and separate intercepts and slopes are estimated for each provider with adjustments for autocorrelation between observations and interindividual differences in intercept and slope.
Population Studied: Approximately 70 primary care providers, with at least one month of data in both the pre- and post-EHR periods (2005-2007), practicing at a large multi-site, federally funded community healthcare center participating in the
PCIP initiative to bring EHRs to practices serving primarily Medicaid and under-served populations.
Principal Findings: Based on an analysis of
Evaluation and Management (EM)) visit codes documented for each patient encounter, a small but
significant decrease in provider productivity trend over time, as measured by RVU (work) total per day was observed, corresponding to a 0.17 ± 0.135 decrease in RVU (work) total per provider during the post-EHR period. However, when EM visit codes were analyzed by type of visit, outpatient consultations (CPT-4: 99241-99245) and outpatient established (CPT-4: 99211-99215) showed small significant increases corresponding to 0.830 ±
0.399 and 0.198 ± 0.111 RVU(work) total per provider respectively, whereas preventive established (CPT-4: 99391-99397) and preventive new (CPT-4: 99381-99387) patient visits showed decreases corresponding to 0.355 ± 0.167 and
0.261 ± 0.171 RVU(work) total per provider in the post-EHR period, and outpatient new visits (CPT-4:
99201-99205) showed no significant changes over time.
Conclusion: These preliminary findings suggest that while there were very small gains and decrements in provider productivity, implementing an EHR with public health functionality had no meaningful impact on provider productivity at the practice-level. Providers at this practice were able to maintain productivity levels during the implementation process.
Implications for Policy, Delivery or Practice:
EHRs with public health functionality have the potential to improve patient quality of care through better documentation and improved measurement of quality of care data, which could potentially be of great benefit to patients in medically under-served communities. One concern providers have is the potential negative effect on productivity that implementing such on EHR can have on their practice. Demonstrating the potential productivity and economic impacts of implementing an EHR with public health functionality could serve to increase provider buy-in and decrease apprehension surrounding the adoption of this technology.
Theme: Quality and Efficiency: Measurement
● An Examination of the Performance &
Characteristics of Health Departments in a
Public Health Practice- Based Research
Network
Angela Dearinger, M.D., M.P.H.; Alex Howard;
Richard Ingram
Presented by: Angela Dearinger, M.D., M.P.H.,
Assistant Professor, Health Services Management,
University of Kentucky College of Public Health,
K512 740 South Limestone Street, Lexington, KY
40536, Phone: (859) 323-6749; Email: angelatackett@uky.edu
Research Objective: The purpose of this study is to compare the performance and characteristics of health departments in a public health practice- based research network, the Kentucky Public
Health Research Network (K-PHReN), with health departments in the state of Kentucky and in the
United States using the National Public Health
Performance Standards Program (NPHPSP) Local
Assessment Instrument and the National
Association of County and City Health Officials
(NACCHO) 2005 National Profile of Local Health
Departments. K-PHReN is one of five new public health practice- based research networks (PBRN) in the country and holds promise for advancing the field of public health systems and services research and improving the health of our communities.
Study Design: This study will use descriptive analysis to examine the performance and characteristics of local public health departments.
Population Studied: This study is evaluating the performance and characteristics of health departments in the US, Kentucky, and health departments that comprise K-PHReN who completed the NPHPSP local assessment instrument and the 2005 NACCHO profile. K-
PHReN is made up of 18 of Kentucky’s 55 health departments (including solo county, district, and independent departments).
Principal Findings: On average, health departments in K-PHReN performed better than national averages on the 10 Essential Public Health
Services (EPHS), except EPHS #4, “Mobilize community partnerships to identify and solve health problems,” (43.7% versus 52.2%). The percentage of health departments in K-PHReN that have a local board of health is higher than the percentage of departments that have local boards of health in the
NACCHO profile (87% versus 74.3%), and local boards of health associated with K-PHReN departments are more likely to have governing, policymaking, and advising functions. In addition,
87% of K-PHReN health departments have completed a community health assessment within the last three years, whereas only 51% of US health departments have done so.
Conclusion: The mean NPHPSP performance for
K-PHReN departments is higher than that of all departments in the US that completed the instrument. However, a high percentage of K-
PHReN health departments had recently completed a community health assessment, but performed worse than national averages in mobilizing community partners to identify and solve health problems. Public health PBRNs may bridge the gap between local health department’s and other community partners who make up the local public health system, and as a result, may improve the quality of public health services through functions such as facilitating the mobilization of community partnerships. Further longitudinal studies will be necessary to evaluate the role of public health
PBRNs in mobilizing community partnerships.
Implications for Policy, Delivery or Practice:
Public health practice- based research networks may provide an infrastructure that improves community health, particularly through encouraging
and supporting collaborations between public health departments and their community partners.
● Quality & Efficiency of Nursing Homes
Nailya DeLellis, M.H.A.
Presented by: Nailya DeLellis, M.H.A., Student,
Health Administration, Virginia Commonwealth
University, 1504 Westover Hills Boulevard,
Richmond, VA 23225, Phone: (804) 231-1641;
Email: delellisno@vcu.edu
Research Objective: To define factors affecting efficiency and quality of nursing home care;
To define characteristics of nursing homes that provide high quality services and are highly efficient
Study Design: Cross-sectional; efficiency score is calculated by Data Envelopment Analysis.
Structural Equation Modeling (SEM) is used to study the relationship among market, structure, efficiency and quality variables. Logistic regression will be used to assess high quality and high efficiency facilities.
Population Studied: Nursing homes (OSCAR database, 2008) of size 20-360 beds, excluding hospital based units. 10% random sample or 1430 facilities.
Principal Findings: There is no correlation found between quality measures and efficiency for nursing homes. Efficiency score is highly correlated with occupancy rate. SEM data is under analysis now
Conclusion: It is possible to improve quality and efficiency at the same time.
Implications for Policy, Delivery or Practice: It will be useful to know what nursing homes can provide high quality services and be highly efficient.
● Relationship Between Hospital Report Cards
& CMS Quality Measures
Bonnie, Jerome-D'Emilia, Ph.D.; Derek DeLia,
Ph.D.
Presented by: Derek DeLia, Ph.D., Assistant
Professor/Senior Policy Analyst, The Center for
State Health Policy, Rutgers, The State University,
55 Commercial Avenue, 3rd Floor, New Brunswick,
NJ 08901-1340, Phone: (732) 932-4671; Email: ddelia@ifh.rutgers.edu
Research Objective: Public reporting of quality measures is widely viewed as a way of improving hospital care. Today 39 states make hospital quality data available via the internet. It is assumed that this public reporting of quality data has an effect on one or more of these variables: consumer choice, selective contracting by payers, or hospital’s efforts to improve competitive standings, and that these variables will ultimately improve quality of care.
Although legislation continues to be proposed at the state and federal level to require this transparency, the impact of public reporting on the abovementioned variables and quality outcomes is not clear. Further the effect of state report cards on quality given the existence of national data reporting is uncertain. This study evaluates the change over time of hospital quality as expressed in the Centers for Medicare and Medicaid Services
(CMS) database of hospital process measures, to determine whether the introduction of state hospital report cards affects the quality of care delivered.
Study Design: The study uses hospital process measures at the state level for the treatment of heart attack (8 measures, 1 summary score), heart failure (4 measures, 1 summary), and pneumonia (6 measures, 1 summary) collected by CMS from
2004-2006. The relationship between report cards and quality outcomes is identified by a variable that equals 1 if a report card was available for a given state in a given year and 0 otherwise.
Population Studied: 50 states and Washington,
DC.
Principal Findings: For the vast majority of indicators studied, the presence of a statewide hospital report card had no systematic association with hospital quality. From 2004-2006, quality improved (on average) for most pneumonia and heart failure indicators but decreased for many heart attack indicators. Nevertheless, these trends were unrelated to the availability of state hospital report cards.
Conclusion: The availability of hospital report cards does not seem to have an appreciable effect on the quality of care as measured in this study.
While report cards may lead to quality improvements at specific hospitals, this analysis found no evidence of a broader system-wide relationship. It may take time for consumers and payers to become aware and make use of report cards and hospitals may require time to change their processes of care in response to poor scores.
Although some states have made hospital report cards available for several years, others in this study have introduced report cards only recently.
Implications for Policy, Delivery or Practice:
Two possible explanations of these findings are:
1.CMS process measures are inadequate to demonstrate changes in quality related to the use of public report cards, or 2. State report cards do not improve quality. As much effort and money has gone and will continue to go into quality transparency, it is imperative that a clear connection be drawn between hospital performance data and quality of care. It is also important to determine whether state reporting has an impact that is independent of national reporting. If these relationships cannot be established then resources currently used to produce report cards may be better used elsewhere.
Theme: Quality and Efficiency: Policies and
Incentives
● Using a Likelihood Ratio Test to Identify the
Key Determinants of Enrollee Satisfaction with a
Managed Health Care Plan
Allard Dembe, Sc.D.; Bo Lu, Ph.D.; Cynthia Sieck,
Ph.D.
Presented by: Allard Dembe, Sc.D., Associate
Professor & Chair, Health Services Management &
Policy, Ohio State University College of Public
Health, 472 Cunz Hall, 1841 Neil Avenue,
Columbus, OH 43210, Phone: (614) 292-2129;
Email: adembe@cph.osu.edu
Research Objective: Many studies have examined patient satisfaction with health care. Fewer have investigated enrollees’ satisfaction with a managed health plan and only a handful of those have focused on identifying the specific factors that most strongly predict plan satisfaction. Being able to determine the most salient factors allows administrators to make improvements in a way that will be most meaningful and well-received by the plan members. The goal of this study is to identify the strongest predictors of enrollees’ satisfaction with managed health care plans using an analytical approach that features a combination of logistic regression analysis and stepwise application of the likelihood ratio test.
Study Design: Data for this study were obtained from an employee satisfaction survey based on the
CAPHS core questionnaire (version 3.0), along with additional survey items pertaining to plan membership and the plan’s wellness and prevention services. Unadjusted odds ratios with a 95% confidence interval were calculated for 48 potential independent variables predicting the odds of respondents being very satisfied with their health plan (9-10 on a 10-point scale). Of the 48 factors originally considered, logistic regression analyses identified 8 factors that were found to be significantly associated with plan satisfaction, after controlling for enrollees’ gender, age, race, and education. Likelihood ratio testing was then conducted to determine which particular factors had the greatest individual impact on the regression analysis results.
Population Studied: The survey was administered to a sample of 2,978 employees enrolled in a managed health care plan operated by a large university in the Midwestern United States. A total of 1,533 employees in the sample submitted completed questionnaires, a response rate of
51.2%.
Principal Findings: Factors most strongly associated with health plan satisfaction included: satisfaction with the plan’s wellness and prevention services (OR=3.69, CI: 2.60-5.23), having a personal doctor or nurse (OR=2.70, CI: 1.63-4.47), being satisfied with the cost of the health plan
(OR=2.18, CI: 1.52-3.13), and having their claims handled correctly (OR=1.90, CI: 1.31-2.77).
Subsequent stepwise application of the likelihood ratio test indicated that members’ perception of the quality of a plan’s wellness and prevention services was the single most influential factor in determining overall plan satisfaction, followed by whether the plan makes it clear to members how much they will have to pay for a particular service or visit, members’ overall satisfaction with the cost of the plan, and whether they have a personal doctor or nurse.
Conclusion: The most immediate direct implication of our study findings, from the enrollees’ point of view, is that greatest attention needs to be directed at strengthening the prevention and wellness services that are currently offered within managed health plans.
Implications for Policy, Delivery or Practice: The analytical approach we applied allows researchers and managed care officials to identify the strongest predictors of plan satisfaction in a more detailed and precise way than has been done previously. It provides plan managers with specific information to more effectively customize plan benefits and service delivery techniques to meet enrollee preferences, and thereby achieve greater overall satisfaction with the plan.
Funding Source(s): A commercial managed health care plan
Theme: Quality and Efficiency: Measurement
● Impact of Parental Attitudes & School
Mandates on Uptake of HPV Vaccines Among
Adolescents
Amanda Dempsey, M.D., Ph.D., M.P.H.; David
Mendez, Ph.D.
Presented by: Amanda Dempsey, M.D., Ph.D.,
M.P.H., Assistant Professor of Pediatrics &
Communicable Diseases, Pediatrics, University of
Michigan, 300 North Ingalls, Room 6E08, Ann
Arbor, MI 48109-5456, Phone: (734) 615-0398;
Email: adempsey@umich.edu
Research Objective: The clinical impact of human papillomavirus (HPV) vaccination on cervical cancer and other HPV-associated diseases has been predicted using mathematical models. However, a significant limitation of these models is that they assume near-instantaneous and widespread (70-
100%) vaccine uptake among adolescent girls. The objectives of this study were 1) to develop a more realistic mathematical model of adolescent HPV vaccine uptake that incorporated existing data on parental attitudes about the vaccine and adolescent health care utilization patterns; and 2) to use this model to explore the potential impact of school mandates for HPV vaccination.
Study Design: We developed a dynamic, compartmental model of HPV vaccine uptake among 11-17 year old girls that spanned a 50-year time period. Model parameters were derived from existing data on parental attitudes about HPV vaccines, validated models of health behavior, adolescent health care utilization patterns and census data. Markov probability sampling techniques were used to estimate vaccine utilization
at the population level. Sensitivity analyses were performed to determine the impact of parameter assumptions and uncertainties on model outputs.
Population Studied: A theoretical population of 11-
17 year old girls and their parents.
Principal Findings: Under baseline, non-mandate conditions, our model predicted that in the first year of HPV vaccination programs first, second and third doses of vaccine would be received by 25%, 17% and 7% of adolescent girls, respectively. Seventy percent vaccination coverage, the lower threshold value used in previous studies assessing the clinical impact of HPV vaccination, was not achieved until year 23 of the program. Vaccine coverage after 50 years was 79%. Instituting school mandates was predicted to increase HPV vaccine utilization substantially, resulting in 19% of adolescent girls fully vaccinated after the first year of vaccine implementation, and 70% vaccination coverage by year 8 of vaccine availability. Maximal vaccination coverage, 90%, was achieved by year
43 of the program.
Conclusion: Given current parental sentiments about HPV vaccination and existing patterns of adolescent health care utilization, our results suggest that assumptions about HPV vaccine uptake among adolescents may have been too generous in previous models. School mandates have the potential to improve vaccination rates, but even with such policies high levels of vaccine utilization could still take several years to achieve.
Implications for Policy, Delivery or Practice: Our results suggest that without significant changes in adolescent health care utilization patterns and/or parental attitudes, utilization of the HPV vaccine by adolescents will be sub-optimal for many years.
School mandates are one policy that might improve vaccine utilization, and several different types of
HPV vaccine-related school mandates are currently under legislative consideration. However, to ensure that the optimal policies are adopted, the potential impact of these various types of HPV vaccine mandates should be explored further. Modeling is the only viable mechanism to explore these different policy options.
Funding Source(s): NICHD
Theme: Child Health
● Trends in Health Insurance Access in
Pennsylvania: Analysis of the Pennsylvania
Insurance Department Health Insurance Survey
Ronald Deprez, Ph.D., M.P.H.; Amy Kinner, M.S.;
Brian Robertson, Ph.D.; Patrick Madden; Ed Nagel
Presented by: Ronald Deprez, Ph.D., M.P.H.,
Executive Director, Center for Health Policy,
Planning, & Research, University of New England,
170 Stevens Avenue, Portland, ME 04104, Phone:
(207) 221-4560; Email: rdeprez@une.edu
Research Objective: The objective of this study was to research insurance access in the State of
Pennsylvania before and after the introduction of comprehensive insurance reform in order to assess the effectiveness of reforms and identify future challenges.
Study Design: Based on the recently conducted the 2008 Pennsylvania Insurance Department
Health Insurance Survey, a follow-up of a similar survey we conducted in 2004, we evaluated the impact of the Cover All Kids (SCHIP), adultBasic and the enhanced Medicaid program on the uninsured population of Penna. The two surveys provide data on trends in insurance coverage by demographic and geographic characteristics; the effectiveness of recent health insurance initiatives in the State, including the Cover All Kids Program and the adultBasic Program; and information on remaining gaps in insurance.
Population Studied: The study included a telephone survey of more than 20,000 households in all counties throughout the State of Pennsylvania.
Data was collected on a variety of demographic groups and stratified by characteristics such as age, race/ethnicity, income, employment status, and other characteristics.
Principal Findings: Pennsylvania’s efforts to expand insurance coverage have served as a successful safety net for residents with low family incomes and particularly for African Americans. In addition, between 2004 and 2008 there was very little reduction in existing private coverage, an indication that the private insurance market overall was not substantially impacted by economic slowdown or other externalities.
Conclusion: While reforms have been relatively successful, the survey data highlight a variety of future challenges for Pennsylvania. These may include: expanding the adultBasic Program to meet the high level of demand for the program; continuing to maintain employer-sponsored insurance coverage levels in the future in the context of economic challenges; promoting insurance enrollment among young adults; reducing administrative barriers for enrollment in state insurance programs; and continuing to improve health care quality and chronic care management.
Implications for Policy, Delivery or Practice:
Lessons learned from Pennsylvania may provide insight both to Pennsylvania policymakers wishing to improve reform programs and address future challenges and to policymakers in other States seeking to implement health reforms that are similar to those implemented in Pennsylvania.
Funding Source(s): State of Pennsylvania
Theme: Medicaid, SCHIP and State Health Reform
● Outcomes Following Bariatric Surgery: Data from a National Plan-Provider Collaborative
Andrea DeVries, Ph.D.; Marc Gottlieb, M.P.A.;
Likun Hou, M.S., M.A.; Patricia Martin, R.N., B.S.N.;
Al Gamroth; Joseph Singer, M.D.
Presented by: Andrea DeVries, Ph.D., Research
Operations Director, HealthCore,
WellPoint/HealthCore, 800 Delaware Avenue,
Wilmington, DE 19801, Phone: (302) 352-1766;
Email: adevries@healthcore.com
Research Objective: In 2006, the Blue Cross Blue
Shield Association created a program to identify and promote use of facilities with the best outcomes for bariatric surgical procedures. Working in collaboration with expert physicians and medical organizations, a survey tool was created requesting data on structure, process, and outcome measures such as procedure volume, mortality, care team, and patient follow-up care. The process is transparent, and the measures are publicly available. Facilities meeting selection criteria are awarded the designation of Blue Distinction®.
Facilities not meeting selection criteria are given feedback regarding suggested areas for improvement. Over 250 facilities have received this designation for bariatric procedures. The purpose of this study was to assess episode level outcomes for designated facilities as compared to nondesignated facilities.
Study Design: We compared 30-day, 90-day, and
6-month outcomes for patients undergoing bariatric procedures in 2007 at 136 designated and 301 nondesignated facilities, taking into account age, gender, and co-morbid conditions. Chi-square tests and Wilcoxon-Mann-Whitney tests were performed to measure differences in outcomes for a range of complications including surgical, digestive, hemorrhage, and infections.
Population Studied: This study reports on data from 14 Blue Cross Blue Shield plans. A national cohort of commercially insured members received services from 437 facilities representing all regions of the United States (136 designated and 301 nondesignated facilities). There were 2,310 patients treated at designated centers and 2,169 treated at non-designated centers. Procedures of interest include laparoscopic band, laparascopic gastroenterostomy, open gastroenterostomy, open gastric bypass, and laparoscopic gastroplasty.
Principal Findings: Designated facilities had complication rates that were 26% lower than nondesignated facilities (6.7% vs. 5.0%; p<.02) in the
30-day period following surgery. The open gastroenterostomy procedures had the greatest risk for complications (18.6% for non-designated centers and 12.2% for designated centers in the 30day period following surgery). The greatest difference occurred in the category of infections.
The designated facilities had 60% lower rates for infections treated in either an inpatient setting or outpatient setting subsequent to discharge (2.6% vs. 1.0%; p = .0001). Costs during the follow-up period were also lower for patients at designated centers (p<.05), with the greatest difference observed for open procedures.
Conclusion: We observed significant variation in complications following bariatric surgery by procedure type and statistically significant differences for designated vs. non-designated facilities. This population-based study validates the use of a plan-provider collaborative to identify facilities with the highest probabilities for good outcomes following bariatric surgery.
Implications for Policy, Delivery or Practice: By working closely with providers, collaborative efforts such as this one offer credible, clinically relevant information, helping members and employer groups make informed decisions regarding their healthcare.
Participating providers are given recognition for offering high quality services and an incentive to maintain or improve quality outcomes for patients.
Increasingly, employer groups are incorporating program components into benefit designs. For example, members may have lower out-of-pocket costs when selecting a designated center. This effort is part of a growing trend to proactively use health data to help members choose the best provider to meet their needs.
Funding Source(s): WellPoint, Inc.
Theme: Quality and Efficiency: Policies and
Incentives
● Linguistic Access Services in the United
States: Do they Measure up to the National
Culturally & Linguistically Accessible Services
(CLAS) Standards?
Lisa Diamond, M.D., M.P.H.; Amy Wilson-Stronks,
M.P.P; Elizabeth Jacobs, M.D., M.P.P.
Presented by: Lisa Diamond, M.D., M.P.H.,
Healthcare Research Fellow, Health Policy
Research, Palo Alto Medical Foundation Research
Institute/University of California, San Francisco
Institute for Health Policy Studies, 795 El Camino
Real, Ames Building, Palo Alto, CA 94301, Email: diamondl@pamfri.org
Research Objective: To assess US hospitals' compliance with the National Culturally and
Linguistically Accessible Services (CLAS)
Standards. Federal regulations require health care organizations to provide language services to patients with limited English proficiency (LEP). Four of the National Standards on Culturally and
Linguistically Appropriate Services (CLAS) in Health
Care address provision of linguistically accessible services: (1) Timely provision of interpreter services, (2) Patient notification of their right to receive language services, (3) Utilization of competent interpreters, and (4) Availability of written materials in languages commonly encountered within a hospital. It is not known how
US hospitals have incorporated the CLAS
Standards into practice. The objective of this study was to evaluate compliance with these 4 standards in a national sample of hospitals.
Study Design: Cross-sectional study of US hospital interpreter services managers.
Population Studied: The sample was selected from all US hospitals based on demographic and geographic characteristics; 239 hospitals were selected randomly from within these counties.
Interpreter services managers from 221 of the 239
(93%) hospitals were identified and they received a link to the survey by email. The survey asked for information about demographics of the patient populations served, the types of language services available, barriers to providing services that hospitals have encountered, and adherence to the
4 language related CLAS standards.
Principal Findings: One hundred and thirty-five interpreter services managers responded to the survey (61% response rate). Seventy-eight percent of hospitals were able to provide interpreters for their most common language within 15 minutes during business hours, but only 48% could provide interpreters in that time frame for their third most common language. Hospitals were more challenged to inform patients of their right to linguistically accessible services. Less than 50% did so via
Patients’ Bill of Rights forms and posters in other languages and in English (48% and 44%), forms or brochures in other languages and in English (40% and 29%), multilingual “I speak” posters (32%), verbally in other languages (33%), through interpreter services outreach (4%) and media campaigns (2%). Most hospitals did not meet the standard regarding use of competent interpreters either: 62% reported that family members or friends of patients were used as interpreters, even though
70% of these hospitals also reported having a policy prohibiting this practice. Most hospitals required that staff (79%) and contract (63%) interpreters undergo interpreter training, but ad-hoc interpreters, such as volunteers and bilingual staff usually did not. Finally, hospitals made the following translated documents available in their most commonly requested language: Advance directives
(65%), patients’ rights (61%), discharge instructions
(58%), informed consent (57%) and hospital signage (51%). Less than a third of hospitals had these written documents available in two or more non-English languages. Very few hospitals could provide all of these forms in their most common language (17%). Cost and lack of insurance reimbursement for interpreter services was identified as the biggest challenge hospitals face in providing adequate language services for their patients with LEP.
Conclusion: Many hospitals are not in compliance with the CLAS standards, and risk violations of federal laws requiring linguistically accessible care for patients with LEP.
Implications for Policy, Delivery or Practice:
This data provides insight into which standards should be targeted for intervention and enforcement in order to improve the care provided to LEP patients across the nation.
Funding Source(s): The Robert Wood Johnson
Clinical Scholars Program
Theme: Coverage and Access
● Use of Carotid Revascularization among
Medicare Beneficiaries, 2003-2006
Lisa DiMartino, M.P.H.; Melissa Greiner, M.S.;
Manesh Patel, M.D.; Kevin Schulman, M.D.; Lesley
Curtis, Ph.D.
Presented by: Lisa DiMartino, M.P.H., Clinical
Research Manager, Center for Clinical & Genetic
Economics, Duke Clinical Research Institute, P.O.
Box 17969, Durham, NC 27715, Phone: (919) 668-
8458; Email: Lisa.DiMartino@Duke.edu
Research Objective: In 2004, Medicare issued a
National Coverage Determination (NCD) for carotid artery stenting (CAS) as an alternative to carotid endarterectomy (CEA) for the treatment of carotid artery disease, but it is unknown whether CAS has had an impact on CEA use. The objective of this study was to examine geographic variation in rates of CEA and CAS among Medicare beneficiaries from 2003-2006.
Study Design: Inpatient, outpatient, and carrier claims for all beneficiaries who underwent CEA or
CAS for the period 2003-2006 were obtained. In addition, we obtained denominator files for 100% of
Medicare beneficiaries from 2003-2006. We identified patients undergoing carotid revascularization by searching carrier claims for evidence of CEA (HCPCS 35301) or CAS (HCPCS
37215, 37216). The U.S. average of CEA was calculated for 2003-2004 and 2005-2006. Since
Medicare coverage began in 2004, the U.S. average of CAS was calculated only for 2005-2006.
We extracted beneficiary Hospital Referral Region
(HRR) from the 2005 Dartmouth HRR boundary files, and mapped beneficiary ZIP code to 306
HRRs. Age-adjusted rates of carotid revascularization by HRR were calculated using a direct standardization method. We then calculated the ratio of each HRR rate to the U.S. average.
Population Studied: Medicare beneficiaries aged
65 and older living in the U.S. who underwent CEA or CAS between 2003 and 2006. Only beneficiaries enrolled in fee-for-service Medicare were included in the analysis.
Principal Findings: There were 174,047 CEAs
(3.18 per 1000) performed in 2003-2004; 146,307
CEAs (2.69 per 1000) in 2005-2006; and 19,444
CAS (0.36 per 1000) performed in 2005-2006. In
2003-2004, there was nearly a nine-fold difference between the highest (7.17 per 1000 in Beaumont,
Texas) and lowest (0.82 per 1000 in Honolulu,
Hawaii) CEA rates. In 2005-2006, the age-adjusted rate of CEA ranged from 5.5 per 1000 in Beaumont,
Texas to 0.79 per 1000 in Honolulu, Hawaii. More than half of HRRs had CAS rates that were at least
50% below the U.S. average. The highest CAS rate observed was 2.73 per 1000 in Saint Joseph,
Michigan. Five HRRs that had CEA rates at least
50% above the U.S. average in 2003-2006 also had high CAS rates (Hattiesburg, Mississippi; Houma,
Louisiana; Joplin, Missouri; Kalamazoo, Michigan; and Lawton, Oklahoma). In contrast, several
HRRs with high CEA rates in 2003-2006 had lower
CAS rates than the U.S. average. Two HRRs with the largest decrease in CEA rates from 2003-2004 to 2005-2006 had CAS rates markedly higher than the U.S. average (Elyria, Ohio and Munster,
Indiana).
Conclusion: There was considerable geographic variation in CEA use both before and after Medicare issued a NCD for coverage of CAS. In some regions, CEA rates remained constant, suggesting that CAS was adopted slowly or that it was used in beneficiaries who were not eligible for CEA. In other regions, CAS appears to be used as a substitute for CEA, as indicated by a decrease in
CEA rates.
Implications for Policy, Delivery or Practice: Our findings signify that it will be important to monitor outcomes in order to better understand the effect of
CEA and CAS in the treatment of carotid artery disease.
Funding Source(s): AHRQ
● An Innovative Approach to Address the
Nursing Faculty Shortage: Initial Findings from the First Cohort of the VA Nursing Academy
Aram Dobalian, Ph.D., J.D.; Candice Bowman,
Ph.D., R.N.; Tamar Wyte, D.P.T., M.P.H.; Lynn
Soban, Ph.D., R.N.; Barbara Simon, M.A.; Jack
Needleman, Ph.D.
Presented by: Aram Dobalian, Ph.D., J.D.,
Research Health Scientist, COE Healthcare
Provider Behavior, Department of Veterans Affairs,
16111 Plummer Street (152), Sepulveda, CA
91343, Phone: (818) 891-7711 x 7182; Email: aram.dobalian@va.gov
Research Objective: HRSA has estimated that by
2015 all 50 states will experience a shortfall of nurses. By 2010, 22,000 Registered Nurses (RNs) will be eligible for retirement within the Department of Veterans Affairs (VA). It has been suggested that much of this shortfall is due to a lack of qualified nursing faculty. Consequently, the VA established the VA Nursing Academy, a 5-year pilot program that will fund faculty positions at 14 competitively selected partnerships of VA facilities and nursing schools. We are conducting a national evaluation of the program to provide evidence to support the long-term growth and dissemination of successful models for expanding nursing faculty, enhancing professional development, increasing student enrollment, promoting innovations in education and practice, and increasing recruitment and retention of nurses within VA.
Study Design: We conducted semi-structured interviews that assessed the partnerships’ structures and the consequences of their activities during the Academy’s first year (2007-2008).
Population Studied: Nursing leadership, local project directors, nursing faculty, nurse managers, and nursing staff at the first four partnerships.
Principal Findings: Most interviewees expressed significant support and enthusiasm for their local programs. Each partnership increased faculty size and student enrollment. Interviewees noted that having clinical experiences at the VA allowed students to observe nurses in a variety of roles, including advanced practice and managerial positions, thus highlighting a range of career opportunities. We identified two major barriers to the program’s implementation: (1) a “culture clash” that defined various challenges in assimilating faculty into the broader organizational structure of the VA and the academic culture of the schools; and (2) challenges around integrating new faculty that related to difficulties in maintaining effective communication among stakeholders within and between the partners. Significant resources were devoted to addressing these challenges. In contrast, we identified a “spill-over” effect in that the
Academy strengthens existing ties between the partners, and “seeds innovation” in a manner that would not have otherwise occurred. Variations in the expertise, commitment, and role stability of the primary stakeholders at each of the partnerships in combination with the degree of support from the local bureaucratic structures, appear to affect the partnerships’ ability to efficiently address barriers and leverage facilitators.
Conclusion: Discovering innovative ways to negotiate the intersection of such disparate organizational structures and cultures seems to have been a major focus of the partnerships’ first year. The program’s efforts to promote the recruitment of newly licensed RNs through a focus on clinical teaching at VA hospitals with similar organizational structures allows students to spend less time learning how different hospitals function administratively and more time studying clinical care. We expect to see the “seeds” of innovation to grow in successive years.
Implications for Policy, Delivery or Practice: The
VA Nursing Academy may potentially serve as a model for how best to design and target policies to expand clinical training capacity in the fastest and most cost-effective manner, structure school-clinical site relationships to enhance the use of evidencebased practice, and improve recruitment and retention of hospital staff. This model may be adaptable to address shortages in other health professions.
Funding Source(s): VA
Theme: Health Care Workforce
● Understanding Variations in the IRB Review
Process Among VA & University-Affiliated IRBs
Aram Dobalian, Ph.D., J.D.; Brian Mittman, Ph.D.;
Barbara Simon, M.A.; Laura York, M.A.; Cynthia
Gammage, B.A.; Deborah Mittman, M.P.S.
Presented by: Aram Dobalian, Ph.D., J.D.,
Research Health Scientist, COE Healthcare
Provider Behavior, Department of Veterans Affairs,
16111 Plummer Street (152), Sepulveda, CA
91343, Phone: (818) 891-7711 x 7182; Email: aram.dobalian@va.gov
Research Objective: The existence of variation among IRB review processes and outcomes is welldocumented. However, the underlying structures and processes that lead to variation are understudied. This project was designed to improve our knowledge regarding underlying sources of variation among IRBs that review research funded by the Department of Veterans Affairs (VA).
Study Design: We undertook three main activities:
(1) literature review to assess the range of variations in IRB processes and outcomes; (2) development of instruments for qualitative interviews based on the literature review and consultation with experts; and (3) interviews with
Principal Investigators (PIs) and IRB staff and members to determine their views regarding IRB variations and factors contributing to variations.
Population Studied: We identified eligible studies at 6 VA sites and contacted PIs to obtain their agreement to participate in this study. We conducted semi-structured interviews with the PIs to assess their perspectives on the IRB process for the identified studies. We also conducted in-depth qualitative interviews with (1) IRB chairs and members, and (2) IRB administrators and coordinators. Eligible IRBs included both VA-based
IRBs and University-based IRBs that are affiliated with VA facilities. In total, 22 PIs across 3 groups
(clinical, health services research, quality improvement), and 5 IRB staff and members were interviewed. The study team compiled a metamatrix that captured the collected data for analysis of themes. Data reduction and coding were performed as a team.
Principal Findings: Respondents identified a number of sources of variations in IRB review processes and outcomes, including: a lack of expertise on the IRB committees for certain types of research (e.g., social sciences, health services research, surveys, quality improvement, qualitative research, exploratory research); the lack of a systematic approach to weigh risks and benefits that is consistent across studies and across time; that IRB decisions may be influenced by institutional risk aversion; that IRBs are more concerned with the content of the consent document rather than the process or the readability of the consent; that regulations sometimes provide conflicting, inadequate or no guidance; unclear or duplicative forms or instructions; inconsistencies in the level of review (exempt, expedited, full board); and limited resources within the IRB, a problem that some indicated had been exacerbated by recent changes in data security reporting requirements within VA. These findings were generally endorsed by all groups.
Conclusion: This project represents the first attempt to evaluate sources of variations in the IRB review process in VA settings. The consequences of variations in IRB review (e.g., delays in study timelines) are significant for all research, but are particularly important in studies addressing growing, high-priority treatment challenges facing VA (e.g., polytrauma, TBI, PTSD) for which the current treatment evidence base is still developing.
Implications for Policy, Delivery or Practice: IRB variations can adversely affect research designs and procedures, add unanticipated costs, introduce protocol variability that may affect response rates, cause delays in project initiation and completion
(including delays in implementing interventions designed to improve care), and may impair rather than support the protection of human subjects.
Programs to reduce inappropriate variations among
IRB reviews are warranted.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Clinical vs. Corporate Accountability for
Performance: A Multiple Case Study of
Comprehensive Cancer Centers in the United
States & Canada
Mark Dobrow, Ph.D.
Presented by: Mark Dobrow, Ph.D., Scientist /
Assistant Professor, Cancer Services & Policy
Research Unit / Health Policy, Management &
Evaluation, Cancer Care Ontario / University of
Toronto, 620 University Avenue, Toronto, M5M
1S8, CA, Email: mark.dobrow@utoronto.ca
Research Objective: The quest to improve quality of care and health care system performance has mobilized a burgeoning health services research community. Through parallel efforts to establish clinical practice guidance (what should be done) and measure performance (what is done), health services research has provided an increasingly informative and revealing knowledge base for assessing a wide variety of health services. This has contributed to an increasing emphasis on accountability, a natural extension of better informed expectations and enhanced understanding of results, as a key component of performance improvement efforts. However accountability, while fundamental to performance improvement efforts, is a complex and elusive concept that is often given only perfunctory attention. While efforts to define high performance are important, complementary efforts are needed to examine mechanisms for improving accountability for performance. The main objective of this study was to improve understanding of accountability for
health system performance by comparing/contrasting the defining characteristics of clinical and corporate accountability structures/systems in comprehensive cancer centers and examine how these accountability structures/systems are aligned and/or integrated to achieve performance improvement objectives.
Study Design: Embedded multiple case study design employing document analysis and key informant interviews.
Population Studied: Cases represent four comprehensive cancer centers in the US and two in
Canada. Key informants include senior clinical and administrative leaders from each cancer center.
Principal Findings: Clinical accountability reflects the consequential relationships of individual clinicians with individual patients, clinical colleagues, professional regulatory colleges and associations, health care organizations, employers, public/private payers, legal systems and/or governments. Clinicians can be held accountable for professional competence, clinical and financial performance, adherence to best practice and/or legal and ethical conduct. Corporate accountability reflects the multiple accountabilities administrators have within health care systems, including fiscal, legal, ethical and political responsibilities amidst a diverse group of public and private sector stakeholders, like clinicians, managers, payers, suppliers, boards, unions and the research community. The presentation will provide further insights on the nature of, and inter-relationships between, corporate and clinical accountability as they relate to performance improvement objectives in the various comprehensive cancer center contexts as well as the implications of different health system models in the US and Canada.
Conclusion: Despite the longstanding use of the term accountability, there is remarkably limited understanding of the concept, including how it is defined, how it is operationalized in practice, and how it can support or potentially constrain policymakers and clinicians in their efforts to improve quality of care and health system performance.
While a one-dimensional focus on either clinical or corporate accountability is inadequate to account for the myriad factors influencing performance within complex health care systems, more attention is needed to improve understanding of this fundamental component of performance improvement.
Implications for Policy, Delivery or Practice: The study findings are directed at, and should be relevant to, several specific audiences, including clinical and administrative cancer/health system leaders at regional, national and international levels tasked with the responsibility of improving health system performance. The study should inform future research and the design, structure and implementation of emerging performance management systems.
Funding Source(s): CWF
Theme: Organizational Performance and
Management
● Availability of Data to Measure Disparities in
Leading Health Indicators at the State & Local
Level
Allison Hedley Dodd, Ph.D.; Marsha Gold, Sc.D.;
Melissa Neuman, M.S.
Presented by: Allison Hedley Dodd, Ph.D., Senior
Health Researcher, Mathematica Policy Research,
600 Maryland Avenue SW, Suite 550, Washington,
DC 20024, Phone: (202) 484-4687; Email: ahedley@mathematica-mpr.com
Research Objective: Healthy People 2010 identifies the elimination of heatlh disparities as a critical national goal. Efforts by states and localities are critical to achieving this goal. Our research examined whether data to measure disparities are available at this level of aggregation to support work to eliminate disparities.
Study Design: We assessed state data availability for the 10 leading health indicators (LHIs), a set of
26 measures defined as priority in Health People
2010. The analysis involved a systematic review of information available on federal and state web sites.
Analysis is current as of mid 2007.
Population Studied: We systematically reviewed which indicators are available for each state, the data source and timeliness, and the population and geographic subgroup estimates they support.
Principal Findings: Federal data sources allow aggregate state estimates for 24 of the 26 LHIs, though some are not available for all states or vary from the federal definition. National sources capture some but not all of the subgroups defined as national priorities and data by geographic subdivision often are lacking. States also vary in how aggressively they use these data and make them available externally. Work to eliminate disparities within states is at an early stage though some states are in the lead.
Conclusion: While relevant state-level disparities data exist, major gaps remain, local estimates are limited, and some states make better use of data than others.
Implications for Policy, Delivery or Practice:
Federal leadership has been critical to state capacity to assess LHI disparities. Such leadership and support is likely to be even more critical in the future in supporting effective federal-state-local collaboration to achieve national disparities goals.
Funding Source(s): RWJF, HCFO Program
Theme: Public Health
● A More Complete Picture on Access to
Medical Care by U.S. Patient Groups
Martey Dodoo, Ph.D.
Presented by: Martey Dodoo, Ph.D., Chief
Economist, Robert Graham Center, 1350
Connecticut Avenue, NW, Suite 201, Washington,
DC 20036, Phone: (202) 331-3360; Email: mdodoo@aafp.org
Research Objective: Use nationally representative data to identify patient groups reporting relatively low levels of access to medical care, and determine whether those levels are dwindling. Most past attempts to measure access to medical care services analyzed aggregate insurance groups, ignored cost, affordability, and access to auxiliary medical services like prescription medications. This study aims at assessing the consequences of these omissions, by carrying out a more comprehensive analysis
Study Design: Analyzed data from 2003 CTS in attempt to replicate earlier studies, and 2001-2006
NHIS to confirm results from CTS and assess if there are any time trends. Created two measures of access in 12 months prior, (1) “Delayed getting medical care because of any of 5 reasons?” (2)
Delayed getting care or not get care due to cost.
Needed mental healthcare or counseling, or prescription medicine but could not afford them.
Analyzed 7 patient insurance groups, trends (2001-
2006), and assessed odds of access problems using regression techniques.
Population Studied: Two nationally representative health surveys (2001-2006 NHIS and 2003 CTS) of
U.S. population.
Principal Findings: In 2006, 18.4% disabled
Medicare beneficiaries, and 14.4% Medicaid beneficiaries reported barriers to getting needed care, compared to 8.6% privately insured elderly
(55-64 years old). 39.1% disabled Medicare beneficiaries, and 21.9% Medicaid beneficiaries also report they could not afford needed auxiliary medical care services like prescription medication, compared to 7.4% privately insured elderly.
Similarly 9.9% aged Medicare beneficiaries report having cost and affordability problems. Trends in the proportions lacking access were mostly flat
(2001-2006) for the patient subgroups. Odds of someone facing care access barriers are increased almost two fold for the poor, and more than two and half fold for those in poor health.
Conclusion: Despite results from earlier studies that there are generally no patient access problems, we found many patient groups from two nationally representative samples reporting significant problems in gaining access to medical care.
Implications for Policy, Delivery or Practice: :
Even though past studies apparently found no access problems for some patients, our more comprehensive analysis unearthed significant barriers for many patient sub-groups.
Comprehensive analysis of medical access like that undertaken here is clearly needed for appropriate policy
Theme: Coverage and Access
● A Positive Deviance Approach to
Understanding the Cancer-Related
Psychosocial Distress Literature
Caroline Carney Doebbeling, M.D., M.S.; Nadine,
Najjar, M.S.; Kathleen Abrahamson, R.N., Ph.D.;
Katherine Schilling, M.L.S., Ed.D., A.H.I.P.; Lori
Losee, M.A.
Presented by: Caroline Carney Doebbeling, M.D.,
M.S., Research Scientist at Regenstrief Institute,
Inc. & Director, Healthcare Evaluation, Research,
Outcomes, & Quality; Office of Medicaid Policy &
Planning, Healthcare Evaluation, Research,
Outcomes & Quality, Office of Medicaid Policy &
Planning, 402 West Washington Street, Room
W374, MS07, Indianapolis, IN 46204-2739, Phone:
(317) 233-2945; Email:
Caroline.CarneyDoebbeling@fssa.IN.gov
Research Objective: Psychosocial distress resulting from cancer diagnosis is common, occurring in up to half of patients depending on malignancy type and stage. Mental distress (e.g. depression, anxiety) as well as social distress (e.g. transportation, job loss, family discord, economic compromise) affect many aspects of a cancer patient’s life. As the cancer-related psychosocial distress literature has not been organized into a comprehensive model, this study applies a positive deviance (PD) model to the literature. PD is an approach based on the idea that solutions to optimize resources already exist, and can be applied more broadly to organizations that do not provide comprehensive psychosocial distress screening nor services.
Study Design: We undertook a systematic review of the medical literature from 2000 to 2008 using
MEDLINE, CINAHL and Psyc INFO. Articles were collected under thematic categories including
Distress, Psychosocial Interventions, Complexity
Theory Application, and Recommendations based on the Positive Deviance Model.
Population Studied: The population studied included individuals in the medical literature who suffered from cancer.
Principal Findings: Psychosocial distress is complex, and ranges from emotional to spiritual, to practical matters of daily living. Specific populations are at greater risk for distress, and the same methods of recognition and treatment may not fit all patients. Less than 2/3 of providers screen for distress, rarely using standardized tools.
The consequences of failure to recognize and treat distress can be serious and expensive; resulting in failure to receive needed treatment, over utilization of emergency services, and diminished quality of life. Psychosocial interventions have been shown to be effective in decreasing cancer-related distress for some patients. Few studies exist regarding the dissemination of best practices to a variety of cancer treatment settings. Organizational
mechanisms to successfully overcome common barriers have rarely been reported.
Conclusion: Despite strong evidence about the prevalence and outcomes of distress, the literature is bereft of best practices and positive deviance solutions. There exists an on-going challenge to discover these practices at the clinical and organizational level which will benefit the cancer population in a cost-effective manner.
Implications for Policy, Delivery or Practice:
There is to date no evidence regarding the application of national policies or funding models for psychosocial care. Next steps should include the application of the PD to community interviews of cancer care providers. Hence, practitioners will ultimately be provided with cost effective strategies that can potentially be transferred to a wide variety of settings and patients.
Funding Source(s): Regenstrief Foundation
Theme: Behavioral Health
● The Aging of the Generalist Physician
Workforce: Are Rural Locations at Higher Risk?
Mark Doescher, M.D., M.S.P.H.; Meredith Fordyce,
Ph.D.; Susan Skillman, M.S.
Presented by: Mark Doescher, M.D., M.S.P.H.,
Director, Washington, Wyoming, Alaska, Montana,
& Idaho Rural Health Research Center, Family
Medicine, University of Washington, 4311 Eleventh
Avenue, Northeast, Suite 210, Seattle, WA 98105,
Phone: (206) 616-9207; Email: mdoesche@u.washington.edu
Research Objective: This study quantifies the extent to which rural generalist physician shortages may be exacerbated by impending physician retirement. Generalist physicians are an essential component of rural health care delivery, but many rural communities struggle to recruit and retain sufficient numbers of generalists to meet health care needs. As fewer young physicians choose generalist careers, the retirement of older physicians may place additional strain on the supply of rural generalists.
Study Design: Cross-sectional assessment of allopathic and osteopathic physician supply employing data from the American Medical
Association and American Osteopathic Association
2005 Masterfiles.
Population Studied: The US population of rural generalist physicians (family/general practitioners, general internists, general pediatricians) was identified. This study focused on those nearing retirement, defined as physicians aged 56 years or older who were clinically active (n=27,746).
Federally-employed physicians and those in residency training were excluded from analyses.
Proportions and generalist/population ratios were determined at the national-, state-, and county-level for metropolitan (“urban”) and nonmetropolitan
(“rural”) locations. Counties falling into the upper quartile of generalist age distribution were termed
“aging generalist” locations, because a large proportion of their practicing physicians were nearing retirement age. Counties lacking generalist physicians were termed “no generalist” locations.
Principal Findings: Nationally, rural areas had a higher percentage of generalists nearing retirement than urban locations (27.6% versus 25.6%, p<.0001). This percentage increased to 29.1% in remote rural locations. At the state-level, the proportion of generalists nearing retirement varied, ranging from a high of 42.1% in Massachusetts to a low of 20.9% in South Dakota. Three particularly high-risk states were identified: Florida, Nevada and
Oklahoma. These states had both a high proportion of older, rural generalists (33.0%, 31.0% and
32.5%, respectively) and a low overall rural generalist/population ratio (46, 46 and 50 per
100,000, respectively). Of the 2,052 rural counties in the US, 176 (8.6%) were “aging generalist” locations and 167 (8.1%) were “no generalist” locations. Compared to all other rural counties,
“aging generalist” counties had more minority group members, greater poverty, lower levels of education, lower population density and a lower generalist/population ratio, while “no generalist” counties had more elderly persons, lower population density and greater population loss.
Many of the no generalist counties were located in the Great Plains.
Conclusion: Rural generalist physician shortages will be exacerbated in the coming years by impending physician retirement. Furthermore, retirement-related attrition will have the largest impact on locations in which the burden of illness is high, including areas with substantial racial/ethnic minority populations and high rates of poverty.
Implications for Policy, Delivery or Practice: To reduce the impact of retirement-related attrition, public and private investment to bolster the rural generalist physician workforce supply could be directed preferentially to locations currently without generalist physicians and locations with high proportions of generalists who are approaching retirement.
Funding Source(s): HRSA
Theme: Health Care Workforce
● Where Do Veterans with Problem Drinking
Receive Health Care? A National Study
Mark Doescher, M.D., M.S.P.H.; J. Elizabeth
Jackson, M.A.
Presented by: Mark Doescher, M.D., M.S.P.H.,
Director, Washington, Wyoming, Alaska, Montana,
& Idaho Rural Health Research Center, Family
Medicine, University of Washington, 4311 Eleventh
Avenue, Northeast, Suite 210, Seattle, WA 98105,
Phone: (206) 616-9207; Email: mdoesche@u.washington.edu
Research Objective: The extent to which veterans who drink heavily or binge drink receive health care in VA facilities as opposed to non-VA sites has not been reported. This national study examines the prevalence of any drinking, heavy drinking, and binge drinking among Veterans Administration
(VA)-eligible adults, positing that the prevalence of each of these behaviors is higher among VAeligible adults than their non-veteran counterparts.
It subsequently explores the extent to which VAeligible adults who drink heavily or binge drink receive health care at VA facilities as opposed to non-VA settings.
Study Design: This cross-sectional prevalence study used national data from the 2003 and 2004
Behavioral Risk Factor Surveillance System
(BRFSS) to examine the likelihood of VA-eligible adults reporting: any drinking; heavy drinking (daily average of more than two drinks for men and more than one for women); and/or binge drinking
(consuming at least five drinks at one sitting).
Among VA-eligible adults, the proportion with problem drinking behaviors being seen at VA health care facilities vs. other sites of care was assessed.
Analyses controlled for sociodemographic characteristics and health status. Standard errors were adjusted to account for the complex sample design of BRFSS. Weights were employed to produce nationally representative estimates.
Population Studied: This study examined a nationally representative sample of noninstitutionalized adults aged 18 and older
(n=554,630), of which 73,113 were retired or discharged veterans who were eligible for VAservices.
Principal Findings: Among VA-eligible adults, the unadjusted prevalence of drinking was, as follows: any drinking- 59.2%; heavy drinking - 5.4%; and binge drinking - 13.6%. In the adjusted analyses comparing VA-eligible respondents to non-veterans, a significantly higher prevalence of any drinking
(Adjusted OR [AOR] 1.05; 95% Confidence Interval
[CI] 1.01, 1.09) was observed; the difference was of borderline significant for heavy drinking (AOR 1.08;
CI 1.00, 1.17), and not significant for binge drinking
(AOR 0.96; CI 0.91, 1.01). Among VA-eligible adults, the unadjusted proportion that received some or all health care at a VA facility was 17.3%.
Among the subset reporting heavy drinking, this proportion was 15.9% and among those reporting gbinge drinking, it was 14.3%. However, no significant relationship between site of care (VA facility or not) and heavy drinking (AOR 0.96; CI
0.79, 1.16) or binge drinking (AOR 0.94; CI 0.80,
1.06) was observed after adjusting for covariates.
Conclusion: After adjustment for sociodemographic factors, a higher proportion of
VA-eligible adults reported any drinking and heavy drinking than their non-veteran peers. However, the prevalence of binge drinking did not vary significantly by veteran status. Most VA-eligible adults, including those with problem drinking behaviors, receive care outside the VA system.
Implications for Policy, Delivery or Practice: VAinitiated alcohol treatment efforts should address the needs of VA-eligible adults, the majority of whom are seen outside of the VA clinical system.
Partnership with non-VA health care entities might help veterans receive needed alcohol treatment.
Funding Source(s): Funded in part through the
Department of Veterans Affairs
Theme: Military and Veterans Health Care
● Are Rural Cancer Patients Accessing
Recommended Radiation Therapy?
Mark Doescher, M.D., M.S.P.H.; Laura-Mae,
Baldwin, M.D., M.P.H.; Shilpen Patel, M.D.; C. Holly
Andrilla, M.S.; Roger Rosenblatt, M.D., M.P.H.
Presented by: Mark Doescher, M.D., M.S.P.H.,
Director, Washington, Wyoming, Alaska, Montana,
& Idaho Rural Health Research Center, Family
Medicine, University of Washington, 4311 Eleventh
Avenue, Northeast, Suite 210, Seattle, WA 98105,
Phone: (206) 616-9207; Email: mdoesche@u.washington.edu
Research Objective: This research examines the degree to which rural cancer patients meet recommended radiation therapy guidelines for common cancers. Nearly 40% of colorectal cancer patients in small and isolated small rural areas and
30% in large rural areas travel over 50 miles to receive radiation oncology consultation, raising concerns about the adequacy of cancer treatment for rural populations requiring radiation therapy.
Study Design: Cross-sectional comparison of unadjusted radiation therapy use rates between rural and urban patients. Logistic regression analysis to examine whether rural/urban differences persisted after controlling for regional variation
(state) and cancer type, as well as demographic
(age, sex, race/ethnicity, marital status) and health system factors (e.g., county-level radiation oncologist availability). Breast cancer patients dominated the samples, and were analyzed separately.
Population Studied: 117,722 urban and 16,591 rural cancer patients included in 10 states’
Surveillance, Epidemiology, and End Results
(SEER) cancer registries between 2000 and 2004.
Study patients were diagnosed with one of seven cancers for which the National Comprehensive
Cancer Network guidelines recommend radiation therapy –DCIS and T1, T2, T3, and T4 breast cancer treated with lumpectomy; stages IB2, II, and
III cervical; stage III non-small cell lung; stages IA,
II, and III small cell lung; stages I, II, and III pancreatic; and stages II and III rectal cancer; stages II and III anal cancer.
Principal Findings: Overall, 68.1% of breast cancer patients with lumpectomy received radiation therapy. After adjustment, urban breast cancer
patients had a statistically significant 40% (C.I.
33%-48%) higher odds of receiving radiation therapy than rural patients. About a quarter of this difference was explained by county-level radiation oncology service availability (adjusted odds of urban breast cancer patients receiving radiation
1.31 times that of rural patients [C.I. 1.23-1.39]).
Overall, 60.3% of patients with other types of cancers (range from 30.4% to 89.4% depending on cancer type) received recommended radiation therapy. After adjustment, urban patients with other study cancers had a statistically significant 12%
(C.I. 3%-21%) higher odds of receiving radiation therapy than rural patients. About a third of this rural-urban difference was explained by countylevel radiation oncology service availability
(adjusted odds of urban patients with other study cancers receiving radiation 1.09 times that of rural patients [C.I. 0.99-1.20]).
Conclusion: Rural cancer patients are less likely than their urban counterparts to receive guidelinerecommended radiation therapy. Availability of radiation oncology services only partially explains this disparity.
Implications for Policy, Delivery or Practice:
Policy changes that address geographic access will manage a portion of the rural-urban disparity in radiation therapy receipt. Further research to increase our understanding of the role of patient preference as well as other health systems, provider, and contextual factors in explaining this rural-urban disparity is needed to identify policy, health delivery, and practice changes that will improve cancer care for rural populations.
Funding Source(s): HRSA
Theme: Disparities
● Does Higher Quality of Care in Family
Practices Reduce Emergency Hospital
Admissions? Evidence from Diabetes
Management in the UK’s Quality & Outcomes
Framework
Tim Doran; Mark Dusheiko; Hugh Gravelle; Martin
Roland
Presented by: Tim Doran, Clinical Research
Fellow, National Primary Care Research &
Development Centre, University of Manchester,
Williamson Building, Oxford Road, Manchester,
M13 9PL, UK, Phone: +44 161 275 7658; Email: tim.doran@manchester.ac.uk
Research Objective: The Quality and Outcomes
Framework (QOF), introduced in 2004, links substantial financial rewards for UK family practices to a range of quality of care indicators. This pay-forperformance scheme, which includes three quality indicators for glycaemic control in diabetics, was intended to improve outcomes for patients with chronic conditions, including reducing admissions to secondary care. We aimed to investigate the association between achievement of quality indicators for glycaemic control and emergency admissions to secondary care for short term complications of diabetes.
Study Design: We analyzed data extracted from the clinical computing systems of family practices in
England on achievement of quality indicators during the first three years of the pay-for-performance scheme (2004-05 to 2006-07). We also analyzed hospital admissions data for patients registered with these family practices between 2000-01 and 2006-
07. We performed panel data multiple regressions of practice yearly emergency admission rates for short term diabetic complications on the proportions of diabetic patients with good (HbA1c = 7.5%), moderate (HbA1c = 10%), and poor (HbA1c > 10%) glycaemic control; diabetes prevalence; baseline hospital admission rates; and attributed practicelevel socio-economic, demographic, and geographic characteristics.
Population Studied: 2.0 million diabetic patients attending 8288 English family practices between
2000-01 and 2006-7.
Principal Findings: Over the first three years of the pay-for-performance scheme (2004-05 to 2006-
07), the percentage of registered diabetics whose
HbA1c levels were measured increased from 91.1% to 93.9%. Of these, the percentage with good glycaemic control increased from 51.3% to 59.2%, and the percentage with poor control fell from
16.8% to 13.3%. Over the same period recorded prevalence of diabetes increased from 3.4% to
3.7%. The rate of emergency diabetic admissions increased from 5.1 to 5.9 per 10,000 person years between 2000-01 and 2003-04 (pre-QOF), and to
6.8 per 10,000 person years in 2006-07 (post-
QOF). Post-QOF, higher emergency admission rates were associated with higher rates of risk factors (including smoking and obesity); a higher proportion of patients over age 75; income deprivation in the practice locality; and greater average distance of patients from the family practice. On regression modeling, emergency admission rates for all short term diabetic complications were significantly lower in practices with higher proportions of patients with good and moderate glycaemic control. A 10% increase in the proportion of diabetic patients with good control was associated with a reduction in emergency hyperglycaemic admission rates of 11.5%. There was no statistically significant effect of tighter control on hypoglycaemic complications.
Conclusion: The UK’s pay-for-performance scheme was introduced against a background of rising diabetes prevalence and an increasing rate of emergency diabetic admissions to secondary care.
Rates of good glycaemic control generally improved under the scheme, and this was associated with reductions in emergency admissions for short term complications of diabetes. There was no evidence that the financial incentive to reduce HbA1c levels led to an increase in hypoglycaemic admissions.
Implications for Policy, Delivery or Practice:
Calculations of the cost-effectiveness of financial incentive schemes in primary care should take account of reduced secondary care costs.
Funding Source(s): UK Department of Health
Theme: Quality and Efficiency: Policies and
Incentives
● Influence of Physician Advice on Weight Loss
Behaviors: Racial & Ethnic Difference Among
Obese Women in the United States
Rashida Dorsey, Ph.D., M.P.H.; Lorna Haughton
McNeill, Ph.D., M.P.H.
Presented by: Rashida Dorsey, Ph.D., M.P.H.,
Program Analyst, Office of the Assistant Secretary for Planning & Evaluation, Department of Health &
Human Services, 200 Independence Avenue,
Southwest, Hubert Humphrey Building, Room
446F.7, Washington, DC 20201, Phone: (202) 690-
5925; Email: rashida.dorsey@hhs.gov
Research Objective: Non-Hispanic black (NHB) and Hispanic (HISP) women are more likely to be obese than non-Hispanic white (NHW) women and less likely to engage in weight loss behavior. This study examined the association between physician advice and the adoption of weight loss behaviors.
Study Design: A cross-sectional analysis of the
2006 National Health Interview Survey was performed. All data were self-reported. Body mass index (BMI) was calculated from height and weight
(kg/m2). Weight loss behaviors included losing weight, increasing exercise, and changing diet over the past year. Physician advice included recommendation to lose weight, increase exercise, and change diet over the past year. Logistic regression was conducted with models stratified by physician advice and adjusted for age, education, health insurance coverage and number of physician visits.
Population Studied: Our sample consisted of obese (BMI =30) NHW, NHB, and HISP women
(N=3,250).
Principal Findings: In this nationwide sample of obese women, 78%NHW, 73% NHB, and 66%
HISP women reported losing weight in the previous year; 57%NHW, 57% NHB, and 53% HISP women reported increasing exercise in the previous year; and 69% NHW, 62% NHB, and 62% HISP reported changing their diet in the previous year. Additionally
49%NHW 49%NHB and 47%HISP women reported their physician advised them to lose weight;
50%NHW, 52%NHB, and 48%HISP women reported their physician advised them to increase exercise; and 41%NHW, 47%NHB, and 45%HISP women reported their physician advised them to change their diet. Overall, obese women who received physician advice were more likely to report weight loss behaviors. However, compared to NHW women, NHB and HISP women who did not have physician advice to lose weight, increase exercise, or change diet were less likely to lose weight
(AOR=0.7, 95% CI:0.5-0.9 for NHB; AOR=0.6, 95%
CI:0.5-0.8 for HISP), increase exercise (AOR=0.7,
95% CI:0.5-1.0 for NHB; AOR=1.0, 95% CI:0.7-1.2 for HISP), and change diet (AOR=0.7, 95% CI:0.5-
0.9 for NHB; AOR=0.7, 95% CI:0.4-0.8 for HISP), respectively. Among obese women given advice from physicians, there were no racial and ethnic differences in weight loss behaviors.
Conclusion: The data from this study suggest that not receiving weight loss behavior advice from physicians may be more disadvantageous for obese NHB and HISP women than for obese NHW women.
Implications for Policy, Delivery or Practice:
This study highlights potential need to further investigate the role of physician advice on racial and ethnic differences weight loss behavior among obese women.
● The Medicare Modernization Act & Medicare
Part B Drug Utilization & Spending
Jalpa Doshi, Ph. D.; Andrea Puig, B.S.; Pengxiang
Li, Ph. D.
Presented by: Jalpa Doshi, Ph. D., Research
Assistant Professor of Medicine, General Internal
Medicine, University of Pennsylvania, Blockley Hall,
Room 1222, Philadelphia, PA 19104, Phone: (215)
898-7989; Email: jdoshi@mail.med.upenn.edu
Research Objective: The Medicare Prescription
Drug Improvement and Modernization Act (MMA) of
2003 not only established an outpatient prescription drug benefit or Part D program but significantly changed the reimbursement of physician dispensed outpatient drugs already covered under Part B.
Starting in January 1, 2004, reimbursement for most Part B drugs was reduced from 95% average wholesale price (AWP) to 85% AWP. Beginning in
January 1, 2005, reimbursement was further reduced to 106% of the average sale price (ASP).
This study examines trends in Medicare Part B drug utilization and spending across 2002 to 2006 given the changes in Part B drug reimbursement over time (2003-2005) and availability of alternative outpatient drugs through Part D (2006).
Study Design: The study used data from the 2002 to 2006 5% Medicare standard analytical files (SAF) which include Medicare claims and enrollment data for a 5% random sample of the Medicare population. Part B drugs were identified from
Medicare carrier SAFs using annual Health Care
Procedure Classification Codes lists from CMS.
Nationally representative estimates of Part B drug use (percent users, number of claims per user, number of units per claim) and expenditures were estimated from 2002 to 2006. Trends in expenditures for Part B drugs that had a direct replacement drug under Part D were also examined. Recently available Medicare Part D data
will be linked to further examine substitution of Part
B drugs with alternatives available under Part D.
Population Studied: Annual cross-sectional samples of Medicare beneficiaries with full year feefor-service Part A and B coverage and alive during the entire year in 2002 to 2006 (unweighted n= 1.88 to 1.96 million, weighted N=37.5 to 39.3 million).
Principal Findings: Medicare Part B payments for drugs and biologics increased from $7.9 billion in
2002 to $10.9 billion in 2006, yet the rate of increase in expenditures was remarkably slowed down after the passage of the MMA. The sharpest annual increase (24.9%) in Part B drug payments was in the pre-MMA period from 2002 to 2003.
Total payments decreased by 8.9% between 2004
(85% AWP) to 2005 (106% ASP). This decline was observed despite the fact that the number of users and total Part B drug claims increased by 6.7% and
2.9%, respectively, in the same period. Throughout this period the top 20 drugs comprised over 80% of total spending; however, the mix of drugs in this list changed after MMA reimbursement changes under
MMA. Part B drugs with substitutes under Part D comprised 14% of total spending in 2005; however, this share declined by only 3.7% after Part D in
2006.
Conclusion: The changes introduced by the MMA were associated with a remarkable slow down in the rate of increase in Part B drug expenditures despite increases in utilization rates.
Implications for Policy, Delivery or Practice:
Understanding how changes introduced by the
MMA influenced Part B drug utilization and spending is an important step towards insuring rational reimbursement policies and efficiency of the
Medicare program. The findings from this study will also bear direct implications for federal policymakers debating further changes in Part B drug payment policy and whether Part B drugs should be eventually covered under Part D.
Funding Source(s): NIA, Leonard Institute of
Health Economics
Theme: Medicare
● Impact of Psychiatric Comorbidity on
Rehospitalization, Mortality & Functional
Outcomes in Stroke Patients Following Inpatient
Rehabilitation
Almas Dossa, Ph.D.
Presented by: Almas Dossa, Ph.D., Postdoctoral
Health Services Research Fellow, Edith Nourse
Rogers Memorial VA Hospital, Center for Health
Quality, Outcomes, & Economic Research, 200
Springs Road, Bedford, MA 01730, Phone: (781)
687-2642; Email: adossa@bu.edu
Research Objective: Existing research shows that significant predictors of rehospitalization, mortality, and decreased function include patients with psychiatric diagnoses. Studies on stroke patients with these diagnoses are limited. This study explores associations between psychiatric comorbidities and important clinical and functional outcomes in stroke patients who have undergone rehabilitation in a hospital setting.
Study Design: We used the 2001 VA Integrated
Stroke Outcomes Database which included clinical, administrative, and mortality data. Outcome variables included 3 and 6-month mortality postdischarge, 3 and 6-month rehospitalization postdischarge, and functional change post inpatient rehabilitation. We defined psychiatric comorbidity as having any mental health or substance abuse diagnosis. Our models examined associations between the presence of outcomes and: psychiatric comorbidity; any mental health diagnoses or substance abuse; depression, psychosis, anxiety; and psychiatric diagnoses > 1 versus 1 diagnosis and none, as independent variables. Covariates included the Charlson index, demographics, discharge setting, length of stay, rehabilitation setting, and admission functional status.
Population Studied: The sample consisted of
2545 stroke patients with a mean age of 68 years.
Principal Findings: Logistic regressions showed that having any psychiatric comorbidity predicted 6- month rehospitalization (OR: 1.25, p = .04).
Having any mental health diagnosis was significantly associated with 6-month rehospitalization (OR: 1.09, p =.04) but substance abuse was not significant. Depression was significantly associated with 6-month rehospitalization (OR: 1.47, p =.005). Patients without psychiatric comorbidity were significantly less likely to be rehospitalized at six months compared to those with > 1 psychiatric comorbidity
(OR: .69, p = .02). Logistic regressions showed that anxiety was significantly associated with 3 and
6-month mortality (OR: 2.12, p =.02; OR: 2.37, p =
.001). Generalized linear models showed that patients with anxiety tended towards more functional change than those without anxiety (p =
.06).
Conclusion: We found weak associations between psychiatric variables and rehospitalization and mortality. Functional change was not impacted by psychiatric variables possibly because stroke patients accepted into rehabilitation programs may not have severe behavioral issues or because the
VA has successfully intervened for these problems.
Implications for Policy, Delivery or Practice:
Since rehospitalizations can impact function, quality of life, and cost of care in stroke patients, we need to continue working on successful interventions to address psychiatric comorbidity in this population.
Further research is needed on impact of psychiatric comorbidity and specific psychiatric diagnoses on functional change in older veterans with different diagnoses and with high disability risk.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Assessing the Premier - CareScience Risk
Adjusted Mortality Model in the Newborn
Population
Michael Duan, M.S.; John Martin, M.P.H.
Presented by: Michael Duan, M.S., Sr. Research
Analyst, Premier Research Services, Premier
Research Services, Premier, Inc., 3600 Market
Street, 7th Floor, Philadelphia, PA 19104, Phone:
(215) 689-2224; Email: michael_duan@premierinc.com
Research Objective: Value based purchasing is becoming more reliant on risk adjusted outcomes measures. Models have been successfully developed for use only in the newborn and infant populations; however, it is not clear if general riskadjustment methodologies for the entire inpatient population are able to predict mortality as well in these populations. To work well, inpatient risk adjustment methodologies need to appropriately account for the additional risk in the premature and low birth weight populations. To that end, the
Premier - CareScience Risk Adjusted Inpatient
Mortality Model (PCRIM) segments newborns.
Newborns with ICD-9 CM diagnosis codes 764.xx
(slow fetal growth and fetal malnutrition), 765.xx
(disorders relating to short gestation and unspecified low birth weight) and V21.3x (low birth weight status) are modeled separately from babies without those codes. To that end, the objective was to test the predictive power of a method that segregates newborns into defined subgroups to generate expected mortality outcomes using regression modeling.
Study Design: A linear probability model was used to predict mortality outcomes for the two groups.
The step-wise regression model contained patient characteristics and socioeconomic factors; however only those factors that were significant (alpha =
0.10) are included in the model. It was designed to account for patient risk of mortality based on status at the time of admission; therefore, it includes birth weight, sex, race, income, relative distance, secondary diagnoses understood to be comorbidities (e.g., respiratory failure), admission source, admission type, and payer class. The
Receiver Operating Characteristic Curve (ROC) and its accompanying c-statistic, which is often used to measure the discriminatory power of model for binary outcomes, were used to assess the model.
Population Studied: In the database used for this study (Premier inpatient data base), the low birth weight population consisted of 158,559 cases, of which, 5,430 (3.24%) died. The normal birth weight population consisted of 1,385,508 newborns, of which, 1,051 (0.08%) died. Because over 80% of mortalities in newborns occurred among low birth weight population, the predictive power of the
PCRIM model was tested on that group.
Principal Findings: Results indicated that the
PCRIM was a robust tool to predict mortalities of low birth weight newborns. The C-Statistic was as high as 0.903. The C-Statistic for a model populated with normal birth weight babies was 0.80.
Records from low birth weight newborns in a single hospital system were used to test the robustness of the model in an out-of-sample population (n=8,102).
Results indicated that the model performed well with a C-statistic = 0.921.
Conclusion: Results indicate that the PCRIM methodology was able to accurately predict mortality in the newborn inpatient population
Implications for Policy, Delivery or Practice:
Value based purchasing initiatives tie performance to payment. When outcomes measurements are used to assess performance they should be appropriately risk adjusted to account for variations in patient severity. The model tested in this study demonstrated that a general population based model sufficiently adjusts for variations in the newborn inpatient population such that it could be considered for a value based purchasing initiative.
● Race, Income, or Race & Income: Health,
Behavior & Health Care Disparities in Adults
Lisa Dubay, Ph.D., Sc.M.; Lydie Lebrun, M.P.H.
Presented by: Lisa Dubay, Ph.D., Sc.M., Associate
Professor, Department of Health Policy &
Management, Johns Hopkins Bloomberg School of
Public Health, 624 North Broadway, Hampton
House 488, Baltimore, MD 21205, Phone: (410)
502-0985; Email: ldubay@jhsph.edu
Research Objective: To examine race/ethnicity and income disparities in health, behavior, and health care use among adults.
Study Design: We used descriptive and multivariate analyses to assess race disparities within high- and low-income populations, as well as income disparities within race groups. Outcomes examined included general health status, overweight or obesity, health behaviors (e.g., exercise, alcohol consumption, tobacco use), usual source of care, and cancer and health screening patterns.
Population Studied: Nationally representative sample of US adults (n=30,852) from the 2003
National Health Interview Survey.
Principal Findings: Among low-income adults
(less than 200 percent FPL), Blacks were more likely and Hispanics less likely than whites to be in fair/poor health. Among high-income adults, however, both Blacks and Hispanics were more likely to be in fair/poor health than whites. Whites were also significantly less likely to be overweight or obese. Within high- and low-income groups, whites were more likely to exercise at recommended levels than Blacks and Hispanics.
However, whites were much less likely to be never smokers and more likely to be heavy drinkers
compared with Blacks and Hispanics. No significant differences were found between Blacks and whites at either income level in having a usual source of care or a doctor visit in the past 12 months, but Hispanics showed less access and use compared to whites. Whites were more likely to use dental care than Blacks or Hispanics. The odds of colorectal screening, receiving a pap test, and receiving a mammogram were similar for whites and Hispanics but were highest for Blacks.
White adults were more likely to receive blood pressure and cholesterol checks than Hispanics and significantly less likely to do so than Blacks.
These results were consistent across descriptive and multivariate analyses. While racial/ethnic disparities within income groups exist, they are much smaller, by an order of magnitude, than the income-based disparities found within racial/ethnic groups. For each outcome considered, large and statistically significant differences in odds were found between high- and low-income populations by race.
Conclusion: This analysis provides clear evidence that race/ethnicity differences exist within both low- and high-income populations, however, the patterns are not consistent across indicators of health, health behaviors, and health care use. In particular,
Blacks are more likely to be in poor health than whites despite their protective behavior regarding smoking and drinking and their greater use of health care screening. Results for Hispanics compared to whites are more consistent, with
Hispanics generally being worse off along all dimensions. Large and consistent disparities exist across income levels within all racial/ethnic groups, higher income populations were in better health, engaged in more protective behaviors, and reported greater use of health care.
Implications for Policy, Delivery or Practice: The exclusive policy focus on eliminating racial/ethnic disparities in may be shortsighted. There is clear evidence that low-income adults, regardless of race, are worse off than their higher income counterparts. While a focus on racial/ethnic disparities is important, it is critical that policy makers recognize the scope and size of income disparities and realize that their elimination will contribute greatly to the reduction of racial/ethnic disparities in the US.
Theme: Disparities
● PHRs & the Needs of the Medicare Population:
Findings from the Field
Prashila Dullabh, M.D.; Dan Gaylin, M.P.A.; Alison
Muckle; Casey Hogle; Randy Horton, M.I.L.S.
Presented by: Prashila Dullabh, M.D., Health IT
Program Manager, National Opinion Research
Center, Email: dullabh-prashila@norc.org
Research Objective: In 2007 the Assistant
Secretary for Planning and Evaluation (ASPE) contracted with the National Opinion Research
Center (NORC) to conduct an evaluation of a
Personal Health Record (PHR) pilot offered to CMS
Fee-for-Service (FFS) beneficiaries. The intent of the evaluation is to: 1) determine the usability/utility components of PHRs generally and specifically for
Medicare beneficiaries; 2) identify standard features and best practices in the PHR development and deployment; and 3) analyze key consumer issues and perceptions related to PHR use for Medicare
Beneficiaries.
Study Design: NORC is conducting a largely qualitative evaluation of a CMS-sponsored PHR pilot for Medicare fee-for-service beneficiaries. Key components of the evaluation include , 1) an environmental scan and literature review, 2) meetings with beneficiaries who are repeat users of the PHR, 3) a series of semi-structured interviews with non-users of the system, 4)discussions with providers in South Carolina, 5) , observational studies of Medicare beneficiaries, and 6) quantitative analysis of utilization data from the
PHR vendor.
Population Studied: The population studied included beneficiaries that have Medicare health coverage in South Carolina and signed up through
Medicare to use My Personal Health Record South
Carolina—also known as MyPHRSC
Principal Findings: We present findings on user and non-user discussions, observational studies and usage data. Beneficiary Discussions: Results from discussions with beneficiaries suggest that
Medicare beneficiaries have unique desires for a
PHR, including a need for inclusion of more comprehensive, historical information; less selfentry; medication interactions checking; and an easy login process. Strong technical support may be essential to beneficiaries’ continued use of
PHRs. Beneficiaries, although concerned about privacy, emphasized the utility of sharing information from the PHR with key trusted sources.
Analysis of PHR Usage: To gain a fuller understanding of how the PHR pilot members were interacting with the PHR and to complement the various qualitative data collection methods used in the study, NORC collected and is analyzing usage data, including a basic demographic profile of each member (including their medical conditions) and data about their usage of the PHR (logins and page views.) This analysis will identify significant longitudinal trends and patterns in the data, such as differences between heavy users of the PHR and light/non-users of the PHR in terms of their demographics (e.g. age and gender) and medical conditions. Similarly, NORC is seeking to identify and possible correlations between member demographics, health conditions and the usage of specific pages/features in the PHR. User
Observations: For the final data collection activity in this project, NORC is conducting observational studies of Medicare beneficiaries that will provide a rich set of context-based insights into how PHR
records fit into the beneficiaries’ home lives, relationships with their providers and overall healthcare management process. By conducting these observations of the beneficiaries in a "natural setting", NORC will be able to better understand the ways in which the PHR does and does not fit into the workflow of the beneficiaries’ daily lives.
Conclusion: Medicare beneficiaries have unique requirements for PHRs and early results have provided important insights into the specific design and functional components.
Implications for Policy, Delivery or Practice:
Further, larger scale pilots are suggested to better understand the needs of Medicare beneficiairies with respect to PHR usability an utility components.
Funding Source(s): Assistant Secretary for
Planning and Evaluation
Theme: Health Information Technology
● Liability Insurance for Regional Health
Information Organizations: Lessons from the
AHRQ-Funded State & Regional Demonstration
Projects & Other Community Efforts
Prashila Dullabh, M.D.; Maria Molfino, B.A.
Presented by: Prashila Dullabh, M.D., Health IT
Program Manager, MD, Email: dullabhprashila@norc.org
Research Objective: In 2008 the Agency for
Healthcare Research and Quality funded NORC as part of its work on the National Resource Center for
Health IT to gather and synthesis lessons from the field on the key considerations, issues, and challenges associated with liability insurance for regional health information exchange organizations
(RHIOs). As hosts and conveyers of clinical information, RHIOs must worry about issues such as data theft, accidental disclosure of patient information, data errors and omissions, and technology failures. Specific areas covered in this report include 1) how do RHIOs distribute liability among partners; 2) which partnering entities currently take on liabilities; 3) what are the liability concerns of RHIO partners; 4) what levels of liability coverage are appropriate; 5) which factors affect these levels of coverage; 6) how do RHIOs find and manage brokers and underwriters and 7) what are the impacts of law and government on a RHIO’s liability?
Study Design: The information in this report was gathered through an environmental scan of published and unpublished materials on liability insurance and HIE in the last five years. The NRC held discussions with the six State and Regional
Demonstration projects—the Delaware Health
Information Network (DHIN), the Rhode Island
Quality Institute (RIQI), the Indiana Patient Care
Network (INPC), the Colorado Regional Health
Information Organization (CORHIO), the MidSouth eHealth Alliance (MSeHA), and the Utah Health
Information Network (UHIN)—and two other wellestablished RHIOs in the industry: the Taconic
Health Information Network Community (THINC) in
Hudson Valley, New York, and HealthBridge in
Cincinnati, Ohio. The NRC then used a classic
“snowballing technique” to identify additional resources and/or respondents with experience and expertise in this area, as well as a core set of respondents to interview. The NRC also asked each respondent to identify other individuals who would benefit the report; respondents referred the
NRC to attorneys, insurance agents, and other fellow experts in the domain of liability insurance for discussions.
Population Studied: The focus of the study was the experiences of several regional health information organizations, in various stages of maturity. Specifically the AHRQ-funded SRD projects and two other large RHIO initiatives.
Principal Findings: When shopping for coverage:
Hire an insurance broker and reach out to your community: Several respondents discussed their approaches to finding the appropriate insurance company, agent/broker, and underwriter. As HIE continues to grow, more and more insurance companies are becoming increasingly acquainted with the amount of liability insurance required to cover a RHIO. Plan ahead with the application process. Like any insurance application, the application process can be long and onerous. In addition to its length, the application often asks for a level of detail that will require consulting both the appropriate documentation and other individuals.
Insurance should be considered at the very beginning of RHIO efforts because this will allow the roles and responsibilities of all participating organizations—i.e., data sources, users, owners— to be clearly delineated. Be prepared to educate the insurance brokers and/or underwriters. Once
RHIOs have researched their options and decided on a particular underwriter, they also need to educate the underwriters on the way that RHIOs work
Know state and federal regulations. It is important that the RHIO’s board of directors and management are well versed and briefed in the current state of the industry. In particular, understanding the RHIO’s state regulations, as well as state interpretations of federal regulations, is important to tracking the possible impact of these trends on RHIO activities.
Identify gaps in coverage. It is important to understand that underwriters and insurers may not capture all the pitfalls and loopholes of coverage.
Expect Some Administrative Burdens. Most respondents reported that managing current coverage—i.e., annual renewal of policy—is not an administrative burden.
Conclusion: In summary, the current RHIO liability landscape is affected by legal uncertainty, a lack of standard coverage methodologies, and a lack of experienced brokers and underwriters. In general the importance and weight of liability issues varies among RHIOs, the process of obtaining liability
insurance is oftentimes a long one and while insurance options are growing they remain quite limited.
Implications for Policy, Delivery or Practice:
Further work is required on the role of states in limiting liability insurance. Additionally developing good risk assessment models to assist HIEs in quantifying risk would be very helpful as they engage in discussions with their brokers.
Theme: Health Information Technology
● Enrollee Satisfaction in a Changing Medicaid
Program
R. Paul Duncan, Ph.D.; Allyson Hall, Ph.D; Lorna
Chorba; Babette Brumback, Ph.D; Jianyi Zhang,
Ph.D.
Presented by: R. Paul Duncan, Ph.D., Professor,
Department of Health Services Research,
Management & Policy, University of Florida,
P.O.Box 100195, Gainesville, FL 32610, Phone:
(352) 273-6065; Email: pduncan@phhp.ufl.edu
Research Objective: Florida Medicaid is currently operating a demonstration program in two Florida two counties.. One element of the demonstration is that enrollees choose the health plan in which their care will be managed. Participating plans are required to accept the enrollees who choose them, and enrollees can change plans only in specified circumstances. Approaches to medical care that require enrollees/consumers/patients to make choices among a range of potential providers of care presume that levels of satisfaction will influence those choices. This analysis explores the degree and manner in which enrollee satisfaction differs in the demonstration from that measured in comparable enrollees prior to the onset of the pilot program.
Study Design: The study is a pre-post design comparing findings from two cross-sectional,
CAHPS-like enrollee satisfaction telephone surveys, one conducted just prior to the demonstration and the second conducted approximately one year into the demonstration.
The “pre” sample was a stratified random sample
(n= 5,767) representative of enrollees potentially eligible to participate in the demonstration. They were queried about their experiences with care six months prior to the onset of the demonstration. The
“post” sample (n=6,209) was a stratified random sample selected a year later and representative of enrollees actually participating in the demonstration.
They were queried about their experiences with care during the preceding six month period, generally meaning the early stages of the demonstration. Instrumentation and other methodological aspects of the two surveys were virtually identical. Pre-post comparisons were made for: plan satisfaction, satisfaction with care, personal doctor rating, and specialist rating. These comparisons were done by county, race/ethnicity, and Medicaid eligibility category.
Population Studied: Florida Medicaid enrollees who were eligible to participate in the demonstration
(prior to its onset) and enrollees participating in the demonstration at about one year after its initiation.
Principal Findings: In both counties combined, the percent of individuals giving high ratings for overall satisfaction with care declined (66.54 vs 59.63 p<.0001), while the percent rating their personal doctor highly increased (70.19 vs. 73.41). In
Broward County, for the overall population, the non-
SSI group, Hispanic individuals, and White individuals, the proportion giving their personal doctor the highest rating increased. In Duval, with a few exceptions ratings did not significantly differ pre vs post.
Conclusion: Enrollee satisfaction with certain aspects of their care differed somewhat pre vs post reform. However, no definitive conclusions can be drawn about the success or failure of the reforms
Implications for Policy, Delivery or Practice:
Legislators and other policy participants should not look to enrollee satisfaction for strong and clear evidence that the demonstration is accomplishing
(or failing to accomplish) its objectives, at least in the early stages. Similarly, these analyses provide very limited support for the positions of advocacy groups, both those favoring and those opposed to the approaches being tested in the demonstration.
Funding Source(s): State of Florida
Theme: Medicaid, SCHIP and State Health Reform
● Does Organizational Culture Predict Job
Satisfaction & Surgical Outcomes ?
Edward Dunn, M.D., M.P.H.; Aartee Ignaczak,
M.P.H.; Scott McKnight, Ph.D.
Presented by: Edward Dunn, M.D., M.P.H.,
Director of Policy & Clinical Affairs, National Center for Patient Safety, Veterans Health Administration,
24 Frank Lloyd Wright Drive, Lobby M, Ann Arbor,
MI 48106, Phone: (734) 930-5872; Email: edward.dunn@va.gov
Research Objective: To examine the relationship between organizational culture with professional staff job satisfaction and patient outcomes in surgical services.
Study Design: Observational study employing cross-sectional cohort analysis of existing data.
Linear regression modeling: Depenedent variables - aggregate surgical morbidity and mortality O/E ratios for National Surgical Quality Improvement
Program (NSQIP) 2006; Independent variables: aggregate scores for Org culture and the Job
Satisfaction Index from the 2006 VA All Employee
Survey (AES); and Factor Analysis of AES org culture scores.
Population Studied: All veterans receiving surgical care in 135 VA facilities during 2006, and all VA facilities with 20 or more professional staff members
in surgical services who completed the VA All
Employee Survey in 2006. The unit of analysis was the VA Medical Center.
Principal Findings: 1. Teamwork and Innovatve organizational culture had a significant association with higher job satisfaction in surgical services at a p < 0.05 level. 2. The linear gap between bureaucratic culture (higher) and innovative culture
(lower) organizational culture scores had a significant association with higher surgical morbidity and mortality at a p < 0.05 level.
Conclusion: 1. The Competing Values Framework
(CVF) 4-culture model could be replaced by a 2culture model as a better fit for org culture data from health care organizations. 2. Teamwork and innovative culture in surgical services was associated with higher professional staff job satisfaction. 3. Surgical services with higher bureaucratic and lower innovative organizational cultures have less favorable outcomes of surgical care.
Implications for Policy, Delivery or Practice: 1.
Consolidate the CVF to a 2-culture model for survey application in health care organizations. 2.
Implement Crew Resource Management training in health systems. 3. Implement programmed and feedback coordination processes in surgical services of h. 4. Mgmt. reward teams rather than individuals. 5. Establish stable surgical teams. 6.
Product line organizational of surg. Services.
Funding Source(s): VA
Theme: Organizational Performance and
Management
● Patient Misidentification in Laboratory
Medicine
Edward Dunn, M.D., M.P.H.
Presented by: Edward Dunn, M.D., M.P.H.,
Director of Policy & Clinical Affairs, Natioinal Center for Patient Safety, Veterans Health Administration,
24 Frank Lloyd Wright Drive, Lobby M, Ann Arbor,
MI 48106, Phone: (734) 930-5872; Email: edward.dunn@va.gov
Research Objective: To identify system vulnerabilities in laboratory specimen collection, processing, analysis, and reporting that led to patient misidentification in clinical lab, anatomic pathology and blood transfusion services
Study Design: Qualitative data analysis using multipe case study design from Root Cause
Analysis (RCA) reports in the national VA database containing more than 15,000 RCAs submitted from
2000 through 2008. Data organized by three by three stages of the laboratory test cycle.
Population Studied: All veterans undergoing laboratory tests from 2000 through Septmber 30,
2008. Unit of analysis is the RCA case report from a
VA Medical Center.
Principal Findings: Patient misidentification events occurred in all three stages of the lab test cycle:
58% pre-analytic, 28% analytic and 14% postanalytic. Vulnerabilities leading to these events included patients with similar names and social security numbers, "batching" specimens and printed labels, relabeling specimens with accssion numbers, manual labeling, microscopic slides with non-unique labels from multiple patients in the same folder, and failure to use two-source identification or two-person verification in blood transfusion process.
Conclusion: Patient misidentification is a significant risk in all three phases of the laboratory test cycle. Yhr consequences of misidentification include patients with delays in necessary treatment, patients undergoing unecessary testing and surgical procedures, and reporting laboratory data to the wrong medical record with consequences for patients.
Implications for Policy, Delivery or Practice: 1.
All patients in a health system should be labeled with a unique identifier visible on their person and in their medical record. 2. Wireless bar code technology applications coupled with automated electronic processes should be considered for adoption by all health systems. 3. Introduce forcing function for two-source identification and twoperson verification in blood transfusion services process. 4. Two pathologist review of all microscopic slides with cancer diagnosis
Funding Source(s): VA
Theme: Quality and Efficiency: Organized
Processes
● An Elephant in the Room: The Role of Race in
Nurse-Physician Work Relationships
Karen Dunn Lopez, Ph.D., M.P.H., R.N.
Presented by: Karen Dunn Lopez, Ph.D., M.P.H.,
R.N., Research Associate, Medicine, The University of Chicago, 131 East Maple Street, Hinsdale, IL
60521, Phone: (630) 455-1026; Email: kdlop1@comcast.net
Research Objective: To gain understanding of the potential role that racial imbalance between nurses and physicians has in their work relationships.
Study Design: Qualitative methodology was used on two medical surgical units in an urban academic medical center. Attitudes about nurse-physician work relationships were obtained via one-on-one, face-to-face semi-structured interviews. Interviews lasted about 60 minutes (range of 25 to 150 minutes), were transcribed verbatim and analyzed inductively and deductively using directed content analysis methods.
Population Studied: Ten inpatient nurses, four hospitalists and six non-hospitalist physicians participated. Ninety- five percent of the nurse population on the study units were African
American; in contrast, 91% of the physician sample population was Caucasian.
Principal Findings: Both physicians and nurses felt that the racial imbalance between the professions negatively impacted communication and work relationships. African American nurses used race as a lens to examine unpleasant interpersonal interactions with Caucasian physicians. Nurses reported that they considered whether physician negativity might be related to the physician being “somewhat prejudiced” as the first possible cause of the negative interaction. If the nurse determined that the physician held prejudicial feelings, this strongly influenced and limited future discretionary communication between them.
Caucasian physicians viewed the racial imbalance in this institution as a source of tension in the workplace. They described awareness that they were perceived by nurses to be prejudiced if they expressed any negativity in their communication and felt powerlessness in their ability to change this perception. As a result, many physicians avoided communication with nurses that they deemed not to be emergent or essential. In addition, three forms of racism as described by Barabee (1993) were found in this research. First, “denial” in which some of the participants avoided questions about race by not discussing. Second, “color blind” is the point of view expressed by some participants that interpersonal relationships are not influenced by group membership. Third, aversive racism is characterized by ambivalence: the conflict between value of an egalitarian system and the unacknowledged negative feelings about another racial group.
Conclusion: The results of this study are consistent with previous research that has found that African Americans commonly experience both blatant and subtle forms of racial discrimination that shape interactions as well as findings that demographic diversity in the workplace is associated with more communication problems. If there are raced based communication problems in hospitals that serve a greater proportion of minorities, these communication difficulties may be one of the root causes of disparities in outcomes for hospitalized minority patients.
Implications for Policy, Delivery or Practice: The suggestion that provider race plays a role in patient centered communication and thus might impact patient outcomes merits thoughtful attention in future research. It is possible that cultural sensitivity training may be benficial in hospitals where there is a racial imbalance between physicians and nurses.
● The Feasibility of a Community Pharmacy-
Based Intervention to Reduce Barriers to
Antidepressant Adherence - Lessons Learned from a Pilot Study
Stacie Dusetzina, B.A.; Richard Hansen, Ph.D.;
Bradley Gaynes, M.D., M.P.H.; Anthony Pudlo,
Pharm.D., M.B.A.
Presented by: Stacie Dusetzina, B.A., NRSA Pre-
Doctoral Fellow, Eshelman School of Pharmacy,
University of North Carolina at Chapel Hill, 19
Beard Hall, CB#7360, Chapel Hill, NC 27599,
Phone: (919) 475-2615; Email:
Stacie_Dusetzina@unc.edu
Research Objective: Improving medication adherence can improve depression treatment outcomes. However, interventions are generally complex and are difficult to implement and sustain in clinical practice. Accruing evidence supports the role of pharmacy-based interventions in depression management. We conducted a pilot study to assess the feasibility of a community pharmacy-based intervention to improve patient adherence to antidepressants.
Study Design: We used prescription refill records to identify patients with gaps in medication supply of
7 or more days to target them for a pharmacybased intervention. The study pharmacist conducted phone interviews using a protocol designed to identify barriers to adherence (e.g., lack of efficacy, side effects, cost, or inadequate knowledge about the disease or its treatment). The pharmacist then intervened based on patients’ selfreported reason(s) for non-adherence. Feasibility of the intervention was assessed using patient participation rates, days needed to contact the patient, and time needed to conduct the intervention.
Population Studied: We included patients of a single community pharmacy who were 18 years or older, obtained prescription drugs from the study pharmacy during the previous year, and had a new prescription for a second-generation antidepressant during the previous 30 days. We excluded patients who did not speak English or who filled a prescription for an anti-dementia or anti-psychotic medication during the previous year.
Principal Findings: Twenty-four patients experienced a gap in medication supply of 7 or more days, making them eligible for the intervention. We excluded 15 patients; 12 patients who could not be contacted using the phone numbers provided to the pharmacy, 2 who declined participation, and 1 who denied filling the prescription listed in the pharmacy record, leaving 9
(37.5%) who participated in the study.
Among the patients contacted, the most commonly reported barriers to adherence were drug side effects (44%) and forgetting to take the medication
(22%). For these 9 patients, the median number of days needed for the pharmacist to contact the patient once was 14 (range: 2 - 23 days). The median time required for the pharmacist to conduct the barrier-specific intervention protocol was 18 minutes (range: 10 - 30 minutes).
Conclusion: The feasibility of our pharmacy-based intervention among community pharmacists was limited. Notably, the feasibility could be enhanced by having more accurate phone information (only
50% had correct numbers). Although the time to conduct the intervention was consistent with our expectations (10-30 minutes), the delay in making initial contact (median = 14 days) was longer than anticipated. This was primarily due to the pharmacy records lacking patient daytime phone numbers, which resulted in their having to wait until nonbusiness hours to contact the patient.
Implications for Policy, Delivery or Practice:
Interventions that demand substantial investments of health care personnel time will need to overcome time and other logistical barriers. While identifying patients at risk for non-adherence may be appropriate in a community pharmacy setting, alternative approaches to the conduct or delivery of interventions need to be explored.
Funding Source(s): AHRQ
● The Determinants of Hospitalist Efficiency: A
Qualitative & Quantitative Study
Linda Dynan, Ph.D.; Rebecca Stein, Ph.D.; Guy
David, Ph.D.; Laura Cancilla Kenny, M.D.; Mark
Eckman, M.D.; Amy Short, M.H.A.
Presented by: Linda Dynan, Ph.D., Assistant
Professor, Economics & Finance, Northern
Kentucky University, 426 BP, Highland Heights, KY
41099, Phone: (859) 572-5158; Email: dynanl@nku.edu
Research Objective: To understand the underlying processes and activities by which hospitalist physicians achieve efficiencies associated with their practice as documented in the literature.
Study Design: Using qualitative methods, we develop quantitatively testable hypotheses derived from the hospitalists’ understanding of the impact of their activities and processes on the patient care they deliver relative to non-hospitalist physicians.
We conduct our analysis for hospitalists collectively, and at the individual hospitalist level.
Population Studied: Our research focuses on The
University Hospital (TUH) affiliated with University of Cincinnati (and a Health Alliance member).
Specifically we study the physicians in two programs that simultaneously staff the medical wards at TUH— hospitalist teams and teaching teams. The teaching program is long standing while the hospitalist program was initiated in 2002 by the Division of General Internal Medicine at the
University of Cincinnati, in cooperation with TUH, in response to the anticipated implementation of resident work hour restrictions.
Principal Findings: We find that hospitalists are, on average, more efficient diagnosticians and better coordinators of care, lowering charges through reductions in length-of-stay (LOS) and testing, than teaching teams. Much of that benefit is concentrated among patients admitted by intensivists. We find little evidence of quality focus, or higher use of community resources among hospitalists. Hospitalists have a relative advantage in coordinating care in hospitals, yet have little effect on the choice of post discharge outlets.
Further, we document substantial variation among the hospitalists on a number of relevant quality and charge dimensions to understand any potential tradeoffs—between diagnostic testing and reducing length of stay or charges, for example—which physicians might make in delivering care. We find that hospitalists vary in their care delivery: some substitute time for testing (watchful waiting), some are efficient diagnosticians, while others reduce
LOS and are efficient diagnosticians. All three variations in care are characterized by reduced charges in diagnostics and ancillary services relative to teaching teams. They differ with respect to their performance in reducing length of stay—or coordinating care within hospital.
Conclusion: Our findings support the portrait of hospitalists provided in the literature. However, we identify two mechanisms for the reported benefits: reduced LOS and reduced diagnostic testing. We need however, to establish a better understanding of the link between the in-hospital experience (both general and institution specific) of physicians and the benefits (reduced LOS and reduced testing) they provide. This will provide a better understanding of the cost to hospitals of hospitalist turnover.
Implications for Policy, Delivery or Practice:
Hospitalist programs have the potential to reduce costs through two mechanisms: reduced LOS and reduced testing. These benefits counsel for broader adoption of the hospitalist model among hospitals.
Theme: Quality and Efficiency: Organized
Processes
● Presence of Necrotizing Enterocolitis (NEC),
Intraventricular Hemorrhage (IVH) &
Supplemental Oxygen Usage in Pre-term
Newborns & The Predicted Impact on NICU
Costs & Length of Stay
Lisa Edler, B.S.; Shital Shah, Ph.D.; Jeff Canar,
Ph.D.; Kathleen Ferket, M.S.N., C.P.N.P.; Jeremiah
Tennant, M.B.A., M.P.H.
Presented by: Lisa Edler, B.S., Health Systems
Management, Rush University, Rush University
Medical Center, 1700 West Van Buren Street,
126B, Chicago, IL 60612, Phone: (312) 942-7926;
Email: Lisa_W_Edler@rush.edu
Research Objective: To determine the magnitude of direct operational costs (not charges or reimbursement) and length of stay associated with the presence of necrotizing enterocolitis, intraventricular hemorrhage, and supplemental oxygen usage in pre-term newborns.
Study Design: This is a retrospective data analysis of all pre-term newborns born at an academic medical center from May 1, 2007 through
November 30, 2008.
Population Studied: This study will be based on eighteen months of data (May 1, 2007 through
November 30, 2008) for all pre-term births (< 37 weeks gestation) at an urban academic medical center. The sample will be one of convenience defined by the presence of International
Classification of Diseases (ICD-9) codes for gestational age indicating a pre-term newborn
(765.21-765.28) and will be cross referenced with birth weight (765.0-765.19) of accuracy. All newborns that died while in-house will be excluded from the study.
Principal Findings: After the analysis is completed in March 2009, the study may demonstrate that there is significant difference in direct operational cost and length of stay for pre-term newborns with the presence of necrotizing enterocolitis, intraventricular hemorrhage and supplemental oxygen usage.
Conclusion: This study will provide guidance to the hospital administrators and strategic planners with regards to the resources needed to care for the preterm newborns with the above three conditions as well as altering service line projections in light of potential approval of Gestiva (weekly injections during pregnancy to reduce pre-mature births) by the FDA.
Implications for Policy, Delivery or Practice: If the drug (Gestiva) is approved by the FDA, hospital administrators will need to focus on the potential cost savings along with potential revenue loss associated with the widespread usage of this drug.
The study should also provide guidance for strategic planning executives to evaluate market variations associated with the possible decrease in admissions and re-evaluate the need for expansion of NICU services. Indirect effects of usage of this drug can decrease Pediatric Intensive Care Unit
(PICU) admissions along with the long-term costs associated with caring for these children. If this drug is not approved by the FDA, this study will provide valuable insight to the costs associated with clinical conditions that are prevalent in pre-term infants.
Focus can shift to disease management programs that investigate the lifestyle choices behind women whom have previously delivered pre-term and suggest possible intervention programs.
Theme: Quality and Efficiency: Measurement
● Improving Texas Rural Community Healthcare
Through HIT Implementation
Janine Edwards, Ph.D.
Presented by: Janine Edwards, Ph.D., Professor,
Rural & Community Health Institute, Texas A & M
Health Science Center, 3833 Texas Avenue, Bryan,
TX 77802, Phone: (979) 862-5001; Email: edwards@tamhsc.edu
Research Objective: The objectives are to improve the safety, quality, efficiency, and effectiveness of healthcare delivery through implementation of an electronic health record (EHR) and telemedicine in all healthcare agencies within two rural Texas communities.
Study Design: The design consists of two case studies of the two communities utilizing mixed methods (qualitative and quantitative data) with process and outcome measures. Only qualitative process measures are reported here. The external evaluator studied documents critical to the project, such as the criteria for selecltion of the EHR. Semistructured interviews of nine key persons were conducted, audio-taped, and analyzed for recurring categories and themes. Quantitative outcome data consisting of patient administrative claims data of two disease management indicators and one patient safety indicator are being extracted from the
EHR and analyzed during Spring, 2009.
Population Studied: The Texas Office of Rural
Community Affairs selected two frontier communities located in the Texas Panhandle to be demonstration models for a HRSA -funded state initiative. Both communities are Medically
Underserved Areas (MUA) and have Critical Access
Hospitals (CAH). All of the healthcare providers and agencies in each community comprise the population studied; i.e., hospitals, doctors, clinics, pharmacies, nursing homes, home health agencies, assisted living. Implementation of telemedicine from the Texas Tech Health Science Center is also being studied.
Principal Findings: 1. The process of selecting and contracting with a vendor that could provide an
EHR and connect all the healthcare agencies within a community required twice aa much time as the 18 month project allowed. 2. Organization
/management complexity and decentralized decision-making are inversely related to efficiency of time, money and effort. 3. The culture of a rural community plays a critical role in implementing HIT.
Receptivity of physicians and hospital administrators to technology innovation facilitates implementation and use of HIT. 4. Controlling informaztion and lack of transparency about state and federal funds among community partners creates misunderstanding and resistance. 5. In implementing telemedicine, existing physician referral patterns are an important variable.
Conclusion: 1. An EHR can be implemented in rural community healthcare agencies that have had little prior exposure to HIT. 2. A positive culture toward change/innovation within a community facilitates full utilization of HIT to improve quality, safety, efficiency, and effectiveness. 3. Existing physician referral patterns are an important variable to consider in implementing telemedicine. 4. The themes emerging from this study relate to the exogenous processes of the Normalization Process
Model. Reference: May, C. 2006. A rational model for assessing and evaluating complex interventions in healthcare. BMC Health Services Research.
Implications for Policy, Delivery or Practice: 1.
Federal and state agencies need to plan longer
durations for projects to implement HIT in states with rural areas that are not knowledgeable about
HIT. 2. Outcome data can be reported only if projects are of longer duration. 3.Full utilization of expensive HIT is likely to be achieveid only in communities that possess a culture with positive attitudes toward partnerships and technological innovation. 4. In this early stage of national implementation of HIT, selecting communities with a positive culture as demonstration models is likely to result in greater success and sustainability.
Funding Source(s): HRSA
● Hospital Quality Data: Understanding
Decision Making in Vulnerable Populations
Brian Elbel, Ph.D., M.P.H.; Maria, Raven, M.D.;
Collen GIllespie, Ph.D.
Presented by: Brian Elbel, Ph.D., M.P.H., Assistant
Professor, New York University, 423 East 23rd
Street, 15-120N, New York, NY 10010, Phone:
(212) 263-4283; Email: brian.elbel@nyumc.org
Research Objective: Increasingly, health systems—particularly hospitals—are releasing quality information in order to help patients make more informed decision, and improve health system quality. However, very little is known about how patients respond to quality information, particularly vulnerable populations. Recently, the public hospital system in New York City, serving primarily the uninsured and publicly insured, was one of the first public systems to release hospital quality data.
We undertake one of the first studies to understand how vulnerable patients respond to hospital quality measures.
Study Design: A series of focus groups were conducted and supplemented by survey data. All referenced quality data are based on indicators from the Centers for Medicare and Medicaid
Services (CMS) or from the Agency for Healthcare
Research and Quality (AHRQ), and encompass areas of acute and chronic disease treatment and prevention. Focus groups were designed to explore patients’ decision making processes when choosing a hospital and their attitudes toward and perceptions of the actual hospital quality data made publicly available. Qualitative data are coded and analyzed via an iterative grounded theory approach.
Population Studied: The sample consisted of
English-speaking patients ages 18 and older drawn from outpatient clinics of an urban, public, safety net hospital.
Principal Findings: Research is ongoing. Based on the first 3 focus groups (14 patients; 64% female; mean age=44; approximately 2/3 racial/ethnic minority), we have identified emergent themes in two categories. Relating to vulnerable patient choice of hospitals: a) Vulnerable patients highly value quality of care and report that they actively choose where they seek care; b) Patients value both the level of clinical care and interpersonal interactions with physicians and staff.
Relating to perceptions of quality data and its use: a) Patients were universally unaware of the existent of the quality data, but generally were interested in it once it was presented to them; b) There is much heterogeneity in how well patients were able to understand the data; c) While some measures had resonance with patients, many aspects of how patients view quality data were not present in the standard ratings; d) Some patients were clearly distrustful of the data, its presentation and sources.
Conclusion: While no patients were previously aware of the hospital quality data, the focus groups suggest that vulnerable patients are interested in hospital quality measures. A focus on purely clinical measures, while important to patients, does not convey the breadth of information this patient population seeks. Preliminary data are providing much needed insight into a vulnerable population’s perceptions about hospital quality data and the ways it might be optimized to better serve their needs.
Implications for Policy, Delivery or Practice:
Choice based policies need to particularly consider the needs of vulnerable groups.
Funding Source(s): RWJF
Theme: Consumer Choices in Health Care
● Treatment for Elderly Prostate Cancer
Patients: Disparities in a Southern State
Keith Elder, Ph.D.,M.P.H., M.P.A.; Maria Pisu,
Ph.D.; JoAnn Oliver, Ph.D.; Young Kim, Ph.D.;
Michelle Martin, Ph.D.; Lisa Richardson, M.D.
Presented by: Keith Elder, Ph.D.,M.P.H., M.P.A.,
Assistant Professor, Health Services
Administration, University of Alabama at
Birmingham, 1675 University Boulevard, Webb 608,
Birmingham, AL 35294, Phone: (205) 996-9808;
Email: ktelder@uab.edu
Research Objective: An important aspect of quality cancer care is receipt and completion of recommended therapy. In previous studies, African
American (AA) prostate cancer patients were reported to be less likely to receive treatment for prostate cancer than white patients. In the Southern
US, minorities may be more at risk of not receiving treatment. This study describes treatment for elderly prostate cancer patients in Alabama.
Study Design: This is a retrospective analysis of
Medicare administrative claims data. We linked
Medicare claims and Alabama Statewide Cancer
Registry records for AA and white men diagnosed with prostate cancer in 2000-2003. Treatment was identified in the 365 days post diagnosis using
International Classification of Disease-v.9 procedure and Current Procedure Terminology codes. Logistic regression models were constructed to investigate the association between race and the likelihood of receiving: 1) definitive therapy
(prostatectomy or XRT or brachytherapy) and 2)
hormone therapy for individuals who survived at least 9 months post diagnosis. Similarly, models were constructed to examine the association between race and receiving less than 30 days of
XRT and less than 6 months of hormone therapy in individuals who survived = 1 month after their last treatment. In all analyses, we adjusted for age, tumor grade, comorbid conditions, and proportion of persons living below poverty level and with less than high school education in the Census track of residence.
Population Studied: The study sample included
4130 men age >=64 with a diagnosis of diagnosis of prostate cancer anuary 1, 1999, to December
31, 2003, in the Alabama Statewide Cancer
Registry.
Principal Findings: Among 4130 men, 35.7% were
75 or older, 21.7% were AA, 25.5% had =1 comorbid condition, 3.1% had well differentiated tumors, 63.3% moderately differentiated, 20.7% poorly differentiated, and the remainder did not have tumor grade information. Compared to whites,
AA lived in Census tracts with a higher proportion of persons living below poverty level (22.9% vs 12.9,
P <0.0001) and with of people with less than high school education (38.3% vs 35.2%, P <0.0001).
Overall, 6.9% of AA and 4.3% of whites died within
9 months of prostate cancer diagnosis.
Among those who survived 9 months (N = 3930),
24.3% received prostatectomy, 45.5% XRT, 18.2% brachytherapy, 41.8% hormone therapy, and 71.5% received definitive therapy. About 16.1% received none of the treatments. After adjusting for covariates, AA were found to be less likely to receive definitive therapy (62.7% vs 73.8%, adj. OR
0.59, P <0.0001). No difference was found in receipt of hormone therapy (46.8% vs 40.4%, adj.
OR 1.18, P = 0.057). Among those who survived = one month after the last treatment (N =3924), XRT lasted on average 42.7 ±1.32 days and hormone therapy lasted on average 207.5 ± 4.91 days. There were no difference in the likelihood of receiving less than 30 days of XRT (30.8% vs 32.1%, adj. OR
0.98, P = 0.90) and less than 180 days of hormone treatment (46.6% vs 52.3%, adj. OR 0.81, P = 0.11) between AA and white men.
Conclusion: Treatment rates in Alabama elderly prostate cancer patients are comparable to those found in other parts of the US, particularly the
SEER areas. We found that AA men were less likely to received definitive therapy. This difference persisted after adjusting for socio-economic variables. However, once the treatment was started, there were no differences in duration of radiation or hormone treatment. Further research is needed to establish reasons why disparities in type of treatment exist in elderly Alabamians with prostate cancer.
Implications for Policy, Delivery or Practice: The treatment disparities between AA and White men warrant further research due to the elevated morbidity and mortality in AA men.
● Monetary Incentives for Healthy Behaviors:
Early Data from Florida Medicaid’s Enhanced
Benefits Rewards Program
Kimberly Elliott, B.A.; Christy Lemak, Ph.D.; Jianyi
Zhang, Ph.D.
Presented by: Kimberly Elliott, B.A., Ph.D. Student,
Health Services Research, Management & Policy,
University of Florida, P.O. Box 100195, Gainesville,
FL 32610, Phone: (352) 273-6068; Email: kelliott@ufl.edu
Research Objective: Florida Medicaid introduced a new enhanced benefits rewards (EBR) program as part of a larger Medicaid Reform demonstration.
EBR was implemented in two Florida counties in late 2006, with the premise of rewarding healthy or health-seeking behaviors with monetary rewards.
These monetary awards could be used to purchase specific items at participating pharmacies (e.g. diapers). This paper presents a preliminary picture of program participants, focusing on those who earned and spent EBR credits during the first full year of the program.
Study Design: This is a cross-sectional study using
EBR and Florida Medicaid enrollment data covering
September 2006 to November 2007. Data included eligibility information as well as procedure codes and dollar amounts for EBR credits and debits. The data were collapsed into ten earning (credits) and thirteen spending (debits) categories and analyzed per 1,000 member months (PMM).
Population Studied: Florida Medicaid enrollees eligible to participate in the EBR program between
September 2006 and December 2007 (including
SSI, TANF, and other eligibility categories). There were 2,014,448 observations in the dataset, with
1,212,113 member months for Broward County and
802,335 member months for Duval County.
Principal Findings: On average, 0.04 credits were earned per 1,000 member months . Credit earning was 10 times spending by beneficiaries, with
$6,257,850 granted for credits earned and
$673,669 spent on allowed purchases. Office visits
(0.064 PMM) and prevention visits for children
(0.031 PMM) were the highest credit-earning healthy behavior activities. There were no credits earned for disease management, addiction treatment, or weight loss behaviors. Spending was greatest on baby related items and cold and sinus relief products. HMO members earned almost twice as many credits as members of Provider
Service Networks (0.081 PMM vs. 0.047 PMM), and had greater spending activity (0.0061 PMM vs.
0.0047 PMM).
Conclusion: Credit earnings were significantly higher than debits spent, which is likely related to the newness of the EBR program. It is too soon to tell if members are not utilizing more active healthseeking behaviors, such as weight loss, or if
reporting problems may be preventing these activities from being included in the EBR data.
Implications for Policy, Delivery or Practice:
Financial incentive programs may be effective for encouraging utilization behavior among Medicaid populations. Further research is needed to fully understand the types of behaviors, the implications for outcomes and health, and how to quickly overcome initial program start-up challenges.
Funding Source(s): State of Florida
Theme: Medicaid, SCHIP and State Health Reform
● Colorectal Cancer Screening
Recommendations in Primary Care
Jennifer Elston Lafata, Ph.D.; Sue Flocke, Ph.D.;
Greg Cooper, M.D.; Kurt Stange, M.D.; Laura
Siminoff, Ph.D.; Tracy Wunderlich
Presented by: Jennifer Elston Lafata, Ph.D.,
Research Scientist, Center for Health Services
Research, Henry Ford Health System, One Ford
Place, 3A, Detroit, MI 48202, Phone: (313) 875-
5454; Email: jlafata1@hfhs.org
Research Objective: Adherence to physician recommendation for colorectal cancer (CRC) screening is poor and less than two-thirds of eligible patients have been screened. We describe the content of CRC screening discussions in primary care. Of particular interest is whether a shared decision-making process is used.
Study Design: We use audio-recordings of patientphysician discussions during routine annual check ups to describe the frequency and content of CRC screening discussions. A coding manual was developed based on the 5 As framework and a grounded approach in which 30 cases were used to discover how screening recommendations occur in practice. Audio-recordings were transcribed, allowing the use of both audio and transcripts when coding. Coded data were linked with that from 6month post-visit claims data to allow evaluation of the association of CRC screening discussion content and subsequent CRC screening use.
Population Studied: Physician participants (n=75) are salaried primary care physicians practicing in an integrated delivery system in southeast Michigan.
Patient participants are insured, aged 50-80 years and due for CRC screening at time of scheduled visit. Results are presented for the first n=74 (of over 400) visits for which coding of the audios/transcripts is complete.
Principal Findings: Participation rates are 46% for physicians and 43% for patients. Patients were on average 59 years of age, 68% female, 24% African
American. 95% of visits contained CRC screeningrelated talk and 88% of those conversations were initiated by the physician. Among visits with CRC screening talk, almost half (46%) have evidence of prior CRC screening discussion. Among these,
79% have evidence the patient was non-adherent to prior CRC screening recommendation. Among those non-adherent, 48% have the reasons for nonadherence explored. 68% of visits with CRC screening talk included discussion of patient’s eligibility for screening (Assess). 97% included a recommendation for screening (Advise), but only
30% offered the patient a choice regarding screening and 10% a choice among screening modalities (Agree). 73% include an explanation of how to schedule an endoscopy (Assist) and 5% discuss how results will be relayed (Arrange). 45% of patients receiving a recommendation for screening have completed screening within 6 months. 61% of previously non-adherent patients have completed screening within 6 months.
Conclusion: CRC screening decisions are often not shared. Over a third of the time the patient and physician have previously discussed screening and no screening occurred. The reasons why the patient did not adhere are explored less than half the time. As 61% of these non-adherent patients are screened following this visit, it appears that some patients can ultimately be persuaded, while others remain persistent in their decision not to be screened.
Implications for Policy, Delivery or Practice:
Despite the availability of screening tests that can detect CRC early as well as prevent it, many patients are not screened. Our data illustrate that more than one conversation about CRC screening may be necessary before patients seek screening.
Whether different types of recommendations are more or less likely to result in CRC screening remains to be determined.
Funding Source(s): NIA
Theme: Quality and Efficiency: Measurement
● Disaggregating Racial & Ethnic Differences in
Patient Perceptions of Access to Care: Using
MEPS & CAHPS Data to Look Through the Eyes of the Beholder
Kimberly Enard, M.B.A., M.S.H.A.
Presented by: Kimberly Enard, M.B.A., M.S.H.A.,
Ph.D. Student, Health Services, University of
California, Los Angeles School of Public Health,
3770 Keystone Avenue #203, Los Angeles, CA
90034, Phone: (323) 627-8418; Email: kenard@ucla.edu
Research Objective: The primary research objective was to investigate potential root causes of racial/ethnic disparities in patient perceptions of access to care in order to identify negative perceptions that may be addressed through appropriate interventions.
Study Design: Employing a cross-sectional study design, multivariate analyses were conducted using a two-part model with logistic regression to estimate unconditional outcomes (utilization) and conditional outcomes (patient perceptions, conditional on utilization having occurred). Likelihood of utilization estimated whether the patient had appointments or
visits during the reporting period. Likelihood of patient perceptions measured appointment timeliness, perceived/evaluated need for care, problems receiving care and patient-provider communication. Relative risks by race/ethnicity were also estimated for each outcome. The study controlled for factors associated with healthcare access disparities, including income, type/continuity of insurance and usual source of care provider.
Population Studied: The study population
(N=21,281), classified by race and ethnicity, included a nationally representative sample of adults aged 24 to 64 years from the Medical
Expenditure Panel Survey Household Component data files for 2005 and 2006. Race classifications included white, black, Asians and Hawaiian/Pacific
Islanders (Asian/PI), American Indians/Alaskan
Natives (AI/AN) and multi-race. Ethnic classifications included non-Hispanic and Hispanic.
Principal Findings: Compared to whites and non-
Hispanics, racial/ethnic minorities were associated with lower probabilities of utilization and higher probabilities of negative perceptions of access to care. Racial/ethnic minorities were associated with lower probabilities of receiving appointments as soon as they wanted (RR black: 0.94; Asian/PI:
0.78), lower probabilities of perceived/evaluated need for healthcare (RR black: 0.85; Asian/PI: 0.91;
AI/AN: 0.84; Hispanic: 0.88) and higher probabilities of perceived problems getting healthcare (RR black:
1.20; Asian/PI: 2.15; Hispanic: 1.32). With few exceptions, patient-provider communication estimates revealed no significant differences between minorities and non-minorities. For most outcomes, results for AI/ANs and the multi-race group were insignificant, which may be a function of small sample sizes for those groups.
Conclusion: Potential negative perceptions from patients, healthcare workers and clinicians may impact whether care is sought, whether a need for care is identified during the clinical encounter and, ultimately, whether healthcare is delivered. Even after controlling for relevant covariates, race/ethnicity are significantly associated with less utilization and more negative patient perceptions on standardized measures of healthcare access.
Implications for Policy, Delivery or Practice: The findings from this study reprise potentially disturbing questions given that racial/ethnic minorities have disproportionately high rates of disease, compared to whites and non-Hispanics. First, are perceived signs and symptoms, or even the motivation to seek preventive care, different for racial/ethnic minorities than for whites and non-Hispanics? Patient perceptions may result in delaying necessary care until minor health problems develop into major health crises. Second, how effective is communication during the clinical encounter? In order to appropriately diagnose and recommend treatment, clinicians must rely on information received from the patient, in addition to physical exams/tests. Notwithstanding the findings of this and previous studies, the actual effectiveness of patient-provider communication and the way patients perceive it are not necessarily the same.
Does ineffective communication disproportionately affect racial/ethnic minorities? These questions should be a focus of future research and improved education and outreach efforts.
Theme: Disparities
● Overcoming Barriers to Care-Seeking for
Depression: Recognizing, Naming & Explaining
Patients' Distress
Ronald Epstein, M.D.; Paul Duberstein, Ph.D.;
Richard Kravitz, M.D., M.P.H.; Jennifer Becker,
M.P.H.; Patricia Bamonti, B.S.; Debora Paterniti,
Ph.D.
Presented by: Ronald Epstein, M.D., Professor,
Family Medicine, Psychiatry & Oncology, University of Rochester Medical Center, 1381 South Avenue,
Rochester, NY 14620, Phone: (585) 506-9484;
Email: ronald_epstein@urmc.rochester.edu
Research Objective: Because patients often do not disclose their depressive symptoms to their physicians, we sought to gain a deeper understanding of what enables patients to disclose depression-related symptoms and seek care for depression in primary care. The results are intended to guide development of media messages to improve depression care-seeking in primary care.
Study Design: Fifteen focus groups, stratified by gender and income, were audio-recorded, transcribed and analyzed using qualitative editing/coding methods.
Population Studied: 116 English-speaking men and women (56%); 25-64 years of age (mean = 46 years); Caucasian, Latino (14%), or African
American (16%) race; low- (<$30,000, 55%) and mid-income (>$30,000, 44%); who reported a personal history (88%) of and/or close family member with (40%) depression in 3 mid-sized cities
(Rochester, NY, Austin, TX, Sacramento, CA). Online postings, physician office flyers and neighborhood recruiting strategies were used to achieve income and racial/ethnic diversity.
Principal Findings: We identified 4 major themes: knowing, naming, explaining and overcoming.
Knowing: Participants described being unaware that something was wrong or not normal, often for years. Awareness resulted from a growing realization, prompts from others or a sentinel event.
Naming the illness was complicated by confusion between colloquial and technical uses of the words
“depression” and “depressed”, and heterogeneity of patients’ experiences of depression -- a vast array of somatic, cognitive, emotional, perceptual, and social phenomena. Nonetheless, participants affirmed the value of giving a name to their distress.
Questionnaires (in clinical settings or via the media) helped patients both identify ill-defined symptoms
(“knowing”) and categorize and validate their
experience (“naming”). Explaining: The most common explanations involved character weaknesses or personal flaws, often associated with self-blame; physical explanations (e.g., chemical imbalance); and situational explanations that referred either to past events or present social circumstances. Some participants held two or more explanations simultaneously, albeit rarely integrated into a coherent model. Participants frequently felt less stigmatized when they abandoned explanations related to character weaknesses and endorsed physical causation. Overcoming:
Knowing, naming and explaining were often facilitated by family members, friends and health professionals. However, in other situations, health professionals, friends and family presented social barriers -- stigmatization, minimization, misguided efforts to help and social pressure not to talk about depressive symptoms. In addition, depression itself resulted in isolation, fatigue and cognitive distortion
(personal barriers), limiting opportunities for positive social support and care-seeking.
Conclusion: People with depressive symptoms often are unaware that anything is wrong. In order to engage in care for depression, they must a) become aware of ill-defined and heterogeneous symptoms, b) name their experience to bring symptoms to the attention of family and health professionals, c) arrive at meaningful explanations for their distress and d) neutralize the effects of potent negative social messages.
Implications for Policy, Delivery or Practice: The media, physicians, and families and friends of persons with depression can prompt patients to seek care by engaging patients around knowing, naming and explaining their distress and overcoming social and personal barriers to care.
Funding Source(s): NIMH
Theme: Behavioral Health
● The Impact of Social Capital on Clinical Risk
Management in Nursing
Nicole Ernstmann; Oliver Ommen, M.D., M.P.H.;
Elke Driller; Christoph Kowalski; Melanie Neumann;
Holger Pfaff, Ph.D.
Presented by: Nicole Ernstmann, Research
Assistant, Center for Health Services Research,
University of Cologne, Eupener Street 129,
Cologne, 50933, DE, Phone: +4922147897106;
Email: nicole.ernstmann@uk-koeln.de
Research Objective: The aim of our study is to examine the impact of social capital on clinical risk management among nursing staff of hospitals. It is assumed that high social capital is associated with increased clinical risk management.
Study Design: Cross-sectional postal survey.
Population Studied: 959 nurses of four German hospitals.
Principal Findings: Nurses age, care level of the hospital and social capital are highly significant indicators of clinical risk management (p = 0.001).
Social capital is the most important predictor of clinical risk management.
Conclusion: With an effective clinical risk management many medical errors and adverse events are preventable. The results of this survey suggest that efforts to improve social capital in hospital among nursing staff can maintain effective clinical risk management strategies.
Implications for Policy, Delivery or Practice:
Following our results, an established atmosphere of trust and a feeling of common values and convictions can help nurses to integrate clinical risk management into their daily work. Increasing social capital in hospitals requires in-house strategies for reinforcing a culture of trust.
Funding Source(s): German Federal Ministry of
Education and Research, Grant no 01HWO112.
Theme: Organizational Performance and
Management
● High-Performance Work Systems in
Healthcare
Jason Etchegaray, Ph.D., M.A.; Eric Thomas, M.D.,
M.P.H.
Presented by: Jason Etchegaray, Ph.D., M.A.,
Assistant Professor, Internal Medicine, The
University of Texas Medical School at Houston,
6410 Fannin, UTPB 11.42, Houstonq, TX 77030,
Phone: (713) 500-6782; Email: jason.etchegaray@uth.tmc.edu
Research Objective: The research objective of the present study is to identify the key dimensions of high-performance work systems (HPWS) in hospitals. While there is a growing body of literature in other industries about HPWS, there is a dearth of literature about HPWS in the healthcare field. As a result, identifying the key HPWS dimensions is a necessary first step in creating a reliable and valid HPWS measurement instrument.
Study Design: To achieve the primary aim, one step will be implemented. A sample of at least four hundred healthcare executives (two hundred in the
U.S. and two hundred in the U.K.) will be sent a cover letter and a link to an electronic survey. One hundred hospitals in the U.S. and one hundred hospitals in the UK will be randomly selected from the hospitals in these respective locations. The Sr.
VP of Human Resources (or similar position) will be surveyed along with at least one of the following individuals at each hospital: CEO, CMO, VP of
Quality/Chief QualityOfficer, and/or CNO (to the extent that several of these individuals can be identified, all indentified individuals will be contacted). Assuming a 15% response rate, it is anticipated that 60 executives (across the U.S. and
U.K.) will respond to this survey request. The results will be examined to determine common
HPWS dimensions.
Population Studied: Healthcare executives in the
U.S. and U.K. who have human resource and/or clinical experience.
Principal Findings: This study is currently being conducted. The findings will reflect the themes identified from participant responses as it relates to
HPWS dimensions.
Conclusion: The present study is an initial step in creating a valid and reliable HPWS measure that can be used in healthcare in multiple countries.
Implications for Policy, Delivery or Practice:
Based on other industries, it appears that HPWS is related to safety outcomes. To the extent that a valid and reliable HPWS measure can be created for healthcare and then related to safety and quality outcomes, this research will eventually provide evidence for or against a HPWS-quality/safety relationship.
Funding Source(s): AHRQ, K02 HS17145
Theme: Organizational Performance and
Management
● JROTC: Reforming Obesity Through
Crunches
Donna Ettel, Ph.D.; Anthony, Blenke; Arnaldo
Mejias, Jr., M.A.; Wayne McGlamry, B.A., M.A.;
Michael Braccio, B.A.
Presented by: Donna Ettel, Ph.D., Adjunct
Professor, Honors College, University of South
Florida, Boca Ciega High School, 4202 East Fowler
Avenue, Tampa, FL 33602, Phone: (727) 368-
4966; Email: etteld@pcsb.org
Research Objective: To investigate following physical changes in students after being enrolled in the Army Junior Officers' Reserve Training Corp
(JROTC) program. The mission of JROTC is to motivate young people to become better citizens.
Study Design: Student researchers partnered with academic researchers and administered a questionnaire to capture information relevant to physical activity and weight loss of students enrolled in the JROTC program, at a public high school in South Florida. Questions were adapted from a previously validated instrument developed by the Center for Disease Control. The independent variables included student characteristics such as grade level, age, and gender. The dependent variables include the cadet's responses to the following survey questions: weight loss, clothing size reduction, and physical endurance. Analysis included descriptive statistics and multivariate analyses of variance were conducted to identify significant trends across groups.
Population Studied: This study analyzed the differences among high school students enrolled in the JROTC program and their associated survey responses. Grade levels were restricted to grades
9 through 12. Cadet population consisted of (51%) males and (49%) females with a racial makeup that was 89.9% Caucasian, and 4.1% African American.
Twenty one students reported themselves Hispanic and the remainder were either Native American or
Asian.
Principal Findings: Of the 183 students enrolled in
JROTC, 168 returned their surveys representing a response rate of 91%. Overall, over 40% of cadets reported an average weight loss of twenty pounds or more after completing one semester in the program (p<.03), in addition to a significant improvement in physical stamina (77%) as measured by improvement in the amount of distance they are able to run, and improved performance in exercise such as crunches, sit/push-ups. Furthermore freshmen recruits had the greatest amount clothes size reduction (15%), and reported improvements with hydration (70%), and weight loss (52%).
Conclusion: The mission statement of JROTC is to motivate young people to become better citizens.
Clearly, overweight teens are becoming more dominant because they are abandoning physical activity. Unfortunately, traditional outdoor sports are being repackaged as video games, and teens no longer need to leave their rooms to play them.
Regular participation in sound physical activity programs such as JROTC have direct impact on cadets' productivity and quality of life.
Implications for Policy, Delivery or Practice:
This information is the first of its kind in the area of child health policy and may assist policymakers and other key stakeholders in Florida – and nationally in identifying, designing, and implementing strategies to incorporate physical activity into their programs.
● Debunking the Bully
Donna Ettel, Ph.D.; Samantha Dedrick; Alicia Isaac,
Ph.D.; Julia Rose Knoll; Wendy Summers, R.N.
Presented by: Donna Ettel, Ph.D., Adjunct
Professor, Honors College, University of South
Florida, 4202 East Fowler Avenue, Tampa, FL
33602, Phone: (727) 368-4966; Email: etteld@pcsb.org
Research Objective: Bullying is a serious issue, and efforts to eliminate violence at school is of paramount concern. Children confronted with bullying are subject to become antisocial, abuse substances, undertake the actions of their aggressors, or seek revenge in future relationships.
Adults should promote healthy relationships between students, and children should be taught how to identify when they are being victimized and also when they are acting out as the aggressor.
Obviously adults must play a crucial role in the vicious cycle of bullying and participate to eradicate negative interactions amongst high school students.
The purpose of study is to examine what methods high school students believe is most effective in making bullies stop.
Study Design: Students partnered with academic researchers and administered a questionnaire to capture information relevant to stopping bullying behaviors at a public high school in south Florida.
The independent variables are student characteristics including grade level and race/ethnicity. The dependent variables are student responses to survey items including: school consequences, the victim fighting back, involvement of law enforcement, parental involvement, peer pressure, or discontinuing contact between the bully and the student. Analyses include descriptive statistics, and a Multi-variate Analyses of Varience to identify significant trends across groups.
Population Studied: Although these findings are preliminary, the study will ultimately include data from the entire student body population. All participants were given the same survey and the grade levels were restricted to 9 through 12.
Initially, the sample population consisted of 499 students (51% males) with a racial makeup that was 89.9% Caucasian, 4.1% African American.
Twenty one students considered themselves
Hispanic and the remainder were either Native
American or Asian.
Principal Findings: Overall the majority of students (48%) reported that the most effective way to put a stop to bullying is for the victim to fight back. Few students (less than 15%) find that school consequences, parental involvement, or peer pressure to be effective. African American and Caucasian students agree (between 50 and
60%) that the most effective method is for the victim to fight back. Twenty percent of Hispanics report that if there are school consequences bullies will discontinue their behavior.
Conclusion: Overall the majority of students (48%) reported that the most effective way to put a stop to bullying is for the victim to fight back. Few students
(less than 15%) find that school consequences, parental involvement, or peer pressure to be effective. African American and Caucasian students agree (between 50 and 60%) that the most effective method is for the victim to fight back.
Twenty percent of Hispanics report that if there are school consequences bullies will discontinue their behavior. Approximately 40% of Native Americans suggest that bullying will cease if contact between the students is eliminated.
Implications for Policy, Delivery or Practice:
Future directions for effective bullying interventions must pay closer attention to the cultural differences of youth that impact choices for how youth interact with each other and respond to bullying. Monolithic policies that drive programs devoid of cultural diversity may fail to provide appropriate interventions because of racial, religious, generational and cultural influences that affect victim behaviors. Practitioners have knowledge of and frequently implement services in the field buoyed by the cultural differences of youth but this is often subjective and not supported by policy mandates. Future education and policy endeavors and program strategies must openly embrace the need for relevant and direct alliances between practice and diversity as a remedy to curb bullying.
● Bullying: An Adolescent Power Trip
Donna Ettel, Ph.D.; Samantha, Dedrick; Alicia
Isaac, Ph.D.; Wendy Summers, R.N.; Rafael
Guimaraes Adames; Amita Kaur Glamour
Presented by: Donna Ettel, Ph.D., Adjunct professor, Honors College, University of South
Florida, 4202 East Fowler Avenue, Tampa, FL
33602, Phone: (727) 368-4966; Email: etteld@pcsb.org
Research Objective: Some of the many pressures students face in high school include worrying about grades, applying to colleges and their prowess at sports. Unfortunately, becoming a victim of bullying is yet another element students need to be cognizant of and deal with. Bullying behaviors are multi-faceted and may consist of name calling, being tripped in the hallway, or even becoming the subject of nasty rumors, and is defined as anything that mentally or physically damages a student.
Study Design: This study analyzed the differences among a sample of high school students enrolled in a public high school and their responses to survey items identifying damages inflicted by bullies .
Students partnered with academic researchers and administered a questionnaire to capture information relevant to bullying behaviors encountered among students enrolled at a public high school in south
Florida. All participants were provided the same survey and the grade levels were restricted to 9 through 12 in the medical magnet program. The independent variables were student characteristics including grade level, age, and race/ethnicity. The dependent variables include the students reported accounts of bullying behaviors that they have encountered including: name calling, stealing, assault threats, and public or private embarrassment. Analyses include descriptive statistics and a Multivariate Analyses of Variance was conducted to identify significant trends across groups.
Population Studied: Although these findings are preliminary, the study will ultimately include data from the entire student body population. All participants were given the same survey and the grade levels were restricted to 9 through 12.
Initially, the sample population consisted of 499 students (51% males) with a racial makeup that was 89.9% Caucasian, 4.1% African American.
Twenty one students considered themselves
Hispanic and the remainder were either Native
American or Asian.
Principal Findings: Overall, the top three harassment tactics inflicted include public embarrassment (60%), name calling (30%) and threats of assault (18%). The junior class reported
name calling as the number one tactic used, in contrast to the remainder of the classes who reported observing victims being publicly embarrassed. There were significant differences noted by race. The Asian students reported observing public embarrassment more frequently than the other groups. African American (60%) students reported name calling, stealing and threats to be the most common. Hispanic and Native
American students rarely identified stealing or threats to be an issue.
Conclusion: Many students have reporting observing a variety of reprehensible behaviors at school including unwarranted name calling, faultfinding or public humiliation. Therefore, it is essential to understand what potential victims may encounter in the high school setting and “damask” this monster once and for all by examining and correcting what exactly students report that they encounter.
Implications for Policy, Delivery or Practice: As educators, juvenile justice officials, and mental health practitioners are called upon to address youth bullying, they are frequently confronted with the idea that some forms of bullying are more pervasive and detrimental than others. This concept of “levels of harm” and the ensuing consequences have been continuous themes for attempting to ascertain the long range impact of attacks on youth self-esteem and functioning. Do students really see some forms of bullying as more problematic and how should programs differentiate?
When should adults intervene? One key may be the infusion of cutting-edge programs that utilize students as interveners, counselors, and developers of consequences.
● Implications of Medicaid Non-Payment for
Pediatric Inpatient Complications
Gerry Fairbrother, Ph.D.; Richard Smith, Ph.D.;
Joseph Schuchter, M.P.H.; Lisa Simpson, M.B.,
B.Ch., M.P.H., F.A.A.P.
Presented by: Gerry Fairbrother, Ph.D., Professor
& Associate Director, Child Policy Research Center,
Cincinnati Children's Hospital Medical Center, 3333
Burnet Avenue MLC 7014, Cincinnati, OH 45229-
3039, Phone: (513) 636-0189; Email: gerry.fairbrother@cchmc.org
Research Objective: The objective of this study is to estimate the prevalence and costs of three common and potentially avoidable complications of hospitalized children: lacerations and puncture wounds, selected infections due to medical care, and decubitus ulcers.
Study Design: Data come from the 2005
Nationwide Inpatient Sample (NIS) from the
Healthcare Cost and Utilization Project (HCUP), sponsored by the Agency for Healthcare Research and Quality (AHRQ). National estimates of patientsafety rates, mean length of stay, and costs were estimated using the patient-level weights in the NIS data. To estimate costs, we used the 2005 NIS supplemental Cost-to-Charge file, a dataset containing hospital-specific factors allowing for the conversion of hospital charges to costs.
Population Studied: Our sample consists of all discharges of children, age 0 to 17, identified through AHRQ’s Pediatic Quality Indicators (PDI) software as at risk for experiencing at least one of the three complications.
Principal Findings: PDI rates are significantly higher for Medicaid than privately-insured children
(2.8 vs. 2.5; p<.05) for these three adverse events.
There is a substantial increase in the use of hospital resources if an adverse event occurs, ranging from more than doubling the length of stay for decubitus ulcer (2.3 increase), to about an eight-fold increase
(8.5 times as long) for infection. Overall, looking at the combined pool, Medicaid children spend, on average, about 5 more days in the hospital with a complication than privately insured children.
Estimates of the excess costs of treatment, due to extended visits and more intensive care, for these three potentially avoidable complications are approximately $160 to $200 million annually. These figures represent about one-half to 1 percent of the annual hospital cost of care for these Medicaid children.
Conclusion: While potentially preventable complications are relatively rare, they can result in significant and substantial excess costs. Costs for
Medicaid children are higher than for privately insured children.
Implications for Policy, Delivery or Practice:
Medicaid seems ready to follow Medicare and private insurers in adopting a policy of withholding reimbursement for avoidable complications in the hospital. Medicaid pays 40% of pediatric inpatient visits, and adoption of such a policy could serve as an important spur for more serious effort on the part of hospitals to improve quality and ensure safety.
Theme: Quality and Efficiency: Organized
Processes
● Poor Experiences with Health and
Prescription Plans Incompletely Explain
Coverage Changes among Elderly Adults
Alex Federman, M.D., M.P.H.; Ethan Halm, M.D.,
M.P.H.; Salomeh Keyhani, M.D., M.P.H.; Albert Siu,
M.D., M.S.P.H.
Presented by: Alex Federman, M.D., M.P.H.,
Assistant Professor of Medicine, Medicine, Mount
Sinai School of Medicine, 1 Gustave L. Levy Place,
Box 1087, New York, NY 10029, Phone: (212) 824-
7565; Email: alex.federman@mssm.edu
Research Objective: Expectations that competition in health insurance markets can produce efficiencies are predicated on the assumption that consumers make rational choices, based in part on their experiences with coverage. We sought to
examine the association of experiences with health insurance with changes in coverage among older adults.
Study Design: Longitudinal, prospective study of adults recruited from residential and community centers in New York City, NY. Subjects were interviewed in English or Spanish at baseline, and
12-month follow-up interviews occurred after the
Medicare open-enrollment periods. The current analysis is based on data for the subgroup of
Medicare beneficiaries (n=301; follow-up rate,
81%). We analyzed the association of health and prescription plan changes (added, dropped, or switched health or prescription coverage) at followup with baseline ratings of health or prescription plans (rated coverage as poor to excellent on a 5point scale) and avoidance of medical care or prescription drugs because of cost, as well as with demographic and health status characteristics.
Population Studied: Independently living adults aged 60 years and older.
Principal Findings: Mean age at follow-up was 76 years, 34% were men, 29% black non-Hispanic,
32% Hispanic; 50% had no college education, 26% had Medicaid, and 40% reported poor-fair general health. At baseline, 17% rated their coverage as fair-poor and 50% of those individuals considered making a change in coverage (chi-square, p<0.0001). At follow-up, 19% had changed plans.
In univariate analysis, changing coverage was more likely with poor baseline ratings of coverage,
Medicaid coverage, Hispanic ethnicity, low educational attainment, poor English speaking ability, and poor-fair health. Variables not significantly associated with coverage change included age, gender, cost-related medication avoidance, health literacy, and cognitive impairment. In multivariable analysis, changing coverage was more likely among those with
Medicaid (AOR 3.6, 95% CI 1.8-7.2), low ratings of coverage (AOR 2.3, 95% CI 1.0-5.3), and poor
English speaking ability (AOR 2.1, 95% CI 1.0-4.7).
Conclusion: Low ratings of insurance coverage are associated with changing plans, partly supporting market-based strategies to optimize coverage for individuals. However, Medicaid coverage and poor
English speaking ability are also associated with coverage changes, independent of coverage ratings.
Implications for Policy, Delivery or Practice:
Vulnerable elders may not be equipped to appropriately choose health and prescription plans and may require more assistance with insurance coverage decision making.
Funding Source(s): NIA
● Severity of CHF at Entry to Specialized Heart
Failure Clinic in Men & Women
Debbie Feldman, Ph.D.; Louise Pilote, M.D. M.P.H.
Ph.D.; Julie DesLauriers, M.A.; Hassan Behlouli,
Ph.D.
Presented by: Debbie Feldman, Ph.D., Professor,
Rehabilitation, Université de Montréal, Pavillon du
Parc, c.P. 6128, Succ. Centre-Ville, Montreal,
H3C3J7, CA, Phone: (514) 343-6111,1,1252;
Email: debbie.feldman@umontreal.ca
Research Objective: Multidisciplinary clinics are recommended for management of patients with congestive heart failure (CHF). Women may be admitted to these clinics with more severe disease than men.The objective of this study was to compare severity of congestive heart failure between men and women at entry to the multidisciplinary heart failure clinic based on three measures: New York Heart Association (NYHA)
Functional Class, norm-referenced Six Minute Walk
Distance, and health related quality of life (score on the Minnesota Living with Heart Failure
Questionnaire).
Study Design: In this cohort study, data were collected from the common clinical database used at the participating multidisciplinary clinics, from questionnaires that were administered to patients, and tests administered by clinic nurses. We compared men and women with respect to the three severity indicators at entry to the clinic.
Population Studied: 531 newly-admitted patients to one of six CHF multi-disciplinary clinics were enrolled.
Principal Findings: Only 26% of patients were female and women were older than men (p<0.01).
In adjusted analyses, women had a lower health related quality of life (p=0.04) but did not have lower norm-referenced six minute walk distance or lower
NYHA Functional Class upon entry to the clinic.
Having more comorbid conditions, being on medication and having visited the emergency or admitted to hospital in the past 6 months were associated with higher severity at entry according to all three indicators.
Conclusion: Only about a quarter of patients newly admitted to the clinic were women. Women report a lower health related quality of life at entry to the clinic, although functional capacity does not seem to differ between women and men.
Implications for Policy, Delivery or Practice:
Reasons for discrepancies between subjective quality of life and objective functional measures will need to be elucidated so as to better address women’s health needs. Since prevalence of heart failure is similar in women and men and our study confirms the low proportion of women admitted to these clinics, it may be important to publicize the benefits of these clinics for women with CHF and ensure that women are referred.
Funding Source(s): Canadian Institutes for Health
Research
Theme: Gender and Health
● Inequities in Access to Specialists in Heart
Failure: Results from Two Cohorts
Debbie Feldman, Ph.D.; Louise Pilote, M.D.,
M.P.H., Ph.D.; Sasha Bernatsky, M.D., Ph.D.
Presented by: Debbie Feldman, Ph.D., Professor,
School of Rehabilitation, Université de Montréal,
Pavillon du Parc, C.P. 6128, Succ. Centre-Ville,
Montreal, H3C3J7, CA, Phone: (514) 343-6111-1-
1252; Email: debbie.feldman@umontreal.ca
Research Objective: It is recommended that persons recently diagnosed with heart failure consult with a specialist in heart failure. In addition, specialized multidisciplinary clinics are effective in management of this disease and implementation of newer, more effective treatments. The objective of our research was to determine factors associated with access to specialty care in persons with congestive heart failure.
Study Design: We conducted two cohort studies to investigate access to specialty care in CHF. First, we sought to determine whether patients who were diagnosed with new-onset CHF by a noncardiologist consulted with a cardiologist using a physician claims database. Second, we conducted a cohort study of patients newly admitted to multidisciplinary heart failure clinics and measured time since diagnosis until admission.
Population Studied: The first study was population-based. Physician reimbursement administrative data were obtained for all adults who had suspected new-onset CHF in the year 2000 in
Quebec, defined operationally as a physician visit for CHF (based on ICD-9 diagnostic codes), with no prior physician visit code for CHF in the preceding 3 years. Among those first diagnosed by a noncardiologist, we used Cox’s regression modeling to identify patient and physician characteristics associated with time to cardiology consultation.
The second study was a multi-centered cohort of newly admitted patients to one of six CHF multidisciplinary clinics in the province of Quebec,
Canada. Data were collected from the common clinical database used at these clinics as well as questionnaires that were administered to the patients.
Principal Findings: In study 1, of the 13,523 persons coded as incident CHF, 54.9% consulted a cardiologist within the next 2.5-3.5 years. Older patients, females, those with lower comorbidity, and lower socioeconomic status had significantly longer times to cardiology consultation. In study 2, we enrolled 531 patients with a mean age of 65.9 years. Only 26% were women. The median duration of disease prior to admission to the CHF clinic was 1.2 years. The majority of patients
(62%) were referred by either a cardiologist or an internist, while 24% were referred by other specialists, and 14% by general practitioners. Onefifth of patients had not been receiving regular follow-up for their CHF prior to being admitted to the clinic. Factors associated with shorter disease duration at admission to the clinic were referral by a specialist, not having had regular medical follow-up for CHF, having higher income and having visited the emergency room for CHF .
Conclusion: Our data suggest that many patients with suspected new-onset CHF do not receive prompt cardiology care, as stipulated by current recommendations. Equity of access for women and those with lower socioeconomic status appears problematic and needs to be addressed so that outcomes may be optimized for all patients with
CHF. Dissemination of information to general practitioners regarding availability and benefits of
CHF clinics may need to be improved and criteria for referral better defined.
Implications for Policy, Delivery or Practice:
Equity of access for women and those with lower socioeconomic status appears problematic and needs to be addressed so that outcomes may be optimized for all patients with CHF. Dissemination of information to general practitioners regarding availability and benefits of CHF clinics may need to be improved and criteria for referral better defined.
Funding Source(s): Canadian Institutes of Health
Research
Theme: Disparities
● Association of Primary Health Care
Organizational Models & Personal Factors with
Outcomes in Patients with Chronic Disease
Debbie Feldman, Ph.D.; Jean Frédéric Lévesque,
M.D., Ph.D.; Houda Ouchène, M.S.
Presented by: Debbie Feldman, Ph.D., Professor,
School of Rehabilitation, Université de Montréal,
Pavillon du Parc, C.P. 6128, Succ. Centre-Ville,
Montreal, H3C3J7, CA, Phone: (514) 343-6111-1-
1252; Email: debbie.feldman@umontreal.ca
Research Objective: Persons with chronic disease often require close follow-up by their primary health care providers. We sought to identify whether outcomes (improvement in health status, health related quality of life and lower hospital admissions and ER visits) were associated with certain models of primary health care (PHC) and with personal characteristics (e.g. socioeconomic status, age, social support, gender, etc.) .
Study Design: Longitudinal cohort study
Population Studied: 776 patients with one of four chronic diseases (diabetes, heart failure, arthritis, chronic obstructive pulmonary disease) were recruited in 34 primary health care (PHC) clinics.
We conducted face to face interviews at baseline followed by three follow up phone interviews at 6 month intervals, using validated questionnaires.
PHC organizations were classified according to administrative type and according to a previously developed taxonomy. Logistic regression models assessed associations between improvement(or no change vs. deterioration) in outcomes and type of
PHC organization and personal characteristics.
Principal Findings: PHC organization type was not associated with improved outcomes. However, personal characteristics were related to outcomes.
Notably, improvement in SF36 was associated with higher socioeconomic status, identifying as being
Canadian (vs another ethnicity), having diabetes (vs
COPD, CHF, or arthritis) and having fewer comorbid conditions. However, perception of chronic illness process of care was associated with
PHC organizational model, with significantly less favorable perception of care associated with the professional contact model (e.g. walk-in clinic).
Conclusion: PHC organizational model does not appear to influence outcomes over one year in persons with chronic disease. Nevertheless, perception of care is associated with PHC model.
Implications for Policy, Delivery or Practice:
There is no evidence that organizational model influences health outcomes in the short term.
Theme: Prevention and Treatment of Chronic
Illness
● Hospital Quality Performance as Assessed by
Appropriate Care Measures
Beth Feldpush, M.P.H., Dr.P.H.
Presented by: Beth Feldpush, M.P.H., Dr.P.H.,
Senior Associate Director, Policy, American
Hospital Association, 325 7th Street, NW, Suite
700, Washington, DC 20004, Phone: (202) 626-
2963; Email: bfeldpush@aha.org
Research Objective: This study used patient-level data from the Hospital Compare dataset to examine hospital performance using appropriate care measures, which assess the completeness of care delivered to patients. Hospitals' performance on appropriate care measures was compared to performance on the individual measures and a global "opportunities" composite score.
Study Design: The study examined hospital performance on 19 Hospital Compare quality measures of heart attack, heart failure, and pneumonia care for discharges occurring from
January through June 2007. Patient-level appropriate care scores were developed and then aggregated at the hospital level. Hospital performance on appropriate care measures was compared to performance on individual quality measures and global composite scores using ttests. To assess the relationship between hospital characteristics, such as bed size and teaching status, and performance on appropriate care measures, multiple regression models were tested.
Correlations were used to compare hospitals' performance across the three health conditions.
Population Studied: The study examined performance data for 170 hospitals in three states,
Georgia, Indiana, and New York.
Principal Findings: On average, patients were eligible for a surprisingly low number of individual measures. The average number of measures per patient was 3.3 for heart attack care, 2.3 for heart failure care, and 3.4 for pneumonia care.
Approximately 18% of heart attack, 13% of heart failure, and 16% of pneumonia patients were not eligible for any measures. Hospitals with at least 25 heart attack or heart failure patients had significantly higher appropriate care scores than hospitals with fewer than 25 patients (T value =
6.21, p=<.0001 and T value = 5.23, p=<.0001, respectively), and for hospitals that were not part of a hospital system, bed size was associated with pneumonia appropriate care scores (F value = 7.17, p value = 0.0089). Appropriate care scores were generally lower than scores on individual measures and global composites, and there was more variation in appropriate care scores among the sample hospitals. Appropriate care scores for heart attack and pneumonia care and for heart failure and pneumonia care were correlated, but scores between heart attack and heart failure care were not.
Conclusion: On average, the appropriate care score for individual patients was based on a small number of individual measures. For those patients not eligible for any measures, there is no information available on the quality of their care.
Appropriate care scores were lower than scores on individual measures and showed more variation among hospitals, suggesting that there is still room for improvement on the Hospital Compare measures. Low-volume hospitals had significantly lower performance on the appropriate care measures.
Implications for Policy, Delivery or Practice: The findings from this study have implications for hospital public reporting and pay-for-performance programs. Appropriate care measures are a promising option for evaluating hospital quality, as they show more room for improvement and more variation in performance. However, they may have some limitations, as many patients are eligible for only a few individual measures. Because hospitals with small patient volumes had significantly lower scores, policymakers should determine a minimum volume threshold for participation in pay-forperformance programs to avoid creating unintended consequences for small hospitals and their patients.
Theme: Quality and Efficiency: Measurement
● Implementing ICD-10 in the U.S.: Implications for Health Services Research
Susan Fenton, Ph.D., R.H.I.A.; Sue Bowman,
R.H.I.A., C.C.S.
Presented by: Susan Fenton, Ph.D., R.H.I.A., HIT
Institute Management Director, Texas State
University, San Marcos, TX 78666, Phone: (210)
332-3525; Email: sf36@txstate.edu
Research Objective: The Secretary of Health and
Human Services issued a Final Rule proposing that modified versions of the International Classification
of Diseases, 10th revision, be used in place of ICD-
9-CM in the United States beginning on October 1,
2013. The use of these systems for all diagnostic coding and inpatient procedural coding is expected to have an impact across the healthcare industry.
This is a synthesis of publications regarding ICD-10 implementation with recommendations for health services researchers preparing for the transition from ICD-9 to ICD-10.
Study Design: This project is a review of lessons learned from countries which have previously implemented ICD-10 and results from conversion impact studies conducted in the U.S.
Principal Findings: In the 1970’s ICD-9-CM codes began being used for reimbursement purposes. As a result complete, accurate, and detailed coding became much more important. Today, uses of ICD-
9 coded data throughout the nation’s healthcare system include measuring the quality, safety, and effectiveness of care; designing payment systems; processing claims for reimbursement; conducting research, epidemiological studies, and clinical trials; and setting health policy, among others. CMS is now proposing to adopt ICD-10-CM for all diagnostic coding and ICD-10-PCS for inpatient procedural coding. The ICD-10-CM system consists of more than 68,000 codes, compared to approximately 13,000 ICD-9-CM diagnosis codes.
ICD-10-PCS consists of 87,000 codes, compared to approximately 4,000 ICD-9-CM procedure codes.
The greater level of specificity and clinical detail and the improvements in the capture of medical technological advances in ICD-10-CM and ICD-10-
PCS are expected to result in the improved ability to analyze trend and cost data; a reduced need for manual review of health records to perform research and data mining and adjudicate reimbursement claims; and the improved ability to ascertain disease severity for risk and severity adjustment. Benefits of the changes researchers can expect are more detailed data and improved data quality. Challenges of the implementation for researchers will include difficulties conducting trend analyses, complexities in applying maps between the legacy and new coding systems, and the potential effect of an increase in coding errors.
Conclusion: The implementation of ICD-10-CM for all diagnostic coding and ICD-10-PCS for inpatient procedural coding represents an unprecedented change in healthcare data collection in the United
States. Health services researchers will need to be prepared for the impact especially for longitudinal data analyses.
Implications for Policy, Delivery or Practice: The implementation of ICD-10-CM and ICD-10-PCS will impact payment policies, including quality measurement and value-based purchasing programs. Researchers will need to participate in any organizational ICD-10 transition plan by becoming educated on ICD-10, identifying all areas where ICD-9 is currently utilized, determining if information system changes will be required and planning accordingly, understanding how the ICD-9 to ICD-10 maps are implemented and performing data quality analyses to understand the impact of the transition upon the outcomes, risk-adjustment and all information derived from ICD-9-CM coded data.
Funding Source(s): AHIMA and the AHIMA
Foundation of Research and Education
● Using a Systems Dynamics Framework to
Improve State Health Policymaking
Rachel Ferencik, M.P.A.; Karen Minyard, Ph.D.;
Mary Ann Phillips, M.P.H.
Presented by: Rachel Ferencik, M.P.A., Research
Associate, Georgia Health Policy Center, Georgia
State University, 14 Marietta Street, Suite 221,
Atlanta, GA 30340, Phone: (404) 413-0307; Email: rachel@gsu.edu
Research Objective: State legislators tasked with making state health policy lack the external resources (advisors, staff) to help improve their understanding of health policy issues. In response to legislative requests, the Georgia Health Policy
Center created the Legislative Health Policy
Certificate Program for lawmakers who want to improve their understanding of health and health care. The objectives of the program were to develop a cadre of legislators with more in-depth knowledge and understanding of health policy issues and to use systems dynamics and systems thinking to encourage broader and more systemic approaches to policymaking.
Study Design: Legislative leadership from the key health committees guided the program’s design.
The program consisted of four core sessions and four issue-specific sessions. Sessions provided legislators and their staff with evidence-based information, training on systems concepts and practice bringing a systems thinking “lens” to each policy issue. A six-question framework was used consistently throughout the program for weighing the complex variables—such as trends, time delays, and unintended consequences—involved in making policy decisions.
Population Studied: State lawmakers serving on health-related committees and their staff.
Principal Findings: Eleven legislators and four legislative staff completed the full certificate program, which required participation in the first four “core” sessions and at least two of four issuespecific sessions (six sessions total). Twenty-six legislators and ten staff attended individual sessions. Participants included both major political parties in the state Senate and House of
Representatives. Preliminary evaluations are positive and legislators are beginning to use the skills developed in this program. Committee chairs are interested in examining current policy initiatives from a systems perspective. Legislative leadership in the state House of Representatives is asking for
an analysis of bills using the six-question framework.
Conclusion: State legislators are willing to invest in improving their understanding of the health care system in a non-ideological seminar series.
Systems thinking is a valuable tool that can be adopted by legislators to study challenging state health policy issues. The systems framework provides tools to help legislators develop bigpicture, long-term perspectives on health policy issues, to identify higher leverage policy interventions, and to better understand and anticipate “policy resistance” (the tendency for policies or interventions to be weakened by the system’s response to the policy change itself).
Through the Certificate Program, legislators learn to apply systems thinking to difficult health policy problems such as reducing health care costs, expanding insurance coverage and improving access to medical care.
Implications for Policy, Delivery or Practice:
This program is creating a cadre of state legislators who approach challenging health policy problems with an eye toward high leverage, sustainable solutions and reform. Especially for states without full-time legislative staff, a systems thinking educational curriculum can provide lawmakers with valuable tools to examine policy issues. Systems thinking enables policy makers to expand their conceptualizations of health reform, encouraging legislators to consider the health impacts of social, economic, environmental and other policies. This curriculum can be adapted and used by states to improve policymakers’ understanding of various complex health policy issues.
Funding Source(s): Robert W. Woodruff
Foundation
Theme: Medicaid, SCHIP and State Health Reform
● Physician Views on Incentives-for-Adherence in Childhood Asthma
Evan Fieldston, M.D., M.B.A.; Andrea Puig, B.A.;
Judy Shea, Ph.D.; Joshua Metlay, M.D., Ph.D.;
Susmita Pati, M.D., M.P.H.;
Presented by: Evan Fieldston, M.D., M.B.A.,
Robert Wood Johnson Clinical Scholar, Robert
Wood Johnson Clinical Scholars & Pediatrics,
University of Pennsylvania School of Medicine, 423
Guardian Drive, 1303A Blockley Hall, Philadelphia,
PA 19104, Phone: (215) 573-2585; Email: fieldsto@upenn.edu
Research Objective: Assess physician views regarding appropriateness & effectiveness of I4A in childhood asthma. Asthma is the most common chronic illness of childhood & leading cause of hospitalization, with costs of billions of dollars.
Adherence rates to medications & disease maintenance activities are very low. Multiple approaches have been tried to improve adherence, with mixed results. Providing incentives-foradherence (I4A) to parents has had limited consideration.
Study Design: Cross-sectional, web-based anonymous survey of 1200 Pennsylvania physicians in general pediatrics, pediatric pulmonology, pediatric emergency medicine, pediatric critical care, and allergy.
Population Studied: Pennsylvania physicians who care for children with asthma.
Principal Findings: 329 physicians (30%) responded, reflecting demographics and practice patterns of state's physicians. Overall, 61% said I4A would be appropriate and 77% effective. Half said
I4A would be appropriate & effective; 12% said I4A would be inappropriate & ineffective (and did not answer all follow-on questions). Majority favored linking I4A to 4 activities (asthma well check-ups, flu shot, refill of controller medicines, and proven adherence via electronic home monitoring) and favored only 1 incentive-delivery method (refund of copayments). A system of accruing points to cash in for gifts was the next most favored delivery method (35%). 41% said incentives should be valued at no more than $20/month and 66% said all patients with asthma should be eligible. Majority of respondents did not think incentives would threaten patient/parent autonomy, undercut social fairness, or interfere with patient-doctor relationships. 56% said I4A would improve asthma outcomes; 33% were uncertain. Bivariate and regression analysis did not reveal any substantial differences in reported attitudes by physician demographic or practice characteristics.
Conclusion: Majority of physicians viewed I4A in childhood asthma as appropriate & effective, but some had reservations about the specific design.
Certain approaches were favored over others, particularly those emphasizing prevention, but not those that could lead to adverse outcomes, such as avoiding emergency care.
Implications for Policy, Delivery or Practice:
Incentives for adherence have support among physicians who care for children with asthma and are viewed as an appropriate and effective means to improve adherence and outcomes. A parallel survey will be undertaken to understand parents views to aid in design of a pilot program of I4A.
Funding Source(s): RWJF
Theme: Child Health
● Midnight Census is a Misleading Metric for
Hospital Capacity-Planning
Evan Fieldston, M.D., M.B.A.; Bhuvaneswari
Jayaraman, M.P.H.; Mahesh Narayan, M.B.,
M.S.E.; Kelly Allebach, B.A.; Susmita Pati, M.D.,
M.P.H.
Presented by: Evan Fieldston, M.D., M.B.A.,
Robert Wood Johnson Clinical Scholar, Robert
Wood Johnson Clinical Scholars & Department of
Pediatrics, University of Pennsylvania School of
Medicine, 423 Guardian Drive, 1303A Blockley Hall,
Philadelphia, PA 19104, Phone: (215) 573-2585;
Email: fieldsto@upenn.edu
Research Objective: Determine how one large urban tertiary children s hospital's occupancy data compare over time of day. Health system capacity and hospital crowding are significant concerns to clinicians, administrators, policymakers, and patients. Appropriate inpatient capacity is needed to handle the regular patient flow and to be available in case of surges in demand. The traditional measure of hospital capacity is based on midnight census, often expressed as monthly or annual averages. Many hospital beds have closed in the last two decades, while length-of-stay has also shortened. The former would lead to more crowding overall and the latter to great inaccuracy of the midnight census, particularly at children's hospitals that have a large number of short-stay patients.
Study Design: Retrospective admission-dischargetransfer (ADT) database analysis.
Population Studied: ADT data for 22,320 inpatients from fiscal year 2007 were abstracted.
Each record included dates/times of patient arrival to an inpatient unit and departure from it, as confirmed by registrar staff. Using the timestamps, a statistical inpatient hospital census was generated for every hour of each calendar day; midnight, peak, and minimum census figures were extracted for each day. Hospital occupancy was calculated as
(census / mean beds available for the month).
Principal Findings: Mean midnight occupancy was generally closer to mean minimum occupancy than mean peak occupancy, particularly when excluding weekends/holidays. On 2/3 of days, midnight census was >2.5%-points lower than daily peak occupancy. On average, peak occupancy was 4%points higher than midnight occupancy, but the difference was as large as 13%-points. Daily peak census usually occurred between 10AM-12PM.
Conclusion: Midnight census does not accurately reflect demand for hospital beds, but is the routinely used metric for capacity-planning, budgeting, and staffing purposes.
Implications for Policy, Delivery or Practice:
Hospitals and ADT-system vendors should review their census data and build systems that automatically report out peak census in each 24hour period and/or a routine daytime census figure
(i.e. 12 PM census) in order to optimally plan for and manage hospital occupancy.
Funding Source(s): RWJF
Theme: Organizational Performance and
Management
● Do Childrens Hospitals Respond to
Predictable Fluctuations in Patient Volume?
Evan Fieldston, M.D., M.B.A.; Matthew Hall, Marion
Sills; Anthony Slonim, M.D., Dr.P.H.; Angela Myers;
Courtney Cannon; Susmita Pati, M.D., M.P.H.;
Samir Shah, M.D., M.S.C.E.
Presented by: Evan Fieldston, M.D., M.B.A.,
Robert Wood Johnson Clinical Scholar, Robert
Wood Johnson Clinical Scholars & Department of
Pediatrics, University of Pennsylvania School of
Medicine, 423 Guardian Drive, 1303A Blockley Hall,
Philadelphia, PA 19104, Phone: (215) 573-2585;
Email: fieldsto@upenn.edu
Research Objective: To quantify how CH respond to high hospital occupancy (HHO), which adversely impacts patient care.
Study Design: Midnight census data from 39 CH in the 2006 Pediatric Health Information System were used to construct cutoffs for 85, 90, 95% occupancy
& for own-hospital %ile occupancy. Dependent variables were: # medical & surgical elective, <24hour, low-severity admissions; # transfers in from & out to other hospitals; observed-to-expected length of stay for ambulatory-sensitive conditions (ASCs)
& for non-ASCs. Generalized linear models quantified the association between occupancy & responses.
Population Studied: Inpatients at 39 free-standing children's hospitals in the Pediatric Health
Information System for calendar year 2006.
Principal Findings: 510,616 admissions to CH frequently result in HHO at midnight: 7.8% at 85-
89%, 4.9% at 90%-94%, 3.2% at 95-100%, and
5.5% at >100%. Annual, hospital-level data show inter-facility variation. Analyses of occupancy by day & month show expected periodic variation in some hospitals but not others. Compared to normative standards of HHO or within-hospital
%iles, few hospitals reacted to HHO with changes in the potential responses modeled.
Conclusion: This first systematic study of children s hospitals showed planning for or response to high hospital occupancy was rare & of small magnitude.
Few are adjusting elective inflow relative to seasonal fluctuations in non-elective admissions.
Implications for Policy, Delivery or Practice:
Hospitals should engage in local review of flow and elective scheduling to maximize safety and quality of care.
Funding Source(s): Child Health Corporation of
American, Pediatric Health Information System
Theme: Organizational Performance and
Management
● ER Visits for Preventable Dental Conditions
Leonard Finocchio, Dr.P.H.; Lisa, Maiuro, Ph.D.
Presented by: Leonard Finocchio, Dr.P.H., Senior
Program Officer, California HealthCare Foundation,
1438 Webster Street, Suite 400, Oakland, CA
94612, Phone: (510) 587-3131; Email: lfinocchio@chcf.org
Research Objective: Assess the extent to which
Californians, uninsured and insured, use the ED for ambulatory care sensitive (ACS) dental conditions, i.e. conditions which good outpatient care can
potentially prevent the need for ED visits or hospitalizations, or for which early intervention can prevent complications or more severe disease.
Study Design: Cross-sectional analysis of
California Office of Statewide Health Planning &
Development patient discharge data (2002-2007) and emergency department data (2005-2007) to ascertain use of EDs and inpatient hospital services for specific dental conditions identified by ICD-9 codes.
Population Studied: Californians using hospital emergency departments and inpatient hospital admissions.
Principal Findings: There were over 80,000 ED visits for dental ACS conditions in California in
2007. ED visits for dental ACS conditions have increased 16% over the past three years. In 2007,
ED visits for dental ACS conditions statewide were more frequent than ED visits for diabetes ACS conditions but less frequent than ED visits for asthma ACS conditions. For some counties, however, ED visits for dental ACS conditions are even more frequent than for asthma ACS conditions. Medi-Cal beneficiaries and the uninsured represent nearly two-thirds of all ED visits. The number of visits for these two groups has risen about 20% over three years. The statewide rates per 100,000 are highest for the 18-34 age range, about 382 per 100,000 in 2007. Women had the highest ED visit rates for ACS dental conditions from 2005 to 2007: 364 per 100,000 for ages 18-34.
Blacks have a disproportionately high number of ED visits for ACS dental conditions.
Conclusion: In some parts of California, visits to the ED for preventable dental conditions are more frequent that visits for other other ambulatory care sensitive conditions. Good oral health can be maintained with access to primary dental care.
Without routine access, however, people may use hospital EDs that are costly and inadequately equipped and staffed to handle dental problems.
Women 18-34 have high rates of ED utilization compared to the rest of the population. Evidence linking poor oral health to adverse birth outcomes suggest that it is important for women of child bearing age have access to appropriate dental care. Ensuring access through dental insurance and an adequate supply of dental providers is critical to both oral health and overall health.
Implications for Policy, Delivery or Practice:
State Medicaid programs should consider higher reimbursements or other incentives to increase dentists' willingness to treat underserved populations. Primary care health practitioners should receive basic training in oral health examination and treatment. States legislatures should consider expanding the scope of practice of dental hygienists and other auxiliary personnel to address unmet needs. Health care reform efforts, both national and state, should include improving oral health and access to dental services as key goals.
Funding Source(s): California HealthCare
Foundation
Theme: Coverage and Access
● Workforce Trends in Occupational & Physical
Therapy
Gail Fisher, M.P.A., O.T.R./L.; Mary Keehn, P.T.,
M.S., D.P.T.
Presented by: Gail Fisher, M.P.A., O.T.R./L.,
Clinical Associate Professor, Occupational Therapy,
University of Illinois at Chicago, 1919 West Taylor
Street, MC 811, Chicago, IL 60612, Phone: (312)
996-4371; Email: gfisher@uic.edu
Research Objective: Identify critical workforce needs and issues in occupational and physical therapy, including current and future trends in supply of and demand for practitioners.
Study Design: A mixed methods study which included telephone interviews of a national sample of 40 key informants/stakeholders. The interviews covered perspectives on the shifts in supply of and demand for occupational and physical therapists and therapy assistants, impact of a supply shortage and future growth areas as well as possible constraints. The interviews were taped and transcribed and analyzed. Twelve themes were identified for more detailed analysis across 7 subgroups of participants. Quantitative time trend data on education programs, enrollment, and applications for both the therapist and therapy assistant level programs were also analyzed to discern trends and compare the two professions.
Population Studied: Study participants were recruited from both fields as well as from outside the professions. Participants were leaders in the professions of physical therapy and occupational therapy, educators, higher education administrators, national level employers, owners and supervisors of occupational and physical therapy clinical practices, representatives of payer groups and practicing therapists.
Principal Findings: A shortage of occupational therapy and physical therapy practitioners exists.
The shortage exists as a result of both increased demand and reduced supply. The shortage has developed in the past several years, following a period of a surplus of practitioners between 1998 and 2002. There is a reduced supply of new graduates, resulting from a significant drop in the applicant pool in the period that began in approximately 1999 and reached its low point in
2002. Study participants cited the effects of the
Balanced Budget Act of 1997 and the change in degree level requirements for therapists as likely reasons for the drop. Changes in funding by federal and state programs and private insurers have limited access to occupational and physical therapy.
Demand has increased due to increased numbers of older adults needing services to maximize
functional independence and rising numbers of children on the autism spectrum.
There are many negative impacts of the therapist shortage for both therapy recipients and practitioners.
Barriers to increasing supply to meet demand include the shortage of faculty, restrictions on the entrance of foreign trained therapists, and the lack of federal funding for educational program expansion.
Conclusion: While the shortage of therapists has not gained as much attention as the shortage of nurses, it is growing and is expected to worsen due to the reduced number of programs and enrollment capacity. This shortage has had the greatest negative impact on acute care hospitals, skilled nursing facilities, and home health agencies, but also affects schools and outpatient clinics.
Implications for Policy, Delivery or Practice: The shortage of physical and occupational therapy practitioners is tied to Medicare changes that restricted payment for therapy in outpatient settings and skilled nursing facilities. Changes in degree levels in both fields also affected the number of programs and capacity. The resulting decreased supply of practitioners limits access to therapy services. Desired action at the federal and state level, similar to the support for expansion of nursing education, would ease the supply shortage.
Theme: Health Care Workforce
● Integration of Health & Social Services for
Veteran & Military Communities: Program
Perspectives & Practices
Michael Fisher, M.H.A.; Christina Ciucci, M.S.;
Christine Elnitsky, Ph.D.
Presented by: Michael Fisher, M.H.A., Senior
Research Associate, Health Financial Systems,
Altarum Institute, 4401 Ford Avenue, Suite 800,
Alexandria, VA 22302, Phone: (703) 927-2461;
Email: michael.fisher@altarum.org
Research Objective: Integration has been identified in the literature as an important approach to enhancing efficacy and efficiency of community services. It has not, however, been well-studied as relates to complex services for veterans and their family caregivers. Important questions remain about the definition of integration and about the factors that affect successful implementation. This paper describes a systematic, retrospective review of leaders’ experiences while managing successful health and social services integration programs for veteran and military populations, the determinants of successful integration, and strategies for successful implementation.
Study Design: This exploratory qualitative study was based on grounded theory methodology. Data were collected through semi-structured individual interviews. All interviews were audio-taped and transcribed verbatim. Constant comparative analysis was done using Microsoft Word.
Population Studied: A purposive sample of six senior-level staff members of programs that integrate community-based programs for veteran or military populations participated in interviews.
Principal Findings: Health and social services integration programs, within community-based veteran or military services environments, respond to the needs of these populations across disciplinary boundaries (i.e., respond to healthcare and social services needs). These programs engage a number of actors at multiple levels of formal organizations and communities. Data suggest that communication between program and organizational leaders and community partners and stakeholders is a core category, or “foundational element,” for these programs. Specific types of communication practices were found across programs. These included: (1) creating program awareness, (2) negotiating goals and objectives, and (3) building trust. Other key categories, or determinants of successful integration, included:
(1) uniting leaders at various levels of organizations and communities; (2) calling communities to action; and (3) obtaining funding for sustainability.
Corresponding to these determinants of successful integration were a number of implementation strategies: (1)(a) using community networks, (b) aligning centralized and local actors, and (c) partnering with families; (2)(a) focusing on the veteran or service member, and (b) measuring and improving performance; and (3)(a) seeking agency and foundation funding, and (b) leveraging resources. Cultural competence on the part of programs, partners, and stakeholders was critical to programs being perceived by veterans as “trusted entities” for obtaining services.
Conclusion: These determinants of successful integration and implementation strategies illuminate the highly complicated process of integrating community health and social services for veteran and military populations. Open communication between actors at multiple levels of formal organizations and communities is likely to help surmount barriers to integration.
Implications for Policy, Delivery or Practice:
Integration of community services for veterans and their family caregivers could be considered an efficacious and efficient intervention. As such, integration models should be well-defined and tested for their usefulness with complex, multiservice community health and social services. The determinants of successful integration and corresponding implementation strategies identified in this study should be explored as components of an integration model.
Funding Source(s): Altarum Institute (self-funded)
Theme: Organizational Performance and
Management
● Applying Grounded Theory to Studying Media
Coverage of Veterans Healthcare
Jonathan Fishman, B. S.; Katerine Osatuke, Ph.D.;
Sue Dyrenforth, Ph.D.
Presented by: Jonathan Fishman, B. S., Research
Assistant, National Center for Organization
Development, Veterans Health Administration,
11500, Northlake Drive, Suite 230, Cincinnati, OH
45249, Phone: (513) 247-4289; Email:
Jonathan.Fishman@va.gov
Research Objective: Media coverage captures important aspects of public perception of organizations that provide healthcare. This paper presents the methodology used for creating a database of Veterans Health Administration hospitals media coverage in national and local news. Our intent was to determine a coding system that could be linked with employee perceptions of their workplace and employee satisfaction in these
VA hospitals.
Study Design: A database was created from briefings of VA news coverage from the year 2006, compiled daily by the Office of Public and
Intergovernmental Affairs. The grounded theory method (Glaser & Strauss, 1967; Glaser, 1978;
Charmaz, 1983; Corbin & Strauss, 1990) was used to determine categories from our initial impressions of the data, and applied these categories to successive media articles in order to further refine them. Media articles for the entire year worth of data were then coded using these categories. A qualitative iterative process was used for monitoring the categories-to-data fit: making refinements to the coding system, and updating previous coding based upon these refinements. After the coding was finished and categories were finalized, we determined which hospitals received the most media coverage, and compared their public perception data with the independently collected data from the annual anonymous census survey of all employees from those hospitals (All Employee
Survey: AES).
Population Studied: VHA hospital employees
Principal Findings: Our finalized categories included the scope, valence, informativeness, format, and content of the media coverage. We report reliability for these categories as determined by their internal consistency. We also present validity data, as demonstrated by connections between the categories of public perception and independently collected workplace climate and employee satisfaction data for these same healthcare organizations.
Conclusion: Media coverage captures aspects of public perception that are relevant to the quality of care employees provide in specific Veterans
Healthcare hospitals, including understanding aspects of organizational functioning, workplace climate, and satisfaction within these hospitals.
Implications for Policy, Delivery or Practice: The categories of public perception that we established offer a useful means of describing aspects of organizational functioning of VHA hospitals. This highlights the importance for healthcare organizations attending to the specific aspects of coverage which we summarized.
Funding Source(s): VA
Theme: Quality and Efficiency: Policies and
Incentives
● Assessing the Value of Diabetes Educators &
Diabetes Self-Management Education/Training
Karen Fitzner, Ph.D.; Dawn Rosenthal, R.D.,
C.D.E.; James Specker
Presented by: Karen Fitzner, Ph.D., Chief Science
& Practice Officer, Science & Practice, American
Association of Diabetes Educators, 200 West
Madison Street, Suite 800, Chicago, IL 60606,
Phone: (312) 601-4854; Email: kfitzner@aadenet.org
Research Objective: To determine how effective diabetes educators are in improving patients’ adherence to indicators of quality care (e.g., Health
Plan Effectiveness Data and Information Set
(HEDIS) measures) and whether the intervention they provide results in positive economic outcomes.
Study Design: In 2007, a multifaceted study was envisioned that would include qualitative and quantitative research. It began with a review of the literature, in-depth case studies and a rigorous analysis of clinical and financial data (e.g. retrospective claims analysis). Researchers at the
University of Missouri and AADE undertook a systematic review of the literature on the clinical and economic effectiveness of diabetes education.
Although the literature review revealed that diabetes education was ill defined and/or differed across the studies, the intervention resulted in positive clinical and economic (cost, utilization, productivity) outcomes. Case studies of DSME/T in diabetes disease management programs supported these findings. Yet, gaps in knowledge remain regarding the proportion of people with diabetes that receive DSME/T and the effectiveness of diabetes educators’ as well as the intervention they provide in achieving some key outcomes.
Accordingly, the second phase of the study aimed to fill in gaps relating to the value of the diabetes educator and retrospective claims analysis of 4 million covered lives (Commercial and Medicare) was undertaken by an actuarial firm in spring/summer 2008.
Population Studied: Diabetes affects more than
7% of Americans and represents more than $116 billion of these expenditures. The overall economic cost of diabetes in 2007 was $174 billion, with reduced national productivity accounting for $58 to
$105 billion. ADA 2007 Interestingly, even with these considerable expenditures, in 2005, CMS
reimbursed only $4.8 million on DSMT codes G108 and G109. AADE analysis of CMS data, 2008.
Diabetes Self-Management Education/Training
(DSME/T) is provided to people with and at risk of diabetes by approximately 30,000 diabetes educators who are licensed as nurses, dietitians, pharmacists, physicians, exercise physiologists, or behavioral health professionals. DSME/T is typically offered by a multi-disciplinary team of healthcare professionals in collaboration with physicians and builds upon the Chronic Care Model. It is generally accepted that diabetes educators make a positive difference in the lives of people with diabetes and both public and private payors cover and reimburse
DSME/T.
Principal Findings: People receiving DSME/T:
Save money and have better clinical behavioral and clinical outcomes; Are more likely to adhere to recommendations for screening/HEDIS measures; and Are younger, more likely to be female, located in more affluent areas, have lower clinical risk, higher adherence to diabetes care recommendations and lower average costs. The studies also show that: Physicians and patients exhibit high variation in their use of DSME/T;
People with diabetes treated by physicians who are high users of diabetes education have lower average cost; and Differences in average costs are entirely driven by reduced inpatient costs; outpatient and pharmacy costs are slightly higher than those who do not.
Conclusion: In conclusion, DSME/T is effective and efficient. Diabetes educators collaboration with physician practices result in improved quality and reducing costs for those with diabetes.
Implications for Policy, Delivery or Practice:
DSME/T provided by diabetes educators is effective in achieving positive clinical results.
● The Long Run Health Returns to College
Quality: Evidence Using Siblings
Jason Fletcher, Ph.D.; David Frisvold, Ph.D.
Presented by: Jason Fletcher, Ph.D., Assistant
Professor, Health Policy & Administration, Yale
University, 60 College Street, New Haven, CT
06510, Phone: (203) 785-5760; Email: jason.fletcher@yale.edu
Research Objective: Examine the long term effects of college quality on obesity in old age
Study Design: We use longitudinal analysis of the
Wisconsin Longitudinal study, which has tracked individuals and their siblings for 50 years to examine the link between college quality and obesity at age 50 and 65. We also use sibling fixed effects to control for family background and other confounding.
Population Studied: A cohort of Wisconsin high school graduates and their siblings between 1957 and 2007
Principal Findings: Our results suggest large effects of college selectivity on reducing obesity for individuals in their 60s. While we are able to explore several potential mechanisms responsible for this link, including occupational characteristics, marital status, income/wealth, later health, and cognition, none of our proposed mechanisms seem to explain the observed effects.
Conclusion: College quality seems to be robustly linked with reductions in obesity rates forty years after college completion. The results are robust to controls for IQ, childhood health, and family fixed effects amoung other controls.
Implications for Policy, Delivery or Practice: The results suggest that increasing education as well as increasing quality of education has long run effects on health and health behaviors such as obesity.
Funding Source(s): Wisconsin Longitudinal Study
Pilot Grant
Theme: Obesity Prevention and Treatment
● Assessing the Effect of Changes in Housing
Costs on Food Insecurity
Jason Fletcher, Ph.D.; Susan Busch, Ph.D.; Tatiana
Andreyeva, Ph.D.
Presented by: Jason Fletcher, Ph.D., Assistant
Professor, Health Policy & Administration, Yale
University, 60 College Street, New Haven, CT
06510, Phone: (203) 785-5760; Email: jason.fletcher@yale.edu
Research Objective: In this paper, we examine whether changes in housing costs lead to changes in rates of food insecurity for economically vulnerable families.
Study Design: We use data on a national, longitudinal sample of young families with children merged with data on housing (rental) costs at the state, MSA, and county levels (2001-2003).
Population Studied: We focus on families near or at the poverty level and use household fixed effects.
Principal Findings: Our preferred results suggest that a $500 increase in yearly rental costs is associated with nearly a 3 percentage point increase in food insecurity rates (10% relative increase).
Conclusion: As food insecurity is linked with poor health and other outcomes for children and adults, our results suggest that housings cost changes or other unexpected expenditure shocks may increase the likelihood of food insecurity for poor families, which may then lead to worse outcomes.
Implications for Policy, Delivery or Practice: Our results suggest the need for further increase in food and housing related social programs such as food stamps and housing subsidies to attempt to cushion the impacts of fluctuations in necessary expenditures for poor families.
Funding Source(s): USDA
● Patient-Provider Communication about Sex
During & Following Treatment for Cancer:
Insights from the PROMIS-Sexual Function
Measure Development
Kathryn Flynn, Ph.D.; The PROMIS Sexual
Function Subcommittee; Kevin Weinfurt, Ph.D.
Presented by: Kathryn Flynn, Ph.D., Assistant
Professor, Duke Clinical Research Institute, Duke
University, P.O. Box 17969, Durham, NC 27715,
Phone: (919) 668-4617; Email: kathryn.flynn@duke.edu
Research Objective: Shared decision-making depends on patients and providers who can talk openly, but this may be particularly difficult when discussing sexual concerns. In focus groups conducted as part of the development of the sexual function questionnaire for the Patient-Reported
Outcomes Measurement Information System
(PROMIS), cancer patients and survivors often mentioned concerns about patient-provider communication. During the psychometric testing of the PROMIS-Sexual Function questionnaire, we included items to assess needs and experiences regarding discussions about sexuality with oncology providers.
Study Design: We conducted telephone focus groups with oncology clinicians and in-person focus groups with men and women in active treatment for breast, prostate, lung, colorectal, gynecologic, and other (mixed) cancer types, and with survivors of breast, prostate, and gynecologic cancers. A trained observer summarized the resulting themes, which were verified against the transcripts by an independent auditor; inter-rater reliability was 91%.
We then conducted a web-based survey (selfadministered or interviewer-assisted over the phone) of cancer patients and survivors. Patients were recruited from the Duke University tumor registry and oncology clinics and the NexCura internet panel, and clinicians were recruited from
Duke. We present descriptive statistics describing survey responses by sex and cancer type.
Population Studied: 5 participants in 2 provider focus groups, 109 participants in 16 patient focus groups, and 491 participants in the PROMIS-Sexual
Function survey (data collection 95% complete).
Principal Findings: In focus groups, patients reported significant unmet needs for information and support regarding sexual issues, even for patients whose cancer or its treatments did not directly affect sexual physiology. For some, personal reluctance to talk to their provider was a barrier, though many described a strong desire for providers to initiate communication about sex.
Oncology clinicians explained that given limited time in visits, they may only discuss those qualityof-life topics that are mentioned by the patient. In the survey, nearly half (49%) of participants reported unmet needs. While 74% of patients thought it was at least a little bit important to have discussions with oncology providers about sexual problems, only 44% reported having received information from an oncology doctor or nurse. This varied greatly by cancer type, with 23% of lung,
29% of breast, 39% of colorectal, and 79% of prostate cancer patients receiving information.
Many more men (41%) than women (13%) had asked their oncology providers about sexual problems.
Conclusion: Sexual problems are a widespread concern among cancer patients and survivors, but there is much variation by sex and across cancer types in patients’ experiences discussing sexual issues with providers. The results from this study support and extend findings from previous studies that indicate infrequent communication about sexual issues in oncology visits and the lack of agreement among patients and providers about who should initiate discussions on this topic.
Implications for Policy, Delivery or Practice:
Providers’ approaches to communication about sexual issues can be at odds with patients’ views, raising concerns about the quality of shared decision making if providers are the sole source of information about sexual issues. An important next step will be understanding how patients use alternative sources of information on sexual health.
Funding Source(s): NCI
Theme: Prevention and Treatment of Chronic
Illness
● Measuring Changes in Care Practices &
Postoperative Outcomes by Ethnicity & Gender
Using a Disease Registry by a Cardiac Surgery
Quality Improvement Consortium
Edwin Fonner, Jr., Dr.P.H.; Alan Speir, M.D.;
Vigneshwar Kasirajan, M.D.; Scott Barnett, Ph.D.;
Clifford Fonner, B.A.
Presented by: Edwin Fonner, Jr., Dr.P.H.,
Executive Director, Virginia Cardiac Surgery Quality
Initiative, 14808 West 81st Place, Lenexa, KS
66215, Phone: (913) 888-2179; Email: edfonner@gmail.com
Research Objective: The goal of this study was to measure improvements in process measures by ethnicity and gender for patients undergoing coronary artery bypass grafts (CABG) procedures in Virginia and to determine the relationship with postoperative outcomes and disparities.
Study Design: Risk-adjusted observed-to-expected
(O/E) ratios for operative mortality and postoperative complications were calculated annually from 2002 to 2008 (first Half). Process measures
(percentages and means) were: total ICU hours, use of internal mammary artery (IMA), preoperative beta blockers, and post-discharge medications
(aspirin and other anti-platelets, anti-lipids, and beta blockers). Study data and risk adjustment algorithms were derived from the Society of
Thoracic Surgeons (STS) database.
Population Studied: A disease registry of the
Virginia Cardiac Surgery Quality Initiative (VCSQI) was used to study all on-pump isolated CABG patients operated on from January 2002 to July
2008 (N=23,197). VCSQI, a voluntary consortium of 17 hospitals and 13 cardiac surgical practices, maintains matched clinical and financial data on
99% of the state’s open-heart procedures. Ethnicity was categorized into Caucasian/Hispanic and
African-American by male and female.
Principal Findings: There were across the board improvements in process measures in Virginia from
2002 to 2008 by gender for Caucasian and African
American patients undergoing on-pump CABG procedures. Related differentials by ethnicity and gender narrowed. Risk adjusted outcomes witnessed some improvements. Related differences by ethnicity and gender were mixed.
Regression analysis is underway to isolate the impact of selected risk factors on outcomes by ethnicity and gender.
Conclusion: Changes in process measures over time are influenced by improved reporting (i.e., data quality), better adherence to standards of care, and selected ‘underperformers’ accelerating their adoption of standards. The extent to which improved adherence to process measures affects postoperative outcomes and narrows differentials based on ethnicity and gender is unclear.
Continued efforts are warranted to determine ethnicity and gender-specific risk factors in cardiac surgery.
Implications for Policy, Delivery or Practice: The change in administration may favorably impact initiatives directed at understanding and eliminating disparities in access, treatment, and outcomes based on inequities between minorities and nonminorities, males and females. In order to formulate provider incentives and to induce improvements, we need to examine more thoroughly risk factors and explain variations across providers. Further work is needed to improve data integrity and understand how standards of care can be tailored to specific risk groups.
Theme: Quality and Efficiency: Organized
Processes
● Ethnicity & Cost Following Isolated Coronary
Artery Bypass Surgery: Results from the
Virginia Cardiac Surgery Quality Initiative
Edwin Fonner, Jr., Dr.P.H.; Scott, Barnett, Ph.D.;
Vigneshwar Kasirajan, M.D.; Alan Speir, M.D.
Presented by: Edwin Fonner, Jr., Dr.P.H.,
Executive Director, Virginia Cardiac Surgery Quality
Initiative, 14808 West 81st Place, Lenexa, KS
66215, Phone: (913) 888-2179; Email: edfonner@gmail.com
Research Objective: The objective was to use a disease registry of all open-heart surgical procedures in Virginia to measure and understand differences in postoperative outcomes and costs by ethnicity and gender.
Study Design: Society of Thoracic Surgeons (STS) data were used to measure CABG outcomes. Cost data, matched with STS patient records, was derived from UB-04 discharge data. Individual hospital cost to charge ratios were applied to normalize total institution cost. A multivariate model was used to control for institution, preoperative and postoperative characteristics after excluding the upper and lower cost deciles.
Population Studied: 10,491 1st time, on-pump, isolated CABG patients operated on between
January 2004 and December 2008 from the
Commonwealth of Virginia were studied. This is nearly 100% of all CABGs performed in Virginia during the period. Ethnicity was categorized into
Caucasian, African-American (AA)and Other.
Principal Findings: Patients were primarily male
(74.8%), average age of 63.8±10.7 years and ethnicity: Caucasian (77.0%), AA (11.7%) and
Other (11.2%). Atrial fibrillation (14.1%) and renal failure were the most frequently occurring postoperative complications. Operative mortality was 1.7%. Although significant differences in postoperative complication rates were observed between ethnic groups, the incidence was always greatest among Caucasian patients. Average cost per case was $32,075±25,466. Unadjusted cost per case for AA patients was the greatest among groups ($33,698 vs. Caucasian [$31,825], Other
[$27,931], p<0.001). After multivariable regression, hospital variation accounted for 25.4% of the cost variance with a baseline case cost of $24,991. After risk adjustment, AA patients resulted in a $148 cost increase (vs. Other, p<0.001); being Caucasian resulted in $461 decrease (vs. Other, p<0.001).
Among females, being AA was significantly more costly after risk-adjustment than being Caucasian
(+$1328 vs. + $532, p<0.001).
Conclusion: In Virginia, small but significant differences in risk adjusted cost per case were observed. Upon initial analysis, a significant ethnicity and gender interaction was observed. The consortium is exploring controllable risk factors specific to ethnicity and gender to improve outcomes.
Implications for Policy, Delivery or Practice: A better understanding of how ethnicity, gender, and other inputs influence postoperative outcomes and drive costs will help heart programs design care pathways that minimize adverse effects of controllable factors. Studying programs the that achieve superior outcomes while controlling costs is helpful in identifying and replicating best practices.
Funding Source(s): Consortium dues
Theme: Disparities
● Downstream Revenue of General Internal
Medicine Patients
Heather Forst, B.S.; Shaun Cooper, M.S.; Tricia
Johnson, Ph.D.; John Rechner, M.B.A., C.M.P.E.
Presented by: Heather Forst, B.S., Administrative
Project Assistant- Corporate Finance, Health
Systems Management, Rush University Medical
Center, 1700 West Van Buren Street TOB Suite
126B, Chicago, IL 60612, Phone: (847) 418-1533;
Email: Heather_Forst@rush.edu
Research Objective: The objectives of this study were to (1) quantify the downstream revenue contribution of a general internal medicine practice to an academic medical center, and (2) examine the extent to which downstream revenue varies by patient characteristics and by the patient’s health condition at the first appointment with the general internist.
Study Design: This was a retrospective, cross sectional study using physician and hospital billing data for all new patients seen in the largest academic general internal medicine practice at
Rush University Medical Center between July and
October 2007. The downstream charges and revenue generated by each patient was collected for twelve months following the initial visit to the practice. Data included information on the patient’s age, gender, primary payer, geographic residence, income, primary diagnosis code and all downstream charges and revenue generated within the organization. Downstream revenue and charges were segmented into three service categories, including routine services, ancillary services and specialty physician services. Multiple linear regression analyses were used to examine the association between patient characteristics and the patient’s initial health condition with downstream charges and revenue.
Population Studied: The number of new patients seen in the general internal medicine practice totaled 888 between July and October 2007. Of these patients, 64% were female and 36% were male. The majority (47%) was between 18 and 34 years old.
Principal Findings: The most common reasons for the initial general medicine visit were for signs, symptoms and ill defined conditions (18%) followed by patients with diseases in their circulatory system
(13%) and respiratory system (10%). Patients seeking preventive care as the primary reason for their visit, such as a wellness visit, were minimal, accounting for 2% of the sample. Patients initially seen in the general medicine practice generated downstream charges and revenue in routine services, ancillary services and specialty physician services. On the physician side, total downstream charges generated for ill defined conditions were
$53.05 per patient compared to $32.24 per patient for circulatory conditions and $22.63 for respiratory conditions. Total hospital downstream charges and revenue are forthcoming.
Conclusion: As academic medical centers search for opportunities to increase revenue it is critical to understand the contribution of general internal medicine practices to academic medical centers.
This study provides the rigorous statistical methodology to estimate the financial impact and importance of supporting academic general internal medicine practices. The large variation across types of patients suggests that patient mix has a substantial impact on the downstream financial contribution of a general medicine practice to an academic system.
Implications for Policy, Delivery or Practice:
Academic general internal medicine practices are often viewed as unprofitable to a health care system due to low reimbursement levels particularly for evaluation and management services. While general internal medicine practices are an important source of specialty referrals, little is known about the financial contribution of these referrals to an academic medical center. With a better understanding of where downstream revenue is generated within an academic medical center and the types of general medicine patients that generate the greatest downstream, hospital systems can justify financially and operationally supporting general internal medicine practices.
Theme: Organizational Performance and
Management
● U.S. National Estimates of Mortality, Length of
Stay & Mortality Attributable to Inpatient
Complications of Care
David Foster, Ph.D., M.P.H.; Janet Young, M.D.,
M.H.S.A.; Sivana Heller, M.D., M.P.H.
Presented by: David Foster, Ph.D., M.P.H., Chief
Scientist, Center for Healthcare Improvement,
Thomson Reuters, 777 East Eisenhower Parkway,
Ann Arbor, MI 48108, Phone: (734) 913-3295;
Email: david.a.foster@thomsonreuters.com
Research Objective: To provide a national assessment of excess mortality, length of stay
(LOS) and costs attributable to medical injuries in
U.S. hospitals
Study Design: The Agency for Healthcare
Research and Quality (AHRQ) Patient Safety
Indicators (PSIs) were used to identify medical injuries from a nationally projected U.S. database of hospital discharge abstracts from federal fiscal year
(FFY) 2007. The nationally estimated volume of injuries was determined by measuring the observed
PSI incidence rates. Risk adjusted estimates of the impact of the PSIs on mortality, LOS and total hospital costs were obtained using standard statistical methods.
Population Studied: More than 21.5 million discharge abstracts from 2,620 acute-care hospitals in FFY 2007 in the Thomson Reuters Projected
Inpatient Data Base (PIDB).
Principal Findings: Almost 30 thousand excess deaths, 3.4 million excess days of stay, and more than $9 billion in additional hospital costs were attributable to PSI adverse outcomes in FFY 2007
in acute-care hospitals in the U.S. Per case attributable costs ranged from around $250 for
Obstetric Trauma to more than $47,000 for
Postoperative Physiologic and Metabolic
Derangement. Attributable increases in LOS ranged from non-existent in some PSIs to as high as 16 days for Selected Infections Due to Medical
Care. Increases in patient-level odds ratios for mortality as a consequence of a medical injury ranged from negligible in some PSIs to as high as
33 in patients who developed Postoperative
Respiratory Failure.
Conclusion: Some potentially preventable medical injuries are responsible for a substantial amount of excess mortality, LOS, and hospital costs.
Considerable variability exists across various patient safety adverse outcomes with regard to the consequences. The total impact of medical injuries on quality of care is a function of both the case-level impact and the incidence of the specific adverse outcome.
Implications for Policy, Delivery or Practice:
These results provide a basis for estimating the consequences of various medical injuries. As such, these results may lead to better quantification of potential benefits resulting from lower PSI incidence rates achieved through improved hospital quality of care.
Funding Source(s): Thomson Reuters
Theme: Quality and Efficiency: Measurement
● Does Making Medical Record Visit Notes
Available Online to Patients Change How
Physicians Document Their Care?
Jinnet Fowles, Ph.D.; Allan Kind, M.D.; Cheryl
Craft, R.N.; Elizabeth Kind, M.S., R.N.; Sara
Richter, M.S.; Scott Glickstein, M.D.
Presented by: Jinnet Fowles, Ph.D., Senior Vice-
President, Research, Health Research Center, Park
Nicollet Institute, 3800 Park Nicollet Boulevard,
Minneapolis, MN 55416, Phone: (952) 993-1949;
Email: jinnet.fowles@parknicollet.com
Research Objective: This study seeks to answer the research question: In what ways, if any, does making visit notes easily available to patients online affect how providers document the visit? The study tests 13 hypotheses of specific content changes that could be anticipated. Hypotheses included increasing use of understandable terminology for: medication names, directions for medication frequency, anatomy, and other terminology; and decreasing use of: negative words to describe appearance and behavior, jargon, acronyms, stories, mental health status, and the term ‘obesity.’
We also hypothesized that parts of the visit note
(assessment, plan) would be lengthened to provide more complete explanations.
Study Design: Content analysis using a pre-post study design with a non-randomized control group.
The content analysis coding schema was developed from a review of the literature, interviews with physicians and with patients, and extensive discussions among the investigators. Following agreed-upon rules, one investigator identified the terms in each note to be coded. Random pairs from the four investigators were then assigned to code each note. Each note was first coded individually by the two investigators assigned to that note. The pair then met to review and come to agreement on the coding for each note.
Population Studied: Medical record visit notes from five rheumatologists and five pulmonologists in
2005 (before notes were available electronically) and 2006 (after notes were available electronically to rheumatology patients only). A total of 400 visit notes were analyzed (20 per physician per time period).
Principal Findings: In the rheumatologists’ notes, of the 13 possible changes that were tested, ten of the hypothesized changes did not occur; only three differed significantly. Two changes were in an unpredicted direction, however: the proportion of acronyms to total coded words increased (2.4% to
3.1%, P = .002); and the proportion of anatomical terms likely to be understood to all anatomical terms used decreased (74.7% to 67.7%, P = .007).
One change was in the predicted direction: the proportion of medication trade names to all medications increased (57.5% to 66.2%, P = .024).
During the same time periods, there were no changes in the pulmonologists’ visit notes. The average length of the notes was unchanged for both rheumatology and pulmonology.
Conclusion: Making physician notes readily available to patients electronically has minimal impact on their content.
Implications for Policy, Delivery or Practice:
Some physicians fear they will be misunderstood and may be forced to alter their dictation style or spend excess time explaining their dictation to patients, but physician resistance to releasing patient visit notes may be unwarranted. Many patients have expressed interest in viewing their visit notes. Releasing visit notes could promote patient engagement and patient safety. In this study, providing online access to visit notes did not significantly affect their content.
Funding Source(s): Curtis L. Carlson Family
Foundation
Theme: Consumer Choices in Health Care
● Holes in the Safety Net: Recent Trends in
Uninsurance Rates for the Disabled &
Chronically Ill
Austin Frakt, Ph.D.; Steven Pizer, Ph.D.; Lisa
Iezzoni, M.D., M.S.
Presented by: Austin Frakt, Ph.D., Health
Economist, Health Policy & Management, Boston
University/VA Boston Healthcare System, 150
South Huntington Avenue (152H), Boston, MA
02130, Phone: (857) 364-6064; Email: frakt@bu.edu
Research Objective: The percentage of nonelderly adult U.S. residents without health insurance has increased steadily for more than two decades. Yet, the risk of being uninsured is not evenly distributed throughout the population. The objective of this study is to examine how the risk of being uninsured depends on health, disability, potential eligibility for
Medicaid, and geographic region.
Study Design: This is a descriptive study of the working-age (ages 25-61) population using personlevel data from the household and medical provider components of the Medical Expenditure Panel
Survey (MEPS), pooled across years 2000-2005.
To these data we merged a variable that reflects the generosity of the Medicaid program available in each respondent’s region.
Population Studied: This study focuses on the civilian, non-institutionalized US population between the ages of 25 and 61.
Principal Findings: Among working-age adult U.S. residents, the risk of being uninsured increased from 2000 to 2005 and varied widely across subpopulations defined by income, health status, and geography. Over all regions and years, uninsurance rates were particularly high for lowincome workers reporting potentially serious health conditions (32%) or disabilities (24%). Rates for low-income workers were highest in the South (42% for persons with health conditions; 39% for persons with disabilities) and West (27% health conditions;
21% disability).
Conclusion: Uninsurance rates approach alarming levels among low-income persons living in the
South, even for individuals with disabilities or potentially serious health conditions. State-level differences in Medicaid eligibility rules may contribute to these high uninsurance rates.
Implications for Policy, Delivery or Practice: The uninsured individuals identified by this study face substantial health risks because of their lack of insurance. Persons without health insurance who have significant health conditions have shorter life spans than do insured individuals. Uninsured working-age adults with potentially serious conditions are much less likely to have seen a physician within the previous year than are insured persons, and they are also much more likely to report unmet medical needs during the past year.
Providing health insurance can improve health, changing downward trajectories of functional declines.
Funding Source(s): AHRQ
Theme: Coverage and Access
● Using GIS Tools to Target “Hot Spots” of Low
Quality Care & Disparities & Foster Shared
Action Among Health Plans & Other
Stakeholders
Allen Fremont, M.D., Ph.D.; Nicole Lurie, M.D.,
M.S.P.H.; Malcolm Williams, Ph.D.; Marc Elliott,
Ph.D.
Presented by: Allen Fremont, M.D., Ph.D., Natural
Scientist & Sociologist, RAND, 1776 Main Street,
Santa Monica, CA 90407, Phone: (310) 393-0411 x
7569; Email: fremont@rand.org
Research Objective: Previous work suggests that
GIS mapping and decision tools can help health plans target disparities by identifying local poor quality “hotspots” and their contributing factors. To date, however, these tools have mainly been used with individual plans and have yet to be applied at a regional level in support of cooperative efforts between multiple plans and other stakeholders. We sought to assess whether and how these tools could help foster and guide shared action among multiple plans and other stakeholders in a given community.
Study Design: We are obtaining member-level
HEDIS diabetes care performance data along with member address and surname from 4 HMO/POS plans in each of 3 states (California, Colorado,
Massachusetts), respectively. We are then providing the plans in each region with brief summary reports highlighting local areas or
“hotspots” where diabetes care quality for their members was poor and minority members were concentrated. The reports also include information characterizing the distribution and potential role of selected sociodemographic factors within and around a given hotspot. Member race/ethnicity information is indirectly estimated using RAND’s
Bayesian surname/geocoding method (reported elsewhere); other sociodemographic information
(e.g., SES, linguistic isolation) is derived from
Census data. Each plan will receive information about its own members, pooled data that shows areas where the 4 plans in a given region had hotspots that overlap, and salient sociodemographic information about the affected neighborhoods. We will conduct meetings with decision makers from the 4 plans and other community stakeholders (e.g., local providers, public health officials, and community leaders) in each pilot region to discuss the patterns highlighted in the reports and possible interventions. Formative evaluation techniques (e.g., focus groups, observation) are being used to assess whether and why participants found the GIS mapping tools to be helpful.
Population Studied: The mapping analyses and related reports provided as part of the pilots are focusing on over 1 million diabetic members distributed across 12 plans and 3 states. Teams of between 2 and 4 decision makers from each of the
12 plans, and an additional 5-10 community stakeholders participating in each of the 3 regional pilots will be the subject of the evaluation.
Principal Findings: Initial feedback from plans recruited to participate in the pilot has been extremely positive and has led to preliminary discussions about cooperative and coordinated efforts between competing plans serving a given area. However, at the time this abstract was submitted pilots were just beginning, so we still have little data on whether plans will actually coordinate their disparities efforts with other plans and stakeholders in a given area. Basic results from the pilot will be ready for presentation by the time of the AcademyHealth meeting.
Conclusion: To be determined.
Implications for Policy, Delivery or Practice: To be determined.
Funding Source(s): RWJF, California Health Care
Foundation
Theme: Disparities
● Time Required to Reconcile Medications at
Hospital Admission & Discharge in a Quality
Improvement Intervention
Kevin Frick, Ph.D.; Linda Costa, Ph.D., R.N.; E.
Robert Feroli, Pharm.D.; Michelle Silas, M.P.H.,
B.S.N., R.N.; Melissa Vista, R.N., B.S.N., O.N.C.
Presented by: Kevin Frick, Ph.D., Associate
Professor, Health Policy & Management, Johns
Hopkins Bloomberg School of Public Health, 624
North Broadway, Room 606, Baltimore, MD 21205-
1901, Phone: (410) 614-4018; Email: kfrick@jhsph.edu
Research Objective: In many academic medical centers, various members of the interdisciplinary team (nurses, physicians, pharmacists) obtain medication histories from patients and document their findings on separate tools. This study documented the time required for a protocol to reconcile the medication history with pharmaceuticals actually being taken to improve patient care and assessed whether patient characteristics were predictive of the time required for the initial interview or the protocol.
Study Design: For all patients consented to participate a reconciliation protocol was implemented that involved nurse medication research tasks, nurse consultation with a pharmacist, other pharmacy activities, nurse consultation with a physician, the nurse providing education to the patient prior to discharge, and nurse tasks post-discharge. The time for each was recorded. A generalized linear model using a log link was used to test for relationships between age, sex, and an indicator of the patient being admitted to one service rather than another and the time for the initial interview as well as testing for relationships between the same exogenous variables and the time for the remainder of the protocol. The association between the time of the initial interview and the time for the remainder of the protocol was also tested.
Population Studied: The study included patients admitted medical services within The Johns
Hopkins Hospital. The primary exclusion criteria is the admission lasting less than 24 hours.
Principal Findings: For the first 341 patients in the study with non-missing protocol time data, the average age was 55.6 years (SD 16.6, range 19-
89), 50.2 percent were male, and the two services nearly equally represented (50.8 percent versus
49.2 percent). The average time to conduct the admission medication interview was 11.2 minutes
(SD 8.0, range 1-68 minutes). 16.1 percent required no nurse research time, 64.5 percent required no consultation with a pharmacist, 72.4 percent required no other pharmacy time, 31.4 percent required no physician consultation, 84.5 percent required no patient education, and 25.5 percent required no post discharge activities. The average total time for the protocol excluding the initial interview was 29.3 minutes (SD 30.2, range 0-425 minutes). The time for the interview was significantly related to being male (a 21.9% decrease, p-value 0.003) and there was a 41.3% difference between patients admitted to the different services (p-value<0.001). In contrast, the time for the protocol increased with the log of the age of the patient (0.39 percent increase in time for each 1 percent increase in age, p-value=0.034), but was not significantly associated with being male (pvalue=0.381) or the service to which the patient was admitted (p-value=0.798). Protocol time was not significantly related to the initial interview time.
Conclusion: A protocol to detect and correct unintended discrepancies between prescriber admission/discharge orders and a comprehensive home medication list for patients admitted to a medical service required an average of less than 30 minutes in addition to a standard medication interview upon admission.
Implications for Policy, Delivery or Practice: The predictors of the time for the initial interview differ from the predictors for the time for the protocol.
Hospitals with older patients should expect the protocol to take longer.
Funding Source(s): RWJF
Theme: Organizational Performance and
Management
● Long-Term Costs of Treatments for
Local/Regional Prostate Cancer
Kevin Frick, Ph.D.; Claire Snyder, Ph.D.; Amanda
Blackford, Sc.M; Robert Herbert; Michael Carducci,
M.D.; Craig Earle, M.D.
Presented by: Kevin Frick, Ph.D., Associate
Professor, Health Policy & Management, Johns
Hopkins Bloomberg School of Public Health, 624
North Broadway, Room 606, Baltimore, MD 21205-
1901, Phone: (410) 614-4018; Email: kfrick@jhsph.edu
Research Objective: Men with prostate cancer have a variety of treatment options, including surgery, radiation, hormonal therapy, combinations thereof, or active surveillance. The objective of this study is to assess how treatment choice is associated with long-term costs and how that varies over time.
Study Design: We examined the inpatient, outpatient, emergency, and other Medicare costs of men diagnosed with local/regional prostate cancer.
Based on the treatments received in the first 9 months from diagnosis, men were assigned to a treatment group: active surveillance, radiation, hormonal, hormonal+radiation, or surgery (which might also include radiation and hormonal therapy).
We matched the prostate cancer cases to noncancer controls on age, race, sex, SEER region, comorbidity, and survival. We estimated the costs of prostate cancer care from -1 to 12 months, 13-24 months, 25-36 months, 37-48 months, and 49-60 months post-diagnosis by calculating the incremental costs of care for surviving cases vs. surviving controls. Univariate comparisons were performed using a Mann-Whitney test. We performed multivariate analyses using a generalized linear model with a log link and a gamma distribution family accounting for repeated observations of the same person when analyzing data from more than one year. Using multivariate analyses we compared the incremental effect of different treatments and the effects of each treatment over time.
Population Studied: From the SEER-Medicare database, we examined men diagnosed with local/regional prostate cancer in 2000 who were
66+ years old and enrolled in the fee-for-service
Medicare program.
Principal Findings: 13,769 prostate cancer cases were matched with 13,769 controls and allocated to these treatment groups: active surveillance
(n=2805), radiation (n=2582), hormonal (n=2190), hormonal+radiation (n=3992), and surgery
(n=2200). The absolute incremental difference in five year expected costs for each treatment was statistically significant (p<0.001 in all cases). The largest absolute differences between cases and controls were for hormonal ($26,608) and hormonal+radiation ($25,546). In multivariate analyses, the largest percentage increment in five year total costs was 59.2% (p<0.001) for surgery, although the difference for surgery was not statistically different from the difference for hormonal (53.5%; p-value for difference 0.347) or hormonal+radiation (58.8%; p-value for difference
0.997). Active surveillance had the lowest percentage incremental cost (20.9%) in comparison with their own controls. The differences in costs between cases and controls were statistically significantly higher for all treatment groups versus active surveillance (all p<0.001). For men undergoing either radiation alone or surgery, the annual incremental costs are similar in each of years two through five. Those who undergo hormonal or hormonal+radiation therapy have higher costs in year two compared to others in years three through five. In contrast, the incremental costs for the active surveillance group peaked in years three and four.
Conclusion: The total incremental costs and trends in incremental costs experienced by men with local/regional prostate cancer over five years vary with the initial choice of treatment.
Implications for Policy, Delivery or Practice:
Incremental costs for more active initial treatments either diminish or remain constant over time.
Incremental costs for the active surveillance group increase as some of the men initially treated this way are switched to more active treatments over time.
Funding Source(s): Johns Hopkins Bloomberg
School of Public Health
● Use of Patient Experience Survey Data by
Massachusetts Physician Groups
Mark Friedberg, M.D., M.P.P.; Gillian Steel-Fisher,
Ph.D.; Melinda Karp, M.B.A.; Eric Schneider, M.D.,
M.S.
Presented by: Mark Friedberg, M.D., M.P.P.,
Fellow in General Medicine, Division of General
Medicine & Primary Care, Brigham & Women's
Hospital, 677 Huntington Avenue, Kresge 409,
Boston, MA 02115, Phone: (617) 432-6814; Email: mark.friedberg@gmail.com
Research Objective: Patient experience survey
(PES) results are an important dimension of health care quality. Massachusetts Health Quality
Partners, a multistakeholder collaborative, recently began fielding a statewide PES and publicly reporting results for ambulatory primary care.
However, little is known about how physician groups are using reports on their own performance.
We evaluated (1) how physician groups have responded to PES results and (2) whether group characteristics (e.g., size, organization, and financial incentives) have been associated with efforts to improve patient experience.
Study Design: Between June and November of
2008 we surveyed leaders of all 122 Massachusetts physician groups 3 or more physicians providing primary care to adults. Statewide PES results were publicly reported for all groups, and each had been provided a detailed supplemental report of its own performance. In 30-minute semi-structured interviews, respondents were asked about awareness of PES results, actions taken for low performers within their groups, and group-wide initiatives to improve patient experience. We classified groups into 3 stages of PES adoption: stage 1 (no awareness of PES results, or if aware, no efforts to change performance); stage 2 (taking action with low performers but not pursuing group-
wide improvement initiatives); or stage 3 (ongoing group-wide improvement initiatives). We also obtained data on group size (number of physicians), group organizational model (“medical group” vs.
IPA), affiliation with multi-group networks, employment of physicians, and exposure to PESbased financial incentives. Bivariate relationships between stage of PES use and other group characteristics were evaluating using Wilcoxon rank-sum and Mantel-Haenzsel tests for trend.
Population Studied: Massachusetts physician groups.
Principal Findings: Seventy-two group leaders
(62%) responded. Median group size was 18 physicians (range, 3-87), 62% had a “medical group” organizational model, 66% were networkaffiliated, and 32% reported financial incentives based on PES results. Most groups were in stage 3 of PES adoption (67%), with 12% in stage 2 and
21% in stage 1. Compared to groups in stage 1 and 2, groups in stage 3 were larger (median physician count 24, compared to 15 in stage 2 and
14 in stage 1; P<0.001) and more likely to be organized as a “medical group” (78% vs. 44% and
20%; P<0.001), employ physicians (68% vs. 38% and 14%; P<0.001), affiliate with a network (73% vs. 56% and 47%, P=0.04), and have PES-based financial incentives (47% vs. 22% and 0%;
P=0.002). While 63% of groups in stage 1 and 2 used the statewide report as their main source of
PES results, only 35% of stage 3 groups did so
(P=0.04). Leaders of the remaining 65% of stage 3 groups used commercial survey products as their major sources of PES results, identifying frequent performance updates and guidance on improvement as advantages of these products.
Conclusion: Larger group size and integrated organizational structures may facilitate the development of PES-based improvement efforts.
Implications for Policy, Delivery or Practice:
Financial incentives based on PES results may encourage improvement efforts. Tailoring reports to the needs of physician groups in different stages of
PES adoption (e.g., providing frequent measurement to groups already engaged in improvement efforts) may enhance the effectiveness of PES data.
Funding Source(s): CWF
Theme: Organizational Performance and
Management
● Qualifying as Patient-Centered Medical
Homes: Will Practices Serving
Socioeconomically Vulnerable Populations Be at a Disadvantage?
Mark Friedberg, M.D., M.P.P.; Dana Safran, Sc.D.;
Kathy Coltin, M.P.H.; Marguerite Dresser, M.A.; Eric
Schneider, M.D., M.S.
Presented by: Mark Friedberg, M.D., M.P.P.,
Fellow in General Medicine, Division of General
Medicine & Primary Care, Brigham & Women's
Hospital, 677 Huntington Avenue, Kresge 409,
Boston, MA 02115, Phone: (617) 432-6814; Email: mark.friedberg@gmail.com
Research Objective: Patient-Centered Medical
Home (PCMH) proposals call for increased payments to primary care practices with specified structural capabilities (e.g., reminder systems, case management). Reducing socioeconomic disparities in health care is a goal of the PCMH, but making the investments necessary to qualify for PCMHbased payments may be particularly challenging for practices serving disproportionate shares of socioeconomically vulnerable patients. Our objective was to evaluate the relationship between the structural capabilities and the socioeconomic case-mixes of Massachusetts primary care practices.
Study Design: We developed a written survey to assess primary care practices’ use of 13 structural capabilities common to PCMH proposals. Between
May and October 2007, we administered the survey to 1 randomly selected physician from each of the
412 primary care practices in Massachusetts with 2 or more physicians, obtaining 310 (75%) responses. Using geocoded U.S. Census data corresponding to patient home addresses, we estimated the socioeconomic profile of each practice’s patients, including prevalence of unemployment, receipt of public assistance, income
<200% federal poverty line (FPL), and education < high school. Prevalence of these 4 variables was highly correlated (Cronbach’s alpha 0.8), and we created a composite score for each practice using the standardized mean of the estimated prevalences. Practices with composite scores in the highest 20% were classified as “highly vulnerable.”
All other practices were defined as “less vulnerable.” Logistic models were used to assess the relationships between practice socioeconomic vulnerability and prevalence of each structural capability. Multivariable models were also constructed, including the following potential confounders: practice size (number of physicians), affiliation with a large physician network, teaching status, multispecialty status, and geographic location (metropolitan Boston vs. other).
Population Studied: Massachusetts primary care practices.
Principal Findings: Responding and nonresponding practices´ patient socioeconomic profiles did not differ. Highly vulnerable practices were more likely than less vulnerable practices to have 4 or more physicians (63% vs. 38%, P<0.001) and less likely to be located in metropolitan Boston
(38% vs. 67%, P<0.001). As expected, highly vulnerable practices had higher median patient unemployment (5% vs. 3%), receipt of public assistance (3% vs. 2%), income <200% FPL (23% vs. 14%), and education < high school (16% vs
10%). In unadjusted analyses, highly vulnerable practices were more likely to have 4 of the 13
investigated structural capabilities: on-site language interpreters (56% vs. 25%, P<0.001), 1 or more clinicians delivering care in a language other than
English (78% vs. 51%, P<0.001), multi-functional electronic health records (EHRs; 40% vs. 31%,
P=0.03), and awareness of patient experience ratings (73% vs 65%, P=0.03). There were no statistically significant associations between practices’ socioeconomic case-mix and the remaining 9 structural capabilities (e.g., case management). After adjustment for confounders, only practice language capabilities remained significantly associated with socioeconomic vulnerability.
Conclusion: At present, Massachusetts practices serving disproportionate shares of socioeconomically vulnerable populations are more likely than other practices to have structural capabilities that could enable them to qualify as patient-centered medical homes.
Implications for Policy, Delivery or Practice:
Practices serving socioeconomically vulnerable patient populations may not be at a disadvantage if payments are tied to PCMH designation.
Funding Source(s): CWF
Theme: Quality and Efficiency: Policies and
Incentives
● Linking Strategy, Performance &
Measurement in Healthcare Organizations
Nurit Friedman, Ph.D.; Nava Pliskin, Prof; Joshua
Shemer; Ehud Kokia, M.D.
Presented by: Nurit Friedman, Ph.D., Director of
Research & Evaluation, Research & Evaluation,
Maccabi Healthcare Services, 27 Hameres Street,
Tel Aviv, 68125, IS, Phone: (972) 514-3663;
Email: friedman_n@mac.org.il
Research Objective: As healthcare expenses spiral, providers face growing pressures to control costs while maintaining high quality care. Most healthcare organizations, however, manage cost more closely than quality due to difficulties in acquiring information to appropriately judge the level of clinical performance. Maccabi (MHS),
Israel's 2nd largest HMO, developed a business strategy aimed at improving the quality of care while containing costs. To implement the strategy an an intranet-based Business Performance Management
(BPM) system was developed and implemented.
Quality and cost performance scores relative to target goals were made available in real time at all levels of the organization. This paper presents the results of a longitudinal case study that evaluates the impact of BPM on the quality and the cost of care.
Study Design: To estimate the impact of BPM we used data reported by the Ministry of Health to compare changes over time in MHS relative to other HMOs. In this analysis outcome variables were 21 performance measures which are part of the BPM and for which nationally reported data during 2002-2005. These measures included diabetes, CVA, mammography, and influenza vaccination. A multivariate logistic regression analysis was used to assess changes over time for all HMOs, changes in MHS in comparison to all other HMOs, and the changes at MHS in 2004-5 after implementation of the BPM system. The results of the multivariate logistic regression analysis are presented by Odds Ratio (OR) that indicates the increase (decrease for OR<1) in odds of the dependent variable being positive, based on a positive independent variable - with 95% confidence interval (CI). The two-sided (Wald) Pvalues, obtained from the logistic regression, were considered as significant under 0.05. Qualitative data were collected via semi-structured interviews with 29 key informants for triangulation purposes and for eliciting perceptions of the major drivers of the improvement in clinical quality.
Population Studied: The target population for quality measures was extracted from 1,592,923
MHS members and 5,010,193 members of all other
HMOs, varying from 298,821 MHS members and
760,095 members of other HMOs measured for primary prevention of CVD to 1,302 MHS members and 13,412 members of other HMOs measured for appropriate medication after bypass surgery
Principal Findings: During 2004-5, MHS and all
HMOs performed significantly better than in 2002 in most measures. MHS improvement rate was larger than that of the other HMOs for 14 of the 21 measures. Average total cost per patient from 2002 to 2005 remained the same.
Conclusion: The rate of improvement post-BPMimplementation was significantly greater for MHS than for all other HMOs. The qualitative analysis clearly pointed to indicators and targets in the BPM as the main reason for the improvement. The visibility, accessibility, and credibility of previously unavailable data was found to have the main effect on performance.
Implications for Policy, Delivery or Practice:
Lessons learned at MHS apply to other organizations in the healthcare industry. This study demonstrates that quality improvement can be achieved in healthcare without incurring excessive costs. It also shows how data can be collected through infrastructural, and routinely used operational IT.
Theme: Organizational Performance and
Management
● Healthcare Use & Expenditure Outcomes of a
Disease Management Intervention
Bruce Friedman, Ph.D., M.P.H.; Brenda Wamsley,
Ph.D., M.S.W.; Ethan Corona, B.A.; Dianne Vliebel,
Ph.D., M.S.Ed.; Gerald Eggert, Ph.D., M.S.W.
Presented by: Bruce Friedman, Ph.D., M.P.H.,
Associate Professor, Community & Preventive
Medicine, University of Rochester, 601 Elmwood
Avenue, Box 644, Rochester, NY 14642, Phone:
(585) 273-2618; Email:
Bruce_Friedman@urmc.rochester.edu
Research Objective: To report the impact on healthcare expenditures, hospital use and expenditures, and permanent nursing home use of a primary care-affiliated disease self-managementhealth promotion nurse intervention for Medicare beneficiaries with disabilities and recent significant health services use.
Study Design: The Medicare Primary and
Consumer-Directed Care Demonstration was a 24month randomized controlled trial that included a disease self-management-health promotion nurse intervention. The intervention included monthly nurse home visits, disease self-management using the Consumer Self-Care Strategies and Healthwise for Life handbooks, the PRECEDE health education planning model to organize patient health behavior change and maintenance, and physician-patientfamily-nurse conferences reimbursed by Medicare.
The present study compares the Nurse (n=382) and
Control (n=384) groups. Two sets of regression models were estimated for three dependent variables measuring healthcare expenditures, seven measuring hospital use and expenditures, and three of permanent nursing home use. The first set of models included intervention group, site, and the number of days the beneficiary was enrolled in the Demonstration, while the second set included these variables plus age, gender, and the baseline value for six variables that differed at p<.20 across intervention group at baseline. Standardized predictions were generated by estimating each regression model and then generating two sets of predictions: one by generating all predictions as being in the intervention group, the second by generating all predictions as being in the Control group. The 95% confidence intervals for the standardized predictions for the Nurse and Control groups were then compared.
Population Studied: 766 persons age 65+ that participated in the Medicare Demonstration. Study participants resided in 19 counties in New York
State, West Virginia, and Ohio, and were required to need or receive help with either 2+ activities of daily living (ADLs) or 3+ instrumental ADLs (IADLs), and must have received recent significant healthcare services use (hospitalization, admission to a nursing home, or receipt of Medicare home health care services within the previous twelve months or 2+ ED visits within the past six months).
The sample had a mean age of 77.4 years, 31% were male, and 3% were non-white. The mean number of ADL dependencies at baseline was 2.3 while the mean number of IADL dependencies was
3.5.
Principal Findings: For all 13 expenditure and service measures the 95% confidence intervals for the standardized predictions overlapped for the
Nurse and Control groups for both models, meaning that the differences between the groups were not statistically significant.
Conclusion: Compared to the Medicare beneficiaries in the Control group, those assigned to the Nurse intervention did not experience lower
Medicare, non-Medicare, or total healthcare expenditures, less acute care hospital expenditures or use, or lower likelihood of or fewer days of permanent nursing home use.
Implications for Policy, Delivery or Practice:
There is considerable skepticism about the likelihood that disease management programs can achieve reductions in healthcare expenditures and hospitalization. The evidence from the present demonstration seems to indicate that for Medicare beneficiaries with disabilities disease management programs may not be able to achieve cost savings in the Medicare fee-for-service system.
Funding Source(s): CMS
Theme: Medicare
● Place Your Bets: Comparing Projections of
Long-Run Health Spending
Bianca Frogner, Ph.D.
Presented by: Bianca Frogner, Ph.D., Postdoctoral
Fellow, Health Policy & Administration, University of
Illinois at Chicago, 2970 North Sheridan Road,
#225, Chicago, IL 60657, Phone: (650) 759-1659;
Email: biancafrogner@gmail.com
Research Objective: Evaluate projections, underlying assumptions, and political implications of long-run health spending as well as projections put forth by the Congressional Budget Office (CBO) and Centers of Medicare and Medicaid Services
(CMS). The paper then presents a model developed by the author that is congruent with economic growth theories. The objective is to compare results and understand the differences.
Study Design: run health care projections are defined as projecting 30 to 75 years into the future.
The author’s model used data from 1970 to 2006 from OECD and US Bureau of Labor Statistics for validation whereupon it was used to predict spending at any point in the future.
Population Studied: The focus is on total health spending in the US.
Principal Findings: Based on data from OECD,
US health spending was 15.3% of GDP in 2006.
CBO developed a model that assumes the nonhealth sectors of the economy will grow at the rate it has been growing in the past with excess growth in the economy attributed to health care growth; they estimate that health care will consume 37% of GDP by 2050. CMS projects health spending will reach
28% of the GDP by 2050 based on the premise that health spending will grow 1% faster than GDP; this exponential growth implies that health care eventually will consume 100% of GDP. Recognizing this unreasonable scenario, CMS revised its estimates to be convergent such that health
spending will grow at the same rate as the GDP, but with little theoretical underpinnings. The literature consists primarily of demand-side models that focus on projecting population growth, aging, and chronic disease morbidity. Estimates range from 21% to 28% by 2050; numerous assumptions are used to create these estimates, most of most of which are based on one time findings using crosssectional data. The projection from the author’s model is 19.9% by 2050. This is on the low end of reported estimates, but it is based on well-studied economic growth theories that focus on supply-side factors such as health care labor and wages, and that account for the growth of other sectors of the economy.
Conclusion: estimates based on demand-side economics results in estimates that are too high because they do not take into account competition for resources in the overall economy. A supply-side economic approach appears to provide more realistic results; however, it does not properly account for the cost of new technologies with the result that the estimates are on the low side. The preferred approach is to combine demand-side and supply-side models.
Implications for Policy, Delivery or Practice:
Projections of long-run growth of health spending are used as the underlying projections of Medicare trust fund bankruptcy. Alarming projections with little theoretical support, particularly as put forth by
CBO and CMS, could lead to hasty policy decisions that lead to funding cuts in health care.
Funding Source(s): NIMH
Theme: Health Care Markets and Competition
● Incorporating New Research in Medicare Risk-
Adjustment
Bianca Frogner, Ph.D.; Gerard Anderson, Ph.D.;
Robb Cohen, M.B.A.
Presented by: Bianca Frogner, Ph.D., Postdoctoral
Fellow, Health Policy & Administration, University of
Illinois-Chicago, 2970 North Sheridan Road, #225,
Chicago, IL 60657, Phone: (650) 759-1659; Email: biancafrogner@gmail.com
Research Objective: This study developed a revised Medicare payment model for Medicare
Advantage (MA) plans and Special Needs
Programs that takes into account the latest research and the availability of pharmaceutical data.
Study Design: This study uses two different datasets: 1) the Medicare 5-percent sample data
(2004-2006) and 2) aged beneficiaries enrolled with a large private national insurer (2006-2007).
Parameters are estimated using a multivariate linear regression approach using a split sample methodology. Predictive ratios for various groups were used to determine if there were incentives for risk selection.
Population Studied: The focus of the study is on a community fee-for-service sample. The study examines three groups of Medicare beneficiaries: 1) new enrollees, 2) beneficiaries who die, and 3) aged beneficiaries alive and enrolled in both parts A and B. New enrollees are defined as aged-in beneficiaries (beneficiaries who became 65 in 2006 and were not previously disabled). Beneficiaries who die are defined as those who died in 2006.
The third group of beneficiaries are age 65 years or older as of January 1, 2004 enrolled for a full 24 months of 2004 and 2005 and at least one day in
2006.
Principal Findings: The current use of only one year of data to set payment rates results in underreporting of persistent chronic diseases. Not reimbursing providers for beneficiaries who die results in a 70% underpayment. The current model underpays for beneficiaries with multiple chronic conditions when it makes the assumption that each additional condition the person has is an additive cost. The incorporation of frailty slightly improves the predictive ratios compared to the basic model.
The use of prescription drug data moves predictive ratios towards 1.0 particularly for psychiatric illness that rely on medications for treatment.
Conclusion: The current risk adjustment system needs to be updated to reflect the current research.
The current system seriously overpays for the healthy and seriously under pays for the beneficiary with multiple chronic illnesses and the person who dies.
Implications for Policy, Delivery or Practice: The
Congress is likely to reduce payments to MA and
SNP plans by up to 13 percent. While doing so
Congress should fix the current system and require that CMS incorporate the most recent research on the characteristics of beneficiaries with high health care costs.
Funding Source(s): XLHealth
Theme: Medicare
● A Data/Information & Analysis Needs
Assessment of Public Health Stakeholders in
Nebraska
Ann Fruhling, Ph.D.; Dhundy Bastola, Ph.D.;
Michael Shambaugh-Miller, Ph.D.; Jason Niven,
BS; Rahman Strum, B.S.; Gregory Hoff, M.S.; Ge
Lin, Ph.D.
Presented by: Ann Fruhling, Ph.D., Associate
Professor, Information Systems, University of
Nebraska at Omaha, 1110 South 67th Street,
Omaha, NE 68182, Phone: (402) 554-4968; Email: afruhling@unomaha.edu
Research Objective: Effective public health practice requires timely, accurate, and authoritative information from a wide variety of private and public sources, including data warehouses, primary and secondary databases using computerized information and surveillance systems and ever
growing information technologies (Linkins,
Kilbourne, Koo, O’Carroll, Yasnoff, 2000). Utilization of these sources allows the development of a knowledge base, which can be disseminated using the tools of information technology. Concurrent with these successes and benefits are barriers. A primary concern is that public health officials and researchers spend significant amounts of time searching multiple and disparate sources to find the needed information. Often the needed expertise and resources to aggregate and analyze such data is not available. One of the major reasons for this situation is lack of clear definitions of public health data needs and the failure to establish a consensus on data and communication standards (Linkins,
Kilbourne, Koo, O’Carroll, Yasnoff, 2000). This deficiency creates a substantial roadblock to researchers, educators, health professionals and service providers in their endeavors to improve public health. In an effort to fill these gaps the intent of our study will be to better understand and prioritize important data needs and analytical tools utilized by public health officials/researchers in public health informatics as well as to determine their familiarity and accessibility to currently available public health information. Therefore, our goal is to conduct a comprehensive public health data needs assessment for the State of Nebraska.
Study Design: The research method we selected is using a web-based survey instrument. The survey includes questions on professional/demographic information, information use/needs, computer information technology skills, as well as querying the participant on their needs, uses and knowledge of various public health data sources and analytical tools. Participants were identified by their affiliations with various public health organizations. The data will be statistically analyzed and provide insights to better understand the needs of the various stakeholders. The data gathering for this study is in progress.
Population Studied: This exploratory research study focuses on public health professionals and students in Nebraska. Participants were identified based on their employer or school (Nebraska
Department of Health and Human services,
University of Nebraska Medical Center, local health departments) and includes health educators, administrators, nurses, epidemiologists, and support staff among many others. We anticipate around 300 respondents.
Principal Findings: We expect to learn what important public health data and analytical tools are needed by stakeholders. These needs will be prioritized and reported. We also hope to better understand the best means to disseminate public health information. In addition, we will identify gaps in the needed data.
Conclusion: The results of this needs assessment study will make a significant contribution in improving public health data for our citizens.
Further, the results will help drive the requirements of a public health informatics portal being developed by the researchers and students.
Implications for Policy, Delivery or Practice: The results of this research will assist the public health workforce to more effectively provide services and information and further the goal of prolonging the quality and quantity of life.
Funding Source(s): Nebraska University
Foundation
Theme: Public Health
● Why Do Hospitals Seem So Crowded When
Occupancy Rates Are So Low? Census
Variability and Its Determinants
Kathleen, Fuda, Ph.D.
Presented by: Kathleen Fuda, Ph.D., Data
Analysis Manager, Program for Management of
Variability in Health Care Delivery, Boston
University, 53 Bay State Road, Boston, MA 02215,
Phone: (617) 353-8900; Email: fuda@bu.edu
Research Objective: Symptoms of hospital overcrowding, such as emergency department boarding, ambulance diversions, and operating room backups, are widespread despite low mean hospital occupancy rates (67%). Variability in daily hospital census levels largely explain this discrepancy. Peaks in hospital census stress staff and reduce hospitals' capacity to admit additional patients from the emergency department, while low census levels represent waste of resources. This study examines the determinants of variability in patient census, in particular whether it is due primarily to random fluctuations in emergent patient arrivals, or to patterns of electively scheduled admissions. While variability in emergency admissions is unavoidable, good management of elective admissions would ensure that those patients would contribute minimally to fluctuations in total census.
Study Design: Discharge data for Massachusetts acute-care hospitals were used to calculate daily census levels at individual hospitals, broken out by admission type (emergency, urgent, elective); results were also aggregated at the state level.
Daily admissions and census patterns were examined. The contribution to total census variability of time-related factors (day of week, month, season, holidays), as well as patient type, was analyzed.
Population Studied: All non-maternity-related inpatient and outpatient observation stay discharges from 70 Massachusetts acute care hospitals in FY 2000.
Principal Findings: Statewide, emergentlyadmitted patients comprise 50.3% of total patient census, while electively admitted patients account for 18.6%. Census levels vary by date for all patient subtypes, but elective patient census was markedly more variable (CV= 15.52) than were census levels of emergent (CV= 6.64) or urgent (CV= 6.33)
patients. Census variability by day of week was especially marked for elective patients, but was significant for all patient subgroups. Seasonal or monthly impacts were most marked for emergency admissions. Major holidays had a marked impact on census, including a rebound effect. Controlling for month and day of week, elective patient census is a stronger predictor of total hospital census statewide than is emergency patient census, despite its smaller size.
Conclusion: Routinely decreased census levels on weekends depress overall occupancy rates in acute care hospitals. Scheduling patterns for elective admissions are a significant contributor to shortterm variability in total census.
Implications for Policy, Delivery or Practice:
Improved management of hospital census, particularly scheduling of elective admissions, could reduce total census variability, improving efficiency, reducing staff stress, and increasing quality of care.
Theme: Organizational Performance and
Management
● How Pharmacists & Physicians Perceive Their
Patients’ Use of Personal Health Records
Kevin Fuji, Pharm.D.; Kimberly Galt, Pharm.D.,
Ph.D.; Andjela Drincic, M.D.; Karen Paschal, P.T.,
D.P.T., M.S.; Mark Siracuse, Pharm.D., Ph.D.; Amy
Abbott, Ph.D., R.N.
Presented by: Kevin Fuji, Pharm.D., Research
Fellow, Office of Research, Creighton University
School of Pharmacy & Health Professions, 2500
California Plaza, Omaha, NE 68178, Phone: (402)
280-4411; Email: kfuji@creighton.edu
Research Objective: To describe physicians’ and pharmacists’ engagement with patients using a personal health record (PHR) and their observations about how patients keep track of personal health information
Study Design: This was a comparative study surveying physicians and pharmacists using a modified Dillman technique. Data was gathered from two surveys and compared: a physician survey about the use of health information technology in practice, and a pharmacist survey about health information technology patient safety issues in practice. Questions regarding PHRs and personal health information were common to both surveys.
Population Studied: All licensed physicians who indicated they have an office-based practice in
Nebraska and were willing to complete the survey
(1,315 physicians), and all licensed pharmacists in
Nebraska (2,195 pharmacists) were invited to participate.
Principal Findings: There were 612 physicians
(47% response rate) and 535 pharmacist responses
(24% response rate). Both groups observed patients keeping track of health information primarily through written means: 87% of pharmacists and 91% of physicians observed patients keeping a written list of medications, while
52% of pharmacists and 85% of physicians observed patients keeping a written list of conditions/disease states. Both groups observed few patients (<5%) keeping track of their personal health information using portable electronic devices
(personal digital assistants, smart cards, or flash drives). Though there has been a proliferation of online PHRs (e.g., Google Health, Microsoft
HealthVault, and the Veteran’s Affairs’
MyHealtheVet) pharmacists and physicians only observed 9% and 3% of their patients using online
PHRs, respectively. Pharmacists reported more engagement with patients’ PHRs (41%) than physicians (5%). when patients brought PHRs in to the physicians office or the pharmacy.
Pharmacists also reported using the information in a patient’s PHR when providing care more frequently than physicians (24% to 7%).
Conclusion: Patients are keeping track of personal health information through written means, primarily medication lists and medical conditions/disease states lists. Though both professions report a similar percent of their patients keeping a written medication list, there is a disparity between the groups for observing patients keeping a written list of conditions/disease states. The explanation for this disparity is unknown and requires exploration.
It may be that patients disclose certain types of health information only to providers they feel are most in need of that information (i.e., medical conditions with physicians, medications with pharmacists), or providers may systematically ask for select types of information.
Implications for Policy, Delivery or Practice: Our findings indicate PHRs may be emerging as a shared tool for patients and pharmacists in medication management, and for patients and physicians in overall disease management. The physician generates the prescription for the patient based on their medical conditions, while the patient and the pharmacist interact on an ongoing basis about medication use. The high percentage of providers who observe patients keeping track of health information shows the potential for a comprehensive health record to aid in the overall management of patient care. It is important to understand how the relationships between patients, their health information, and their providers change with the use of PHRs.
Funding Source(s): AHRQ
Theme: Health Information Technology
● Processes in Care Transitions from Patient &
Provider Perspectives: A Mixed Methods Study
Kevin Fuji, Pharm.D.; Amy Abbott, Ph.D., R.N.;
Joan Norris, Ph.D., R.N.; Kimberly Agalt, Pharm.D.,
Ph.D.
Presented by: Kevin Fuji, Pharm.D., Research
Fellow, Office of Research, Creighton University
School of Pharmacy & Health Professions, 2500
California Plaza, Omaha, NE 68178, Phone: (402)
280-4411; Email: kfuji@creighton.edu
Research Objective: This project describes patients’ experiences in care transitions through one community’s health care delivery organizations.
Their experiences are compared to providers’ views of how care is delivered in the context of patient transitions between facilities and home.
Study Design: Using a mixed methods design, snowball sampling was used to invite professionals from hospital and community facilities, as well as patients, to participate in focus groups about their understanding and experiences in care transition processes. Qualitative themes emerged from the focus groups. Each participant completed a questionnaire to gather quantitative data.
Population Studied: Professionals from hospital and community facilities included: nurses, pharmacists, physicians, occupational therapists, respiratory therapists, social workers, case managers, and representatives of rehabilitation services. Patients had experiences representing transitions across all levels of care.
Principal Findings: Thirteen professionals from four hospitals, six professionals from five community facilities, and nineteen patients participated in focus groups. Professionals were in practice for an average of 16 years. Patients had an average age of 51 years, with 13 patients reporting they or a family member had been hospitalized in the past year. The findings centered around the lack of shared knowledge of patients and professionals and of community-wide care transition processes. The set of responsibilities described by professionals differed from patients’ expectations. Questionnaire data revealed many facilities have a set of policies in place for transitioning patients to a next level of care.
However, qualitative themes suggested that no ideal practice model exists for professionals or patients. As gaps emerged, a need arose to understand processes that presently exist from a community-wide picture. A potential model was constructed that looked sequentially across transitions of care (hospital to community facility, hospital to home, community facility to home) and within a facility at the steps that must be completed in order for a patient to transfer from one level of care to another. Barriers, strategies for improvement, and consequences for the patients and facilities were identified. Within facilities, four processes emerged: pre-admission planning, peritransfer, physical transfer to next level of care, and post-discharge evaluation and follow-up.
Conclusion: The lack of shared understanding between patients and professionals leads to time constraints, knowledge deficiencies, and gaps in communication which delay or prevent effective care transitions between facilities and/or home.
Disconnect between what level of care a facility can provide and the patient’s actual needs is one example of a sequential problem across transitions of care. A coordination barrier, such as who takes responsibility for follow-up care (professional vs. patient), occurs within a facility.
Implications for Policy, Delivery or Practice: The need for a shared vision and understanding between hospital and community providers, and between all providers and patients is evident. The complex nature of care transitions makes it necessary for patients and providers to communicate a shared understanding of roles and responsibilities in a community-wide model. A comprehensive picture of the different processes involved in care transitions was created to potentially identify and help address patient and provider needs.
Theme: Quality and Efficiency: Measurement
● Do Medicare Part D Beneficiaries Anticipate
Reaching the Coverage Gap? Knowledge of and
Responses to Gap Proximity
Vicki Fung, Ph.D.; John Hsu, M.D., M.B.A.,
M.S.C.E.
Presented by: Vicki Fung, Ph.D., Staff Scientist,
Center for Health Policy Studies, Kaiser
Permanente Division of Research, 2000 Broadway,
3rd Floor, Oakland, CA 94612, Phone: (510) 891-
3527; Email: Vicki.Fung@kp.org
Research Objective: Medicare Part D drug benefits include a coverage gap that begins after a total cost threshold ($2,250 in 2006). If beneficiaries are aware of the proximity of their drug spending to the gap threshold during the year, they may be better able to anticipate future drug costs and reduce costs using clinically appropriate methods.
We examined beneficiaries’ awareness of the gap and their proximity to it throughout 2006, and selfreported responses to gap proximity.
Study Design: We conducted telephone interviews in early 2007 about experiences in 2006.
Respondents who reported knowing their proximity to the gap during 2006 were asked how this information affected their drug choices; free-text responses were coded into themes using content analysis.
Population Studied: A stratified random sample of
Medicare Advantage beneficiaries of an integrated delivery system, age 65+, with a coverage gap in
2006 (N=1040, response rate=74.6%). The sample included 5 strata defined by 2006 total drug costs, with three strata below ($0-750, $750-1300, $1300-
2250) and two strata above the gap threshold
($2250-3500, $3500+).
Principal Findings: Among those who did not reach the gap, 63.5% were unaware that their plan included a gap, 19.6% were aware of having a gap but not how close they were to reaching it, and
16.9% were aware of the gap and their proximity to it throughout 2006. Among those who reached the gap, 34.8% were aware of the gap only, and 44.1%
were also aware of their gap proximity.
Beneficiaries reported learning about the proximity of their drug spending to the gap from mailed reports from the health plan (85.5%), keeping track on their own (19.2%), and information on pharmacy receipts (16.2%). Beneficiaries reported a range of self-reported responses to knowledge about gap proximity, including engaging in efforts to reduce drug costs (e.g., seeking generics or samples, decreasing the quantity dispensed per fill), and monitoring expenses. However, beneficiaries also reported reducing medication use (not filling or delaying medication fills); feeling anxious and confused about the gap; and frustration over not knowing how to avoid the gap or lower costs.
Among those who reached the gap in 2006, 13.0% reduced medication use and 14.0% reported other cost-reduction efforts due to gap proximity; among those with costs just below the gap threshold in
2006 ($1301-$2250 in total drug costs), 16.2% reduced drug use and 5.4% reported other costreduction efforts. Fewer than 3% of beneficiaries with $0-$1300 in total drug costs reported any responses to gap proximity.
Conclusion: In this plan, beneficiary knowledge about the gap and spending relative to the gap was limited. Those who were aware of their gap proximity reported both potentially beneficial responses, including switching to generics and anticipating expenses, but more commonly reported potentially adverse responses, including reducing medication use.
Implications for Policy, Delivery or Practice:
Patient knowledge of spending levels relative to the gap may be useful in encouraging cost-effective drug use and reducing unexpected financial burdens; however, efforts are needed to inform beneficiaries of and encourage appropriate drug choices in conjunction with this information to mitigate adverse clinical consequences.
Funding Source(s): Kaiser Foundation Research
Institute
Theme: Medicare
● Pharmacists for Patient Safety: The State of
Patient Safety in Nebraska Pharmacy
Kimberly Galt, Pharm.D., Ph.D.; Kevin Fuji,
Pharm.D.; Alexandra Serocca, Pharm.D.; Ashley
Pham, Pharm.D.; Charles Barr, Pharm.D.; Mark
Siracuse, Pharm.D., Ph.D.
Presented by: Kimberly Galt, Pharm.D., Ph.D.,
Professor, Associate Dean for Research, Pharmacy
Practice, Creighton University School of Pharmacy
& Health Professions, 2500 California Plaza,
Omaha, NE 68178, Phone: (402) 280-4259; Email: kgalt@creighton.edu
Research Objective: The purpose of this project is to understand the patient safety issues Nebraska pharmacists and their patients face across the spectrum of practice and geographic settings. This project supports the next translational research step: to develop evidence-consensus solutions to the specific safety problems and instruct implementation solutions in context-specific settings. A rapid dissemination system is planned to deliver these potential solutions back for pharmacists to incorporate into local practice.
Study Design: A mixed methods, cross-sectional study using a written survey. “Pharmacists for
Patient Safety,” was designed to collect both quantitative and qualitative data from Nebraska pharmacists. The survey explored: 1) different patient safety issues pharmacists face related to medication therapy management, new therapies, and technologies (bar code medication administration, e-prescribing, electronic and personal health records, decision support systems, others); 2) the safety problems their patients shared, and; 3) potential solutions pharmacists have tried. Descriptive and relational statistics, geographic analysis of practices, and thematic analysis of case stories were performed.
Population Studied: All 2,195 licensed pharmacists in the state of Nebraska were invited to participate.
Principal Findings: Twenty-four percent (n = 535) of the state’s pharmacists responded. Respondents are representative of the state’s practitioners by age, gender, practice setting, and average hours worked/week. Results describe the experiences of pharmacists with personal and electronic health records, safety improvements, and problems with practice-specific technologies, culture, workload, training, and new pharmaceutical product safety problems. A story was reported by 250 pharmacists whose patients shared safety concerns with them.
Pharmacists described 362 episodes of errors or near misses in the six months prior to the study.
One-third reported being involved in at least one error in the last six months, and 49% in a "near miss." Half of the respondents reported observing an error, and half a near miss. There were 540 pharmacist stories about types of errors that the use of specific technologies was observed to reduce, and types of errors that emerged with that same technology. In the larger context of patient safety, pharmacists shared 649 stories about safety issues in practice. Overall, 42% of pharmacists indicated a willingness to share safety experiences and suggestions for improving practice through a common database.
Conclusion: Pharmacists want to engage in communication about safety concerns and participate in an information sharing and learning network to improve patient safety in their practices.
The need to continue pharmacists’ reporting of safety issues and to develop a rapid dissemination system to communicate potential solutions back for pharmacists to incorporate into local practice is evident.
Implications for Policy, Delivery or Practice:
This joint effort with the Nebraska State Board of
Pharmacy produced the first comprehensive “State of Patient Safety in Nebraska Pharmacy” report.
The findings are being used to inform pharmacists, patients, health professionals, policy makers, scientists, and other stakeholders the compelling case for establishing an on-going safety reporting and rapid educational feedback system to improve patient safety through a network of communication across the state.
Funding Source(s): Nebraska Department of
Health and Human Services
Theme: Health Information Technology
● Nurse Executive Perceptions of Access & Use of End-of-Life Care in Hospitalized Patients
Kimberly Garner, M.D., J.D.; JoAnn Kirchner, M.D.;
Dennis Sullivan, M.D.; Jean McSweeney, Ph.D.,
R.N.; Ursula Braun, M.D.; Julia Goodwin, Ph.D.,
R.N.
Presented by: Kimberly Garner, M.D., J.D., Staff
Geriatrician/Researcher, Geriatric Research,
Education, & Clinical Center, Central Arkansas
Veterans Healthcare System, 2200 Fort Roots
Drive - 3J, North Little Rock, AR 72114, Phone:
(501) 257-2531; Email: kimberly.garner@va.gov
Research Objective: Most patients wish to end their lives comfortably and surrounded by family and friends. However, approximately 50% of medical patients die in hospitals and 20% die in long-term care facilities with many of them receiving aggressive medical treatments up to the point of death. Thus, the majority of patients do not achieve these wishes. Patients, families, and healthcare providers, including nurse executives, will be navigating these difficult decisions as more and more patients live longer with multiple chronic illnesses and cancers. This pilot study’s goal was to explore nurse executives’ perceptions of facillitators and barriers to providing end-of-life care to hospitalized patients.
Study Design: Qualitative research methodology using analytic content analysis of nurse executives’ perceptions of the facilitators and barriers to end-oflife care provided in their facility. Informants will be recruited from one public academic university, four private local medical centers, and three community hospitals.
Population Studied: Nurse executives from one public academic university, four private local medical centers, and three community hospitals.
Principal Findings: Nurse executives consistently reported that patients, families and caregivers experienced end-of-life issues differently based on previous experiences, level of uncertainty in prognosis and ability to manage symptoms. They identified training and education of nurse executives in the options available for curriculum, policy and protocol development as important for their profession. They also identified that training to individualize care provided at the end-of-life and improved communication skills of all healthcare providers, not just nursing profession, as important to improve care provided at the end of life.
Conclusion: Nurse executives identified many challenges to providing comprehensive end-of-life care for hospitalized patients. A proactive presence of senior management to incorporate training and establish appropriate policies and procedures is, and will be, integral in implementing and maintaining systemic change to improve the care provided to patients at the end of life.
Implications for Policy, Delivery or Practice:
With the aging of the veteran population, end-of-life care is going to be an ever-increasing problem that many healthcare providers will be frequently encountering. This study along with others has indicated that, a proactive presence of nursing senior management is integral in implementing systemic change to improve care provided in hospitals at the end of life.
Funding Source(s): NIH - NINR
● Qualitative Analysis of Hospital Executive,
Physician Administrator, & Legal Counsel’s
Perceptions of End-of-Life Care
Kimberly Garner, M.D., J.D.; JoAnn Kirchner, M.D.;
Dennis Sullivan, M.D.; Ursula Braun, M.D.; Jean
McSweeney, Ph.D., R.N.; Ursula Braun, M.D.
Presented by: Kimberly Garner, M.D., J.D., Staff
Geriatrician/Researcher, Geriatric Research,
Education, & Clinical Center, Central Arkansas
Veterans Healthcare System, 2200 Fort Roots
Drive - 3J, North Little Rock, AR 72114, Phone:
(501) 257-2531; Email: kimberly.garner@va.gov
Research Objective: Most patients wish to end their lives comfortably and surrounded by family and friends. However, approximately 50% of medical patients die in hospitals and 20% die in long-term care facilities with many of them receiving aggressive medical treatments up to the point of death. Thus, the majority of patients do not achieve these wishes. Patients, families, healthcare providers and administrators, and hospital executives will be navigating these difficult decisions as more and more patients live longer with multiple chronic illnesses and cancers. This study will inform the development of educational interventions for hospital executives, physician administrators, and hospital legal counsel that was a first step in “bridging this gap” between actual and desirable utilization of these services.
Study Design: Qualitative research methodology using interviews of hospital executives, physician administrators, and legal counsel to explore perspectives on end-of-life care provided in their facility. Informants will be recruited from all three
VA medical centers within VISN 16 (Alexandria,
Houston, and Little Rock) as well as five private medical centers (Baptist Medical Center and St
Vincent Medical Center in Little Rock and a
community medical center in both Shreveport,
Louisiana and Houston, Texas).
Population Studied: Hospital executives, physician administrators, and legal counsel from three VA medical centers within VISN 16
(Alexandria, Houston, and Little Rock) as well as five private medical centers.
Principal Findings: Hospital executives, physician administrators, and legal counsel consistently reported that patients, families and caregivers experienced end-of-life issues differently based on previous experiences, level of uncertainty in prognosis and ability to manage symptoms. They identified training and education of executives and administrators in the options available for curriculum, policy and protocol development as important for their profession. They also identified that training in ethics and evidence based medicine to determine when end-of-life care as very important. They also felt that effective communication skills of all healthcare providers was essential to improving the care provided at the end of life.
Conclusion: Hospital executives, physician administrators, and legal counsel identified that end-of-life care is fraught with significant medical uncertainty and individual variability. This requires a highly specialized team of compassionate care providers coordinating end-of-life care. To initiate and maintain this care requires a proactive presence of senior management to implement and maintain systemic change in hospital-based end-oflife care.
Implications for Policy, Delivery or Practice:
With the aging of the veteran population, end-of-life care is going to be an ever-increasing problem that many healthcare providers will be frequently encountering. To initiate and maintain systemic change in how care is provided at the end-of-life requires education and training of senior management on end-of-life care.
Funding Source(s): VA, VISN 16 MIRECC -
Clinical Partnership Program
● Consumer Expectations of Physicians &
Health Plans among Middle-Aged & Older U.S.
Adults
Lisa Gary, Ph.D.; Nicole Quon, Ph.D.
Presented by: Lisa Gary, Ph.D., Assistant
Professor, School of Public Health, University of
Alabama at Birmingham, 1530 Third Avenue,
South RPHB 330, Birmingham, AL 35294-0022,
Phone: (205) 975-0506; Email: Lgary@uab.edu
Research Objective: The age distributions in the
U.S. are continuing to change at a significant rate.
Projections from the US Census Bureau suggest that adults who are 65 years and older are one of the fastest growing segments of the US population.
In 2008, older persons made up 12.7% of the US population. Middle-aged adults from 45-65 years are also a growing subset of the American population. Like older adults, middle-aged persons are generally more likely to have greater health care needs than their younger counterparts who are under 45 years. Age differences in health care needs and subsequently, the frequency of interactions with the health care system may be due to the demands of managing chronic illnesses or seeking preventive services to reduce age-related risk for health problems (e.g.- screening for diabetes, various cancers). These population shifts underscore the salience of research on the health care experiences and expectations of adult consumers particularly the elderly and middle-aged.
A better understanding of the patterns, trends and risk factors of these patient-provider issues is essential to guide clinical practice, public policy and future research. The purpose of this study was to determine whether older consumers had different evaluations of how well physicians and health plans meet the social norms/ role expectations for patient trust and medical professionalism.
Study Design: Data are from a nationally representative telephone survey (n=5000) of primarily privately-insured adults. Data were analyzed using multivariate logistic regression models with adjustment for the complex sampling design. The role expectations included patient advocacy, caring, racial fairness, honesty and technical competence of physicians and health plans. These role expectations not only reflect medical professionalism but also dimensions of trustworthiness.
Principal Findings: After adjusting for race, health status, perceived vulnerability, region, and other variables, younger respondents (under 45 years) were nearly twice as likely as middle-aged respondents to indicate low trust in the technical competence of physicians particularly regarding medical error rates (OR=1.85). However for another dimension of trustworthiness, patient agency, age had a negative effect where the elderly were less likely to believe physicians acted as poor patient advocates compared to the middle-aged
(OR=0.50). Respondents with a poor-rated health status were more than twice as likely as their healthier counterparts to expect incompetence
(many medical errors) among physicians. The only trust expectation regarding health plans that had a significant age effect was racial fairness. The older respondents were more likely than the middle aged comparison group to expect racial differences in how health plans to treat their enrollees (OR=1.89).
Respondents who reported having a chronic illness had low trust in health plans on two dimensions: professionalism/caring (OR=1.42) and the ability to provide adequate health care access (OR=1.34).
Conclusion: Exploring the nature of age-related differences in expectations is especially important as these differences may affect 1) the future clinical decision making of patients, 2) patient perceptions of negative health care encounters, and 3) their
responses to negative health care experiences.
These results have significant clinical implications for older persons with chronic illnesses and perhaps minority elders.
Funding Source(s): RWJF
Theme: Disparities
● Do Minority Patients Use Lower Quality
Hospitals?
Darrell Gaskin, Ph.D.; Christine Spencer, Sc.D;
Patrick Richard, Ph.D.; Gerard Anderson, Ph.D.;
Neil Powe, M.D., M.B.A., M.P.H.; Thomas LaVeist,
Ph.D.
Presented by: Darrell Gaskin, Ph.D., Associate
Professor, African American Studies, University of
Maryland, 2169 Lefrak Hall, College Park, MD
20724, Phone: (301) 405-1162; Email: dgaskin@aasp.umd.edu
Research Objective: We explored whether minority patients use lower quality hospitals.
Study Design: Using 3 years of state inpatient discharge data from 11 states we computed individual and composite inpatient quality and patient safety indicators in order to explore whether
Asian, black and Hispanic patients are using lower quality hospitals than white patients. We computed the proportion of discharges in low quality hospitals by race and ethnicity using various measures of hospital quality.
Population Studied: We studied 914 hospitals located in 11 states: AZ, CO, FL, MA, MD, MI, NJ,
NY, PA, TX and WI.
Principal Findings: We found evidence that hospitals serving Asian and black patients had lower quality scores than those serving white and
Hispanic patients. However, these racial and ethnic disparities varied by how quality is measured and the patient’s state of residence. All three minority groups received care in hospitals with higher complication rates of decubitus ulcer and postoperative sepsis. Asians and black patients received care in hospitals with higher complication rates of failure to rescue, postoperative PE and
DVT, and higher inpatient mortality rates for craniotomy. Hospitals serving higher proportions of black patients had more problems with surgical and post surgical care. However, hospitals serving high proportions of Hispanic patients performed well on most patient safety measures. Within state comparisons reveal that minorities were not always more likely to use lower quality hospitals than whites.
Conclusion: We found evidence supporting the claim that minorities used lower quality hospitals than whites. However, a detailed examination of the data reveals that this statement should be qualified by 1) how quality is measured, 2) the state where the minorities reside, and 3) by which minority group.
Implications for Policy, Delivery or Practice:
Remedies should target hospitals serving high proportions of Asian and black patients because they are most vulnerable. However, there were problems for hospitals serving high proportions of
Hispanics for specific services. Future research should examine why hospitals serving high proportions of Hispanics perform well on some patient safety measures.
Funding Source(s): CWF
Theme: Disparities
● Use of the Internet for Health Information among Australians & New Zealanders
Robin Gauld, Ph.D.
Presented by: Robin Gauld, Ph.D., Associate
Professor of Health Policy, Preventive & Social
Medicine, University of Otago, P.O. Box 913,
Dunedin, 9054, NZ, Phone: (643) 479-8632;
Email: robin.gauld@otago.ac.nz
Research Objective: This paper presents descriptive data from surveys that probed the use of health information by Internet users in Australia and
New Zealand. Using this data, the paper also contributes to understanding of factors associated with a series of Internet health information use outcomes.
Study Design: Telephone survey: respondents had used the Internet for health information and were asked about emailing doctors, Internet searching, taking Internet information to the doctor, and perceptions of Internet information reliability.
Regression analyses were conducted to examine associations between key outcomes and independent variables. A bootstrap procedure was performed to select variables used in logistic regression models.
Population Studied: Data was obtained from 406 respondents who had used the Internet for health information. Respondents were divided roughly equally between Australia and New Zealand.
Respondents were randomly selected from published telephone directories.
Principal Findings: 52% of health Internet users had emailed their doctor. Few significant differences between Australian and New Zealand responses were found. Internet use for health information was below levels found in North
American and European studies, but there was similar prevalence of emailing and presenting
Internet information to doctors. Emailers were more likely to believe this improved communication with doctors (OR=3.59; 95%CI 1.95, 6.61). Frequent
Internet searchers were more likely to take Internet information to their doctor (OR=2.41; 1.47, 3.92), and believe the Internet to be a more useful information source (OR=2.47; 1.22, 5.03). Females were less likely to email their doctor (OR=0.47;
0.26, 0.85). More educated respondents were more
likely email their doctor (OR=1.99; 1.16, 3.39) and to check website credentials (OR=2.61; 1.53, 4.43).
Conclusion: This study suggests that there is a digital divide among health care Internet users. It shows that, for males, email could have health care advantages.
Implications for Policy, Delivery or Practice:
Doctors may need to develop strategies to deal with patients presenting with Internet information, including how to source information and check its credibility.
Funding Source(s): University of Otago
Theme: Health Information Technology
● Assessment of New Zealand's Primary Health
Care Reforms, 2002-2008
Robin Gauld, Ph.D.
Presented by: Robin Gauld, Ph.D., Associate
Professor of Health Policy, Preventive & Social
Medicine, University of Otago, P.O. Box 913,
Dunedin, 9054, NZ, Phone: (643) 479-8632;
Email: robin.gauld@otago.ac.nz
Research Objective: To assess the process and patterns of organization resulting from the New
Zealand government's primary healthcare reforms introduced in 2002
Study Design: Qualitative methods, including 30 key informant interviews conducted in 2005 and
2008, and analysis of published and unpublished documents.
Population Studied: 80 government-funded primary health organizations (PHOs) created under the 2002 reforms; 16 pre-existing independent practitioner associations (IPAs); NZ government
Ministry of Health
Principal Findings: The new PHOs, and introduction of new funding for these, has improved patient access to primary medical care and increased the range of primary care services available. However, the primary care reforms have also created considerable organizational complexity through inattention to implementation details in the policy development process, and brought into relief a range of unanswered policy questions including: whether the 80 PHOs should compete with one another or be territorial monopolies; the level of fees that patients should pay to see a primary care doctor on top of government subsidies; how to incorporate a wider range of primary care practitioners within a medical practice; and whether
PHOs should fund and coordinate various hospitalbased services. The government also failed to be inclusive of private medical practitioners and their representative IPAs in planning for the primary care reforms. Instead of succumbing to the government's reform plans and dissolving, the IPAs have grown in strength and now provide crucial managerial and support for many of the new PHOs
Conclusion: New Zealand's primary care reforms represent an ambitious attempt to create 'Alma Ata'influenced primary care arrangements. In this regard, New Zealand is ahead of most developed nations. The scope and pace of the reforms has meant many policy details were left until implementation, creating considerable confusion.
Failure to involve doctors in policy planning, or to build on their organization foundations, increased the prominence of the medical profession bringing into question whether the government sufficiently analyzed the implications of its reform plans before embarking on implementation. As it is, the New
Zealand government could have achieved its reform goals through working with, not against, the medical profession and their pre-existing organizations.
Implications for Policy, Delivery or Practice:
New Zealand's primary care reforms have been innovative, but demonstrate how reform goals could have been achieved through building on existing institutions.
Funding Source(s): University of Otago
Theme: Organizational Performance and
Management
● The Health & Economic Consequences of
Abstinence-Only Programs in the United States
Laura Gaydos, Ph.D.; Zhou Yang, Ph.D.
Presented by: Laura Gaydos, Ph.D., Research
Assistant Professor, Health Policy & Management,
Emory University, 1518 Clifton Road, Northeast,
Atlanta, GA 30322, Phone: (404) 727-6554; Email:
Lgaydos@emory.edu
Research Objective: After declining for over a decade, the teen birth rate in the United States rose
3 percent in 2006, sending an alarming signal to the public health system in the United States to pay attention to the health of teenage women. We hypothesize that one important contributor to this increase may have been the federally-funded abstinence-only education program. From 2000 to
2006, the budget of the federal abstinence-only program nearly tripled from approximately $60 million to $168 million. Because of its requirement of exclusivity of abstinence-only education and state financial incentives, this program likely crowded out other proven, cost-effective interventions, such as condom use. This likely led to increases in teen birth rates and higher teen welfare participation in programs such as Medicaid and WIC. Therefore, we investigated the health and economic consequences of the abstinence-only program at the national level.
Study Design: We used the Pregnancy Risk
Assessment Monitoring System (PRAMS) dataset to conduct a series of logistic regression analyses.
Specifically, we estimated how the variations in the density of abstinence-only programs in different states influenced the probability of unintended births among teens, the birth weights of these babies, the medical care cost covered by Medicaid among both the teen mothers and their babies
during pregnancy and at delivery, as well as the probability of participation in WIC programs.
Analyses control for demographic characteristics, socioeconomic status, health insurance coverage, and religion, as well as state-level variables such as existing abortion laws, poverty levels, unemployment rates, and median household income. A policy simulation was also conducted to evaluate the aggregate total effect of the abstinence-only program on the number of unwanted births and the subsequent cost to
Medicaid and WIC at the state level.
Population Studied: Our study population includes teen mothers, ages 14-19, and the babies born to these teen girls nationwide.
Principal Findings: We found that abstinence-only programs lead to higher unintended teen births, lower infant birth weight, higher health care cost covered by Medicaid among teen mothers and their babies, and higher participation rates in WIC programs in states with more intense implementation of abstinence-only education programs.
Conclusion: Abstinence-only education programs lead to less effective prevention of teen pregnancy, and subsequently negative influences on the health of teen mothers as well as their babies, leading to increased economic burdens for society.
Implications for Policy, Delivery or Practice: The
United States has the highest teenage pregnancy rate among developed countries, with over 800,000 adolescents becoming pregnant each year. Given that preventing unintended teen pregnancies has important public policy and public health implications, this study provided strong evidence to modify or limit the implementation of the abstinence-only programs, and promote sciencebased and cost-effective education and interventions, such as condom use and sex education.
Theme: Gender and Health
● The Role of Faith in Family Planning for
African-American Women
Laura Gaydos, Ph.D.; Audrey Flak, B.S.
Presented by: Laura Gaydos, Ph.D., Research
Assistant Professor, Health Policy & Management,
Emory University, 1518 Clifton Road, Northeast,
Atlanta, GA 30322, Phone: (404) 727-6554; Email:
Lgaydos@emory.edu
Research Objective: African-American women in the Southeast United States face inexplicably higher rates of unwanted pregnancies and negative birth outcomes. Although statistical data indicates that religion does not correlate with whether or not adult women use contraception, our preliminary qualitative research indicates that religious/spiritual/cultural findings may mediate some activities which underlie choices that women make – consciously and subconsciously- about their sexuality, partners, and family formation. To further investigate this issue, we conducted a survey of low-income, African-American women regarding how their faith impacts their reproductive health decision making.
Study Design: We collected data from women attending Women, Infant and Children (WIC) programs clinics at two urban, metro-Atlanta health clinics for three weeks. The survey included questions on contraceptive usage, family planning, religious attendance, religiosity/spirituality scales, and demographics.
Population Studied: 441 women, ages 19-49, participated in the study. Our sample consists of the 365 women who self-reported as African-
American or Black Caribbean, of which 79.45% reported Christian affiliation, 5.48% reported Muslim affiliation, and 15.07% reported other or no affiliation. The modal monthly household income was less than $1250 (with 76.19% reporting income under $2920).
Principal Findings: Supporting our earlier qualitative findings, low-income, African-American women showed a very high level of spirituality and religiosity, which appeared to be tied to their beliefs about contraceptives. 23.68% of respondents indicated that at least one method of contraception was against their religious beliefs. Notably, only
1.11% reported that their religious beliefs actually prevented them from using a contraceptive method suggested by a provider. The majority of respondents, 59.33%, believed that their religious institution would be supportive of their family planning choices; although, 25.50% indicated that they would leave or switch their faith institution if their family planning choices were not supported
(as opposed to not discussing choices publicly), indicating the importance of religious institutional support.
Conclusion: Religion and faith play a mixed, but important role in the reproductive health decision making of low-income, African-American women and should be considered when developing interventions.
Implications for Policy, Delivery or Practice:
Religion can have a two-pronged impact on reproductive health issues such as unintended pregnancy and contraception. On an individual level, religious views and spirituality may influence a person or couple’s decision-making about contraceptives, choices about when and how to have a family, and a myriad of other health decisions faced on a daily basis. At the community level, faith institutions have the potential to be very influential in areas of education, counseling, and establishing community norms – whether from the pulpit, in spoken values shared through adult education classes, or unspoken values shared among a religious community. Given the great importance of religion and spirituality to this population of women, incorporating faith values and institutions should be considered in effective
outreach programs for low-income, African-
American women.
Funding Source(s): Emory Strategic Plan Funds
● Leveraging Public Health Partnerships:
Measuring the Social Return on Investment
Allison Gertel-Rosenberg, M.S.; Sara Olsen,
M.B.A., M.A.S.W.; Andrew Tulchin, M.B.A.
Presented by: Allison Gertel-Rosenberg, M.S.,
Senior Policy & Program Analyst/Special Assistant,
Health & Prevention Services, Office of the Senior
Vice President, Nemours, 252 Chapman Road,
Christiana Building, Suite 200, Newark, DE 19702,
Phone: (302) 444-9171; Email: agrosenb@nemours.org
Research Objective: Nemours Health and
Prevention Services (NHPS) is working with Social
Enterprise Associates and Social Venture
Technology Group to measure the social return on investment (SROI) of its work in Delaware around child obesity prevention. This multi-year, multifaceted and multi-partner community initiative has progressive, incremental milestones. This abstract focuses on the field research process undertaken to measure SROI of a comprehensive initiative to establish “360 Degrees of Child Health Promotion.”
Study Design: The SROI 360 Project developed an integrated model prototype each partner can use to track progress continually, and developed a collective model of all partners’ progress for managing and communicating collective impact.
Prototype creation involved numerous data collection and entry points - surveys, file reviews, and interviews. Data were collected by trained researchers, with multiple iterations of review among partner participants and researchers to ensure shared understanding of terminology and data collection methodologies to ensure data collection and model usage would continue following end of this pilot period.
Population Studied: The study’s population was a convenience sample of NHPS partners in Sussex
County in rural, southern Delaware.
Principal Findings: Initial data collection has allowed for analysis of targeted results. Partners instituted data collection tools; data collection continues in 2009. Preliminary results show children who become aware of 5-2-1-Almost None (the
NHPS health promotion message) do change their behavior. Performance insights underscored through this analysis include: The partners who participated in SROI 360 Project reach 12% of
Sussex County children; Children served by these partners tend to lose track of 5-2-1-Almost None principles regarding diet, and perhaps other aspects as well, when they go home. Therefore, the need to engage children in communicating with their parents is evident, as is educating parents through
NHPS social marketing efforts; and The impact of one partner may be visible at sites of other partner organizations. For example, children eating healthier while in school may be reflected in data collected by youth-serving organizations. Individual partner results tell a different story than the collective measurement needed to truly understand impact.
Conclusion: The collective impact of these partners working together towards a common health outcome is beginning to show. The goal of
360 degrees is to reach children during each hour of their day. With this project, partners now have models to more efficiently track metrics, manage performance and communicate with stakeholders.
Implications for Policy, Delivery or Practice:
This project represents a major step forward in quantifying impact and measuring activity, focused on producing an SROI from NHPS (the lead agency) and its partners (community NGOs, state entities, and private businesses). The process is an affordable community collaboration. The tool created can be used to measure activity and performance, as well as provide important information. It also streamlines information collection to gauge results of their work. With the inclusion of 2009 data in the coming year, the benefits of the groundwork laid will be reaped, with the findings informing future policies and practices.
Funding Source(s): Nemours
Theme: Public Health
● Geographic Variation In Medication
Adherence
Teresa Gibson, Ph.D.; Amber Batata, Ph.D.;
Michael Chernew, Ph.D.
Presented by: Teresa Gibson, Ph.D., Director,
Health Outcomes, Health Outcomes, Thomson
Reuters (Healthcare), 777 East Eisenhower
Parkway, Ann Arbor, MI 48108, Phone: (734) 913-
3000; Email: teresa.gibson@thomsonreuters.com
Research Objective: A considerable amount of the investigation of geographic variation focuses on
Medicare data and excludes prescription drugs.
This study focuses on prescription medications used to treat chronic conditions in the commercial sector and incorporates demand-side information on insurance benefit design (e.g., patient costsharing) to analyze variations in adherence.
Study Design: Data were obtained from the 2005-
2007 MarketScan® Database. Medication
Possession Ratios (MPRs) were calculated annually in 2006 and 2007 for four classes of medications (antidiabetic medications, statins, ACE
Inhibitors, Angiotensin II Receptor Blockers (ARB)) for each patient with at least one prescription fill within the class each year. Multivariate logistic regressions of poor adherence (MPR < 80%) were estimated to adjust for patient, health status, employer, benefit plan, provider supply, and area
(e.g., income, unemployment) characteristics. For each geographic area (hospital referral region,
HRR), we computed the expected adherence
(based on predicted values from the multivariate regressions), aggregated to the HRR level and calculated the ratio of observed to expected adherence. The 95% confidence interval of the observed to expected ratio was calculated using the
Byar approximation of the Poisson distribution.
Population Studied: Patients age 18–64 in employer-sponsored plans with at least two years of continuous medical and prescription coverage and who met diagnostic criteria for diabetes were found
(n=449,127).
Principal Findings: Across HRRs the median rate of subobtimal antidiabetic medication adherence was 32.3% of patients, rates in HRRs varied from
19.2% to 54.6%. The median rate was 44% for
ACE Inhibitors (range 30.2% to 60.4%), 43.9% for
ARBs (range 27.5% to 60.7%) and 59.3% for statins (range 45% to 74.5%). The differences between observed and expected adherence was statistically significant for 8.9% of HRRs for antidiabetic medication, 9.1% of HRRs for ACE
Inhibitors, 6.7% of HRRS for ARBs and 1.3% of
HRRs for statins (p<0.05).
Conclusion: Geographic variation exists in the use of maintenance medications for chronic conditions.
Observed market level traits such as provider supply and unemployment rates explain some of the variation, but after adjusting for covariates, geographic variation remains.
Implications for Policy, Delivery or Practice:
Geographic variation in adherence to medications used to treat chronic conditions exists among the commercially insured. Reductions in rates of poor adherence may improve outcomes and spending.
Funding Source(s): Pfizer Inc
Theme: Prevention and Treatment of Chronic
Illness
● A Budget Impact Analysis of Rapid HIV
Screening in VA Emergency Departments
Risha Gidwani, Dr.P.H.; Matthew Bidwell Goetz,
M.D.; Jack Needleman, Ph.D.; Gerald Kominski,
Ph.D.; Steven Asch, M.D., M.P.H.
Presented by: Risha Gidwani, Dr.P.H., 12105
Avenida Consentido, San Diego, CA 92128,
Phone: (858) 366-2153; Email: rishagidwani@yahoo.com
Research Objective: HIV screening is costeffective but not cost-saving. Concerns therefore exist about the financial feasibility of HIV screening programs. We determined the economic impact of implementing a Rapid HIV Testing program (Rapid
Testing) versus a Usual Care approach (Usual
Care) to HIV testing in a Veterans Administration
Emergency Department.
Study Design: Budget impact analysis was conducted to determine the near-term financial impact of HIV screening. A model was built to project costs for seven years and four budgets of interest: inpatient, outpatient, pharmacy and global.
Financial impact of the Rapid Test program entailed the cost of implementing a screening program and treating patients identified through screening.
Financial impact of the Usual Care program entailed the cost of treating patients who are identified later in this integrated system due to symptoms of HIV/AIDS. Costs were assumed to be dependent upon patient disease severity at diagnosis, indicated by four categories: CD4+ <50,
50-199, 200-350 and >350. Treatment cost was a function of the cost per CD4+ category and the proportion of patients in each category. The proportion of patients in each category was determined using clinical information from the
Veterans Aging Cohort Study (VACS). Cost per
CD4+ category was modeled from utilization data of
112 HIV-positive patients treated in Greater Los
Angeles. All costs in the Usual Care program were adjusted back 1.25 years to reflect the mean 1.25 year delay to diagnosis for these patients. The
Usual Care program therefore had 5.75 years of cost estimates, while costs were projected for all 7 years for Rapid Testing. Due to uncertainly regarding HIV prevalence and test offer rates, a variety of conditions were modeled, reflecting testing rates ranging from 5 tests offered per day to a universal screening offer, and HIV prevalence from 0.5% to 5.4% . Monte Carlo simulations were run with 10,000 iterations to provide an estimate of seven-year treatment costs for each program.
Population Studied: HIV-positive veterans.
Principal Findings: Global seven-year cost estimates indicate Rapid Testing is never significantly more costly than Usual Care. At low disease prevalence rates, Rapid Testing is more costly, due to the expense involved in screening patients. At high disease prevalence rates, implementation costs are offset by the higher costs of treating sicker patients in the Usual Care program, yielding lower overall costs for Rapid
Testing. Treatment costs are heavily driven by pharmacy expenses, which represent 58% of costs for Usual Care and 65 % of costs for Rapid Testing.
Inpatient costs represent a larger percentage of costs for Usual Care (22%) than for Rapid Testing
(14%), reflecting higher disease severity at diagnosis for Usual Care patients. Outpatient costs represent approximately one-fifth of costs for both programs.
Conclusion: While treatment under Usual Care begins on average 1.25 years later than under
Rapid Testing, Usual Care patients incur higher annual costs once identified than patients identified through Rapid Testing and have higher overall costs over the 7 year period modeled. Given that
Rapid Testing never costs significantly more than
Usual Care, and that early detection of HIV and linkage to treatment is associated with better health outcomes, this budget impact analysis provides support for the implementation of Rapid Testing,
even when prevalence of disease is as low as
0.5%.
Implications for Policy, Delivery or Practice:
This work adds new rationale to the body of evidence supporting HIV testing by showing that
HIV screening within the VA is not financially unfavorable.
Funding Source(s): VA
Theme: Public Health
● Provider-Level Effects on Receipt of Mental
Health & Drug Abuse Aftercare Services
Elizabeth Gifford, Ph.D.
Presented by: Elizabeth Gifford, Ph.D., Research
Scientist, Center for Child & Family Policy, Duke
University, Box 90545, Durham, NC 27708-0545,
Phone: (919) 613-9294; Email: beth.gifford@duke.edu
Research Objective: To examine provider-level effects on receipt of aftercare services for children and young adults released from the hospital.
Study Design: This study uses Medicaid Claims data from the state of Tennessee. Data provide information on all inpatient stays from July of 1996 to December of 2003. A multilevel competing risk model is used to examine factors that predict whether patients receive follow-up care, are readmitted or neither.
Population Studied: The sample for this study includes 25,718 admissions by 16,421 to inpatient hospitals for mental health or substance use conditions.
Principal Findings: Preliminary studies show that a very small percentage of youth receive follow-up care in within a week of being discharged from the hospital as recommended by guidelines.
Conclusion: Unlike a previous study of length of stay on inpatient care, provider level effects appear to have a much smaller impact on receipt of aftercare services.
Implications for Policy, Delivery or Practice:
Understanding factors that leads to appropriate use of aftercare services for youth with mental health and substance use disorders is necessary to improve quality of care.
Funding Source(s): NIDA
● A Cost Analysis of the Full Service
Partnerships Implemented in San Diego County
Under the Mental Health Services Act
Todd Gilmer, Ph.D.; Susan Ettner, Ph.D.; Willard
Manning, Ph.D.
Presented by: Todd Gilmer, Ph.D., Associate
Professor, Family & Preventive Medicine, University of California, San Diego, 9500 Gilman Drive, La
Jolla, CA 92093-0622, Phone: (858) 534-7596;
Email: tgilmer@ucsd.edu
Research Objective: To examine mental health service utilization and costs associated with the
“Housing First” Full Service Partnerships (FSPs) implemented in San Diego County under the Mental
Health Services Act. FSP programs provide supported (subsidized) housing and access to team-based services using a rehabilitation and recovery focused model, and do 'whatever it takes' to engage and retain clients in treatment.
Study Design: Encounter data were used to identify FSP clients and a propensity score matched control group. Mental health services costs for outpatient, inpatient, emergency, justice system services, and total services were summarized for one year before and one year after initiation of the
FSP. Incremental costs of the program were calculated as the difference in cost among FSP clients, from pre to post, less the difference in cost among control clients from pre to post.
Population Studied: 615 persons with severe mental illness who were either homeless or at risk for homelessness and receiving public mental health services in San Diego County, California.
Principal Findings: 329 FSP clients and 286 clients in a propensity score matched control group were identified. Among FSP participants, outpatient costs increased by $10,267 from pre to post; inpatient costs declined by $6,262; and emergency service costs declined by $1,020, compared to the contemporaneous control group (P<.001, each).
The standardized difference-in-difference estimate of total costs between FSP clients and the control group was not significant.
Conclusion: Participation in FSPs was associated with substantial increases in outpatient services as well as cost-offsets among inpatient and emergency services. The net cost of services, $2,046 over one year, was substantially lower than the total cost of services ($17,304).
Implications for Policy, Delivery or Practice: On any given night there are over 200,00 persons with severe mental illness living on the streets or in shelters. Housing First programs provide intensive community based services and generate substantial cost-offsets. Although they are not cost saving, their net costs are relatively modest, suggesting that they could be widely implemented.
Funding Source(s): NIMH
Theme: Behavioral Health
● A Medicare Perspective on the Cost-
Effectiveness of Diagnostic Strategies in
Patients with Chest Pain & No Known Coronary
Artery Disease
Amanda Gilmore, Ph.D., M.P.H.; James Min, M.D.;
Matthew Budoff, M.D.; Pankaj Patel, Ph.D.; Ken
O’Day, Ph.D.
Presented by: Amanda Gilmore, Ph.D., M.P.H.,
Associate Director, Applied Health Outcomes,
Xcenda, 6215 Windwood Drive, Pensacola, FL
32504, Phone: (850) 478-0023; Email: amanda.gilmore@xcenda.com
Research Objective: Rates of imaging stress tests increased by almost three-fold between 1993 and
2001, with myocardial perfusion single photon emission computed tomography (MPS) alone accounting for nearly 2% of the annual Medicare budget. No economic analysis has been performed to date from a U.S. payer perspective that assesses
CCTA alone or in addition to other non-invasive imaging strategies using diagnostic accuracy results from a prospective, multicenter trial. This study sought to evaluate the cost effectiveness of different diagnostic strategies for individuals with chest pain without known CAD in the ACCURACY
(Assessment by CCTA of Individuals Undergoing
Invasive Coronary Angiography) eligible population.
Study Design: We performed a decision analysis to compare the cost effectiveness of 5 diagnostic strategies: CCTA followed by ICA for positive, equivocal or indeterminate CCTA (CCTA-only),
CCTA followed by ICA for positive CCTA and MPS for equivocal or indeterminate CCTA (CCTA-first);
MPS followed by ICA for positive, equivocal or indeterminate MPS (MPS-only); MPS followed by
ICA for positive MPS and CCTA for equivocal or indeterminate MPS (MPS-first); and direct invasive coronary angiography (ICA). Long-term cost effectiveness was modeled using a Markov process and expressed in terms of the incremental cost per quality-adjusted life year (QALY). Analyses were conducted from the payer perspective using
Medicare reimbursement rates for all direct medical costs. One-way and multi-way sensitivity analyses were performed varying age, gender, CAD prevalence, baseline test cost, and other model parameters to assess the robustness of the results.
Population Studied: The base case analysis consisted of 55-year old men with 30% prevalence of CAD.
Principal Findings: The CCTA-only strategy demonstrated a favorable incremental costeffectiveness ratio of $20,429 per QALY relative to the least expensive CCTA-first strategy. Both MPSonly and MPS-first strategies were more costly and less effective than either CCTA strategy. Results were most sensitive to CCTA sensitivity, MPS sensitivity and CAD prevalence. A CCTA-first or
CCTA-only strategy remained dominant for evaluation of chest pain patients without known
CAD up to a baseline CCTA test cost of $1,100 and
CAD prevalence of 80%.
Conclusion: Based upon the ACCURACY multicenter prospective trial results and using a
$50,000 willingness-to-pay threshold per QALY, a
CCTA-only approach is the most cost effective diagnostic strategy for evaluation of stable chest pain patients without known CAD with intermediate
CAD prevalence.
Implications for Policy, Delivery or Practice: In light of recent evidence that 8% of U.S. healthcare dollars is spent on imaging services and in the wake of widespread movement towards valuebased initiatives, identifying high quality, low cost imaging strategies is of great interest. Given these results, 64 slice CCTA may provide the greatest value for the diagnosis and prognosis of symptomatic individuals with suspected CAD.
Furthermore, these results indicate that despite the availability of multiple non-invasive diagnostic tests to the practicing clinician, “layered” diagnostic pathways are not cost-effective in individuals with intermediate CAD prevalence and should not be used.
Funding Source(s): GE Healthcare
Theme: Medicare
● Preparing for Health Reform: Findings on
Those Who Need Access, the Benefits of
Access, & the Roadmap to Expanding Access
Shira Gitomer, Michelle Proser, M.P.P.; Bob
Phillips, M.D., M.S.P.H.; Andrew Bazemore, M.D.,
M.P.H.; Falayi Adu; Peter Shin, Ph.D., M.P.H.
Presented by: Shira Gitomer, Research & Data
Specialist, Policy & Programs, National Association of Community Health Centers, 1400 Eye Street,
Northwest, Suite 330, Washington, DC 22201,
Phone: (202) 331-4606; Email: sgitomer@nachc.com
Research Objective: This poster will (1) review the number of medically disenfranchised and where they live; (2) discuss the importance of a medical home; (3) demonstrate the benefits of expanding access; and (4) exhibit the great demand for primary health care staff at health centers.
Study Design: The medically disenfranchised live in primary care Health Profession Shortage Areas
(HPSA) or Medically Underserved Areas (MUA), or who are considered a Medically Underserved
Person (MUP), after subtracting a standard 2000 people for every primary care physician. To view health centers as medical homes, Medical
Expenditure Panel Survey (MEPS) data was used to measure the average and median savings for people whose medical home is a health center.
Clinical workforce needs are based on benchmark values, current median provider-to-patient ratios and staffing patterns that correspond to the physician-to-patient ratios observed in other medical settings.
Population Studied: The medically disenfranchised are those who are at risk of inadequate access to a primary care physician due to local physician shortages. They are a subset of the medically underserved – those facing various, and often times compounding barriers to care. For workforce estimates, we included all (1) physicians, excluding physiatrists, pathologist and radiologists,
(2) physician assistants, (3) nurse practitioners, and
(4) certified nurse midwives.
Principal Findings: 56 million people nationally go without needed care because of primary care physician shortages in their communities, and thus considered medically disenfranchised. 21 states each have over one million medically disenfranchised residents. Patient costs at health centers are 41% lower, thus saving the health care system between $9.9 billion and $17.6 billion a year and producing an overall economic impact of $12.6 billion. Nevertheless, health centers need 1,843 primary care providers to achieve benchmark staffing patterns.
Conclusion: 56 million US residents lack adequate access to a primary care physician due to workforce challenges. Clearly, the number of medically disenfranchised is much higher than the 47 million uninsured Americans. Primary care should be central in the nation’s health reform agenda – thus making health centers a viable option. Still, there are several challenges to overcome in order to expand the federal Health Center Program.
Implications for Policy, Delivery or Practice: As the shortage of primary care physicians worsens, the medically disenfranchised will be adversely affected. Expanding health centers can help resolve access to care problems. There is a reciprocal need to solve the maldistrubtion of primary health care services. In an effort to alleviate the current workforce problem, a redistribution of the current workforce will need to take place by moving primary care providers from areas of high concentration to underserved areas. There are many policy options available in addition to health center expansion that could help, including the growth of loan repayment programs like the National Health Service Corps, payment bonuses for underserved locations, and small business loans and tax incentives for primary care practices.
Funding Source(s): NACHC
● Public Opinion about Fairness of Health Care
Inequalities: Implications for Health Reform
Sarah Gollust, Ph.D.; Julia Lynch, Ph.D.
Presented by: Sarah Gollust, Ph.D., Robert Wood
Johnson Health & Society Scholar, University of
Pennsylvania, 3641 Locust Walk, Room 302,
Philadelphia, PA 19104-6218, Phone: (215) 746-
2771; Email: sgollust@wharton.upenn.edu
Research Objective: Conventional wisdom suggests that the best strategy to persuade
Americans to support health reform is to appeal to
Americans' self-interest— their economic security and concerns about their personal health. Given increasing attention to disparities in health and health care, an alternative strategy might be to frame the problem as a moral one, thus activating the public's underlying attitudes about fairness to mobilize their support for health reform. We aimed to describe Americans' perceptions of the fairness of health inequalities and their influence on health policy opinions.
Study Design: We designed an Internet-based survey to ascertain the public's opinions about health care inequalities and policy. Participants read vignettes about health care inequalities in health care access and health care quality.
Participants were then asked to what extent they perceived these inequalities as unfair, as well as what fairness means to them. The main outcome was support for an expanded role for government provision of health insurance, measured as a scale ranging from private provision of insurance on one end of the scale to a government insurance plan covering all medical expenses on the other. We used multiple regression analysis to determine whether, after taking into account self-interest considerations (i.e., health status, uninsurance, economic insecurity), political orientation, membership in a disadvantaged group, and social values, perceptions of the unfairness of inequalities influence participants' preferences for government provision of health insurance. Data are weighted to represent the national U.S. population.
Population Studied: A nationally-representative sample of U.S. adults, recruited by Knowledge
Networks.
Principal Findings: Those in poorer health, those who had ever been uninsured, Democrats, liberals, and those expressing egalitarian values were more supportive of government provision of health insurance. Even after controlling for these characteristics, those who perceived inequalities in health care quality or access as unfair were significantly more likely to support government health insurance. Self-interest-associated characteristics accounted for a relatively small proportion of the variation in policy preferences. In fact, controlling for all covariates, participants' evaluation of health care inequalities as unfair was more strongly associated with preferences toward government provision of health insurance than was having a personal history of uninsurance.
Conclusion: Attitudes about the unfairness of health inequalities are significant predictors of support for an expanded government role for providing health insurance. These findings suggest, contrary to the conventional policy wisdom, that self-interest considerations are not the only influential determinants of health policy preferences.
Implications for Policy, Delivery or Practice:
Some commentators have observed that health reform advocates in 1993 did not frequently rely on moral appeals in the policy debate. These results are suggestive that if public discourse framed problems in the health care system to emphasize the unfairness of inequalities, this strategy could promote more support for health reform.
Funding Source(s): RWJF
Theme: Disparities
● Working Latinos in California, 2001: An
Analysis of the Impact of Immigration Status on
Latinos’ Take-Up of Health Insurance Coverage
Jaime Gonzalez, M.S.S.W., M.S.H.A.; Kristine
Lykens, A.B., M.P.A., Ph.D.; Sue Lurie, Ph.D.;
Sejong Bae, Ph.D.; Robert Kaman, J.D., Ph.D.,
F.A.W.H.P.
Presented by: Jaime Gonzalez, M.S.S.W.,
M.S.H.A., Doctoral Candidate, Health Management
& Policy, University of North Texas Health Science
Center, School of Public Health, 3500 Camp Bowie
Boulevard, Fort Worth, TX 76107, Phone: (817)
691-2987; Email: jgonzale@hsc.unt.edu
Research Objective: In the US, providing health insurance coverage to Latino immigrants has been, and will remain, a controversial issue for years to come. While policymakers grapple with the issue of making public sector coverage programs available to immigrants, the relative disparity that exists in coverage among Latino immigrants raises concerns about the long-term health consequences of such disparities. This study investigates the relationship between immigration status and health insurance coverage among working Latinos in California and examines the impact immigration status has in determining the take-up of public and private coverage.
Study Design: The study is a secondary data analysis of the Health Insurance Coverage Among
Working Latinos in California, 2001 Survey
(Greenwald, 2001). A telephone survey of 1,000 working California residents of Latino descent was conducted. Interviews were conducted in English or Spanish. The survey asked about respondents’ health status, access to health care, and coverage.
The study assessed respondent attitudes concerning control of one’s fate. The survey gathered demographic and socioeconomic characteristics of the respondents. Multinomial logistics regression is employed in order to allow coefficients to be estimated for all categories of the dependent variable.
Population Studied: The sample consists of working Latinos in California distributed across a variety of coverage types and was categorized by the source of the individuals’ coverage (employersponsored, government-sponsored, or purchased coverage). Approximately 70.1% of the sample held some form of coverage while only 29.9% lacked any form of coverage. Of those who had some form of coverage, approximately 55.9% received it from an employer-sponsored program.
Government-sponsored coverage equated to approximately 9.6% of the sample. Of respondents, 77.1% were of Mexican ancestry.
More than 63% of the sample spoke English “very well” or “well.”
Principal Findings: US immigrant Latino laborers are at great risk of not possessing any form coverage. This disparity is noteworthy in this study’s sample among the construction, manufacturing and agricultural/mining/landscaping industries respectively. Of those laborers who did not possess any form of coverage, greater than 50% worked more than 40-plus hours per week.
Surprisingly, among the sample, the rate of insurance was 70.1% while only 29.9% lacked any form of coverage. Factors affecting insurance status and type are explored further in the multi-nomial regressional analysis.
Conclusion: Healthcare reform policies aimed at alleviating access to healthcare services and programs for the Latino immigrants should take into consideration the varying levels of employment and the fact that many industries that employ immigrant
Latino laborers do not offer coverage. Governmentsponsored coverage must revisit their eligibility requirements to provide programs that offer coverage to Latino laborers and their families.
Implications for Policy, Delivery or Practice: The principal findings of our study provide policymakers with the insight necessary to craft healthcare reform policies to address the high rates of uninsurance among the US Latino immigrant population. Our findings emphasize the need for healthcare reform that address the coverage disparities that exist among the US Latino labor force and the economic impact the lack of such access to coverage will have on the overall US labor force and the economy.
Theme: Coverage and Access
● Predictors of Psychiatric Length of Stay &
Emergency Room Department Utilization
Koren Goodman, M.S.Ed., Ph.D.; John Hudson,
Ph.D.; Margaret Lemaster, B.S.D.H., R.D.H., M.Sc.;
Mohammad Alzahrani, M.S., Ph.D.; Joyce Downs,
B.S.D.H., R.D.H., M.S., Ph.D.; Stacey Plichta, Sc.D.
Presented by: Koren Goodman, M.S.Ed., Ph.D.,
Research Assistant, College of Health Sciences-
School of Community & Environmental Health, Old
Dominion University, 3134 Health Sciences
Building, Norfolk, VA 23529, Phone: (757) 582-
3081; Email: KGoodman@odu.edu
Research Objective: The purpose of this study was two fold: (1) to examine predictors of length of stay (LOS) of patients with a psychiatric diagnosis admitted through the emergency department (ED);
(2) to examine factors associated with having the
ED as point of entry for psychiatric admissions.
Study Design: This is a retrospective cohort study of data from Virginia Health Information
Incorporated. Bivariate tests determined associations between LOS and age, ethnicity, gender, insurance type and ED admission status.
Multiple linear and logistic regression models assessed independent effects on LOS.
Population Studied: The researchers accessed the 2006 database and identified all patients admitted to the hospital within Major Diagnostic
Categories (MDC) 19 (psychiatric). Data were gathered on 4,936 records from the database, which includes psychiatric related admissions into
Virginia hospitals through the emergency department.
Principal Findings: 56% of the patients were female, with a mean age of 40.61 (s.d. 19.864).
The strongest predictor of length of stay for psychiatric patients admitted through the ED was
ED admission status. Other factors that are positively associated with LOS are age, gender, insurance type and those identified as non-white.
The most significant predictor in admission rate of psychiatric patients was self-pay with an unadjusted beta of .124 (p < .000). Other predictors included gender, non-white and medical hospital procedure.
Conclusion: A substantial portion of variance in
LOS is due to socio-demographic variables and payment type; however, only the overall model was significant. Findings suggest the need to further investigate alternative treatment options such as crisis stabilization units for patient with mental illness.
Implications for Policy, Delivery or Practice: By identifying specific needs for the patient with a psychiatric diagnosis, a more attractive alternative can be established. Most patients in this study were female, nonwhite, age 40 and used Medicaid as a source of payment. This study does not address emergency department overuse and the inappropriate use of mental health behavioral services. Further research is needed to determine if the participants were returned visit ED users, as this could not be addressed in this study. However, findings provide information that may be useful in the development of public policy supporting alternative treatment options for patients suffering from mental illness. Additional research is recommended in order to determine in detail which effective community based services can best meet the growing needs of patients suffering with mental illness.
● Vitamin D Status of the U.S. Population by
Age & Race/Ethnicity
Paul Gorrell, Ph.D.; Amy, Galifianakis, M.P.H.
Presented by: Paul Gorrell, Ph.D., Senior
Research Scientist, Social & Scientific Systems,
Inc., 8757 Georgia Avenue, Silver Spring, MD
20910, Phone: (301) 628-3237; Email: pgorrell@s-
3.com
Research Objective: To determine serum 25hydroxyvitamin D (25OHD) levels in the U.S. population by race/ethnicity, age and supplement use to provide information for nutrition counseling and data for policy analysis.
Study Design: Pooled data from six years of the
National Health and Nutrition Examination Survey
(NHANES 2001-2006), a cross-sectional survey of a nationally representative sample of the U.S. population. Vitamin D status (as evidenced by serum levels of 25OHD) were analyzed from 24,407 individuals (10,114 non-Hispanic whites, 6,166 non-
Hispanic blacks, 6,236 Mexican Americans, and
873 Other Hispanics). Weighted N = 275,667,437.
Data were analyzed for the three race/ethnicity groups for which NHANES provides a nationallyrepresentative sample: non-Hispanic Whites, non-
Hispanic Blacks, Mexican Americans. Data were also analyzed for the following age groups: 1-5, 6-
11, 12-17, 18-29, 30-44, 45-64, 65 and older.
Vitamin D serum levels were measured as part of the NHANES laboratory data collection at mobile examination centers (MECs). Weighted population estimates were obtained using a constructed 6-year version of the MEC weight variable per NHANES analytic guidelines. Vitamin D ranges were defined as follows: deficient (<11 ng/mL), inadequate (<20 ng/mL), insufficient (20-30 ng/mL), sufficient (>30 ng/mL).
Population Studied: U.S. civilian population.
Principal Findings: At each age range, non-
Hispanic blacks were significantly more likely to exhibit Vitamin D deficiency (defined as <11 ng/ML) or inadequacy (<20 ng/ML) as non-Hispanic whites and Mexican Americans. Mexican Americans were less likely than blacks to be Vitamin D deficient, but more likely than non-Hispanic whites. Overall 33% of the population had deficient Vitamin D levels
(23% for non-Hispanic Whites, 45% for Mexican
Americans, and 73% for non-Hispanic Blacks); 46% had insufficient levels (50% for non-Hispanic
Whites, 45% for Mexican Americans, and 24% for non-Hispanic Blacks); 21% had sufficient levels
(27% for non-Hispanic Whites, 9% for Mexican
Americans, and 3% for non-Hispanic Blacks). For the three groups Vitamin D deficiency and inadequacy increased during childhood and early adulthood. For non-Hispanic Blacks 18-29 35% had inadequate levels and 84% insufficient levels.
For each group, supplement use (typically 400 IU) decreased deficiency and insufficiency rates although it did not affect between-group differences
(e.g. among non-Hispanic Blacks reporting supplement use 57% had insufficient levels versus
77% for those reporting no supplement use).
Conclusion: Although there is no clear consensus concerning adequate Vitamin D serum levels, the results clearly demonstrate that (i) a large percentage of the overall U.S. population have inadequate Vitamin D levels, (ii) that higher percentages of non-Hispanic Blacks and Mexican
Americans have deficient or inadequate levels; and
(iii) that supplement use is associated with lower percentage of persons with inadequate Vitamin D levels. In addition, for each population studied, childhood and early adulthood show marked increases in percentage of persons with inadequate
Vitamin D levels.
Implications for Policy, Delivery or Practice:
Nutritional recommendations should be revised to increase Vitamin D supplementation, particularly for
African Americans, Mexican Americans and other
Hispanics. Nutrition counseling should be extensive for parents, as well as for teens and young adults. Further, the present findings of this large study reinforce the recent recommendations of the Academy of Pediatrics to increase infant and children’s Vitamin D intake.
● Duration of Child Insurance Coverage as a
Predictor of Health Care Utilization
Anthony Goudie, Ph.D.; Gerry Fairbrother, Ph.D.;
Lisa Simpson, M.B., B.Ch., M.P.H., F.A.A.P.
Presented by: Anthony Goudie, Ph.D., Research
Assistant Professor, Child Policy Research Center,
Cincinnati Children's Hospital, 3333 Burnet Avenue,
Cincinnati, OH 45229, Phone: (256) 343-5121;
Email: Anthony.Goudie@cchmc.org
Research Objective: Recent studies have looked at the association between the number of gaps in child insurance coverage with levels of health care utilization. However, the effect on health care utilization of varying durations without insurance in a homogeneous low income population eligible for public insurance is unknown. The objective of this study is to test the hypothesis that the duration of health insurance coverage will determine the level of health care utilization for office-based physician and hospital emergency department visits. It is anticipated that the continuously uninsured will have fewer physician visits (primarily due to access limitations) but may tend to seek primary health care within an emergency department setting. The child population with longer durations of uninsurance will resemble the continuously uninsured and those with shorter periods of uninsurance will have a health care utilization profile more reminiscent of those with continuous public or private insurance.
Study Design: Data is summarized over 24 months to create a 6 category insurance status variable representing uninsured for 24 months, insured for
1-11, 12-17, and 18-23 months, and full two-year coverage with public or private insurance. The dependent variables of interest are dichotomized
(none or any) to represent office-based physician visits and the use of a hospital emergency department over a two-year period. Logistic regressions incorporating strata and nationally representative population weights are used to model the significance of insurance status
(continuous public insurance as the reference class) on the health care utilization dependent variables. Models are adjusted for demographics and health status.
Population Studied: Children 2-17 years of age in families with household incomes <200% of the FPL with 24 months of complete responses to monthly insurance status from Panel 10 (2005-06) of the
Medical Expenditure Panel Survey (MEPS) were included in this study.
Principal Findings: The white, non-hispanic population and those residing in the Northeast or
Midwest are more likely to have continuous public or private insurance than have any duration of uninsurance. Across constructed insurance categories there is no significant difference in the overall child population across age group, gender, or health status. Continuously uninsured (AOR:
0.307, 95% CI: 0.180-0.526) or those with less than one year of insurance (AOR: 0.369, 95% CI: 0.230-
0.592) are approximately 3 times less likely to have had an office-based physician visit compared to those with continuous public insurance. Also, continuously uninsured (AOR: 0.190, 95% CI:
0.075-0.478) are 5 times less likely to have incurred an emergency department visit in 2005 or 2006 compared to those with continuous public insurance.
Conclusion: Over a two-year period the child population from low income families with no insurance coverage or with coverage for less than one year is more likely to have no contact with an office-based physician. They are not substituting for this lack of health care by accessing primary care in the emergency department. It stands to reason that health care is obtained in another manner or children are going without needed, and especially preventive, care.
Implications for Policy, Delivery or Practice: The implication for policy is to intensify efforts to enroll children into a public health insurance program.
Theme: Medicaid, SCHIP and State Health Reform
● A Profile of Ohio Children with Chronic
Medication Use, 2004-2008
Anthony Goudie, Ph.D.; Lisa Simpson, M.B., B.Ch.,
M.P.H., F.A.A.P.; Gerry Fairbrother, Ph.D.
Presented by: Anthony Goudie, Ph.D., Research
Assistant Professor, Child Policy Research Center,
Cincinnati Children's Hospital Medical Center, 3333
Burnet Avenue, Cincinnati, OH 45229, Phone:
(256) 343-5121; Email:
Anthony.Goudie@cchmc.org
Research Objective: Recent trends indicate that the prevalence of chronic medication use in a commercially insured national child population increased from 2000 to 2005. The main objective of this study is to profile Ohio children to ascertain the secular 2004-2008 trend in chronic medication use.
This study also tests to see if any change in chronic medication use prevalence occurred across children with different insurance status or if demographic factors are associated with trends.
Study Design: Chronic medication use is defined by taking one or more prescription medications to treat a medical, behavioral or health condition for a period of at least 12 consecutive months. Weighted survey prevalence rates and confidence intervals are used to test for a significant change between
2004 and 2008. Stratified logistic regression models
for data from 2004 and 2008 surveys incorporating strata and state representative child weights are used to generate multivariable results. A primary predictor of interest is constructed with 5 categories representing 12 months of child insurance status: uninsured – less than 200% FPL, uninsured – at least 200% FPL, insured part of the year, public, and private (reference class). Demographic covariates include age, race/ethnicity, parent’s education, marital status, and region of residence.
Population Studied: This study includes participants aged 0-17 years in the Ohio Family
Health Survey, 2004 and 2008. Results for 2008 are reflective of 85% complete data entry.
Principal Findings: In 2004 and 2008, children with chronic medication use tended to be proportionally older, male, have parents who are divorced or separated, and maintain full-year public insurance (compared to non-chronic medication users, p-value < 0.05). In contrast to national trends the prevalence of children with chronic medication use has declined (not significantly) from
17.5% (95% CI: 16.7-18.3) in 2004 to 16.7% (95%
CI: 15.7-17.9) in 2008. Children with a full-year of public insurance had twice the odds of being a chronic medication users than children with full year private insurance in 2004 (AOR: 2.00, 95% CI:
1.62-2.48) and 2008 (AOR: 1.80, 95% CI: 1.56-
2.07). Children who are uninsured or insured for only part of the year are less likely to be chronic medication users compared to children with private insurance for the full-year, but this difference is not statistically significant.
Conclusion: At first glance, the decline in chronic medication use in Ohioan children is a positive trend. However, the reasons for this decline need to be further studied to determine whether nontherapeutic treatment of chronic conditions has increased or chronic conditions are going untreated as a result of health care access issues. The significant difference in chronic medication use between low-income children with full-year public and private insurance provides clues that lowincome uninsured and insured part of the year children may be experiencing access and inability to pay for needed therapeutics.
Implications for Policy, Delivery or Practice:
Determine the chronic health care status of uninsured children and aggressively seek to enroll them in public insurance coverage.
Funding Source(s): Ohio Department of Jobs and
Family Services
● Suicidal Ideation Among HIV-Infected &
Uninfected Veterans in Care
Joseph Goulet, Ph.D., M.S.; Steven Crystal, Ph.D.;
Nancy Day, Ph.D.; Cynthia Gibert, M.D.; Robert
Kerns, Ph.D.; Amy Justice, M.D., Ph.D.
Presented by: Joseph Goulet, Ph.D., M.S.,
Research Scientist, Research, VA Connecticut
Healthcare System, 950 Campbell Avenue 11-
ACSLG Building 35A, West Haven, CT 06516,
Phone: (203) 932-5711 ext. 5325; Email: joseph.goulet@va.gov
Research Objective: Suicidal ideation (SI) is highly prevalent among US military veterans. While the majority of veterans with SI may not attempt or complete suicide, it is associated with a variety of medical and mental health conditions, and it may have additional deleterious effects on health. We examined the prevalence of SI, its correlates, persistence, and association with one year all cause mortality in a sample of HIV-positive and
HIV-negative veterans in care.
Study Design: We examined cross sectional associations of SI with: demographics; health risk factors (e.g., smoking, obesity); Short Form-12 (SF-
12) pain and general health items; psychiatric and medical conditions (e.g., major depressive disorder
[MDD], PTSD, alcohol, and HIV); and period of service (e.g. Vietnam). SI was defined as an affirmative response to item 9 of the Patient Health
Questionnaire (PHQ-9), assessed for the preceding
4 weeks. We also examined the persistence of SI at one year follow-up survey, and the association with one year all-cause mortality. Multivariable analyses used logistic and Cox regression.
Population Studied: We analyzed self-report and administrative data on 5,998 US veterans participating in the Veterans Aging Cohort Study
(VACS), a prospective multisite observational study of HIV-positive and HIV-negative veterans in care at eight VA facilities.
Principal Findings: 95% of the sample was male, the mean age was 50, 28% were white, and 48% served in the Vietnam era. Overall, 962 (16%) participants reported SI at baseline, of whom 159
(17%) reported it ‘nearly every day.’ In the multivariable logistic model, SI was associated with: younger age; white race; male gender; MDD;
PTSD; drug disorder; smoking; higher levels of pain; lower levels of self-reported health; and HIVinfection (p<0.05 for all). At one year follow-up, 53% of patients reporting baseline SI continued to report
SI, and 7% of veterans without baseline SI reported it at follow-up. Correlates of persistence were similar to correlates of baseline SI. One year allcause mortality was 3.2% among veterans with SI versus 1.8% among those without (p<0.05). In multivariable Cox models, SI ‘nearly every day’ was associated with a higher mortality risk (hazard ratio=2.42, 95% CI=1.12-5.24).
Conclusion: SI is highly prevalent, and it is associated with health risk factors, and psychiatric and medical conditions. SI is persistent, and is associated with higher all-cause mortality.
Implications for Policy, Delivery or Practice:
Compared to non-VA samples, the prevalence of SI is extremely high (e.g., lifetime prevalence in the
National Comorbidity Survey was 8.7%), especially among older veterans. Our findings underscore the need to provide frequent SI screening, outreach,
and services to veterans, and highlight subgroups at elevated risk that warrant special attention in screening efforts.
Funding Source(s): NIAAA, Department Of
Veterans Affairs
Theme: Military and Veterans Health Care
● Patient Centered Counseling: Understanding the Processes
Anusha Govinda Raj, M.Sc.; Charis Ng, B.A.; Wai
Fung Chong, M.B.A.; Loo See Yeo; Doris Liew
Presented by: Anusha Govinda Raj, M.Sc.,
Research Analyst, Health Services & Outcomes
Research, National Healthcare Group, 6
Commonwealth Lane, #04-01, 149547, Singapore,
Phone: (656) 496-6934; Email: anusha_govinda_raj@nhg.com.sg
Research Objective: Previous management of chronic disease patients involved didactic education and authority in helping them adopt changes necessary for better health. However, a new process, patient centered counseling (PCC) was introduced in February 2007. The counseling process incorporated interviewing skills such as basic principles of motivational interviewing, empowering beliefs, self-efficacy to illicit change talk and goal-setting. The skills applied were dependent upon patients’ readiness level for change. As part of service improvement, the study aimed to understand how nurse care managers
(CMs) applied PCC in their practice and the barriers for implementation.
Study Design: At the study outset, a discussion with the Assistant Director of Nursing, along with the training and implementation tools, allowed for an understanding of the ideal processes of PCC.
CMs from nine primary care clinics where PCC was introduced, were invited to participate in the study.
Three focus groups were conducted in the last two weeks of June 2008. A trained external moderator conducted the focus group with the aid of two observers. The focus group sessions were tape recorded. Transcripts were reviewed by the research team members and analyzed using a grounded-theory approach.
Population Studied: A total of 15 CMs participated in the focus groups with one to two from each of the nine primary care clinics.
Principal Findings: CMs had attended different training sessions due to logistical reasons. A need for a structured and standardized training session was expressed. CMs varied in the time taken to implement the PCC from almost immediately to a year, reflecting their confidence in applying the new skills. However, they expressed a unanimous feeling of fulfillment in PCC, as they felt it facilitated rapport building with the patients, and made patients assume ownership for their own health and well-being. Young and educated patients responded better to PCC, while elderly patients often felt resigned, and being passive and dependent, preferred CMs to tell them what to do.
CMs’ encountered different patient types including motivated patients who were committed to making necessary changes, unmotivated patients who were in denial, depressed, frustrated or in shock, and ambivalent patients who understood the need for change but were not motivated enough to take action. Barriers mentioned included spoken language and time taken to counsel effectively.
Some inconsistencies were noted between the training and implementation materials which may have created some confusion among the CMs in the understanding of PCC.
Conclusion: Despite some initial resistance and challenges, all CMs were applying PCC in their practice a year after training and found it beneficial in improving the health of their patients. A better training structure for continuous learning and sharing will ensure consistency in the delivery of care.
Implications for Policy, Delivery or Practice:
PCC, although well accepted by the CMs, may not be as applicable in some subgroups such as the elderly and uneducated. Hence, other methods need to be explored in these cases. Formalized sessions such as focus groups allow for useful feedback to determine the acceptance of new processes and for ways in which they may be improved.
Theme: Quality and Efficiency: Measurement
● Hospice Utilization Across Sites of Care:
National & State Hospice Rates Among
Community & Institutionalized Settings
Pedro Gozalo, Ph.D.; Susan Miller, Ph.D.
Presented by: Pedro Gozalo, Ph.D., Assistant
Professor (Research), Community Health, Brown
University, Box G-S121-6, Providence, RI 02912,
Phone: (401) 863-7795; Email:
Pedro_Gozalo@brown.edu
Research Objective: Prior research has shown evidence of state variations in hospice penetration rates. However little is known how hospice providers allocate their services across residential settings in their specific local healthcare markets, and across states. The objective of this study is to examine the division of hospice care provided in a community setting versus in institutionalized nursing home (NH) setting in all 50 US states.
Study Design: Retrospective cohort study. A file was constructed linking MDS assessments to
Medicare claims and the denominator file. For each
Medicare beneficiary, utilization dates from all
Medicare claims, MDS assessments, plus the dates of death were used to create a longitudinal
“beneficiary event history file.” This file allowed us to determine to what extent hospice care periods occurred during a NH stay, and allowed us to identify NH and hospice providers who collaborate.
Aggregation of patient-level data was done by hospice provider, NH, local healthcare market area, and by states. Beneficiary zip codes in the denominator file and addresses in the provider of service files were used to determine geographic location of beneficiaries and of sites of care.
Population Studied: Three cohorts were included: all US Medicare beneficiaries decedents, all US
Medicare hospice users and all US NH residents in the year 2000. Secondary data for 2000 included three main sources: the 100% Medicare Part A hospice, inpatient, skilled nursing and home health claims, the 100% Medicare denominator file, and the NH Minimum Data Set (MDS) assessments for all NH residents.
Principal Findings: Nationally, hospice rates for beneficiaries in the community (26.7%) are almost double than among those in NHs (13.8%). There is great inter-state variation in hospice rates, both overall and by setting. The community/NH hospice ratio ranges from 1.08 in Oklahoma to around 10 in
Hawaii. Among hospice providers, almost 89% nationally offered hospice in the NH, ranging from
64% in Wyoming to (close to) 100% in Delaware,
Ohio, Missouri, Iowa and New York. Nationally, each hospice provider served an average of 9 NHs, ranging from none (Alaska and Puerto Rico) or less than 2 (Wyoming and Hawaii), to 26 NHs in Florida or 15 in Illinois and Texas. As far as the beneficiaries they served, nationally 18.14% of all hospice beneficiaries were served (at some point) in a NH. In ten states, less than 10% of all beneficiaries were served by hospice providers in
NHs, while in five states more than 30% of those enrolled in hospice were served in a NH. Nationally, hospice average LOS was 6 days longer among NH residents (45.2 days community versus 51.55 NHs).
However, in 21 states hospice average LOS was longer among community than among NH beneficiaries.
Conclusion: Our findings provide the first comprehensive national study of hospice provision and utilization across settings.
Implications for Policy, Delivery or Practice:
Results can be used by Medicare, and by hospice and NH providers to help in targeting hospice utilization policies in specific states/markets and settings.
Funding Source(s): AHRQ
Theme: Long Term Care
● System Factors Related to Medication Gaps in
VA Medical Centers
Katrina Graham, M.B.A.; Monika Safford, M.D.;
Bonita Agee, Ph.D., M.P.H.; Ellen Funkhouser,
Dr.P.H.; Thomas Houston, M.D., M.P.H.
Presented by: Katrina Graham, M.B.A., Program
Support Assistant - Research, VA Reserve
Educational Assistance Program & University of
Alabama at Birmingham School of Health
Professions, Birmingham VA Medical Center & the
University of Alabama at Birmingham, REAP Mail
Stop 151, 700 South 19th Street, Birmingham, AL
35233, Phone: (205) 933-8101 ext 6076; Email: katrina.graham@va.gov
Research Objective: Diabetes is a major and growing health issue in the United States. Despite guideline recommendations to treat risk factors like blood glucose aggressively, many diabetic individuals are not optimally controlled, and recent evidence suggests that doctors may not manage diabetes aggressively enough. An important barrier to optimal medication management is patient adherence. Diabetes is difficult to manage when patients run out of medications, a problem that could potentially be minimized in health systems with excellent communication among healthcare providers and an emphasis on quality improvement.
In this report, we use results of a factor analysis of the National 2001 VA Quality Manager Survey to examine the relationship between system factors within VA Medical Centers (VAMC) and medication gaps among diabetes patients.
Study Design: The 148-question National VAMC
Quality Manager Survey was conducted in 2001 among 242 quality managers from 129 VA facilities.
It assessed diabetes guideline approaches, level of support for guideline efforts, regional office leadership for guidelines, hospital use of guideline performance data, and hospital culture. For this study, we use results from a factor analysis of the survey that resulted in a 6-factor solution: (1)
Cooperative and flexible culture (Cronbach a=0.89), where team work and interdepartmental coordination is encouraged; (2) Structured implementation process (a=0.90), where key implementation steps are planned, executed, and monitored; (3) Leadership support (a=0.87), where the regional network provides support and fosters competition among facilities; (4) Resources
(a=0.85), where personnel and financial resources are dedicated to implementation; (5) Staff
Resistance (a=0.64); where sources of resistance were identified; and (6) Measurement (a=0.67), where data are collected to measure and evaluate performance. To explore links between these 6 system factors and the outcome of medication adherence, as measured in failure to refill a medication leading up to a visit (“gaps”), we propose using multilevel multivariate modeling.
Population Studied: We studied 645,000 diabetic patients at VA Medical Centers from 2000-2001.
Principal Findings: Associations between each of the system factors and medication gaps for diabetes medications will be presented and discussed in the context of the Donabedian structure-process-outcome paradigm.
Conclusion: We will discuss the relationship between system factors at VAMCs and outcomes of care.
Implications for Policy, Delivery or Practice:
One aspect of medication adherence, running out of
medications, could be readily addressed by systemlevel interventions. Our study will shed light on which system-level factors are particularly related to this outcome of care, permitting quality managers and health system leaders to prioritize their efforts at continuous quality improvement.
Funding Source(s): VA, UAB School of Health
Professions
● Socioeconomic Factor Predict Increased
Incidence for Colorectal, Lung, Stomach, Liver,
Pancreas & Oral Cancers for the Bronx vs. NYC
2001-2005
Dennis Graham, D.N.Sc., R.N., N.P.; Noah
Kornblum, M.D.
Presented by: Dennis Graham, D.N.Sc., R.N.,
N.P., Director of Nurse Practitioner Program,
Nursing & Oncology, Memorial Sloan Kettering
Cancer Center New York, New York & Saint
Barnabas Hospital Bronx, New York, 1275 York
Avenue, New York, NY 10065, Phone: (212) 639-
3028; Email: grahamd@mskcc.org
Research Objective: The purpose of the current study is to assess the differences in the incidence of selected cancers in the Bronx vs. New York City
(NYC) and evaluated the possible association of neighborhood levels of individuals at or below federal poverty level, median household income
(MHI) and levels baccalaureate or higher graduates.
Study Design: The New York Cancer Registry
Cancer Data was queried for seven Bronx neighborhoods and NYC for the average annual incidence for the cancers examine in this study.
This data was entered into a data base as individual cancer by sex. This data were examined for rates/100,000 for the Bronx neighborhoods and compared to rates/100,000 for all of NYC for the selected cancers in this study. Other associated variables were also examined included economic profile as evidenced by the MHI, rates of college graduates and levels of individuals below the federal poverty level by zip codes for these Bronx neighborhoods. Associations were tested using an unadjusted Chi square and logistic regression models. Descriptive statistics were used for other variables
Population Studied: The Bronx is one of the five geographic boroughs making up New York City
(NYC) and has higher rates of crime, poverty, lower median household income (MHI) and fewer college graduates. Other studies have shown that level of poverty and level of education are predictor overall morbidity and mortally. No research has explored the impact of MHI, levels individuals at or below the federal poverty levels and levels baccalaureate or higher degrees on the incidence for common cancers found in Bronx neighborhoods compared to all of NYC.
Principal Findings: The overall incidence of cancer in this study is greatest in Bronx neighborhoods with the lowest MHI, lowest level of college graduates and the higher levels of individual living below the federal poverty level (p < .0001 Chi sq) Regression analysis reveals that for Bronx neighborhoods with higher levels of poverty, lower levels of education and MHI for certain tumors are independent predictors of increased incidence (p <
.0001)
Conclusion: This data demonstrates that the higher incidence of certain types of cancer in Bronx neighborhoods are associated with level of education, poverty and income.
Implications for Policy, Delivery or Practice:
More studies are needed to evaluate other possible reason for the higher incidences of the cancers in
NYC poorest borough.
Theme: Public Health
● Accurately Estimating the Number of Children
Without Access to Healthcare
Roy Grant, M.A.; Sarah Overholt, M.A.; Deirdre
Byrne; Arturo Brito, M.D.
Presented by: Roy Grant, M.A., Director, Applied
Research, Medical Affairs, The Children's Health
Fund, 215 West 125th Street, New York, NY 10027,
Phone: (212) 535-9400; Email: rgrant@chfund.org
Research Objective: The degree to which children in the US have inadequate access to health care has generally been measured by the percent and number of children without health insurance for a full year. While an important step towards assuring adequate healthcare for children, this measure understates the extent of the problem. The methodologies to determine the number of full-year uninsured are inconsistent, and non-economic factors also affect access. We set out to contrast the methodologies of the federal data sets used to estimate the problem of uninsured children
(including MEPS, CPS, NHIS) to determine the reasons for inconsistency among these sources. To more accurately estimate the percent and number of children with inadequate access to care, we supplemented this calculation with estimates of children with transportation barriers to access.
Study Design: Starting with a literature review, we ascertained the inconsistencies and inaccuracies of federal data sources commonly used to estimate uninsured children. We applied a methodology developed by previous investigators to adjust the number of full-year uninsured children to include the additional number of partial year uninsured children.
To ensure an accurate estimate of the percent and number of children without access to healthcare, we additionally used original data to estimate the extent of the problem of transportation as a barrier.
Population Studied: Focus was on uninsured and medically underserved children.
Principal Findings: Current estimates of uninsured children have two principle limitations: Sampling bias inherent in their survey methodologies including an arbitrary upper age limit (e.g., 17 or 18 years) which excludes older adolescents and young adults through age 19 (federal HHS/HRSA upper limit for pediatrics), 21 (American Academy of
Pediatrics upper limit) or 24 (Society for Adolescent
Medicine upper limit). The result is an artificially low estimate of the percent and number of uninsured children and youth. Including children with interrupted coverage who are uninsured for part of the year yields a 14.4% increase in the number of children for whom insurance is a barrier to access.
Additionally, the percentage of children who are inadequately insured (e.g., covered for catastrophic illness but not routine and preventive care) is difficult to estimate. An estimated 4% of children nationwide do not have transportation access to healthcare regardless of their insurance status, a problem which especially affects children in health professional shortage areas. The magnitude of difference based on definition and sampling methodology is indicated by a 2007 report by the
Congressional Budget Office which found that advocates estimated that up to 6 million uninsured children are eligible for Medicaid or SCHIP each year compared to 1.1 million estimated by the Bush
Administration.
Conclusion: When appropriately calculated, the number of children with significant economic and non-economic barriers to healthcare access increases from approximately 8.1 million (full-year uninsured) to more than 20 million children.
Implications for Policy, Delivery or Practice:
Accurately determining the magnitude of the problem of US children without adequate access to healthcare underscores the importance of developing and implementing a national strategy for universal coverage of children and adolescents.
Having accurate data and understanding the limitations of generally accepted federal estimates helps advocates respond to conflicting points of view.
Theme: Coverage and Access
● Premiums Cut Further into Compensation:
Changes from 2001-2005
Peter Graven, B.S.; Lynn Blewett, Ph.D.; Jeanette
Ziegenfuss, Ph.D.; Michael Davern, Ph.D.
Presented by: Peter Graven, B.S., State Health
Access Data Assistance Center, University of
Minnesota, 2221 University Avenue, Southeast,
Suite 345, Minneapolis, MN 55414, Phone: (612)
298-3072; Email: grave165@umn.edu
Research Objective: To provide estimates of the change in the portion of employee compensation dedicated to health insurance premiums for policyholders with private employer-based health insurance coverage.
Study Design: A retrospective analysis that uses premium data from the Medical Expenditure Panel
Survey linked to income and other data from the
Current Population Survey. Data are pooled over the two year periods, 2001/2002 and 2004/2005, to improve reliability of the estimates and show recent changes.
Population Studied: Privately employed individuals and families in the US with employer based health insurance in 2001/2002 (n=86,119) and 2004/2005 (n=79,690).
Principal Findings: Total premiums in 2004/2005 were $4,071 for individuals and $10,882 for families. Individuals contributed 13.5 percent of employee compensation to premiums in 2004/2005 compared to 11.7 percent in 2001/2002, a 15.4 percent increase in the premiums’ share of compensation. For families, the share of compensation increased from 19.8 percent to 22.8 percent over the same period generating a 14.9 percent increase. These increases occur while earned income for individuals and families decreased or remained relatively flat (2.7 percent decrease and 0.5 percent increase, respectively).
Conclusion: The share of total compensation dedicated to premiums increased significantly for individuals and families from 2001/2002 to
2004/2005. Premiums absorbed most of the employee’s total compensation increases over the period.
Implications for Policy, Delivery or Practice:
Concerns about increasing premiums having an impact on the budgets of individuals and families are very real. Part of the reason take home incomes are stagnating for employees is that total compensation increases are absorbed by the increases in health insurance premiums.
● The Quality of Inpatient Care Provided for
Patients with Pneumonia: Findings from the
National Healthcare Quality Report (NHQR) of the U.S. Agency for Healthcare Research &
Quality (AHRQ)
Darryl Gray, M.D., Sc.D.
Presented by: Darryl Gray, M.D., Sc.D., Medical
Officer, Center for Quality Improvement & Patient
Safety, Agency for Healthcare Research & Quality,
540 Gaither Road, Rockville, MD 20850, Phone:
(301) 427-1326; Email: darryl.gray@ahrq.hhs.gov
Research Objective: To describe the quality of care provided for US inpatients with pneumonia.
Study Design: Retrospective cross-sectional study.
AHRQ’s annual National Healthcare Quality Report
(NHQR) includes hospital-reported data from
Medicare’s Quality Improvement Organization
(QIO) Program on the receipt of widely recommended care components not contraindicated in individual patients 18+ years of age. For pneumonia inpatients meeting specific criteria, care components currently assessed
include: 1) collecting blood cultures before starting antibiotics (in patients who were admitted to
Intensive Care Units within 24 hours of hospital arrival and received antibiotics); 2) if antibiotics were administered, starting them within 4 hours of the patient’s arrival at the hospital; 3) if antibiotics were administered, using initial antibiotic regimens consistent with current guidelines (excluding patients with health care-associated pneumonia); 4) among patients 50+ years of age, screening for influenza vaccine status (with vaccination if indicated) before October-February discharges; and
5) among patients 65+ years of age, screening for pneumococcal vaccine status (with vaccination if indicated). Composite measures assess the proportion of instances where inpatients receive individual care components for which they are eligible. Data on inpatient mortality among adult pneumonia patients not transferred to other acute care hospitals came from the Nationwide Inpatient
Sample (NIS) of AHRQ’s Healthcare Cost and
Utilization Project. NIS administrative data include demographic information, ICD-9-CM diagnosis and procedure codes and discharge disposition collected on all discharges from a 20% weighted sample of US non-federal hospitals. Weights based on hospital characteristics are applied to the sample to generate nationwide estimates. Mortality rates for adults with principal discharge diagnoses of bacterial or viral pneumonia (excluding transfers to other acute-care hospitals) were adjusted by age, gender, age-gender interactions, and by All Patient
Refined-Diagnosis Related Group (APR-DRG) risks of mortality.
Principal Findings: Principal Findings:
Aggregated state-specific data indicate that blood cultures were drawn before starting antibiotics in
82.5% (standard error [SE]=0.6) of cases from
2005, versus 90.0% (SE=0.5%) for 2006. If provided, antibiotics were started within 4 hours in
76.4% (SE=1.1%) in 2005, versus 79.7%
(SE=1.0%) in 2006. Initial use of antibiotics consistent with current guidelines rose from 80.4%
(SE=0.5%) to 85.3% (SE=0.5%). For 50+ yearolds, viral influenza vaccination management as described above rose from 56.9% (SE=1.6%) in
2005 to 68.1% (SE=1.5%) in 2006. For pneumococcal vaccination in 65+ year-olds, analogous figures rose from 62.2% (SE=1.5%) to
73.9% (SE=1.5%). In 2005, patients received
93.5% (SE=1.9%) of all recommended pneumonia care components, versus 95.2% (SE=1.9%) in
2006. Individual state-equivalent jurisdictions varied considerably in the provision of recommended care components; state-wide composite scores ranged from 50.6% (SE=0.4%) to
88.1% (SE=0.3%) in 2006 alone. Nationwide adjusted mortality fell from 49.0 (SE=0.2) per 1,000 pneumonia discharges in 2005 to 44.9
(SE=0.2)/1,000 in 2006 (p< .0001).
Conclusion: For all listed process measures, the quality of inpatient pneumonia care improved from
2005 to 2006. While concurrent reductions in inpatient pneumonia mortality seen in other nationwide data may have multiple causes, these results are still encouraging.
Implications for Policy, Delivery or Practice: The variability of results seen across measures and across states identifies some areas of high-quality performance, along with potential opportunities for improvement.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Measurement
● The Mystery of the Health Insurance
Experiment: An Extended Attrition Analysis
Jeremy Green, B.A.
Presented by: Jeremy Green, B.A., Ph.D. Student,
School of Public Health, Yale University, 60 College
Street, New Haven, CT 06510, Phone: (617) 861-
7886; Email: jeremy.green@yale.edu
Research Objective: The Health Insurance
Experiment was the most expensive social experiment ever conducted in the United States.
The experiment found that health insurance plans can save money through the use of cost sharing in their plan designs without causing deleterious impacts on the health of their enrollees. Following the experiment a quarter century of public and private health policy has aimed to impose cost sharing on individuals through copayments and deductibles in order to avoid the use of potentially unnecessary health services that plan enrollees may seek if all of their health care services were paid for by their health insurance. Several recent articles in the health policy literature have suggested that the experimental results may in truth be a spurious artifact of an attrition differential between the experimental cost sharing treatment groups and the free fee-for-service plan control group.
Study Design: In this paper I first discuss the recent arguments surrounding the potential problem of attrition and then proceed to re-analyze the original experimental data. I examine some reasons that participants gave for their attrition. I then use multiple imputation (MI) to fill-in the missing data and re-estimate treatment effects using multiple imputation estimation.
Population Studied: Adult population of the health insurance experiment.
Principal Findings: Of the 176 attrition cases in the treatment group, 25 participants left the experiment voluntarily primarily because they were returning or changing to other insurance because it provides better or less expensive coverage, or is more convenient than the experimental plan, and
23 participants left the experiment voluntarily primarily because they were returning or changing to other health insurance because of difficulty paying for their health care under the experimental plan. Diagnostic analysis of the multiple imputation
procedure finds that hospital admissions are very difficult to predict which may be because illness is a largely random event.
Conclusion: The best information we have regarding the reasons for attrition, are in those reasons for attrition provided by the individuals leaving the experiment. Behavioral economic theories may suggest that people do not like the very idea of trading of health and dollars; individuals may have left the experiment because they were worried about being able to pay for their health care, even if they would have been able to. It is not possible to know whether or not this benign explanation for attrition is true.
Implications for Policy, Delivery or Practice:
Ideally, one would want to examine aspects of the data that are consistent with Nyman’s hypothesis.
For example, one would want to see if the attrition cases had hospital visits just before leaving the experiment, and if they were diagnosed with bad things just before leaving the experiment.
Unfortunately, this information is only available in the public use files for the non-attrition cases, and is not available for the attrition cases. This lack of information on attrition cases after they left the experiment is a potentially serious design flaw in the RAND Health Insurance Experiment. In the
Negative Income Tax Experiments of the 1980s, researchers were concerned with the potential problem of attrition when designing the experiments, and steps were taken in the design phase to avoid weaknesses of the experiments that may result from attrition. Specifically, their outcome variables of interest were employment, wages, and welfare participation rates. In the Negative Income
Tax Experiments, these outcome measures were observed for attrition cases after leaving the experiment because they were available in administrative records of the Unemployment and
AFDC programs. By contrast, in the Health
Insurance Experiment, hospital admissions were not observed for those leaving the treatment groups after they left the experiment, because their health care claims were no longer in the experiment's claims system.
Funding Source(s): AHRQ
● Health Service Usage Patterns among
Michigan Adult Benefit Waiver Health Plan
Enrollees: Use of Office Visits vs. Emergency
Room Admissions
Theresa Green, A.A.C., M.B.A., Ph.D.; Kieran
Fogarty, Ph.D.
Presented by: Theresa Green, A.A.C., M.B.A.,
Ph.D., Director of Community Health Planning,
Berrien County Health Department, 769 Pipestone,
P.O. Box 706, Benton Harbor, MI 49085, Phone:
(269) 927-5607; Email: tgreen@bchdmi.org
Research Objective: The purpose of this study is to determine if the Adult Benefit Waiver program
(ABW) is effective in getting enrollees to seek nonemergent and preventive health services with their primary care provider rather than through the emergency room. This study is a preliminary exploration of enrollee usage patterns. The Adult
Benefit Waiver program is funded through an
SCHIP HIFA 1115 Demonstration Waiver. An objective analysis is necessary to demonstrate the sucess or failure of the SCHIP waiver to increase access to care among adults in Michigan. Michigan is on the forefront in providing coverage to childless adults in poverty through a diversion waiver of unspent funds from the SCHIP program, or State
Childrens Health Insurance program. Through
SCHIP and Medicaid, Michigan insures 95% of the state's children with substantial funds remaining unused. The excess money supports ABW which provides coverage to 62,000 adults annually, since
2004. The program covers individuals with incomes less than 35% of the Federal Poverty Level, and provides several outpatient services, including emergency room visits and pharmaceuticals at minimal or no cost. This research focused on answering the following questions: 1. Is enrollment in the Adult Benefit Waiver program associated with enrollees receiving preventive care with a primary care provider? 2. Are there office visit or emergency room usage pattern differences related to demographic characteristics? and 3. Among Adult
Benefit Waiver program enrollees, does visiting a primary provider correlate with the number of emergency room visits?
Study Design: My research was conducted through secondary analysis of the billing data collected on ABW enrollees and housed by the
Michigan Department of Community Health. I requested and received billing data for those clients throughout Michigan who were enrolled in ABW for a 12 month continuous period from June 06 - July
07, leaving a one month lag time for late billing.
Consequently, the data set had 1,137,289 billing charge entries for 18,111 clients.
Population Studied: Enrollees of the ABW program are adults ages 19 - 64 who are at or below 35% of the Federal Poverty level. Although data was collected and briefly analyzed for the entire state of Michigan, this study currently focuses on Region 3, one of nine regions in Michigan, which includes the counties of Southwest Michigan
(Berrien, Cass, Van Buren, Allegan, Barry,
Calhoun, St. Joseph and Branch). Region 3 represents 1,632 billing enrollees, and 9% of the billing data. (Other regions are currently being analyzed and could be completed by the tiem of the conference)
Principal Findings: The MDCH dataset contained demographic information on gender, race and age of the enrollees as well as region of residence. Of the 1,616 enrollees studied, 48% were female and
52% were male. Race had 6 categories, with
65.3% of enrollees classified as ‘Caucasian’. The age of the enrollees was normally distributed with
ages ranging from 20 – 64 years. The mean age was 43.6 ± 10.3 years. When examining the dependent variables ‘count of ER visits’ and ‘count of office visits’ analysis showed that the data is not normally distributed for either, but rather is severely positively skewed, despite attempts at transformations. For ER, the minimum value was zero and the maximum was 40 ER visits with 91% of enrollees having 4 or less visits. The mean is
1.62 ± 3.22 visits, with a Median of 1 and a Mode of
0.
For ‘number of office visits’, the minimum value was zero and the maximum was 80. The mean is 9.34 visits, with a Median of 6 and a Mode of 2. To answer Q1, simple proportions were calculated.
Question 2 assessed demographic influences.
Because the independent variables were not normally distributed, the non-parametric Mann-
Whitney U and Kruskal- Wallis tests were used.
Both tests are appropriate for ordinal dependent data with independent group, and are the nonparametric analogs for the t-test and the
ANOVA, respectively. For the age correlation with visits, and also for Q3, the Kendall Tau test was calculated. Kendall’s Tau is a nonparametric alternative to the Pearsons r for measuring relationships when data seriously violate the assumptions of r. My initial research question asked whether enrollment in the ABW program would improve enrollees visiting their primary care provider. Results showed that 32% of continuous enrollees still had no routine checkup during the year of observation. This proportion is within the confidence interval of the generalized, and very low income, Michigan population, when asked if they had a routine checkup in the last year during the
2005 BRFS survey. Demographics were examined to answer Question 2. The Mann-Whitney U demonstrated a significant difference based on gender for office visits but not for ER visits, with females having more office visits than males.
(females: average 9.90 ± 9.91 office visits annually, with a Median of 7 and a mode of 6. Males average
8.82 ± 9.80 office visits annually with a Median of 6 and a Mode of 2. The Kruskal-Wallis test showed significance differences based on race in the number of office visits but no differences in emergency visits. Post-Hoc tests found statistically significant results only when comparing Caucasians to Blacks - Caucasians having more visits – even after the Bonferroni correction. For age, the
Kendall’s Tau statistical test showed a positive significant correlation between age and office visits and a negative significant correlation between age and ER visits.Kendall’s Tau yielded a highly significant positive correlation between visiting the office and visiting the ER. The statistical significance implies that a relationship does exist and is not chance, however, the correlation is quite small. R2, which tests the meaningfulness of r, is only .01 which means the variation in ‘office visits’ accounts for only 1% of the variance of ‘emergency visits’.
Conclusion: A summary of the statistics shows several interesting points. First, 32% of all Region
3 ABW enrollees had not billed for an office visit although of those who did bill, there were 5.7 times more office visits than ER visits. There were statistically significant differences in office visit usage based on each of the demographics studied.
There is a correlation between number of office visits and number of ER visits. This relationship is positive, but not very strong.
Implications for Policy, Delivery or Practice:
This study was one of the first attempts to quantify the effectiveness of Michigan ABW program, a unique, groundbreaking program nationally. The future of the Adult Benefit Waiver program is uncertain as the funding waiver is currently up for renewal. Literature would suggest that since enrollment in ABW provides health coverage and a medical home, that enrollees would have increased preventive office visits and decreased ER usage.
However, enrollees in the ABW program were equally likely than the general population to have not visited the doctor within the year and still averaged 1-2 visits to the ER annually. Equally surprising, increased office visits did not correlate with decreased ER visits in this population. To find a more accurate model additional work is necessary including a state-wide analysis utilizing the remaining 8 regions of the data set. Using diagnostic codes, which are included in my data, might give a clearer picture of high vs. low users and true emergency ER visits versus misuse. A prospective study with a direct control group, or a repeated measure study of clients before and after
ABW enrollment, would be a more stringent research examination of usage. And finally, it is extremely important to examine the financial implications. Is it cost effective to enroll clients in the ABW program?
Theme: Medicaid, SCHIP and State Health Reform
● The Course of Chronic Pain & Disparities
Course over time among Patients at a
Multidisciplinary Pain Clinic
Carmen Green, M.D.; Tamera Hart-Johnson, M.S.
Presented by: Carmen Green, M.D., Associate
Professor, Anesthesiology, University of Michigan,
UH1H247, P.O. Box 0048, 1500 East Medical
Center Drive, Ann Arbor, MI 48109, Phone: (734)
936-4240; Email: carmeng@med.umich.edu
Research Objective: Disparities in the healthcare experience based upon social determinants exist.
Minorities and the poor are disproportionately affected by the chronic pain experience and are at risk for suboptimal chronic pain care. We sought to examine the longitudinal course of pain outcome disparities once enrolled at a Multidisciplinary Pain
Center in order to determine the impact on pain
care disparities when pain complaints are addressed by a multidisciplinary approach.
Study Design: This six-month cohort survey study prospectively evaluated the impact of race and neighborhood socioeconomic status on health status among Black and White patients (18-50 years) being seen at a tertiary care pain center.
Subjects were compared on pain characteristics and pain associated outcomes (Pain Disability
Index, Beck’s Depression Inventory, SF-36, and
Brief Symptom Inventory) over six months.
Participants were asked to complete baseline, three- and six-month surveys.
Population Studied: Blacks presenting for chronic pain management at the Multidisciplinary Pain
Center were recruited in the study as was the next white patient of matched age and sex (N=182; 92% response).
Principal Findings: At baseline, most were women
(63%), educated (65% > high school), in a longterm relationship (52%), and employed (58%).
Racial disparities existed in three disability measures (family, p=.06; self care, p=.05; and life support, p=.07), and five psychiatric symptom measures (interpersonal sensitivity, p=.004; depression, p=.06; phobic anxiety, p=.09; paranoid ideation, p<.001;and psychoticism, p<.001) at baseline, although few quality of life measures differed. Disability disparities disappeared over time while the psychiatric disparities persisted. At least one socioeconomic measure, in most cases more, was significantly associated with all disability measures except sexual behavior at baseline
(p<.05) and with most quality of life measures, although not with the psychiatric symptoms. The relationships between low education levels and unemployment and disability reduced over time, where the relationships with poverty did not. The quality of life measures did not consistently change over time.
Conclusion: The increasing prevalence of chronic pain has significant public health and policy implications. Overall, we show that some disparities reduced over time indicating the potential for disparate care at a primary care level.
Furthermore, a multidisciplinary pain management approach may reduce disparities. Specifically, disparities in disability reduced over time while mental health and quality of life disparities persist.
These preliminary findings support the need for testing more multidisciplinary interventions to reduce disparities in pain care and to optimize pain care for all.
Implications for Policy, Delivery or Practice:
Referral to a pain care specialist and a multidisciplenary treatment approach may help diminish disparities in pain related disability, but more research is necessary to addres mental health and quality of life disparities.
Funding Source(s): AETNA Foundation
● Trends in Medicare Part D Prescription Drug
Plan Benefits: What Do Beneficiaries Get for their Money?
Leslie Greenwald, Ph.D.; John Kautter, Ph.D.;
Gregory Pope, M.S.; Jenya Kaganova, Ph.D.
Presented by: Leslie Greenwald, Ph.D., Principal
Scientist, Division for Health Services & Social
Policy Research, RTI International, 5104
Wetheredsville Road, Baltimore, MD 21207,
Phone: (410) 448-2611; Email: lgreenwald@rti.org
Research Objective: To understand how benefits have been changing as the Medicare Part D program moves from initial implementation to a more mature program. The research looked at trends in the multitude of Medicare Part D benefits and premiums between 2006 and 2008, and examined how plans differ.
Study Design: We compared benefits and premiums among the different basic and enhanced products available to beneficiaries in 2006 through
2008. Benefit and premium data analyzed were available from the CMS Health Plan Management
System (HPMS). Benefit variables analyzed included total premiums for Part D, other cost sharing, gap coverage, and utilization management.
Interpretation of the analysis was aided from the findings of discussions we conducted with many private insurers.
Population Studied: Medicare beneficiaries in Part
D PDPs and MA-PDPs.
Principal Findings: Beneficiaries continued to be faced with an increasing number of Part D plan options in 2008 compared to 2006 and 2007. Much of the growth in 2008 occurred within MA-PDs. In general, there was a marked trend toward enhanced plans. Despite the increased competition, we found that premiums increased among many plans between 2006 and 2008.
Availability of some benefits (such as gap coverage) also increased; still, only about 50% of plans offer gap coverage and most of gap coverage is limited to generics or a subset of formulary drugs.
Other coverage elements were less generous on average in 2008; for example, fewer plans offer $0 deductibles. We found that MA-PDs offered slightly better benefits (including more gap coverage) for generally lower premiums compared to PDPs. This is an expected finding given that MA-PDs had the potential to subsidize additional benefits through
Medicare Part A and B rebates. To compare the combined effect of benefits and premiums among various plans, we constructed a basic generosity index that compared the overall out of pocket spending for a representative standardized beneficiary. We found that total beneficiary out of pocket costs can often be lower under enhanced plans compared to basic plans, even taking into account the higher premiums charged.
Conclusion: We found that the number of Part D benefit plans continues to increase. This can mean
more choice for beneficiaries, but the increased competition has not resulted in lower premiums and only somewhat better Part D benefits. There have been some improvements in gap coverage, but also decreases such as the number of plans offering $0 deductibles. The increased prevalence of enhanced benefit plans in 2008 – now the dominant plan type – suggests a shift in the type of plan most likely to be offered to beneficiaries in future years.
Implications for Policy, Delivery or Practice:
Medicare beneficiaries have a growing variety of plan choices for their Medicare Part D coverage.
More of these options are enhanced plans, which may be a reasonable financial choice for those who chose them. But beneficiaries have also shown a preference for low cost plans. It remains to be seen whether benefits in enhanced plans will improve sufficiently in the future to lure beneficiaries from low cost, basic plans.
Funding Source(s): CMS, RTI International
Theme: Medicare
● Do Prescription Drug Advertisements
Facilitate Informed Decision Making?
Jerry Grenard, Ph.D.; Visith Uy, B.S.; Jose Pagan,
Ph.D.; Dominick Frosch, Ph.D.
Presented by: Jerry Grenard, Ph.D., Postdoctoral
Scholar, Health Services, University of California,
Los Angeles, 1990 Polaris Drive, Glendale, CA
91208-2425, Phone: (818) 543-7476; Email: grenard@ucla.edu
Research Objective: Prescription drug advertising has remained controversial since it first became more prominent in 1997. The pharmaceutical industry argues that advertising helps consumers make informed decisions about prescription drugs, an assertion that has been disputed by opponents of the practice. The present study examined how consumers interpret advertisements for cholesterollowering drugs and whether these facilitate informed decision-making.
Study Design: Participants were recruited from the general population to watch nine prescription drug advertisements recorded from television, including ads for four cholesterol-lowering drugs. At the time of recruitment, participants were told that they would be viewing general health programming, in order to avoid topical bias. Participants answered open-ended questions designed to probe the messages and impressions they took away from the ads, as well as their preferences with respect to pharmacological and lifestyle treatment options.
Interviews were audio-recorded and transcribed.
Open-coding methods were used to explore the range of responses and identify recurring themes.
Population Studied: Participants (N=30) were adults from the general population between the ages of 25 and 65, balanced by gender (53% male,
47% female) and evenly divided between those with health insurance and the uninsured. Participants were predominantly Caucasian (67%), followed by
African-American and Latino individuals (13% each), with the remaining participants reporting mixed ethnic background (7%).
Principal Findings: While some participants felt that certain ads were credible and had some educational value, participants were more likely to identify the ads as marketing tools and often found the information untrustworthy, incomplete or misleading. Participants sometimes expressed a desire for more information on the drugs, but despite this lack of information, most were willing to infer that the drugs being advertised were effective.
Information presented in some ads was misleading about the sources of cholesterol. Participants generally thought that the influence of genetics was equal to or greater than diet on blood cholesterol.
Some participants expressed concern about the side effects that reflected the lack of data about frequency or magnitude of side effects. A few respondents identified positive reminder messages about healthy lifestyle choices while others heard messages contrary to public health recommendations that prioritize prevention. These included that medication could substitute entirely for lifestyle change or might enable continuation of unhealthy lifestyle choices without consequences.
Conclusion: Our findings indicate that the ads currently have very limited value as a method for helping consumers make informed decisions.
Implications for Policy, Delivery or Practice:
Policy and regulatory changes may be necessary to improve the educational performance of prescription drug ads.
Funding Source(s): RWJF
● Building Rural Public Health Preparedness
Capacity - An Evaluation of Engaging Public &
Private Community Partners
Jennifer Griffith, Dr.P.H., M.P.H.; S. Kay Carpender,
B.S.; Cara Pennel, M.P.H.; Barbara Quiram, Ph.D.
Presented by: Jennifer Griffith, Dr.P.H., M.P.H.,
Associate Research Scientist, Texas A&M Health
Science Center School of Rural Public Health,
Office of Special Programs, Mail Stop 1266,
College Station, TX 77843-1266, Phone: (979)
458-4504; Email: jgriffith@srph.tamhsc.edu
Research Objective: Rural communities frequently lack resources and sufficient formal government structures, requiring these communities to engage a diverse group of stakeholders to facilitate effective preparedness, response, and recovery processes.
The Pandemic Influence Preparedness Planning
Project was an intergovernmental partnership to aid rural counties in creating plans that engaged public and private community partners. The three phase project consisted of numerous components, both interactive and didactic, that were provided regionally in phases 1 and 2 and locally in phase 3.
Our evaluation questions focused on 1) county participation during the project, 2) the effectiveness of regional versus local components, 3) types of stakeholders participating, and 4) effectiveness of interactive versus didactic components.
Study Design: This evaluation used project tracking materials to address our four evaluation questions. Participation rates for counties, types of components, locations of components and stakeholders were calculated. County participation was defined as having at least one stakeholder present for a project component. Stakeholders were categorized into four categories: 1) public health and medical, 2) government (state, county, or local), 3) first responders (law enforcement, fire,
EMS) and 4) non-traditional stakeholders. Nontraditional stakeholders included businesses, the faith-based community, non-healthcare service providers, and the public/private education.
Population Studied: Thirty-three rural counties in northern Texas and stakeholders within those counties participated.
Principal Findings: The number of counties participating in all offered components increased across the three phases, from 19% in phase 1, to
27% in phase 2, and 58% in phase 3. The number of counties who did not participate in at least one component was greater when offered regionally in phases 1 and 2 (9% and 21% respectively), however, all counties participated in at least one component when offered locally. During the project
481 stakeholders participated in at least one component and were categorized as follows: 29% public health/medical; 28% government; 23% first responders, and 20% non-traditional stakeholders.
The proportions of types of stakeholders were similar throughout the project; however the number of each type of stakeholder increased when components were offered locally with 302 stakeholders participating in phase 3 compared with
148 in phase 1 and 31 in phase 2. Interactive components resulted in the higher levels of participation, particularly those held locally which had county participation rates of 76% or greater; while didactic component participation rates ranged from 28% to 52%.
Conclusion: Our evaluation determined that building community capacity takes time, as evident from increased rates of county and stakeholder participation over the course of the project. In addition, activities, held within target communities, that are interactive in nature result in greater participation at both the county and individual stakeholder levels.
Implications for Policy, Delivery or Practice:
Sufficient time is needed to develop partnerships.
Multiple interactive collaborative components, versus didactic components, further develop these partnerships when held locally rather than regionally to engage a diverse group of stakeholders in the process. These factors are key when building capacity for public health preparedness in rural communities that have fewer resources and depend on a broad spectrum of stakeholders.
Funding Source(s): Texas Department of State
Health Services, Health Service Region 2/3
● It Doesn’t Pay to Kill: Using Publicly Available
Safety Measures to Drive Reductions in
Mortality & Improve Efficiency in a Multi-
Hospital System
Stephen Grossbart, Ph.D.; Amanda Renner, M.P.H.
Presented by: Stephen Grossbart, Ph.D., Chief
Quality Officer, Quality Management, Catholic
Healthcare Partners, 615 Elsinore Place, Cincinnati,
OH 45202, Phone: (513) 639-2784; Email: srgrossbart@health-partners.org
Research Objective: Identify the impact of using the Agency for Healthcare Research and Quality’s
(AHRQ) Patient Safety Indicators (PSI) to drive quality improvement in a multi-hospital system.
This study tests the question: are the publicly available and often publicly reported AHRQ PSIs important process improvement tools that can effectively drive hospitals to improve their care delivery processes?
Study Design: The study identifies the impact of preventable harm, as identified by the AHRQ PSIs, in the areas of mortality, hospital cost, and length of stay (LOS) and how this information has been effectively leveraged to launch systematic process improvement initiatives within Catholic Healthcare
Partners (CHP). The study compares a base line year (2007) to performance in 2008 following the implementation of system-wide initiatives to reduce mortality and LOS with targeted interventions focused on avoidance of patient harm.
Population Studied: Acute inpatients discharged from CHP hospitals in 2007and 2008 from 26 hospitals within CHP. Catholic Healthcare Partners is organized into 8 regional service areas in five states and, with recent acquisitions, includes 34 acute care hospitals. The system is the largest healthcare provider in Ohio and also has hospitals located in Kentucky, Pennsylvania, and Tennessee.
Principal Findings: In analyzing the impact of preventable patient harm, CHP identified the annualized effect of patient safety events on its care delivery processes. In a single year, CHP estimates that 2,316 patient safety events (PSIs) accounted for 9,643 excess or avoidable patient days and $14.3 million excess variable cost. In addition, 6.6% of all mortalities in the system had one or more events. The system’s quality department used such analysis to focus the system’s attention on improving patient care and hospital revenues, as well as preparing the system for implementation of CMS’s policy to not reimburse for preventable hospital acquired conditions (HAC).
In 2008, CHP adopted system-wide objectives to reduce mortality and LOS associated with harm.
Senior leadership performance was evaluated based, in part, on these two objectives and oversight was provided by the CHP Board of
Trustees. Throughout the year, the system and its hospitals adopted systematic process improvement initiatives, including three internal process improvement collaboratives, lean-sigma process improvement training, and targeted task forces to identify opportunities and implement effective strategies. Between the baseline year and the implementation year, CHP reduced LOS by 2.2%.
To adjust for variation in patient mix, CHP compared observed LOS to expected geometric mean LOS, as published in the federal register. In
2007, system performance O/E LOS ratio was 1.12 and improved to 1.10 (p-value < 0.001). This improvement represents an avoidance of over
19,000 days and a projected cost avoidance of $7.7 million. While decreasing LOS, the system simultaneously reduced acute inpatient mortality from 1.84% to 1.80%; a 2.13% overall decrease in mortality. Though this was not statistically significant (p-value = 0.174), it represents an annualized savings of 94 lives. The prime focus of
CHP’s ongoing process improvement collaborative are early recognition of sepsis, improved prevention of venous thromboembolism, enhancing rapid response teams by using an early warning system, as well as improving end-of-life care to avoid unnecessary hospital admissions.
Conclusion: Measuring and analyzing patient harm has provided significant momentum for patient safety and clinical redesign. AHRQ PSIs, as originally intended by the Agency, have focused a large health system on reducing harm and improving patient outcomes. Significant drivers of both public reporting and internal focus on quality improvement included tying senior executive accountability to performance on quality indicators and board oversight of corporate quality objectives and goals, created a strong environment for improvement.
Implications for Policy, Delivery or Practice:
This analysis provides evidence that CMS’s focus on harm in 2008 has had an impact on hospital performance and that the AHRQ PSIs are practical tools to drive quality improvement.
Theme: Quality and Efficiency: Measurement
● Health Insurer-Sponsored Personal Health
Records: Potential Benefits & Limitations to
Patients & Physicians
Joy Grossman, Ph.D.; Teresa Zayas Cabán, Ph.D.;
Nicole Kemper, M.P.H.
Presented by: Joy Grossman, Ph.D., Senior
Researcher, Center for Studying Health System
Change, 600 Maryland Avenue, SW, Suite 550,
Washington, DC 20024, Email: jgrossman@hschange.org
Research Objective: Consumer uptake of personal health records (PHRs) is low and evidence of clinical impact is limited, but advocates believe that
PHRs have the potential to improve quality and reduce costs by engaging patients and enhancing patient-clinician communication. While PHRs sponsored by health insurers and public payers such as Medicare and Medicaid are very new, interest in promoting them has grown because these entities possess claims data—the largest source of electronic patient data—and a large enrollment base. To better understand the potential role of claims-based PHRs, this study identifies key features of insurer-sponsored PHRs, outlines the potential benefits and limitations for insured patients and their physicians and suggests approaches to enhance PHR value.
Study Design: Qualitative analysis of telephone interviews with 35 respondents conducted in July and August 2008, including executives in 12 health insurers offering PHRs and one state Medicaid program, consumer representatives, physicians,
PHR experts and insurance trade association representatives.
Population Studied: Insurers, consumer representatives and physicians were located in the
12 nationally-representative Community Tracking
Study markets. Participating insurers included four national and eight statewide/local insurance companies, which together cover a large proportion of privately-insured individuals in the United States.
Principal Findings: Health insurers have or are actively planning to modify their PHRs by: (1) autopopulating the PHR with claims-based medical care
“event” data, medications and laboratory results; (2) supporting electronic physician access; (3) integrating patient care applications; and (4) leveraging the PHR for care management, such as sending clinical alerts and reminders and identifying patients who are eligible for disease management.
Insurers typically give enrollees control over thirdparty access to patient-entered PHR data but gain consent for insurer access to the data as a term of participation under the PHR privacy agreement.
Enrollees typically must submit a written request for information on disclosures of PHR data. No insurers currently offer PHR portability. Consumer representatives and physicians felt insurer PHRs address a major barrier to PHR use by autopopulating the PHR with claims data, although they identified substantial limitations to using claimsbased data for treatment purposes. Consumer representatives also noted distrust of sharing personal health information with insurers and
Internet security concerns as barriers to enrollee use. Most physicians did not believe that the data’s clinical value was worth the costs of overcoming workflow barriers to accessing and reviewing the data.
Conclusion: Given the substantial challenges to enrollee and physician adoption, it is unclear how widely existing health insurer PHRs will be used.
Implications for Policy, Delivery or Practice: To reduce barriers to use, public and private payers could: (1) develop clear, understandable privacy policies, for example, specifying how the data will and will not be used; (2) provide real-time audit tools that allow enrollees to identify who has accessed the PHR; (3) support portability of PHR data to patient-controlled data platforms and applications, e.g. Google Health and Microsoft
HealthVault; and (4) develop unique insurer-specific tools to help enrollees navigate the health care system, understand insurance benefits and manage out-of-pocket costs.
Funding Source(s): RWJF
Theme: Health Information Technology
● Emergency Department Visits by Long-Term
Care Home Residents in Ontario
Andrea Gruneir, Ph.D.; Paula Rochon, M.D.,
M.P.H.; Geoffrey Anderson, M.D., Ph.D.; Chaim
Bell, M.D., Ph.D.; Susan Brosnkill, Ph.D.; Michael
Schull, M.Sc., M.D.
Presented by: Andrea Gruneir, Ph.D., Postdoctoral Fellow, Kunin-Lunenfeld Applied Research
Unit, Baycrest, 3560 Bathurst Street, Toronto, M6A
2E1, CA, Phone: (416) 351-3732 ext. 2745; Email: agruneir@klaru-baycrest.on.ca
Research Objective: To describe emergency department (ED) visits made by long-term care home (LTCH) residents in Ontario.
Study Design: We used linked population-based administrative databases to create a cohort of
LTCH residents followed for up to six months and counted all ED visits. Data obtained from the provincially mandated National Ambulatory Care
Reporting System (NACRS) were used to describe each resident’s first observed visit. Two visit types were defined: 1) potentially preventable condition
(PPC) visits were those with an ambulatory care sensitive condition listed as the main reason; 2) low acuity visits were those triaged as non-urgent and ended with the resident returning to the LTCH.
Population Studied: All individuals over 65 years reported in an Ontario LTCH on the day of the 2005 provincial facility census. Residents were followed until the first of hospitalization, death, or end of follow-up. The sample consisted of 64,589 residents.
Principal Findings: A total of 21,773 ED visits were counted for a visit rate of 77/100 residentyears; 23.1% of residents visited the ED at least once during follow-up. Before transfer, 12% of residents saw a physician in the LTCH and 86.4% had ambulance transport. The majority of visits included at least one procedure (83.8%). Of 14,884 first-time visits, 24.6% were for a PPC and 11% were low acuity. The most common reasons for
PPC visits were pneumonia (30.1%), kidney problems/urinary tract infections (20.3%), and congestive heart failure (15.9%). PPC visits had a higher frequency of emergent triage (35.3% vs.
25.2%), hospitalization (62.4% vs. 44.2%), and death within 30 days (23.6% vs. 16.9%) than did other visits. Of residents who returned to the LTCH,
81.2% re-visited the ED during follow-up. For low acuity visits, the most common reason was injury
(37% fall-related, 10.4% non-fall-related).
Compared to other visits, low acuity visits were shorter (mean (SD):4.5 hours (4.0) vs. 6.6 hours
(5.6)) and had fewer procedures (72.8% vs. 84.0%).
Five percent of residents with low acuity visits died within the 30 days and 18.1% re-visited the ED.
Fall-related injuries were the most common reason for all other visits (22.7%). No temporal patterns were observed.
Conclusion: LTCH residents made frequent ED visits that tended to be resource-intensive in terms of ambulance transport, time in ED, and volume of procedures. Residents with PPC visits had the highest frequency of urgent triage, hospitalization, repeat visits, and death. Injuries, mostly attributed to falls, were the primary reason for both low acuity and other visits.
Implications for Policy, Delivery or Practice:
Strategies for reducing LTCH to ED transfers are crucial to improving the quality of care provided to vulnerable residents and alleviating burden on overcrowded EDs. Our findings highlight specific areas where interventions may be successful. The severity of visits for PPCs suggests that implementation of early management strategies, such as clinical pathways for pneumonia, may help to reduce exacerbation and ED transfer. The prevalence of fall-related injuries illustrates the need for increased vigilance around fall prevention; that injuries accounted for over one-third of low acuity visits further suggests that access to assessment methods that can be administered by
LTCH staff may reduce non-urgent visits.
Funding Source(s): Canadian Institutes of Health
Research
● Health Service Use, Access & Expenditures among Working Aged Adults with Chronic
Health Care Needs: Bridging Chronic
Conditions & Disabilities
Stephen Gulley, Ph.D., M.S.W.; Elizabeth Rasch,
P.T., Ph.D.; Leighton Chan, M.D., M.P.H.
Presented by: Stephen Gulley, Ph.D., M.S.W.,
Research Associate, Rehabilitation Medicine
Department, National Institutes of Health, Mark O.
Hatfield Clinical Research Center, Building 10,
CRC, Room 1-1469, Bethesda, MD 20892-1604,
Phone: (301) 594-3090; Email: gulley@brandeis.edu
Research Objective: Adults with chronic conditions and those with disabilities are overlapping groups with shared healthcare concerns and are major consumers of healthcare resources. Yet, rarely have these two groups been considered together in
terms of their ongoing needs for healthcare services. While population-level measurement based on elevated service needs over time has been successfully used to bridge these groups among children with special healthcare needs, such an approach has not been implemented among the working aged. Our objective is to provide a preliminary profile of Adults with Chronic Health
Care Needs (ACHCN).
Study Design: We pooled data from the 2002-2004
MEPS. We identified ACHCN on the basis of having one or more health / mental health conditions expected to last at least 12 months and known to require ongoing medical intervention and/or result in some degree of disability. We segmented ACHCN on the basis of reported limitations resulting in three subgroups: A) ACHCN without limitations; B) ACHCN with limitations but not requiring help or supervision with ADL/IADLs;
C) ACHCN with ADL/IADL limitations. For these groups, we used covariate-controlled, two-part models to estimate use of ambulatory and other key services. Using logistic regression models, we estimated the percentage reporting access problems. Finally, we estimated total medical expenditures by group. All analyses were adjusted for the complex survey design.
Population Studied: We studied a nationally representative sample (N=58,408) of the U.S. civilian, community-dwelling population aged 18-64.
Principal Findings: ACHCN comprised 52% of the working aged and reported an average of 2.1 chronic conditions and 2.6 acute conditions during the year. Relative to adults without chronic conditions, ACHCN had significantly (p<.05) higher service use and access problems. Total U.S. health care expenditures for ACHCN were $423 billion compared to $91 billion among other adults.
Of the two disability groups, those reporting need of help or supervision with ADLs/IADLs were the smallest (7% of ACHCN) but had the highest rates of: mean chronic (3.5) and acute (4.4) conditions during the year, service use (all services) and access problems (21% reporting delay or nonreceipt of needed medical care and 20% reporting delay /non-receipt of prescription medications).
Despite being just 3% of the working aged overall,
ACHCN with ADL/IADL limitations totaled $100 billion in medical expenditures, more than was spent on the entire working age population without chronic health care needs.
Conclusion: These results document not only the extent to which disability is associated with an accumulation of chronic and acute health conditions, but the extent to which it is also associated with high service use, access problems, and high costs among persons with one or more chronic conditions.
Implications for Policy, Delivery or Practice: As chronic disease management programs and coordinated care models proliferate, we need carefully articulated definitions and measures to track changes in outcomes among adults with elevated service needs which are expected to persist over time. New approaches are needed to bridge the divide between disability and chronic health conditions among the working aged.
Funding Source(s): NIH Intramural Research
Program
Theme: Prevention and Treatment of Chronic
Illness
● Developing an Analytical Tool for Assessing the Adequacy of State Health Information
Exchange Laws
Margaret Gunter, Ph.D.; Randy McDonald, J.D;
Shelley Carter, B.S.N., M.P.H., M.C.R.P.; Bob
Mayer
Presented by: Margaret Gunter, Ph.D., President &
Executive Director, Lovelace Clinic Foundation,
2309 Renard Place, Southeast, Suite 103,
Albuquerque, NM 87106, Phone: (505) 938-9900;
Email: maggie@lcfresearch.org
Research Objective: To develop and test an analytic legislative tool that provides states with the ability to analyze and propose reform to laws related to the exchange of electronic health information.
Study Design: In examining health information exchange in laws and business practices, the multistate Harmonizing Security and Privacy Law
Collaborative (HSPLC) found numerous barriers.
Barriers can be problematic when they prevent the timely exchange of health information needed for the treatment of patients. There are many inconsistencies in state and federal laws and among state statutes in their definitions, organizational structure, and content. To address these barriers and inconsistencies, HSPLC developed a set of analytical tools and a narrative guide, the Roadmap, to assist states in implementing an effective legal framework for the review and adoption of legislation that supports health information exchange (HIE) and reduces barriers. The tools and Roadmap were created through extensive research to identify best practices for identifying, evaluating, and reforming state laws related to the disclosure of electronic health information.
Population Studied: Some states have adopted new legislation that addresses the exchange of health information that may further exacerbate differences among states and impede interstate exchange of electronic health information. The seven HSPLC states identified, reviewed, and analyzed their own state resources and laws impacting the HIE utilizing the tools developed in the Roadmap. Each of the seven states identified relevant stakeholder groups to review and pilot test the tools and provider feedback on revisions need and usefulness of the tool.
Principal Findings: HSPLC found that various state resources (legal, legislative, healthcare policy, healthcare providers, and consumers) are necessary for successful completion of the
Roadmap to identify opportunities for legislative reform. The seven participating HSPLC states believe that states will have greater likelihood of success in achieving legislative reform if they use the guidelines developed in the Roadmap. For purposes of dialogue and collaboration within and among states, the Roadmap offers states the structure to work with others who are contemplating a change in their HIE legislation. Interstate collaboration and coordination are essential if we are to achieve a national legal and technical infrastructure that facilitates health information exchange.
Conclusion: Legislation in most states does not adequately address the exchange of electronic health information. Drafting of legislation must take into account a state’s unique environment and culture, and the needs and support of stakeholders.
The goal of using the Roadmap and analytic tool is to protect health information while removing barriers that impede the exchange of vital information. The
HSPLC Roadmap provides a step by step process to analyze and reform state legislation.
Implications for Policy, Delivery or Practice: It is difficult to comprehensively address and improve laws and business practices that negatively impact the health information exchange. An essential element for success is to provide tools for stakeholders, including legislators, to identify benefits and barriers and to become fully engaged in the legislative reform process.
Funding Source(s): Office of National Coordinator for Health Information Technology
Theme: Health Information Technology
● The Long-Term Effects of the Chinese Health
Reform in 1978
Jing Guo
Presented by: Jing Guo, Health Studies, University of Chicago, 5118 South Dorchester Avenue,
Apartment 406, Chicago, IL 60615, Phone: (617)
319-3469; Email: guoj@uchicago.edu
Research Objective: examines whether there are long-term effects of the Chinese health system reform in 1978 on adults’ health and socioeconomic status, and tries to quantify the effects
Study Design: Controlling for time trend, gender, and average provincial income of their birth year, this study quantifies the long-term effects of this reform on adult health, education, and socioeconomic status by constructing a differencein-difference estimator.
Population Studied: Chinese urban and rural residents. Using individual-level data -- the China
Health and Nutrition Surveys data -- which cover nine provinces—Heilongjiang, Liaoning, Guangxi,
Guizhou, Henan, Hunan, Jiangsu, and Shandong, and vary substantially in geography, economic development, public resources, and health indicators.
Principal Findings: this study finds out that adults who were exposed to the reform in their early life have significantly shorter height and lower education attainment. In total sample, reformaffected birth cohort lost nearly 1.61 cm in attained height, and have 0.61-year shorter education year than the counterfactual case that the reform had not occurred. Women in particular have experienced more severe negative effects from the reform than men on both health status and education achievement. The true effects of the reform on females are nearly 2.56 cm shorter in height and
0.61 schooling year less than the counterfactual cohorts who were born before the reform.
Conclusion: the findings suggest more attention need to be paid to healthcare reform as a potential mechanism through which the transmission of both health and socioeconomic status takes place, especially for the poor and women, who are more vulnerable in terms of access to health care.
Implications for Policy, Delivery or Practice:
Since China is currently in a process of a new health system reform, this study may draw more attention from policy makers on the potential longterm effects of the current reform, and prevent the uncomfortable inequity to some extent.
● Food Allergy Knowledge, Attitudes, & Beliefs of Primary Care Physicians
Ruchi Gupta, M.D., M.P.H.; Jennifer Kim, M.D.;
Elizabeth Springston, B.A.; Bridget Smith, Ph.D.;
Jacqueline Pongracic, M.D.; Jane Holl, M.D.,
M.P.H.
Presented by: Ruchi Gupta, M.D., M.P.H.,
Assistant Professor, Smith Child Research
Program, Children's Memorial Hospital, 2300
Children's Plaza, P.O. Box 157, Chicago, IL 60614,
Phone: (312) 573-7747; Email: rugupta@childrensmemorial.org
Research Objective: Food allergy is a growing medical concern in the United States, compounded by limited treatment options, the absence of a cure, and the ubiquity of food in our society. Primary care physicians play a pivotal role in the health and wellbeing of affected children as they are often the first and sometimes the only clinicians to diagnose and manage childhood food. Our objective was to to provide insight into food allergy knowledge and perceptions among pediatricians and family physicians in the US.
Study Design: A national sample of pediatricians and family physicians was recruited between April and July of 2008 to complete the validated, webbased Chicago Food Allergy Research Survey for
Primary Care Physicians. Findings were analyzed to provide composite/itemized knowledge scores,
describe attitudes and beliefs, and examine the effects of participant characteristics on participant response.
Population Studied: Our sample was drawn from a population of pediatricians and family physicians across the United States.
Principal Findings: A sample of 407 respondents was obtained. Participants answered 59% of knowledge-based items correctly. Strengths were identified in areas related to triggers/environmental risks, susceptibility/prevalence, and treatment/utilization of healthcare. For example,
80% of physicians knew that the flu vaccine is unsafe for egg-allergic children, 90% recognized that the number of food-allergic children is increasing in the US, and 80% were aware that there is no cure for food allergy. Weaknesses were identified in areas related to definition/diagnosis, symptoms/severity, and triggers/environmental risks. For example, 24% knew that oral food challenges may be used in the diagnosis of food allergy, 12% correctly rejected chronic nasal problems as a symptom of food allergy, and 23% recognized that yogurts/cheeses from milk are unsafe for children with IgE-mediated milk allergies.
Fewer than 30% of participants felt comfortable interpreting lab tests to diagnose food allergy or felt adequately prepared by their medical training to care for food-allergic children.
Conclusion: Knowledge of food allergy among primary care physicians was fair, with little variation between pediatricians and family physicians.
Opportunities for improvement exist as acknowledged by participants’ own perceptions of their clinical abilities in the management of food allergy.
Implications for Policy, Delivery or Practice:
Efforts to increase primary care physicians' knowledge of food allergy in the areas of opportunity identified by this study may improve the quality of care received by food-allergic children and their families.
Funding Source(s): The Food Allergy Initiative
● To Err is Human: Effect of Medical Errors on
Physicians
Gouri, Gupte, M.B.A.,M.H.A.; Warren Martin, Ph.D.
Presented by: Gouri Gupte, M.B.A., M.H.A., Ph.D.
Student, School of Health Professions, University of
Alabama at Birmingham, 1629, 14th Avenue South,
Apartment H, Birmingham, AL 35205, Phone: (205)
789-9197; Email: guptegouri@hotmail.com
Research Objective: To Err is Human: Building a
Safer Health System a book by Institute of Medicine
(2001) struck a right cord by providing a comprehensive strategy to address system errors and not human errors. Nonetheless, no systematic strategy has been developed to assess and deal with the effect of medical errors on physicians. We hence conducted a comprehensive assessment of literature on medical errors and physicians. The assessment was conducted to enable us to design a conceptual framework for better analysis of effects of medical errors on physicians.
Study Design: To assess the literature on effect of medical errors on physicians and nurses, we reviewed the literature. The authors systematically searched the English language literature indexed in
MEDLINE (1995 to January 2008), Pub Med,
ABI/Inform, EBSCO and JSTOR. Keyword, title and abstract information were used. The main search terms were ‘medical errors’, ‘physicians’, ‘doctors’,’ malpractice’, ‘emotions’, ‘stress’. In addition, the lists of references from review papers, book chapters, and articles included in this study and other relevant articles were systematically reviewed. Only English articles were considered; unpublished studies, abstracts, dissertations, theses, book chapters and studies published in non peer-reviewed journals were not.
Principal Findings: The analysis of the literature and theories helped design a conceptual model which would assist in understanding the effects of medical errors on physicians that will help develop a strategic approach to this issue.
Conclusion: To err is human does not seem to be clearly followed in the case of physicians who face adversity after committing medical errors. More research needs to be conducted on effect of medical errors on physicians. The conceptual model designed in this paper could be an important step towards understanding the effect of errors on physicians.
Implications for Policy, Delivery or Practice:
This review will help researchers set priorities for future research on understanding the effect of medical errors on physicians.
Theme: Health Care Workforce
● The Effects of Cost Sharing on Access to Care among Childless Adults
Gery Guy, M.P.H.
Presented by: Gery Guy, M.P.H., Ph.D. Candidate,
Health Policy & Management, Rollins School of
Public Health of Emory University, 1518 Clifton
Road, Northeast, Room 623, Atlanta, GA 30322,
Phone: (404) 712-9603; Email: gguy@sph.emory.edu
Research Objective: Amid high rates of uninsurance among childless adults and new opportunities to expand insurance coverage, many states have taken the lead with expansions in an attempt to increase access to care among low income childless adults. Given the flexibility states have in increasing cost sharing, a number of adult programs provide a leaner benefit package and require significantly more cost sharing than the state’s Medicaid program. The objective of this study is to analyze the impact of these insurance expansions and the use of increased enrollee cost
sharing on receipt of clinically indicated preventive screenings and physician services.
Study Design: This study uses data from the
Behavioral Risk Factor Surveillance System
(BRFSS) from 1997-2007. The BRFSS is well suited for this analysis as it provides detailed information on characteristics likely to affect access to care and utilization (e.g., age, education, employment, health status). The analysis considers access to three recommended preventive screenings: mammography, Pap tests, and cholesterol screenings. Additionally, having a personal doctor or health care provider, and whether adults report costs being a barrier to receiving needed medical care are examined. For the analysis two groups of programs are examined: those with low cost sharing requirements (typical
Medicaid levels) and those with high cost sharing requirements (higher than Medicaid levels). Using a multivariate difference-in-difference regression model the effects of childless adults eligible for insurance expansions are compared with control groups of childless adults not eligible for the health insurance expansions.
Population Studied: The study population includes all childless adults age 19-64 eligible for a public health insurance expansion. The focus is on childless adults due to their high uninsurance rates, historical exclusion from Medicaid eligibility, and the recent insurance expansion efforts at the state level.
Principal Findings: Childless adult expansions regardless of cost sharing levels decreased the probability that cost barriers prevent adults from seeking needed medical care by approximately 3%.
Childless adults eligible for programs with low cost sharing had an increased probability of receiving a recommended mammogram by 4.3%, a pap test by
3.2% and a cholesterol screening by 3.9%; while those eligible for high cost sharing programs did not experience an increased likelihood of receiving any of the three preventive health screenings.
Conclusion: Results indicate that cost sharing requirements do not have an impact on an individual’s ability to obtain medical care when needed. However, results indicate that the level of cost sharing plays an important role in providing access to preventive health services. Eligibility for programs with low cost sharing requirements is associated with an increased likelihood of preventive screening utilization, while eligibility for expansions with higher cost sharing requirements does not result in an increased likelihood of screenings.
Implications for Policy, Delivery or Practice:
Expanding public health insurance coverage to low income childless adults presents a promising policy opportunity. If the goal of public health insurance expansions is to increase access to care when needed, both types of programs are accomplishing that mission. However, if the goal is to increase the use of preventive services, lower cost sharing requirements will help achieve that goal.
Theme: Coverage and Access
● Racial & Ethnic Disparities in Cancer
Treatment: Do They Differ by Minority
Population?
Susan Haber, Sc.D.; Janet Mitchell, Ph.D.; Michael
Halpern, M.D., Ph.D., M.P.H.; Sonja Hoover,
M.P.P.; Kristin Roper, R.N.
Presented by: Susan Haber, Sc.D., Senior
Economist, Division for Health Services & Social
Policy Research, RTI International, 1440 Main
Street, Suite 310, Waltham, MA 02451, Phone:
(781) 434-1721; Email: shaber@rti.org
Research Objective: Previous studies of racial disparities in cancer treatment focused mostly on differences between blacks and whites. This study expands on previous research to examine racial/ethnic disparities in cancer treatment between white, black, Hispanic, and Asian Medicare beneficiaries. We used linked SEER-Medicare data to examine: (1) the likelihood of receiving treatment that complies with stage-specific guidelines; (2) whether the likelihood differs among racial/ethnic groups; and (3) whether treatment disparities differ among cancer sites.
Study Design: We used National Comprehensive
Cancer Network guidelines in effect during the study period to define stage-specific treatment guidelines for breast, colorectal, lung, prostate, and cervical cancer. Based on services identified in
Medicare claims data, we created indicators for whether an individual received treatment in the year following diagnosis that complied with each applicable guideline given their cancer stage and other clinical characteristics and with all clinicallyrelevant guidelines. Logistic regression was used to identify differences among racial/ethnic groups in the likelihood of receiving guideline-compliant treatment controlling for cancer site, cancer stage, and other beneficiary characteristics (age, gender, marital status, dual eligibility for Medicaid, original reason for Medicare entitlement, and residence in a metropolitan area).
Population Studied: Medicare beneficiaries in the
SEER cancer registries who were diagnosed with any of five the study cancers from 2000-2002 and who were eligible for Medicare Parts A and B and enrolled in fee-for-service Medicare for the entire year following diagnosis or until death (N=104,418).
Principal Findings: Rates of compliance with cancer treatment guidelines are low for all racial/ethnic groups for many of the cancers studied. Pooling across all cancer sites, only 54% of
Medicare beneficiaries received treatment that complied with all clinically-relevant guidelines.
Compliance rates varied substantially by cancer site, from 17% and 38% for colorectal and lung cancer, respectively, to 74% and 93% for cervical
and prostate cancer, respectively. Blacks were significantly less likely than whites to receive guideline-compliant treatment for breast cancer, lung cancer, and prostate cancer. Hispanics were significantly more likely than whites to receive guideline-compliant treatment for lung cancer, but less likely for prostate cancer. There were no significant differences between Asians and whites.
Lower rates of receiving guideline-compliant treatment among blacks are generally due to lower receipt of ambulatory services (radiation therapy, chemotherapy, and follow-up surveillance) rather than surgical treatments
Conclusion: Variation in compliance by cancer site partly reflects differences among cancers in the number and type of guidelines. Low compliance with diagnostic imaging guidelines, as well as lung cancer treatment, may reflect physician ambivalence about the value of these procedures.
High compliance rates for Hispanics as a whole may mask below average rates for certain subgroups, based on country of origin and acculturation. More research is needed to identify whether such variation exists within the Hispanic population.
Implications for Policy, Delivery or Practice: The observed differences in disparities among racial/ethnic minorities support the importance of tailoring interventions to specific racial/ethnic populations. The Cancer Prevention and Treatment
Demonstration sponsored by the Centers for
Medicare & Medicaid services is testing the effectiveness of programs to increase use of cancer treatment and screening services designed for specified racial/ethnic minorities.
Funding Source(s): CMS
Theme: Disparities
● Furthering Development of a Practice Change
Intervention Through Incorporation of
Implementation Theory
Hildi Hagedorn, Ph.D.; Nancy Rettmann, M.S.; Eric
Dieperink, M.D.; Janet Durfee, N.P.; Judith Garrard,
Ph.D.; Astrid Knott, Ph.D.
Presented by: Hildi Hagedorn, Ph.D.,
Implementation Research Coordinator, VA
Substance Use Disorders Quality Enhancement
Research Initiativ, Minneapolis VA Medical Center,
One Veterans Drive (116A9), Minneapolis, MN
55417, Phone: (612) 467-3875; Email: hildi.hagedorn@va.gov
Research Objective: The Veterans Health
Administration's Hepatitis C Resource Centers
(HCRC) are tasked with developing best practices in hepatitis C care and disseminating these practices to providers. Based on the literature on the education of medical providers and the dissemination of health care innovations, the
Minneapolis HCRC developed a 6-month preceptorship and has been refining this program since 2002. Recently, the HCRC has joined with the
VA Substance Use Disorders Quality Enhancement
Research Initiative to modify the preceptorship to promote implementation of hepatitis screening, education, vaccination, and treatment referral services in VA substance use disorders (SUD) treatment clinics (The Liver Health Preceptorship).
As part of the evaluation of the Liver Health
Preceptorship, the Promoting Action on Research
Implementation in Health Services (PARIHS) implementation framework was applied to the program to identify strengths and weaknesses and guide refinement of the program.
Study Design: Preceptorship teams include both
SUD clinical staff and gastroenterology staff. The teams complete a baseline needs assessment to determine current clinic practices and gaps between current practice and recommended practices. This is followed by a two-day workshop which provides education on recommended hepatitis services and implementation strategies.
Teams develop an action plan with implementation goals and action steps. Teams receive coaching calls during the six months following the workshop to assess progress and assist in problem solving.
The definitions of high and low ratings on each element and sub-element of the PARIHS framework were reviewed and compared to the design of the preceptorship to determine strengths and weaknesses of the program and identify areas for refinement.
Population Studied: Of 25 teams that applied from
VA Medical Centers across the country, eleven teams were randomly selected to attend. Five teams that were not selected agreed to complete pre- and post-assessment surveys.
Principal Findings: Nine of eleven preceptorship clinics reported implementation of two or more new hepatitis services at six-month follow-up. None of the comparison teams reported implementation of more than one new service. Lessons learned from the application of the PARIHS framework to the program evaluation include: 1) In designing program recommendations, research evidence has been highly valued to the relative neglect of evidence from clinicians' and patients' experiences;
2) Context variables from attending organizations have not been measured in a meaningful way and their measurement would inform interpretation of results, particularly variable implementation among attending teams; and 3) A task oriented style of facilitation appears appropriate for this intervention but the effectiveness of this style may be dependent on an organizational context supportive of innovation.
Conclusion: The Liver Health Preceptorship appears to be effective at promoting implementation of new practices in most clinics that volunteer to attend. Application of an implementation model to assess the program's strengths and weaknesses led to identification of ways to enhance program evaluation and further refine the program in
anticipation of testing the program with nonvolunteer clinics.
Implications for Policy, Delivery or Practice:
This work highlights a potentially promising model for promoting implementation of new clinical practices and highlights the benefits of application of implementation theory to the design of such interventions.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Effectiveness of a Contingency Management
Intervention for Veterans with Substance Use
Disorders
Hildi Hagedorn, Ph.D.; Siamak Noorbaloochi,
Ph.D.; Carl Rimmele, Ph.D.; Daniel Kivlahan, Ph.D.
Presented by: Hildi Hagedorn, Ph.D.,
Implementation Research Coordinator, VA
Substance Use Disorders Quality Enhancement
Research Initiative, Minneapolis VA Medical Center,
One Veterans Drive (116A9), Minneapolis, MN
55417, Phone: (612) 467-3875; Email: hildi.hagedorn@va.gov
Research Objective: Evaluate the effectiveness of adding a contingency management intervention
(CM) to outpatient usual care (UC) on outcomes of veterans entering substance use disorders treatment at two VA clinics.
Study Design: Study participants were randomly assigned to receive either UC or the CM intervention plus UC. All participants consented to breath alcohol and rapid result urine drug screening twice per week for eight weeks. CM participants earned chances to draw for coupons which could be used in the VA cafeteria or gift shop each time they submitted negative screens. Primary outcomes were percent of possible drug free screens submitted during the intervention period, consecutive weeks of treatment attendance, and self-reported percent days abstinent using the Time
Line Follow-Back (TLFB) method at two-month follow-up.
Population Studied: Participants were veterans presenting for treatment of alcohol or stimulant dependence at the Minneapolis or the Seattle VA
Medical Center addictive disorders treatment clinics. To date, 311 participants have completed the intervention and the two-month follow-up.
Participants were predominantly male (98%),
Caucasian (52.5%) or African American (35.2%), with a mean age of 50.
Principal Findings: At study intake, participants reported a mean of 38% abstinent days during the previous 30 days. During the intervention, test results rarely showed recent substance use (4.5% and 2.5% for CM and UC respectively, ns). A generalized linear mixed model demonstrated a significant difference (p<.0002) in pattern of intervention appointment attendance and test results over the 16 visits which was accounted for by higher appointment attendance (11.6 vs. 9.8, p=.001) and more drug-free tests (11.2 vs. 9.6, p=.003) in the CM group compared to the UC group. CM participants earned M=$99 during the intervention. The groups did not differ on consecutive weeks of UC treatment attendance (6.9 for CM and 7.0 for UC). Two-month follow-up was available on 79% of CM and 78% of UC participants. TLFB self-report data rarely disagreed with intervention test results; only five participants self-reported abstinence on the TLFB that was discrepant with intervention tests showing use.
Among participants followed, percent days abstinent at follow-up (95%) was significantly higher than at baseline (38%), but not significantly different for CM (97%) vs. UC (94%) participants. Of the
23% of participants reporting any use on the TLFB at 2-month follow-up, CM participants reported significantly greater percent days abstinent (87%) than UC participants (72%; p=.03).
Conclusion: CM participants had significantly more alcohol and drug free tests during the intervention phase but did not differ from UC participants on consecutive weeks of treatment attendance or percent days abstinent at follow-up, with highly favorable results in both conditions. Of the 23% of patients that reported use at 2-month follow-up, CM participants reported fewer days of use than UC patients.
Implications for Policy, Delivery or Practice: The incentives used in the intervention were low cost
($99 per patient) and reduced the frequency and/or duration of use for the minority of patients that experienced a relapse. Extended follow-up will clarify durability of effects. Subgroup differences by primary drug will also be examined.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Access to Adequate Outpatient Depression
Care for Mothers in the U.S.: A Nationally
Representative Population-Based Study
Erika Hagen, Ph.D., M.S.; Whitney, Witt, Ph.D.,
M.P.H.; Abiola Keller, M.P.A.S., P.A.C.; Carissa
Gottlieb, M.S.; Kristin Litzleman, B.A.; John
Hampton, M.S.
Presented by: Erika Hagen, Ph.D., M.S.,
Postdoctoral Scholar, Population Health Sciences,
University of Wisconsin, Madison, 610 Walnut
Street, WARF 634, Madison, WI 53726, Phone:
(608) 265-3296; Email: ewarkentien@wisc.edu
Research Objective: Maternal depression has important consequences for women’s own health, and may also have a profound negative impact on the long-term health, quality of life, and well-being of their children. Treating maternal depression can lead to remission and reduce children’s mental health symptoms and diagnoses. Therefore, access to treatment for maternal depression is critical for improving maternal mental health and
reducing the impact of depression on the family.
The objective of this study was to determine the factors associated with access to treatment for depression among a population-based sample of depressed mothers in the United States.
Study Design: The adequacy of treatment for depression (none, some, and adequate) was examined by factors related to access to care. Chisquared analyses were used to test for differences in sociodemographic and health characteristics by maternal depression treatment status. Depressed mothers who received some treatment, and depressed mothers who received adequate treatment, were compared to mothers who did not receive treatment with a multinomial logistic regression model.
Population Studied: We examined data on 2,130
(1,882,735 weighted) mothers reporting depression from ten panels from the 1996-2005 Medical
Expenditure Panel Survey, a nationally representative sample of the civilian noninstitutionalized population in the United States.
Principal Findings: Overall, 9.5% of US mothers reported depression. Of depressed mothers, 37.9% did not receive any treatment, 27.3% received some treatment, and 34.8% received adequate treatment for depression over the course of the year. Multivariate analyses revealed that nonworking mothers and mothers with health insurance coverage (either public or private) were more likely to receive both some and adequate treatment than their working and uninsured counterparts, respectively. Black, Hispanic and other minority mothers were less likely to receive adequate treatment for depression than their White counterparts. Mothers with less education and mothers in the Western region of the US were also less likely to receive adequate treatment.
Conclusion: Most depressed mothers do not receive adequate treatment for depression in the
US, underscoring the importance of examining adequate versus any care. This study demonstrates that insurance, employment, racial, ethnic, educational, and geographic disparities exist among mothers for the receipt of adequate treatment for depression. Understanding the barriers to accessing treatment is critical to designing effective interventions, reducing inequities, and ultimately improving the long-term mental health and health of mothers and their families. Future research studies are needed to explore barriers to treatment initiation and adequate treatment among mothers with depression.
Implications for Policy, Delivery or Practice:
Ensuring that depressed mothers have insurance coverage for mental health services is important for obtaining adequate depression care. Additionally, employers may want to expand Employee
Assistance Programs to help with both the identification and treatment of maternal depression.
Mothers and their providers may need to be better informed about symptoms of depression, treatment regimens, and long-term consequences of undertreating or not treating maternal depression.
Funding Source(s): AHRQ, Ambulatory Pediatrics
Association, NIH
Theme: Behavioral Health
● Performance Accountability: A Comparative
Review of Healthcare & Education Programs
David Haggstrom, M.D., M.A.S.; Todd Ellwein,
M.I.A.
Presented by: David Haggstrom, M.D., M.A.S.,
Assistant Professor of Medicine, VA Center on
Implementing Evidence-Based Practice,
Indianapolis VA Health Services Research &
Development Service, 1481 West 10th Street
(11H), Indianapolis, IN 46202, Phone: (317) 988-
2067; Email: dahaggst@iupui.edu
Research Objective: To draw upon the experience of performance accountability in U.S. education to inform the U.S. health care sector
Study Design: Descriptive and comparative review of key health care and education performance accountability programs, searching the academic, policy, and professional literature
Population Studied: Performance accountability programs, with a focus upon federal programs in health care (Medicare) and education (No Child Left
Behind)
Principal Findings: Quality reform efforts gained momentum from major reports in both education
(Nation at Risk, National Commission on
Excellence in Education, 1983) and health care
(Crossing the Quality Chasm, Institute of Medicine,
2001). Teachers’ salary structure has traditionally compensated advanced education and seniority. In health care, quality assessment initially involved accreditation and licensure of health care providers; such assessments provided access to medical practice, but not necessarily increased reimbursement. “No Child Left Behind” mandates schools report locally developed outcome measures to receive federal funding and requires schools report performance among major racial and ethnic subgroups. Medicare provides incentives for hospitals to voluntarily report nationally defined outcomes, processes of care, and patient experiences; public quality reporting of racial/ethnic subgroups has been discussed, but not implemented in Medicare. “Merit pay” programs for teachers date back several decades, and local programs have recently been developed in Denver and Washington DC. The evidence of the benefit of merit pay on student performance is mixed. In the past decade, a wide range of health plans, employers, and government programs, including
Medicare, have sought to link health care spending to quality through pay-for-performance activities.
Medicare’s Physician Group Practice
Demonstration, involving ten large multi-specialty group practices, has provided financial incentives
for practices to improve the care of patients with chronic, high-cost conditions. Performance accountability efforts in both education and health care have faced resistance from teachers and physicians by creating tension between internal institutional improvement efforts and external accountability. Objections are grounded in issues of professional autonomy, as well as criticisms of the validity of performance measures.
Conclusion: Both health care and educational performance measurement programs have faced significant limitations, including a constrained set of measures. Standardized tests of student performance are unable to measure many important cognitive dimensions, and hospital performance measures in Medicare address a narrow range of medical conditions.
Misclassification of “low” and “high” performers based upon small numbers of students and patients is another common problem. Finally, instances of
“gaming” accountability systems have been observed, including the diversion of school resources from students unlikely to increase aggregate performance and physician avoidance of high-risk patients.
Implications for Policy, Delivery or Practice:
Comparisons between U.S. health care and educational programs in performance accountability may provide new policy insights by considering similar issues in different contexts. Federal education policy has more explicitly weighed racial/ethnic disparities in their approaches to financial reward. Stronger punitive measures for disenrollment may be necessary to avoid the gaming that emerges in accountability systems.
The significant limitations of quantitative measurement appear in both health care and education and may be addressed by providing complementary qualitative information in balanced reports. Even though accountability is a universal phenomenon, the means of enacting accountability are not.
Funding Source(s): VA
● Ownership of Electronic Medical Information
Mark Hall, J.D.
Presented by: Mark Hall, J.D., Professor, Social
Science & Health Policy, Wake Forest University,
Email: mhall@wfubmc.edu
Research Objective: To provide greater clarity about who owns medical information, especially in electronic form, and to develop property law principles that facilitate the development of more useful electronic medical record networks.
Study Design: In-depth review of legal sources and public policy literature. Legal analysis and economic policy analysis.
Principal Findings: Medical information ownership was never much in doubt in an age of paper-based records, but now that information content can be easily digitized and freed from any particular storage medium, confusion reigns, creating barriers to investment in medical information networks and systems. The respective property rights of patients, providers and insurers will strongly influence, if not determine, what form of electronic health informatics ends up predominating. Providers own the tangible records they create, but these legal rights can be transferred only for very limited purposes, and they do not include ownership of the information content per se. Patients have access and privacy rights to their own medical information, but those rights do not amount to ownership. There are competing claims of owernship and obligations to protect privacy that overlap and conflict, creating confusion and uncertainty.
Conclusion: Patients can and should be allowed to monetize their rights to access and control rights by assigning them to a trusted intermediary who may then place these rights in a stream of commerce that determines their value and best use. The funds generated can then be distributed both to patients and providers to encourage their adoption and use of interconnected electronic records.
Implications for Policy, Delivery or Practice:
How ownerhips is resolved can have huge impacts on how or whether massive anticipated developments in electronic and personal health records will take shape. Whether rights to access and use medical information can be commercialized may determine whether effective, comprehensive medical information networks can emerge at all, absent overt government mandate.
Theme: Health Information Technology
● Underinsurance in a State High Risk Pool & a
Possible Alternative for Coverage
Jean Hall, Ph.D.; Janice Moore, M.A., M.S.W.,
M.B.A.
Presented by: Jean Hall, Ph.D., Assistant
Research Professor, CRL - Division of Adult
Studies, University of Kansas, 1122 West Campus
Road, Room 517, Lawrence, KS 66045-3101,
Phone: (785) 864-7083; Email: jhall@ku.edu
Research Objective: State high risk insurance pools are the coverage of last resort for individuals who do not have access to group plans and are medically uninsurable in the individual market.
However, most pools require high out-of-pocket spending, which may discourage appropriate management of serious conditions and lead to preventable disability. We sought to discover whether increasing enrollees’ access to health care could prevent or forestall their progression to full
Social Security disability.
Study Design: The study, funded by the Centers for Medicare and Medicaid Services and the State of Kansas, is longitudinal over a four-year period, with random assignment of participants to intervention and control groups. Intervention group
members receive significantly enhanced, Medicaidlike coverage, and reduced out-of-pocket costs.
Outcomes are tracked through surveys, analyses of claims data, and focus group discussions. The primary outcome variables of interest include: health status, employment and earnings, health care utilization, and disability status.
Population Studied: The study population of approximately 400 people is selected from participants in the Kansas high risk health insurance pool who are aged 18 to 60, employed for at least 40 hours per month at baseline, and have a potentially disabling condition based on
Social Security disability criteria.
Principal Findings: Study participants experience a wide range of serious physical and mental health conditions. Their average predicted health care costs and disease burden are four times that of the general population, based on Adjusted Clinical
Group (ACG) case mix scores. Fifty-five percent report difficulties with at least one activity or instrumental activity of daily living. Ninety-four percent meet common criteria for being underinsured and 26% have medical debt. Most report delaying or forgoing care due to costs. Risk pool members historically have transitioned to
Social Security disability at a rate 8 times that of the general population. With the intervention of
Medicaid-like coverage and greatly reduced out-ofpocket costs, appropriate use of medical services such as diagnostic testing and prescription drugs is increasing.
Implications for Policy, Delivery or Practice:
Preliminary findings support the hypothesis that this intervention is slowing the rate of disability and helping participants maintain employment.
Currently, 34 states operate high risk pools.
Twenty-eight states with risk pools also operate, or will soon implement, Medicaid Buy-In progams, which allow people with disabilities to work and purchase Medicaid coverage. With federal cooperation, states could expand Medicaid Buy-In eligibility guidelines reach many members of the high risk pool population. Medicaid coverage would likely be more comprehensive and less costly to beneficiaries—and it would have the potential to offset greater federal and state costs in the long term. Viewed from the standpoint of overall costs to society, the potential costs of lost productivity and greater long-term health care needs for this population far outweigh the short-term costs of better prevention and upfront investment in care.
Funding Source(s): CMS, State of Kansas
Theme: Coverage and Access
● Government-Sponsored Reinsurance:
Purposes & Prospects
Mark Hall, J.D.
Presented by: Mark Hall, J.D., Professor, Social
Science & Health Policy, Wake Forest University,
Email: mhall@wfubmc.edu
Research Objective: State and federal officials are considering proposals for various types of government-sponsored reinsurance for health care to help lower costs and spread risks. This study evaluates a variety of government-sponsored reinsurance programs to determine how well they achieve their purposes, and at what costs.
Study Design: In-depth case studies were conducted in five states where governmentsponsored reinsurance programs have operated for several years.
Population Studied: Three of the states
(Connecticut, Nebraska, Wyoming) have industryfunded reinsurance pools for small groups, and one state (North Dakota) has terminated such a pool.
Healthy New York (HNY) is a unique program that subsidizes individual and small group insurance for low-income people by reimbursing insurers for highcost claims.
Principal Findings: 1) Industry-funded prospective reinsurance pools are no longer needed to serve their original purpose, which was to keep insurers from leaving small group markets in response to regulatory reforms in the 1990s. Whether or not a state has a reinsurance pool bears no detectible relation to the performance of insurance market reforms that aim to improve access to health insurance. However, these pools may have initially helped some insurers to transition into market reforms. 2) There is broad and deep consensus among independent experts, actuaries, and market observers that industry-funded reinsurance pools do not, and inherently cannot, reduce the overall or average price of health insurance. However, if insurers use these pools extensively, they can help to spread risks more evenly across the market. 3)
The Healthy New York (HNY) program enrolls about 130,000 people at a total annual cost to the government of roughly $850 per person. This enrollment reflects only a small percentage of those in New York who are potentially eligible, and a significant portion of enrollees had other insurance available to them. 4) In addition to these subsidies,
HNY reduces costs in the individual market by combining individuals and small groups into the same risk pool to create cross-subsidies, and it excludes some benefits that may attract more healthy enrollees to HNY than the regular individual market. These factors help explain why threequarters of enrollees in HNY are individuals and sole-proprietors.
Conclusion: The reinsurance programs profiled in this study that are not coupled with government subsidies have not reduced costs or expanded coverage, and, by themselves, they are not capable of doing so. In Healthy New York, factors beyond the traditional risk-spreading function of reinsurance, such as substantial government subsidies and a streamlined benefit package, are
likely responsible for lower prices and program enrollment.
Implications for Policy, Delivery or Practice:
Reinsurance is not a panacea. Only reinsurance with substantial government subsidies has the potential to stimulate more people to purchase health insurance.
Funding Source(s): Blue Cross and Blue Shield
Association
Theme: Coverage and Access
● Measuring Safety Culture: Psychometric
Properties of a Questionnaire for Medical
Directors in German Hospitals
Antje Hammer, Oliver Ommen, M.D., M.P.H.; Julia
Röttger; Yvonne Pfeiffer; Tanja Manser, Ph.D.;
Holger Pfaff, Ph.D.
Presented by: Antje Hammer, Research Assistant,
Center for Health Research Cologne, University
Cologne, Eupener Straße 129, Cologne, 50933,
DE, Phone: +49 221 478 97108; Email: antje.hammer@uk-koeln.de
Research Objective: During the past years the importance of measuring patient safety culture in health care has grown worldwide. From the management perspective it is necessary to assess how the top-management of a hospital, especially the medical directors, assesses the patient safety culture in their own organization. Until now, there is no tool known for measuring safety culture from the medical directors’ point of view in Germany. The
German version of the well-known Hospital Survey of Patient Safety Culture (HSOPS) was still now used only for survey of the hospital employees. The aim of the study was to test if the German version of the HSOPS could be used in a medical directors survey and if it shows the same psychometric properties as the HSOPS for hospital employees.
Study Design: In 2008 a cross-sectional representative study was conducted with 1224 medical directors from all German hospitals with at least one internal medicine and one surgery unit.
The standardized written questionnaire included the
Hospital Survey of Patient Safety Culture (HSOPS).
The Swiss-German version of the HSOPS has been adapted for medical directors in German hospitals.
It consists of 43 Items, 12 safety culture dimensions including 2 outcome dimensions and 1 single item outcome measure. Safety Culture is measured by scales like ‘Hospital Management Support for
Patient Safety’, ‘Supervisor, Manager Expectations
& Actions Promoting Safety’, ‘Teamwork within units’ and ‘Communication Openness’. Subjective outcomes are measured by the dimensions like
‘Overall Perceptions of Safety’.
Population Studied: A total number of 551 medical directors filled out the questionnaire. The response rate was 45.2 %.
Principal Findings: Reliabilities were examined for each of the 12 dimensions. Also the construct validity was tested by calculating correlations between all 12 safety-culture dimensions after calculating a composite score. The reliability measured by Cronbach´s Alpha ranged from .62 to
.86. Correlations for all 12 safety culture dimensions ranged from .14 (between ‘Staffing’ and ‘Frequency of Event Reporting’) to .64 (between ‘Hospital
Management Support’ and ‘Overall Perception of
Safety’. Therefore the most dimensions measure different constructs.
Conclusion: The results show that the dimensions have an acceptable level of reliability. Tests of the construct validity also prove that the instrument is eligible for measuring safety culture in German hospitals from the medical directors’ point of view.
The psychometric properties are comparable to the
German version of the HSOPS and the original version.
Implications for Policy, Delivery or Practice:
Until now the HSOPS was translated and used in
27 countries. We have been able to show that the
HSOPS could also be used to assess the patient safety culture from the medical directors’ point of view. The HSOPS could therefore also be used in nationwide hospital surveys to assess the topmanagement view of patient culture in the own hospital.
Funding Source(s): Bundesärztekammer
● Direct & Indirect Effects of Teenage Body
Weight on Adult Wages
Euna Han, Ph.D.; Edward Norton, Ph.D.; Lisa
Powell, Ph.D.
Presented by: Euna Han, Ph.D., Post Doctoral
Research Fellow, Institute for Health Research &
Policy, University of Illinois at Chicago, 1747 West
Roosevelt Road, Chicago, IL 60608, Phone: (312)
996-6824; Email: eunahan@uic.edu
Research Objective: This paper examines the direct effect of BMI on wages and the indirect effects operating through education and occupation choice, particularly for late-teen BMI and adult wages. In addition to the direct effect of BMI in the late teenage years between age 16 and 20 on longterm wages in the early thirties, late teen BMI is hypothesized to also affect wages in the early thirties indirectly through its effect on education and occupation choice in the early thirties.
Study Design: We present an empirical model that predicts wages as a function of BMI (the direct effect), education and occupation choice (indirect effects of BMI), and other factors. The effect of a unit change in late teen BMI on the logarithm of wages in the early career stage is the full derivative of the logarithm of wages in the early career with respect to late teen BMI, taking into account the indirect effect of late teen BMI through education and occupation choice in the early career. To calculate the indirect effects, we estimate the effect of late teen BMI on the stock of education
accumulated by the time an individual reaches their early 30s using OLS. We specify education in the early thirties as a function of late teen BMI and other factors in the early thirties. We then estimate reduced form models of the effect of late teen BMI on occupation choice among white-collar, service, sales, managerial or professional specialty jobs, and blue-collar jobs (based on Census occupational codes), and choice of jobs requiring social interactions with colleagues or customers (based on the Dictionary of Occupation Titles).
Population Studied: Data drawn from the National
Longitudinal Survey of Youth 1979 (NLSY79) are used for this study. The final estimation sample consists of 2,145 observations for women and
2,052 observations for men after restricting our sample to those observations with BMI available in their late teenage years and excluding those in their early 30s who were pregnant within a year from the time of the interview including pregnant at the time of the interview, were in military service at the time of the interview, or had missing data in other covariates in the estimation models. Only observations for employed persons with occupation information were used for log hourly wages (1,514 women and 1,721 men).
Principal Findings: Our results suggest that higher
BMI for both women and men in their late teenage years decreases the stock of education acquired by the early thirties which, in turn, affects occupation choice. We also find that the wage penalty in the early thirties for higher late teen BMI does not vary with any particular vocation path and the effect of late teen BMI on occupation choice in the early thirties is statistically indistinguishable from zero except for blue-collar occupations for women.
Finally, we show that previous studies on the BMI wage penalty for women that condition on employment and occupation underestimate the total effect by about 16%. For men, although no observed direct BMI wage penalty is found, we find a 0.11% decrease in hourly wages for a unit increase BMI in late teenage years, mainly stemming through the indirect effect of BMI on education.
Conclusion: Overall, we show that late teen BMI for women and men affects their choices for occupation in the early thirties mainly through investments in education. We also argue that the past literature on BMI and labor market outcomes may have miscalculated the BMI wage penalty because the calculation was done conditional on education and occupation.
Implications for Policy, Delivery or Practice:
These results suggest that in order to fully capture the relationship between BMI and labor market outcomes researchers need to account for the early part of the life-cycle when individuals make important choices such those related to human capital formation.
Theme: Obesity Prevention and Treatment
● The Effect of Food Stamp Program
Participation on the Probability of Obesity
Euna Han, Ph.D.; Lisa Powell, Ph.D.
Presented by: Euna Han, Ph.D., Post Doctoral
Research Fellow, Institute for Health Research &
Policy, University of Illinois at Chicago, 1747 West
Roosevelt Road, Chicago, IL 60608, Phone: (312)
996-6824; Email: eunahan@uic.edu
Research Objective: This study explores the effect of Food Stamp Program participation, one of the most prevalent policy tool to defeat food insecurity, on body mass index (BMI) for individuals with high school or less education.
Study Design: We improve the previous literature by estimating changes in the dispersion of conditional BMI following Food Stamp Program participation through a quantile regression model.
Propensity score matching method is used to control for the potential selection into Food Stamp
Program participation in addition to indvidual fixed effect model. We control for state-wide yearly variations in nutrition education plan participation and the amount of federally funded budget for the nutrition education plan.
Population Studied: Data from six years (1992,
1994, 1996, 1998, 2000, and 2002) drawn from the
National Longitudinal Survey of Youth 1979 were pooled to create the samples for this study. The final estimation sample is defined as individuals whose stock of education is less than high school or high school.
Principal Findings: The effect of Food Stamp
Program participation on BMI increases at a higher quantile of the conditional distribution of BMI for women. Specifically, the effect of Food Stamp
Program participation is not statistically significant the 10th quantile, but is associated with a 0.8 unit increase of BMI at the 25th quantile. The magnitude of the statistically significant effect of Food Stamp
Program participation becomes more than doubled at the 75th and 90th quanitles, reaching slightly more than a 2 unit increase of BMI at the 90th quantile for the Food Stamp Program participants compared to non-participants. For men, no statistically significant effect of Food Stamp
Program participation is found in both OLS and quantile regression estimations. Overall, quantile estimators for men show less variation across the conditional distribution of BMI compared to women.
Adding the propensity score in the parametric estimations barely changes the estimated results for both genders.
Conclusion: Study results indicate that OLS estimator of Food Stamp Program participation for women is overestimated below the median but underestimated above the median of the conditional distribution of BMI. For men, neither OLS estimator nor quantile regression estimators are statistically significant. We also do not find a strong evidence for selection into Food Stamp program participation.
Implications for Policy, Delivery or Practice:
Future studies should be directed to configure pathways linking Food Stamp Program participatin and the increased body weight to help to develop policy measures to curb the excess body weight problems among food stamp participants.
Theme: Obesity Prevention and Treatment
● Are Occurrences of Unsafe Events
(Medication Errors, Patient Falls with Injuries,
Nosocomial Infections, Staff Work-Related
Injuries & Verbal Abuse of Nurses) in Inpatient
Psychiatric Hospitals Settings Associated with
Organizational Factors?
Nancy Hanrahan, Ph.D.
Presented by: Nancy Hanrahan, Ph.D., Assistant
Professor, Center for Health Outcomes & Policy
Research, University of Pennsylvania, Fagin Hall;
418 Curie Boulevard, Philadelphia, PA 19104,
Phone: (215) 387-0661; Email: nancyp@nursing.upenn.edu
Research Objective: Inpatient psychiatric care in general hospitals is the primary service available to persons who present a grave danger to themselves or others due to psychiatric illness. Numerous concerns have been expressed by mental health consumers and professionals themselves over the quality of the care environment and the safety of patients and staff. In this study, we examined the relationship of organizational factors to unsafe outcomes (medication errors, patient falls with injuries, nosocomial infections, staff work-related injuries, and verbal abuse of nurses) for inpatient psychiatric care in general hospitals.
Study Design: We used a cross sectional descriptive and correlational study. We used 1999 nurse survey data from Pennsylvania registered nurses about the quality of patient care, unsafe events, and organizational performance in their hospital care environments. We measured organizational performance using the Practice
Environment Scale-Nurse Work Index (PES-NWI).
The effect of the PES-NWI subscales
(organizational factors) on unsafe events (wrong medication, infections, falls, injuries, verbal abuse) were examined using robust logistic regression models clustered at the hospital level.
Population Studied: The population studied included 67 hospitals in Pennsylvania with licensed adult psychiatric beds and RN respondents, both psychiatric RNs and non-psychiatric RNs. The total number of RN respondents in our sample was
5,804 of which 357 were psychiatric RNs and nonpsychiatric 5,447 RNs. More than half the general hospitals had between 100 to 250 beds and 46% were teaching hospitals. The 67 hospitals comprised 74% of all general hospitals in
Pennsylvania with dedicated psychiatric beds. The mean number of RNs per hospital was 98
(SD=69.5). Psychiatric nurses per hospital averaged 5.4 (SD=2.8).
Principal Findings: We found a higher occurrence of nosocomial infections, patient falls with injuries and work-related injuries inversely related to a significant reduction in the nurse’s perception of adequate staffing and resources, unit manager skill, quality in the nurse-physician relationships, and the presence of a formal quality of care program.
These relationships were significant at the p<0.010 level.
Conclusion: The high rate of infection and patient falls was an unexpected finding in this study.
Typically, three quarters of patients admitted to acute inpatient psychiatric units have a serious mental illness. In the wake of deinstitutionalization, general hospitals play a primary role in the management of mental illness. People with serious mental illness are at a high risk for infections due to poor primary care, homelessness, and poverty.
Patient falls may be explained as a side effect from antipsychotic and sedative medications that can cause slower mentation and orthostatic hypotension. General hospital organizational managers can significantly reduce the occurrence of these unsafe events by improving unit manager’s leadership skills, supporting nurse and physician collaboration, providing adequate staffing and resources, and developing foundations for quality of care monitoring. Further study is needed to explain these relationships.
Implications for Policy, Delivery or Practice:
This study provides some of the first estimates unsafe events and the quality of the psychiatric inpatient care environment occurring in general hospitals. This information is essential to promote targeted development and implementation of evidence-based strategies to improve inpatient psychiatric care in general hospitals.
Funding Source(s): National Institute of Nursing
Research
Theme: Organizational Performance and
Management
● Use of Evidence-Based Strategies to Reduce
Door-to-Balloon Time in the Setting of the D2B
Alliance
Luke Hansen, M.D.; Christina Yuan, M.P.H.; Ingrid
Nembhard, Ph.D.; Susan Busch, Ph.D.; Harlan
Krumholz, M.D., S.M.; Elizabeth Bradley, Ph.D.
Presented by: Luke Hansen, M.D., Robert Wood
Johnson Clinical Scholar, Robert Wood Johnson
Clinical Scholars Program, Yale University School of Medicine, P.O. Box 208088, New Haven, CT
06520, Phone: (774) 219-9204; Email: luke.hansen1@gmail.com
Research Objective: To examine changes in the use of recommended strategies to reduce door-toballoon time and to measure use of quality improvement campaign resources among enrollees
and non-enrollees in the D2B Alliance, a national quality improvement campaign.
Study Design: We conducted a prospective study of hospitals concerning their use of strategies for treatment of acute myocardial infarction.
Approximately three quarters of these hospitals joined the D2B Alliance and one quarter did not.
Both D2B Alliance enrolled and non-enrolled hospitals reported use of strategies in 2005 and
2008 and use of D2B Alliance resources in 2008.
The majority of D2B Alliance resources were available to both enrolled and non-enrolled hospitals via the D2B Alliance website. Logistic regression was used to compare enrolled hospitals to non-enrolled hospitals according to strategy use and Alliance resource use.
Population Studied: 365 hospitals performing percutaneous coronary intervention for acute myocardial infarction.
Principal Findings: Odds of strategy use in 2008 were not significantly different when enrolled hospitals were compared with non-enrolled hospitals and adjusted for 2005 strategy use (single call activation OR .80 [95%CI 0.61 to 1.94], emergency department activation OR 1.08 [95%CI
0.42 to 1.53], 30 minute arrival expectation OR 0.75
[95%CI .23 to 2.43], prompt data feedback OR 1.38
[95%CI 0.74 to 2.59], pre-hospital ECG use OR
1.35 [95%CI 0.73 to 2.47]). Both enrolled and nonenrolled hospitals experienced significant increases in strategy use between 2005 and 2008 in bivariate analysis (single call cath lab activation increased from 14% to 35% [p <.001], emergency department cath lab activation from 23% to 59% [p <.001], a 30 minute arrival for cath lab staff from 68% to 91%
[p<.001], prompt data feedback from 42% to 73%
[p<.001], and pre-hospital ECG transmission use from 9% to 41% [p<.001]). 85% of enrolled hospitals reported use of any D2B Alliance resource compared with 52% of non-enrolled hospitals. In logistic regression analysis use of D2B Alliance resources was significantly higher among enrolled hospitals for most resources (D2B Alliance website use OR 3.79 [95%CI 1.87 to 7.71], Alliance newsletter use OR 7.62 [95%CI 3.40 to 17.18], D2B
Alliance webinar use OR 3.09 [95%CI 1.50 to 6.38], use of the D2B Alliance online community OR 7.36
[95%CI 3.10 to 17.67], use of the D2B Alliance toolkit for strategy implementation OR 2.82 [95%CI
1.36 to 5.84]).
Conclusion: Significant increases in the use of performance improvement strategies recommended by the D2B Alliance have been observed in both enrolled and non-enrolled hospitals. Use of D2B
Alliance resources was also observed in all hospitals but to a significantly greater extent in enrolled hospitals.
Implications for Policy, Delivery or Practice:
Because non-enrolled hospitals have been shown to seek out campaign resources without formal campaign membership, designers of quality improvement campaigns should consider open accessibility of campaign resources when designing campaign dissemination. For some hospitals, formal membership may not be seen as an important step toward learning and improvement.
Future work will need to measure how hospitals that elect not to join national quality campaigns learn best practices. Particularly for quality improvement goals which have achieved national prominence, optimal campaign effectiveness may be undermined by a strictly bounded membership organization and may be best achieved by broad accessibility of resources.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Organized
Processes
● Community Level Uninsurance & its Affects on the Insured
Rachel Hardeman, M.P.H.
Presented by: Rachel Hardeman, M.P.H., Doctoral
Student, Health Policy & Management, University of
Minnesota, 110 1st Avenue, Northeast #601,
Minneapolis, MN 55413, Email: hard0222@umn.edu
Research Objective: Recent research argues that the proportion of the uninsured population in a community not only has a financial impact on the insured population but also on the quality of care available in the local health care market. Health care providers and the insured population may have to absorb the financial burden of providing uncompensated care to the uninsured population.
The uninsured population also demands a lower quantity and quality of health care services than the insured population, which implies that the average quality in high uninsurance communities is likely to be relatively low.
This research study attempted to capture the different dimensions of community uninsurance spillovers by exploring the price, access and quality of health care services available to the insured population in Minneapolis, Minnesota (a low uninsurance community) and McAllen, Texas (a high uninsurance community). We hypothesized larger negative spillovers for the insured population in McAllen compared to Minneapolis.
Study Design: Focus groups were conducted to obtain the perspectives of insured mothers living in
Minneapolis, Minnesota and in McAllen, Texas.
These communities were selected because they represent substantially disparate local settings with regards to health insurance coverage rates (9% uninsured in Minneapolis versus 48% in McAllen).
Population Studied: The study population included insured Latino mothers who indicated they were currently caring for at least one child in their household. This allowed us to better capture the health care perspectives and experiences of both mothers and their families. Two focus groups were
conducted in each community (n=28 in Minnesota and n=30 in Texas).
Principal Findings: A common perception among insured women in Texas was that the costs of premiums, copayments and deductibles were extremely high. Often times, the expenses were so significant that they felt they could not afford to include other family members on their health insurance plans. High costs were also related to delayed care more often in Texas than in
Minnesota. Focus group participants in Minnesota described the quality of care they receive as adequate and often times as superior while participants in Texas were more often to perceive their quality of care as merely adequate or inadequate. Most participants in Texas expressed frustration with the patient-provider relationship and they experienced greater difficulty in obtaining referrals to specialists.
Conclusion: The findings suggest that the insured population faces substantial differences across these communities in the price, access and quality of health care services. Prices are relatively high in the low uninsurance community while quality and access are relatively low in the high uninsurance community, particularly for high fixed cost services
(e.g., specialized care).
Implications for Policy, Delivery or Practice:
Uninsurance is a problem not only for the uninsured but also for the insured population. High community uninsurance may result in substantial pecuniary and non-pecuniary spillovers on the insured population.
The potential benefits of covering the uninsured population are underestimated if these spillovers are not considered. The form/magnitude of these spillovers has the potential to inform different policy options to cover the uninsured population.
● Perceptions of Bias in Medical Care among
Patients Living with Diabetes
Lee Hargraves, Ph.D.; Celeste Lemay, R.N., M.P.H.
Presented by: Lee Hargraves, Ph.D., Research
Associate Professor, Family Medicine & Community
Health, Univesity of Massachusetts Medical School,
55 Lake Avenue North, Worcester, MA 01655,
Email: lee.hargraves@umassmed.edu
Research Objective: The objectives of this project were to conduct qualitative interviews in groups or individually with patients seeking care at community health centers, to understand patients’ ability to describe unfair treatment, and to develop patient reported measures of perceived bias in their encounters with health care professionals and health care organizations.
Study Design: We examined reports of bias among patients with diabetes utilizing group and individual interviews. Focus groups were stratified by race/ethnicity and language spoken, i.e., English or Spanish. We asked participants to describe good experiences with health professionals and other health center staff. Participants described what makes them feel well cared for as well as about bad experiences, including unfair treatment, being disrespected, or bias. To assess understanding of new measures related to bias and questions from a modified version of the CAHPS Group and Clinician
Survey, we conducted cognitive interviews.
Population Studied: : African American, Latino, and non-Hispanic white patients living with diabetes receiving care from 2 Community Health Centers in
Central Massachusetts who agreed to attend a group or individual interviews.
Principal Findings: In group interviews, patients easily described poor or unfair treatment. However, it was difficult to elicit negative comments about physicians who have established relationships with patients. In describing good experiences, patients focused on how well doctors and nurses understand them. Participants typically described bad experiences related to access problems.
Language congruence between patients, health professionals, and clinic staff was the most important contributing factor related to perceptions of unfair treatment and bias. Speaking to participants in the language of their choice was a key finding from English and Spanish focus groups.
It is difficult for patients to attribute unfair treatment to specific attributes, e.g., the nature of their condition, ethnicity, gender, or ability to communicate. In cognitive interviews, participants understood questions related to being treated unfairly and were able to articulate why they responded as they did. None of the participants responded that they believed that their doctor had treated them unfairly because ethnicity. Most patients did not understand questions asking about doctor’s understanding of their background and culture, assuming that these questions asked about medical history. Racial/ethnic congruence of doctors and health care staff, including front office staff was important to several participants. Many participants reported unfair treatment by front office staff, yet were unable to attribute unfairness to race/ethnicity.
Conclusion: Initial findings from focus groups and cognitive interviews with adult patients seeking care at community health centers include little direct reporting of being systematically treated by physicians and nurses with disrespect, bias, or discrimination. Many patients express extreme gratitude in being treated for their chronic condition at low cost. Patients did report unfair treatment and bias from health center staff, including front office personnel.
Implications for Policy, Delivery or Practice:
Patients who perceive bias and discrimination from health care workers may be less adherent to treatment and less likely to seek follow-up care.
Policymakers may want to consider both direct and indirect measures of unfair treatment in selfreported measures of patient experiences.
Funding Source(s): NIMH
Theme: Disparities
● Medicaid per Member per Month Expenditures
Before and After Implementation of Reforms in
Florida
Jeffrey Harman, Ph.D.; Haichang Xin; R. Paul
Duncan, Ph.D.
Presented by: Jeffrey Harman, Ph.D., Associate
Professor, Health Services Research, Management
& Policy, University of Florida, P.O. Box 100195,
Gainesville, FL 32607, Phone: (352) 273-6060;
Email: jharman@ufl.edu
Research Objective: The state of Florida implemented Medicaid reforms for enrollees in
Broward (Ft. Lauderdale) and Duval (Jacksonville) counties beginning in September 2006. The primary reform involved a requirement that enrollees choose a health plan from among a list of
HMOs and Provider Service Networks (PSNs). This study assesses the impact of these reforms on per member per month (PMPM) Medicaid expenditures.
Study Design: A difference-in-difference approach was used to study changes in total Medicaid PMPM expenditures associated with Medicaid reforms.
The difference in average total PMPM expenditures of eligible enrollees in reform counties during the two fiscal years prior to implementation of reform
(FY0405, FY0506) and PMPM expenditures of enrollees during the first fiscal year of reform
(FY0607) is compared to the difference in average total PMPM Medicaid expenditures over the same time period of Medicaid enrollees in Orange
(Orlando) and Hillsborough (Tampa) counties (nonreform). Differences-in-differences in average
PMPM expenditures between each fiscal year are calculated for two eligibility groups: TANF and SSI.
A multivariate panel data regression was also conducted to determine if any observed differences remained significant in a multivariate context that controls for age, gender, race, and time trend.
Population Studied: Florida Medicaid enrollees in
Broward, Duval, Orange, and Hillsborough counties who received services between July 1, 2004 and
June 30, 2007. Enrollees had to be eligible for
Medicaid through SSI or TANF and could not be enrolled in separate waiver programs. There were
6,215,690 person-months from Broward and Duval and 7,176,510 person-months from Orange and
Hillsborough counties.
Principal Findings: Average PMPM expenditures for SSI enrollees in reform counties were $101 lower in the first year of reform compared to
FY0405 and $31 lower when compared to FY0506.
In the control counties, average PMPM expenditures for SSI enrollees were $126 higher and $75 higher in FY0607 compared to FY0405 and FY0506, respectively. Thus, relative to the control counties, the reform counties saved $227
PMPM and $106 PMPM during the first year of reform compared to FY0405 and FY0506. For
TANF enrollees in reform counties, average PMPM expenditures were $6 lower in FY0607 compared to
FY0405, but $1 higher compared to FY0506.
However, for TANF enrollees in control counties, average PMPM expenditures were $18 higher in
FY0607 compared to FY0405 and $17 higher compared to FY0506. Therefore, relative to control counties, the reform counties saved $24 PMPM and
$16 PMPM during the first year of reform compared to FY0405 and FY0506 respectively. In a multivariate context, Medicaid reform was not associated with significant reductions in expenditures among SSI enrollees (p=.198), but was associated with significant reductions in expenditures among TANF enrollees (p<.001)
Conclusion: It appears that Medicaid reforms in
Florida have resulted in lower PMPM expenditures compared to pre-reform. However, it is not known whether these costs savings were achieved through more efficient provision of care or decreased access to needed care.
Implications for Policy, Delivery or Practice:
Medicaid reforms such as those provided in Florida, may result in lower average PMPM costs, although more needs to be known about how these cost savings were achieved.
Funding Source(s): Florida Agency for Health
Care Administration
Theme: Medicaid, SCHIP and State Health Reform
● Assessment of the Mental Health Funding
Marketplace in Rural vs. Urban Settings
Jeffrey Harman, Ph.D.; John Fortney, Ph.D.; Fran
Dong, Ph.D.; Stan Xu, Ph.D.
Presented by: Jeffrey Harman, Ph.D., Associate
Professor, Health Services Research, Management
& Policy, University of Florida, P.O. Box 100195,
Gainesville, FL 32610-0195, Phone: (352) 273-
6060; Email: jharman@ufl.edu
Research Objective: Rural individuals with mental health (MH) problems are significantly less likely to receive any MH care for their disorder than urban individuals. Reduced access to MH care in rural areas is partially due to an inadequate supply of MH specialists. It is necessary to understand how MH services are financed in rural areas relative to urban areas to determine the best way to develop financial incentives for MH specialists to practice in rural areas. The objectives of this study are to assess the impact of rurality on the source of payment for MH disorders and determine whether rural-urban differences in source of payment vary by MH condition, defined here as seriously mentally ill (SMI) and all other MH conditions.
Study Design: Data from the 2004 Medical
Expenditure Panel Survey (MEPS) are used to identify the proportion of expenditures for all MH services that are paid for by private insurance, public insurance, and self-pay by Rural-Urban
Continuum Codes (RUCC). Generalized linear
models which incorporated survey weights to account for the complex sampling strategy and produce nationally-representative estimates were used to assess the impact of RUCC on proportion of MH services paid by funding source.
Population Studied: The study population includes all individuals included in the 2004 MEPS sample
(n=34,403) who identified themselves as having a mental health disorder (ICD-9 codes 290-314; n=5,174). Additional analyses compared individuals with SMI (schizophrenia, bipolar disorder, and major depression; n=541) to all other individuals with mental health disorders.
Principal Findings: On average, 42% of expenditures for MH services were paid for by private insurance in the most urban areas
(RUCC=1) compared to 37% in the most rural areas
(RUCC=7,8, or 9). An average of 21% of expenditures were paid for by public sources in the most urban areas compared to 25% in the most rural areas. An average of 37% or expenditures for
MH services were paid for by self-pay in the most urban areas compared to 38% in the most rural areas. Statistically significant differences in the proportion of MH services paid for by private insurance (p=.032) and by public insurance
(p=.033) by RUCC were found in the multivariate analyses, with no significant difference in the proportion paid by self-pay (p=.682). Among the
SMI population, a larger proportion of expenditures were paid by self-pay in rural compared to urban areas (37% vs. 28%), while a smaller proportion was paid for by both private insurance (22% vs.
25%) and public insurance (41% vs. 47%).
Conclusion: Overall, individuals living in rural areas are more likely to have services paid for by public insurance and less likely by private insurance than individuals living in urban areas. For individuals with SMI, a greater proportion of MH services were funded through public insurance and self-pay than individuals with SMI in urban areas.
Implications for Policy, Delivery or Practice:
Significant differences in funding source were found between urban and rural areas, suggesting that approaches to providing financial incentives and insurance-based programs to improve access to mental health care need to be tailored specifically for rural vs. urban settings.
Funding Source(s): HRSA
Theme: Behavioral Health
● The Hospitalist Model: Does it Enhance
Healthcare Quality?
Jeffrey Harrison, Ph.D., M.B.A., M.H.A., F.A.C.H.E.;
Lorrie Curran, M.H.A.
Presented by: Jeffrey Harrison, Ph.D., M.B.A.,
M.H.A., F.A.C.H.E., Associate Professor, Public
Health, University of North Florida, 1 UNF Drive,
Jacksonville, FL 32224, Phone: (904) 620-1440;
Email: jeffrey.harrison@unf.edu
Research Objective: Strategic management theory emphasizes the importance of positioning the hospital relative to its environment in order to achieve its objectives and assure its survival. Mick and Wyttenbach (2003), in their use of strategic adaptation theory, believe the relationships between physicians and healthcare organizations are changing due to the shift towards cost effective and process-focused care. Previously, hospitals were often passive players with physicians having complete control of their clinical practices. More recently, hospitals are under increasing pressure to improve the care provided by physicians. This supports physicians networking with hospitals by implementation of the Hospitalist Model which facilitates innovation of inpatient clinical processes.
This study evaluates the relationship of the
Hospitalist Model to hospital quality and efficiency.
It hopes to demonstrate that the Hospitalist Model improves healthcare value by enhancing quality and improving efficiency.
Study Design: This study uses logistic regression to evaluate those hospitals that utilize the
Hospitalist Model in contrast to those hospitals that do not. The independent variables of interest include market factors, operating performance, organizational factors and quality. Data was obtained from the AHA annual survey, the Area
Resource File (ARF) and the CMS hospital
Compare Quality data. The AHA survey provides extensive organizational data including ownership, bed size, hospital services and utilization. The ARF has extensive information on hospital market structure, including demographics, economics, and other measures of the hospital environment. As part of the Hospital Compare quality reporting initiative, the CMS tracks quality measurements for acute care hospitals.
Population Studied: We examined data on 1,376 hospitals utilizing the Hospitalist Model in contrast to 2,129 hospitals not using the Hospitalist Model for the year 2005. As stated previously, Quality data was obtained from the CMS hospital Compare
Quality database.
Principal Findings: Our research clearly indicates the hospitalist model is an opportunity to increase hospital efficiency and quality. We determined that hospitals using the hospitalist model had higher occupancy rates, a shorter length of stay and a higher CMS Hospital Compare quality score.
Conclusion: Our study shows the Hospitalist
Model leads to improved quality. The results also suggest that improved quality and efficiency is due to improved management. As a result, this study would support strategic management theory as an effective method to improve organizational quality and efficiency.
Implications for Policy, Delivery or Practice: As policy makers seek to improve the healthcare delivery system, the Hospitalist Model is an opportunity to enhance the quality and efficiency of
U.S. hospitals. The implementation of the
hospitalist model is most appropriate in hospitals with inefficient operations and poor quality. Also, as healthcare planners monitor the allocation of hospital beds, they should consider the use of the hospitalist model to maximize the use of the existing hospital infrastructure. Finally, our study found that the hospitalist model reduces cost among the elderly population. As a result, implementation of the hospitalist model is of particular value in communities with a growing elderly population where the Hospitalist Model will utilize evidence based research to treat chronic diseases among the elderly leading to improved quality and efficiency.
Theme: Quality and Efficiency: Organized
Processes
● Strategies for Providing Culturally Competent
Care in Family Planning Clinics
Jamie Hart, Ph.D., M.P.H.; Amy Brown, M.P.H.,
B.A.; Sandra Silva, M.M., B.A.
Presented by: Jamie Hart, Ph.D., M.P.H.,
Research Director for Health Equity, Policy,
Planning, & Evaluation Practice Area, Altarum
Institute, 1200 18th Street, NW Suite 700,
Washington, DC 20036, Phone: (202) 828-5100;
Email: jamie.hart@altarum.org
Research Objective: Altarum is conducting an evaluation to examine specific strategies that are being created and adapted to provide culturally competent care within Title X-supported clinics for the Office of Family Planning. The project objectives are to: assist OFP in defining cultural competency within the Title X context; identify current services, activities, and gaps; highlight innovative strategies and best practices; and inform the development of a national strategy to support
Title X providers in addressing culturally competent service delivery.
Study Design: This evaluation involves several key components, including: A comprehensive review of the literature pertaining to definitions of cultural competency, research on the impact of cultural competence in health care settings, and strategies for enhancing cultural competency among Title X providers and clinic staff; Telephone interviews with OFP Federal and regional staff and researchers that are experts in the fields of family planning and cultural competence; Site visits with
Regional Training Centers and Title X-funded clinics that have developed and pursued innovative trainings or have implemented cultural competency initiatives; A final report that presents the major findings from a content analysis of the themes and patterns emerging from the qualitative data collected evaluation and recommendations based on the evaluation that may be considered in the development of a national strategy to support Title
X providers in addressing culturally competent service delivery.
Population Studied: This evaluation collects information from OFP Federal and regional staff, researchers that are experts in the fields of family planning and cultural competence, Title X Regional
Training Center staff that have pursued innovative cultural competence trainings, and the staff and clients at Title X-funded clinics that have implemented cultural competence initiatives.
Principal Findings: Many cultural and linguistic competence guidelines and resources have not been tailored specifically to family planning settings.
However, many clinics have been able to adapt these materials to implement strategies in three areas identified as critical for delivering more culturally and linguistically competent care: (1) policy and regulation, (2) systems development, and (3) organizational structure.
Conclusion: This project is unique because it examines the adaptation and implementation of cultural competence frameworks within previously underexplored settings, family planning clinics. This session will offer lessons learned and concrete strategies for adaptation at a variety of levels, including policy, administrative, practice and service delivery, client and family, and community.
Implications for Policy, Delivery or Practice: As the only federal program dedicated solely to providing family planning and reproductive health care, Title X provides services to diverse clients, many from historically underserved populations.
Title X therefore represents an important venue for the delivery of culturally and linguistically competent services and is well positioned to reduce disparities in reproductive health care. Initial research has indicated that while many Title X-funded programs are providing a range of culturally competent services, it is unclear how these programs have been consciously addressing disparities and cultural competency and how OFP can help to further support these activities. This evaluation will help move the field forward by examining specific strategies that are being created and adapted to provide culturally competent care within Title Xsupported programs.
Funding Source(s): DHHS Office of Population
Affairs, Office of Family Planning
Theme: Disparities
● Self-Management Support of Vulnerable
Populations: Reviewing Evidence on the
Perspective of Elderly Living Alone with
Chronic Conditions
Joerg Haslbeck, Dr.P.H., M.Sc.N.; Ruth McCorkle,
Ph.D., F.A.A.N.
Presented by: Joerg Haslbeck, Dr.P.H., M.Sc.N.,
Postdoctoral and Visiting Harkness-Fellow, School of Nursing, Yale University, 100 Church Street,
South, New Haven, CT 06536, Phone: (203) 737-
1351; Email: joerg.haslbeck@yale.edu
Research Objective: Elderly people living alone with one or more chronic conditions constitute a particularly vulnerable population. A shifting demographic and societal landscape has left an increasing number of elderly individuals with few resources, family and/or social support that have been recognized as important for successful chronic illness management. So far, this particularly vulnerable group has received only limited attention in research on self-management support in chronic illness. Started in October 2008, the purpose of this one-year-study is to fill this gap by providing a deeper understanding of specific self-management needs and abilities of elderly people living alone with chronic conditions.
Study Design: The study is fusing two different perspectives. (I) Based on an integrative literature review, available evidence on the perspective of elderly living alone with chronic conditions has been systematically examined. Using an initially developed search strategy and database-specific concept tables, Medline, CINHAL, PsycINFO and
GEROLIT have been searched in November and
December 2008 followed by an extensive hand search and further reference mining. In addition, a
(II) series of expert interviews with roughly twenty
U.S. health care experts and providers is about to be finished by Summer 2009. A content analysis approach will be used to analyze collected data.
Population Studied: Part 1: review of empirical studies on elderly living alone; Part 2: interviews with health care professionals and experts located nationwide.
Principal Findings: Preliminary findings from the review process support the lack of evidence on the perspective of elderly living alone as well as on their needs and strategies in chronic illness selfmanagement. A limited number of studies have so far been identified, also due to the fact that existing evidence is spread across disciplines and appears to be poorly indexed in databases.
Conclusion: Early results from the literature review suggest that the vulnerable situation of elderly living alone is not only related to difficult living situations, limited resources, a lack of social support, or feelings of loneliness and isolation. It might be complicated by the hidden occurrence of research evidence on this particular topic.
Implications for Policy, Delivery or Practice: In order to increase consumer engagement in chronic illness care, addressing the needs of elderly living alone is an important goal. As health policy decisions are usually based on evidence, further research on the perspective of elderly living alone should be supported as well as intensified in order to provide empirical support for the development of appropriate intervention strategies for chronic illness self-management support.
Funding Source(s): CWF
Theme: Consumer Choices in Health Care
● The Impact of Medicare Supplement Coverage on Access, Utilization & Cost of Health Care and on Compliance with Recommended
Pharmacuetical Treatment
Kevin Hawkins, Ph.D.; Jaclyn, Marshall, B.A.;
Ronald Ozminkowski, Ph.D; Michael Anderson,
Pharm.D.
Presented by: Kevin Hawkins, Ph.D., Senior
Research Director, Ingenix Consulting, Ingenix, 137
Bedofrd, Brooklyn, MI 49230, Phone: (517) 414-
4276; Email: kevin.hawkins@ingenixconsulting.com
Research Objective: The literature was reviewed to determine the impact of Medicare Supplement
(i.e., Medigap) coverage on the Medicare program with respect to access, utilization, outcomes, costs, and drug compliance. Additionally, we discuss a new program to improve pharmaceutical utilization for Medigap insureds.
Study Design: We conducted a literature search using PubMed, looking for peer-reviewed articles that compared access, utilization, outcomes and costs between those with Medicare fee-for-service coverage alone to those with Medigap. Additional searches focused on differences in pharmacological compliance, for those with and without drug coverage, among the two groups. Finally, we searched for pharmacy compliance programs offered through Medigap plans.
Population Studied: 27 articles met our search criteria.
Principal Findings: The literature suggests
Medigap coverage can be cost-effective, that it is correlated with better access to health care services, and may result in higher utilization of preventive services than would be the case without such coverage. Also, the type of supplement coverage did not significantly influence prescription drugs utilization among Medicare insureds. No articles found discussed any current efforts to manage the pharmaceutical treatment of Medigap insureds.
Conclusion: Medigap programs have not historically managed their insureds like Medicare
Advantage plans have done. In particular, the literature suggests there is much room for improvement in pharmacy management for all
Medicare populations, including those enrolled in
Medigap plans. This has led UnitedHealth Group to offer a pharmacological compliance program, with disease management and case management programs, for their AARP Medigap insureds, beginning in 2009. Results from this care management effort will help tailor models to better manage the care for fee-for-service Medicare insureds with supplement coverage.
Implications for Policy, Delivery or Practice:
Pharmacy management for those enrolled in
Medigap plans is expected to lead to better health status, as pharmaco-compliance is improved.
Funding Source(s): UnitedHealth Group
Theme: Coverage and Access
● Disparities in Major Join Replacement Surgery among Insureds with AARP Medicare
Supplement Coverage Underwritten by
UnitedHealth Group
Kevin Hawkins, Ph.D.; Kamisha Hamilton Escoto,
Ph.D.; Ronald Ozminkowski, Ph.D.; Gandhi
Bhattarai, Ph.D.; Richard Migliori, M.D.
Presented by: Kevin Hawkins, Ph.D., Senior
Director, Ingenix Consulting, Ingenix, 137 Bedford,
Brooklyn, MI 49230, Phone: (517) 414-4276;
Email: kevin.hawkins@ingenixconsulting.com
Research Objective: Determine if disparities in hip and knee replacement surgery exist among osteoarthritis patients with Medicare Supplement
(i.e., Medigap) coverage.
Study Design: Data were obtained from United
Health Group’s database of insureds who have
Medigap coverage via AARP Medicare
Supplement insurance. Patients were selected for the study if they had one or more medical claims with a diagnosis of osteoarthritis from July 1, 2006-
June 30, 2007. Logistic regression analyses tested for age-, gender-, race-, or income-related differences in the likelihood of receiving a hip or knee replacement surgery. The regression models controlled for socioeconomics, health status, type of supplement plan, and residential location.
Population Studied: Of the 2.2 million Medigap insureds who were eligible for the study, 529,652
(24%) had osteoarthritis. Of these, 32,527 (6.1%) received a hip or knee replacement surgery.
Principal Findings: Males were 6% (p<0.001) more likely than females to have a replacement surgery. Patients residing in minority or lowerincome neighborhoods were less likely to receive a hip or knee replacement surgery. The surgery rate decreased with age. Medigap plan type was not a strong predictor of the likelihood of hip or knee replacement. Disparities were much greater by comorbid condition and residential location. Obese patients were 1.79 (p<0.001) times as likely, whereas patients with COPD were only 0.69 times
(p<0.001) as likely to have a hip or knee replacement. Patients in rural areas were 14%
(p<0.001) less likely than those in urban areas and patients residing in South Dakota, Idaho, and
Alaska were 3.24 (p<0.001), 2.58 (p<0.001), and
2.49 (p<0.001) more likely than those in New
Jersey to have replacement surgery.
Conclusion: Disparities in hip and knee replacement surgery existed by age, gender, race, and income levels. Larger disparities were found by residential location and comorbid condition.
UnitedHealth Group is designing interventions to address these disparities; these interventions will begin in mid-2009.
Implications for Policy, Delivery or Practice: In this Medigap population, age-, gender-, race-, and income-related disparities were minor. Larger disparities existed by location and comorbid condition. Disparities that are related to location and health status deserve attention as well.
Funding Source(s): UnitedHealth Group
Theme: Disparities
● Disparities in Coronary Artery Disease Care among Insured with AARP Medicare
Supplement Coverage Underwritten by
UnitedHealth Group
Kevin Hawkins, Ph.D.; Kamisha Hamilton Escoto,
Ph.D.; Ronald Ozminkowski, Ph.D.; Gandhi
Bhattarai, Ph.D.; Jaclyn Marshall, B.A.; Henry
Harbin, M.D.
Presented by: Kevin Hawkins, Ph.D., Senior
Research Director, Ingenix Consulting, Ingenix, 137
Bedford, Brooklyn, MI 49230, Phone: (517) 414-
4276; Email: kevin.hawkins@ingenixconsulting.com
Research Objective: Discover if age-, gender-, race-, or income-related disparities in care for coronary artery disease (CAD) exist among the elderly with Medicare Supplement (i.e. Medigap) coverage.
Study Design: Data were obtained from United
Health Group’s database of insureds who have
Medigap coverage via AARP Medicare Supplement insurance. Patients were selected for the study if they had one or more medical claims with a diagnosis of CAD from July 1, 2006-June 30, 2007.
Logistic regression analyses tested for age-, gender-, race-, or income-related differences in the likelihood of receiving an office visit, coronary angiography, or surgical intervention. The regression models controlled for socioeconomics, health status, type of supplement plan, and residential location.
Population Studied: Of the 2.2 million Medigap insureds eligible for the study, 25.4% (570,711) had
CAD.
Principal Findings: Males were 60% (p<0.001) more likely than females to have an office visit, but gender was not a significant predictor for the other services. Patients residing in high-minority neighborhoods were about 8% (p<0.001) less likely to receive any services for CAD. Older individuals were significantly less likely (p<0.001) to have invasive procedures (angiography and surgery).
Patients residing in lower-income areas were about
9% (p<0.001) more likely to receive any of the CAD services. Patients with mental health problems were about 45% (p<0.001) less likely to receive any
CAD services. Additionally, CAD-related care varied significantly by state of residence and urban versus rural location. Members with policies that covered more out-of-pocket costs were more likely
to receive an office visit, but policy type was not a significant predictor for invasive CAD procedures.
Conclusion: Disparities in CAD-related care existed by age, income, and race, but the magnitude was relatively small (about 10%). Larger disparities were found by residential location and for those with mental health problems. UnitedHealth
Group is designing interventions to address these disparities; such interventions will begin in mid-
2009.
Implications for Policy, Delivery or Practice:
Disparities related to the existence of mental health problems, and those related to location and other health problems, deserve attention.
Funding Source(s): UnitedHealth Group
Theme: Disparities
● Integrated Disease, Case & Depression
Management for Those with Medicare
Supplement Coverage
Kevin Hawkins, Ph.D.; Kamisha, Escoto, Ph.D.;
Ronald Ozminkowski, Ph.D.; Cynthia Hommer,
M.S.W., L.I.C.S.W.; Cynthia Barnowski, R.N.,
M.B.A.; Richard Migliori, M.D.
Presented by: Kevin Hawkins, Ph.D., Senior
Research Director, Ingenix Consulting, Ingenix, 137
Bedford, Brooklyn, MI 49230, Phone: (517) 414-
4276; Email: kevin.hawkins@ingenixconsulting.com
Research Objective: To better understand depression among the elderly, highlight the need for depression management, discuss disease management programs for Medicare beneficiaries, and provide details about new initiatives being developed for those with Medicare Supplement
(i.e., Medigap) coverage.
Study Design: The literature was reviewed to summarize and highlight major issues in diagnosing and treating depression in older adults. We discuss disease management programs available for
Medicare beneficiaries, noting the lack of care management (and especially depression management) available for beneficiaries with supplement coverage. Finally, we describe a new integrated care management and depression program for those with Medigap coverage.
Population Studied: We conducted a literature search using PubMed, looking for peer-reviewed articles pertaining to disease management, care management, and depression in the elderly.
Principal Findings: Depression is a substantial and increasingly recognized public burden among older adults and is commonly present among those who also suffer from other serious chronic illnesses.
While comprehensive depression management programs have been shown to improve outcomes in academic trials, systematic integration of depression management into clinical programs has not occurred widely for beneficiaries in Medicare with or without Medigap coverage.
Conclusion: The new integrated disease management, case management, and depression management program developed by UnitedHealth
Group will address the critical needs of older
Americans suffering from chronic illnesses and depression.
Implications for Policy, Delivery or Practice:
This is the first known effort to improve the health of
Medigap patients via disease and care management.
Funding Source(s): UnitedHealth Group
Theme: Prevention and Treatment of Chronic
Illness
● Interventions to Improve the Quality of
Ambulatory Care in Aligning Forces for Quality
Communities
Seiji Hayashi, M.D., M.P.H.; Christal Ramos,
M.P.H.; Janet Pagan-Sutton, Ph.D.; Marsha
Regenstein, Ph.D.; Mariza Hardin, M.P.H.
Presented by: Seiji Hayashi, M.D., M.P.H.,
Assistant Research Professor, Health Policy, The
George Washington University, 2021 K Street, NW,
Suite 800, Washington, DC 20006, Phone: (202)
416-0774; Email: sphash@gwumc.edu
Research Objective: The primary purpose of this study is to identify common models of ambulatory care quality improvement that are being implemented in various communities in the United
States. Understanding of these models will assist communities of all sizes to recognize options for advancing ambulatory quality.
Study Design: Information gathered for this study, including descriptions of the design of the initiative, key measures collected, summary of the impact of the initiative and barriers to implementation, was collected through key informant interviews.
Interviews were conducted with members of the
Aligning Forces for Quality Leadership Teams, a
Robert Wood Johnson Foundation initiative to improve the quality of care in selected communities.
Individuals interviewed consisted of hospital, physician and nursing leaders, representatives of consumer organizations, public and private payers, and other organizations committed to improving the quality of care in their community. A snowball technique was used to identify quality improvement initiatives as well as individuals who could provide the most comprehensive information about these initiatives.
Population Studied: The focus of this study was the health care system within the different communities and the major ambulatory health care quality stakeholders that include consumers, providers, employers/purchasers, and payers/health plans. To the extent that they are also engaged in the development and implementation of quality improvement initiatives directed at enhancing the quality of ambulatory care, organizations outside of these stakeholders were also examined.
Principal Findings: Three dominant models emerged from the dozens that were identified.
These included models designed to promote the use of health information technology (e.g., electronic medical records adoption, health information exchange, public reporting of measures), models of provider education that use group learning and peer-to-peer techniques, and quality improvement practice models (e.g., medical homes, chronic care model). Most of the models identified were led by multi-stakeholder organizations with the potential to effect change in provider practices of all sizes – from practices with one physician to those with several hundred physicians. The majority of models identified were led by multi-stakeholder organizations with a moderately high level of integration.
Conclusion: Fueled largely by multi-stakeholder organizations, communities across the country are effectively meeting identical quality objectives using approaches that have been customized to meet local needs.
Implications for Policy, Delivery or Practice:
Regional quality improvement initiatives sponsored by multiple stakeholder groups have the potential to realize national health policy goals by affecting change in the delivery of health services and fostering consumer engagement. This study represents a first step in identifying effective and efficient ambulatory quality improvement initiatives.
Findings are expected to assist communities throughout the country to identify and develop models to meet regional and national ambulatory quality improvement objectives.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Organized
Processes
● Medicare Race/Ethnicity Data Quality: A VA
Medicare Comparative Study
Thomas Haywood, M.P.H.; Elizabeth Tarlov, Ph.D.;
Kevin Stroupe, Ph.D.; Arika Owens, M.P.H.;
Thomas Weichle, M.S.; Denise Hynes, Ph.D.
Presented by: Thomas Haywood, M.P.H., Social
Science Analyst, VA Information Resource Center,
Veterans Health Administration, 5000 South 5th
Avenue, Hines, IL 60141, Phone: (773) 202-8387;
Email: thomas.haywood@va.gov
Research Objective: Race-based healthcare disparities among Medicare enrollees is receiving increased attention from policymakers, program administrators, and other stakeholders. Medicare administrative and claims data, including information on enrollees’ race/ethnicity, are used by the Centers for Medicare and Medicaid Services
(CMS) and other researchers to evaluate disparities and their outcomes. In the Veterans Health
Administration, researchers often use race information obtained from Medicare records to supplement VA self-reported race data when patient race information is missing in the VA record.
Since information about the concordance between
Medicare and VA race data is lacking, implications of this practice for race data quality in and results of
VA disparities studies are not known. This analysis investigated the quality of Medicare race/ethnicity data using self-reported VA data as the comparison.
Study Design: As part of a larger study for which we selected a random sample of patients who received VA healthcare between 1997 and 2005, we focused in this analysis on those who had used
VA health services in 2004 and/or 2005, were age
65 or older, and had non-missing race and ethnicity values in VA data. Race/ethnicity information in the linked Medicare record was compared to selfreported VA data. We calculated sensitivities, specificities, and positive and negative predictive values, using Medicare data as the “test” against self-reported VA race/ethnicity data, as the gold standard. The kappa statistic evaluated agreement between the two data sources.
Population Studied: The study population comprised Medicare enrollees age 65 and older who had used VA healthcare during fiscal years
2004-to-2005.
Principal Findings: The sample included 141,409 individuals. Among those self-identified as Non-
Hispanic White (NHW) or Non-Hispanic Black
(NHB) in VA data, 98% and 95%, respectively, had concordant values in the Medicare record. For the
7,347 Non-Hispanic Asian (NHA), Non-Hispanic
American Indian (NHAI), and Hispanic (any race) patients, the Medicare data had sensitivities of
53%, 25%, and 25%, respectively. The Medicare record correctly excluded non-NHWs from that category 78% of the time, while specificities for the other race groups was 99% or higher. Kappa statistics ranged from 0.31 for NHAIs to 0.94 for
NHBs. VA minorities who were misclassified in
Medicare were most often identified as NHW except for Asians, who were most often identified as
“Other”.
Conclusion: Using self-reported race data from the
VA as a comparison, Medicare data were shown to have high sensitivity and overall agreement for patients who were NHW or NHB. In contrast,
Medicare race data under-identified non-Black minorities, misclassifying them most often as NHW or “Other”.
Implications for Policy, Delivery or Practice: Our analysis confirms results of other studies, demonstrating that researchers and policymakers can rely on Medicare data to identify NHW and
NHB but not individuals from non-black minority groups. When using Medicare race data to supplement missing data in other databases such as the VA, the most accurate categorization and comparisons will be obtained by using a dichotomous NHB / “Other Race” classification scheme. CMS’ recent development and implementation of an imputation algorithm for
improving the identification of Hispanic and Asian or
Pacific Islander beneficiaries in Medicare data shows promise for resolving this limitation.
Funding Source(s): VA
● National Trends of Physician Assistant
Workforce, 1980 to 2007
Xiaoxing He, M.D., M.P.H.; Ellen Cyran, M.S.; Mark
Salling, Ph.D.
Presented by: Xiaoxing He, M.D., M.P.H.,
Assistant Professor, Health Sciences, Cleveland
State University, 2121 Euclid Avenue, HS 122,
Cleveland, OH 44115, Phone: (216) 687-5447;
Email: xiaoxing.he@jhsph.edu
Research Objective: To examine the structure of nation’s healthcare workforce, with a focus on identifying national trends of physician assistant
(PA) workforce among American communities.
Study Design: Population data from the 1980 to
2007 Census Public Use Microdata Sample
(PUMS) were used to analyze how the demographic distribution of PA workforce has changed over time, and to describe PA-topopulation relationships nationwide. An interactive map was developed to provide an intuitive graphical display of the data. Analysis was supplemented with the Occupational Employment Statistics (OES) from U.S. Department of Labor. All analyses were adjusted for the complex census design and analytical weights provided by the Census Bureau.
Population Studied: Nationally-representative sample in all 3,141 counties (or county equivalents) in the U.S.
Principal Findings: Preliminary analysis indicates wide variation among states in the numbers of PAs and hourly wages. In 1980, California employed the highest numbers of PAs (3,120), and North Dakota,
Vermont, and Wyoming employed the lowest numbers of PAs (20). This has been changing over time. In 2007, New York employed the greatest numbers of PAs (9,010), and North Dakota stayed at the bottom (106). The 2007 salary profile shows that Connecticut’s PAs earned the highest hourly mean wages ($43.8), and the lowest were $20.3 in
Mississippi. An increase in the total numbers of
PAs, especially females, was identified. In 1980, there were about 29,120 PAs, of which, over 63.5% were males. By contrast, there was an estimated total of 97,721 PA and more than 66% were females in 2007. In 1980, Nevada had the highest rates (40), and North Dakota had the lowest rates
(3.1) of PAs per 100,000 individuals. The corresponding rates in 2007 were 84.7 at New
Hampshire, and 10.4 at Mississippi. Additionally, the levels of PA education have increased from less than 21% of PAs with four or more years of college in 1980, to more than 65% with four or more years of college in 2007. Furthermore, there was an increase in minority (Non-White) PA. While less than 17% PA were minority in 1980, this figure raised to 23% in 2007. Nevertheless, the PA workforce is aging. While nearly 70% of PAs were less than 35 years old in 1980, this percentage down to 38% in 2007.
Conclusion: Among American communities, a trend of increase in PA workforce, especially the increase of female PA workforce was identified nationwide. However, geographic variation exists among states in both the rates of PA workforce per
100,000 population and the rates of hourly or annual wages. The levels of education, the percentage of minority, and the aging of PA workforce have increased steadily over time.
Implications for Policy, Delivery or Practice:
There is tremendous variability across the nation in terms of the estimated PA workforce meeting the population’s primary care needs. Further research will identify the potentially underserved areas, the strategies in practice development, as well as how the current PA supply-demand chain impacts clinical practice and cost outcomes for primary care populations.
Funding Source(s):
Theme: Health Care Workforce
● The Role of Identity, Trust & Place in Access to Pediatric Primary Care among Poor, Urban
Adolescent Mothers
Jennifer Hebert-Beirne, Ph.D., M.P.H.; Michele
Kelley, Sc.D., M.S.W.
Presented by: Jennifer Hebert-Beirne, Ph.D.,
M.P.H., Director of Research, Women's Health
Foundation, 632 West Deming Place, Chicago, IL
60614, Phone: (773) 305-8206; Email: jennifer@womenshealthfoundation.org
Research Objective: Despite the presence of multiple safety net providers, disparities in access to care exist for disenfranchised, special populations such as adolescent mothers and their children. Consumer perceptions of the health care system may play a role in these disparities as they shape health seeking behavior and health care utilization. The aims of this study were: (1) To describe adolescent mothers’ perceptions of and experiences with pediatric primary care within a community context ( 2) To identify the possible influences of community context and sense of place on navigating resources for child health care.
Study Design: An exploratory, qualitative study using semi-structured interviews and two group discussions was conducted.
Population Studied: Adolescent mothers were recruited from a small, alternative educational program for parenting teens in a multicultural, medically underserved, urban community with a significant health safety net. All 21 mothers who were students at the time agreed to be interview participants. The participants’ ethnic breakdown was: 8 Mexican; 7 Puerto Rican; 1 “other” Hispanic cultural groups; and 5 African-American.
Principal Findings: Through maternal narratives of health care utilization, perspectives of the health care system were discerned suggesting findings with respect to identity, trust and place. Negative interactions with health care providers where the young mothers experienced an awareness of social distance and limited power, threatened these women’s sense of maternal and social identity.
Dimensions of distrust and skepticism of the health care system emerged in the data, likely contributing to a strong reliance on maternal social networks for health advice with selective adherence to provider recommendations. Place played a significant role in health-seeking behavior as adverse community influences including poverty, gangs and rapid sociodemographic change undermined the establishment of a pediatric medical home. Identity threats, distrust and community context seemed to influence erratic patterns of health care utilization including deliberate, consistent use of multiple providers
(“stand alone” clinics, E.Rs.), and use of providers outside of the community.
Conclusion: Despite the relative availability and affordability of local heath care services, poor adolescent mothers lacked the recommended medical home for their children and were largely disengaged with the health care system, as evidenced by variable utilization and significant disinterest in provider recommendations.
Nevertheless, the young women exhibited agency in their determination to make sense out of a disarray of health services and varied experiences.
Their rich narratives provided insights into possible influences on their health-seeking behaviors, including deliberate use of disparate medical care resources and lay consultation - information unavailable from routine public health systems data.
Place influences access. Future research should further investigate place effects on access to care, as well as how issues of identity and trust are especially significant for adolescent mothers with regard to provider interactions, access and choice.
Implications for Policy, Delivery or Practice:
Public health strategies to ensure a relevant medical home for all children need to be attentive to consumer’s developmental needs, culture, community and life experiences. Community-level education is needed to increase family and peer health literacy and health care system knowledge.
Cultural sensitivity of community providers must be improved, with community involvement in local reforms.
● Public vs. Private Hospitals & Hospital
Performance: Correlation Versus Causation
Guenther Heller, Ph.D., M.D.; Svetlana Rasch;
Antje Hammer; Christoph Wagner; Christian
Günster; Holger Pfaff, Ph.D.
Presented by: Guenther Heller, Ph.D., M.D.,
Health Service Researcher, Research Institute of the Local Healthcare Insurance (WIdO),
Rosenthaler Street 31, Berlin, 10178, Germany,
Phone: +49 30 34646 2121; Email: guenther.heller@wido.bv.aok.de
Research Objective: Numerous studies have addressed the topic whether private hospitals provide different quality of care compared with nonprivate. However, only few analyses have analyzed whether hospital privatisation is followed by a increased or decreased quality of care, which could be expected when thinking of a causal effect, or whether hospital performance of private hospitals is different from the start with no clear trend after privatisation (correlation). Thus, we analysed whether hospital performance is different for private
(for profit) hospitals and if different trends in
Hospital performance can observed for private vs. non-private Hospitals in a longitudinal analyses after privatization.
Study Design: We performed trend analyses using nationwide hospital reimbursement data describing
Hospital episodes from patients discharged in 2003-
2007 who were insured within Germany’s largest health insurance (AOK). Risk adjusted mortality 30day-mortality, 90-day-mortality and 1-year-mortality were analysed using an adoption of the Charlson
Score as well as the Elixhauser Score to our data
(Med Care 2005;43:1073-1077). Estimates were derived using logistic regression analyses and are expressed as Standardized Mortality Ratios
(SMRs).
Population Studied: The AOK covers health insurance for approximately. 30 % of the German
Population. Altogether 27,2 Million hospital cases were analyzed
Principal Findings: 85% of patients were treated in hospitals with non-private provider status for the entire study period. SMRs in this group were only slightly above the average (e.g. 90-Day Elixhauser adjusted SMR = 1.01 for all years). In contrast for hospitals, which were private during the entire study period, SMRs were decreased (e.g. 90-Day
Elixhauser adjusted SMR = 0.94 in 2003), without showing an clear upward or downward secular trend. For hospitals privatized after 2003, SMRs were closer to the risk adjusted average. Again no clear trends could be observed.
Conclusion: Our results may be interpreted in a way that the observed association between type of hospital provider and hospital performance measured here is rather correlational than causal.
Implications for Policy, Delivery or Practice: It may be suspected that the used risk adjustment method for global mortality indicators alone is not able to fully adjust for case mix differences between private vs. non private hospitals. Thus, further analyses should aim to explain the observed differences in order to obtain improved risk adjusted hospital comparisons in the future. Meanwhile, our analyses imply, that hospital comparisons should not be based on risk adjusted global mortality indicators alone.
Funding Source(s): Federal Board of Physicians,
Germany
Theme: Quality and Efficiency: Measurement
● Moving Care from Hospital to Home: The
Challenges of Implementing Remote Care
Services
Jane Hendy, Ph.D.; James Barlow, Ph.D.
Presented by: Jane Hendy, Ph.D., Associate
Professor/ Research Fellow, Health Management
Group, Imperial College Business School, South
Kensington, SW7 2AZ, UK, Phone: +44 207 594
5935; Email: j.hendy@imperial.ac.uk
Research Objective: A key agenda for governments worldwide is how best to care for older people, whilst limiting the demand for expensive hospital beds. One solution is the development of technologies that support care remotely. Remote care allows patients with chronic conditions to monitor their health at home. Faith in the potential of this technology has led to the UK to take a world lead. Approximately £150 million has been allocated by the UK government to introduce remote care into 180,000 homes during 2006 –
2011. Our aim is research the organisational implementation of this innovative method of care delivery.
Study Design: We employed a variety of ethnographic methods to study service implementation. Over two years we attended project and strategic meetings and observed the projects (51 hours of observations). We analysed over 20 project documents and conducted interviews (n = 115).
Population Studied: We identified five case study sites in England most likely to be able to implement a mainstream remote patient care service. Cases were Local Authorities and their associated Primary
Care Trusts. Inclusion criteria included having at least two years prior experience of running remote care trials, previous evaluation of remote care schemes, an organisational commitment to mainstream remote care services over the next two years, and having a dedicated remote care project manager. Participants were senior managers strategically involved in implementation and included executive officers, project managers, directors of services, commissioners and senior nurse practitioners.
Principal Findings: Three of our 5 case studies moved forward in developing sustainable remote care services, with greatly increased numbers of remote care users. Two case studies did not progress and had decreased numbers by end of the study (Jun. 2006-Jun. 2008).
Conclusion: At an individual level, we found good support for remote care as a new model of service delivery. The hurdles to delivering this innovation were at a system level; ensuring that remote care could be organisationally integrated into existing health and social care services. The successful cases moved from project-based initiatives to integrated mainstream services with senior level leadership that supported a culture of innovation, consistent marketing of all relevant stakeholders and with robust evaluation. Divides were dismantled by moving away from project based strategies and thinking, towards an inclusive organisation-wide roll-out, that sought to position remote care as
‘normal business’ from inception.
Implications for Policy, Delivery or Practice: The mixed picture our case studies present, with two of the five experiencing a decline in the number of remote care users, suggest more research is needed if we are going to address the organisational challenges of mainstreaming remote care services. Research is also needed on how to stimulate uptake by using existing levers within the care system for payment and reimbursement, and service commissioning. However, a central government policy shift from the current situation where remote care services are optional to one where they are an integral part of a care package, unless there is good reason for exclusion, would also do a great deal to smooth the way forward.
Funding Source(s): EPSRC
Theme: Long Term Care
● Regranex (Becaplermin) use in Patients with
Non-Healing Lower Extremity Ulcers: A
Population-Based Cohort Study
Velma Henry, B.A.; Jeffery Talbert, Ph.D.
Presented by: Velma Henry, B.A., Pharmacy
Intern, Pharmacy Practice & Science, University of
Kentucky College of Pharmacy, 1245 Centre
Parkway, Apartment 617, Lexington, KY 40517,
Phone: (859) 227-7554; Email: vellhenri@uky.edu
Research Objective: To evaluate the incidence of cancer in Regranex (becaplermin) users with nonhealing diabetic ulcers compared to matched comparators. To determine the mortality rate from malignancies, and to quantify the absolute risk.
Study Design: A retrospective cohort study was performed using Kentucky Medicaid recipients diagnosed with uncontrolled diabetes (ICD-9
250.X1, 250.X2) during the period January 1, 1998 to December 31, 2006. Patient demographic data, diagnostic data, and prescription claim files were obtained from the administrative files of the
Kentucky Medicaid program.
Population Studied: Cases were identified as (1)
Medicaid recipients with continuous eligibility, (2) recipients 16-85 years with at least one reimbursement claims record for the drug regranex® within the period 1998-2006 and (3) had claims data available for at least 1 year before and
1 year after the initial regranex reimbursement claims record. Study subjects were excluded if they
(1) had a prior history of cancer and (2) had missing demographic data. Each case was matched by age,
sex and cancer history with five controls with a diagnosis of uncontrolled diabetes and no reimbursement claims record for the drug regranex.
Cancer incidence was determined by insurance claims for malignant neoplasms (ICD-9 140-208.9).
Cancer mortality was identified by matching study subjects with individuals in the rosters of the
National Death Index (NDI).
Principal Findings: The medical histories of the
707 regranex users and the 3,510 matched comparators were not similar. The regranex group had a higher incidence of all associated comorbidities. There were 83 confirmed cancers among the regranex users and 744 confirmed cancers among the comparators. (RR=0.55, 95%
CI=0.45-0.68). Analysis indicated no elevation in risk of cancer with 1 regranex dispensing or =3 regranex dispensing compared to those with none
(RR=0.19, 95% CI=0.13-0.28 and RR=0.24, 95%
CI=0.17-0.33) respectively. The average time for the development of cancer with regranex use was
39 months with incidences as early as three months post treatment. The history of amputation in the regranex group was 12 cases per 100 and comparators 2 cases per 100.
Conclusion: The results of this study suggest no relationship between regranex exposure and cancer incidence. It is possible that regranex may be reserved for patients that already have complicated diabetes and have already had one or more amputations. In practice regranex may be used as a last resort when all else has failed. Further research on the cancer mortality rate is required.
Implications for Policy, Delivery or Practice: A possible policy implication of this project would be for medical institutions to reassess the use of regranex and initiate it earlier in the treatment of non-healing diabetic foot ulcers. This study suggest no added risks of cancer with this drug and regranex may add the benefit of lower amputation rates, and hence lower overall treatment costs.
Funding Source(s): University of Kentucky:
Pharmacy Practice and Science
● One Size Does Not Fit All: Understanding
Hispanic Subgroup Differences in Health Care
Access & Quality in NYC
Susan Hernandez, M.P.A.; Michelle Doty, Ph.D.;
Anne Beal, M.D., M.P.H.
Presented by: Susan Hernandez, M.P.A., Program
Associate, Health Care Disparities, The
Commonwealth Fund, One East 75th Street, New
York, NY 10021, Phone: (212) 606-3861; Email: seh@cmwf.org
Research Objective: Hispanics are a heterogeneous group several of countries.
Differences in socioeconomic status, acculturation, foreign-born status, and access to insurance affect their ability to access high quality health care. By aggregating Hispanics into one homogenous group, previous studies have masked the diversity that exists among Hispanic subgroups and the different factors affecting subgroup variations in access to care. This study investigates whether Hispanic subgroups in New York City experience differential access to health care and quality of care, and examines what factors could reduce subgroup disparities.
Study Design: This study uses data from the 2004
NYC Community Health Survey; a stratified random sample telephone survey of non-institutionalized adults aged 18 and older who live in a household with a landline telephone in New York City.
Population Studied: We limit the analyses to the
Hispanic population and categorize individuals to the following groups according to their country of origin: Puerto Rican (N= 779), Dominican (N= 513),
Mexican (N=218), and Central and South American
(N=422). The key outcomes include having a regular provider, receiving routine checkups and other preventive health services.
Principal Findings: Compared to other Hispanics,
Mexicans stand out for having lower incomes and lacking insurance coverage or access to a regular provider. More than three-quarters (77%) of
Mexicans were uninsured during the year, but uninsured rates are still high among Central-South
Americans (42%), Dominicans (35%), and Puerto
Ricans (23%) relative to uninsured rates for white
New Yorkers (16%). Only 34% of Mexicans have a regular provider compared to majorities of Puerto
Ricans (79%), Dominicans (65%), and
Central/South Americans (65%). Mexicans have the lowest rates of receiving preventive health care such as routine annual check-ups and Pap Smears.
When they have a regular provider and insurance, however, most differences between this group and other Hispanics in receiving these services are eliminated. However, Mexicans are less likely to have a regular provider after controlling for insurance.
Conclusion: Despite NYC’s extensive network of safety net providers, and the availability of free or low cost services, Hispanics continue to face barriers to obtaining health care services, particularly Mexican Hispanics.
Implications for Policy, Delivery or Practice:
Given that having a regular provider and insurance are important mediators for receiving preventive health care services, programs targeting Mexican
Hispanics that connect them to a regular provider would help improve the delivery of preventive health to a very underserved population in New York City.
In addition, although insurance coverage attenuates disparities in accessing a regular provider, coverage alone is not enough in eliminating the barriers to accessing a regular provider. Health reform proposals for expanding coverage will not be enough to eliminate Hispanic disparities without considering other, non-financial, barriers to accessing care.
Funding Source(s): CWF
Theme: Disparities
● Laparoscopic Versus Open Gastric Bypass
Surgery: Are There Differences in Patient
Demographics, Safety & Outcomes?
Tina Hernandez-Boussard, Ph.D., M.P.H.; Gaurav
Banka, B.S.; John Morton, M.D., M.P.D., F.A.C.S.
Presented by: Tina Hernandez-Boussard, Ph.D.,
M.P.H., Biostatistician, Department of Surgery,
Stanford University, 300 Pasteur Drive, Stanford,
CA 94305, Phone: (650) 725-5507; Email: boussard@stanford.edu
Research Objective: The objective of this study was to determine differences between laparoscopic and open gastric bypass surgery for patient outcomes using nationally representative data and
Patient Safety Indicators to identify preventable adverse events.
Study Design: The Nationwide Inpatient Sample
(NIS) was queried for all patients undergoing open gastric bypass (ICD9 procedure codes 44.31 and
44.39) and laparoscopic gastric bypass (ICD9 procedure codes 44.38). Both demographic and outcomes variables were compared by either t-test or chi-square analysis. Logistic regression analyses indicated potential predictors of gastric bypass inpatient mortality and complications after gastric bypass surgery.
Population Studied: In 2005 and 2006 NIS reported 34,380 gastric bypass admissions, with
71% laparoscopic.
Principal Findings: In this observational study, laparoscopic bypass was more commonly performed than open bypass (71.6% vs. 28.4%, p<.0001). The majority of patients were white
(75.8%) and female (82.5%). There were no significant differences in severity of illness between open vs laparoscopic bypass patients. A higher percentage of open compared to laparoscopic patients were Medicare (8.9% vs. 7.4%) and
Medicaid (10.4% vs. 5.9%) beneficiaries, p <.01.
Open gastric bypass cases were less likely to be performed at high volume hospitals (53 vs. 61%, p<.001). A greater proportion of open gastric bypass patients were discharged with non-routine dispositions (7.7 vs. 2.4%, p=.0001), died ( .2% vs.1%, p=.04), and had one or more complications
(18 vs. 12.3%, p<.0001). All selected Patient Safety
Indicator rates were higher for open compared to laparoscopic gastric bypass. Open gastric bypass patients also had longer average lengths of stay
(3.49 vs. 2.44, p<.0001) and average total charges
($34,012 vs. $31,546, p<.05) than laparoscopic gastric bypass patients.
Conclusion: There remain large sociodemographic differences between those qualifying for and those receiving laparoscopic bariatric surgery. Medicare and Medicaid patients were more commonly accessing open gastric bypass, which had higher rates of adverse events, as exemplified by
PSIs,than laparoscopic gastric bypass surgery.
Implications for Policy, Delivery or Practice: In our study, laparoscopic gastric bypass patients had improved outcomes compared to open gastric bypass, regardless of patient and hospital demographics. These results help demonstrate the differences between low volume and high volume hospitals can be due to the surgical approach, not strictly the volume effect.
Funding Source(s): Stanford University
Theme: Obesity Prevention and Treatment
● Medicare Beneficiaries Have Improved
Bariatric Surgery Outcomes Following the
Medicare National Coverage Expansion
Tina Hernandez-Boussard, Ph.D., M.P.H.; John
Morton, M.D., M.P.H., F.A.C.S.
Presented by: Tina Hernandez-Boussard, Ph.D.,
M.P.H., Biostatistician, Department of Surgery,
Stanford University, 300 Pasteur Drive, Stanford,
CA 94305, Phone: (650) 725-5507; Email: boussard@stanford.edu
Research Objective: To assess the potential implications of the Centers for Medicare and
Medicaid Services expanded coverage of bariatric surgery, we compared outcomes of Medicare beneficiaries undergoing bariatric surgery 1 year before and after the CMS expanded coverage.
Study Design: We performed an observational study using the Nationwide Inpatient Sample (NIS) database.
Population Studied: Bariatric surgeries in
Medicare beneficiaries were identified using ICD-9 codes 44.31, 44.39, 44.38, 44.95, and 46.68 and
DRG code 278 for obesity in the NIS between May
2004 to December 2006, 1 year before and after
CMS implemented expanded coverage for bariatric surgery for Medicare procedures.
Principal Findings: 10,789 bariatric procedures were performed on all Medicare patients, 6,439 prior and 4,350 after the CMS expansion. This distribution of these cases changed, with the number of total cases decreasing by 30%. Patients undergoing bariatric surgery after the CMS expansion rule benefited from a shorter length of hospital stay (2.9 to 2.6 days) and lower complications (23% to 19%) with no differences in in-hospital mortality. There was a significant decrease in sleep apnea as a diagnosis (37% to
27%, p=.006). There were no significant differences between patient and hospital demographics before and after the decision, although we saw a 2% increase in patients greater than 65 years of age and a 8% increase in laparoscopic cases.
Conclusion: The decision by CMS to expand coverage of bariatric surgery to Medicare beneficiaries resulted in improved outcomes for these patients, however volume did decrease
noting the need to provide appropriate access to care.
Implications for Policy, Delivery or Practice:
Center of Excellence designation by CMS leads to improved outcomes. However, access to care needs to be monitored.
Funding Source(s): Stanford University
Theme: Obesity Prevention and Treatment
● Emerging Federal & State Activities on the
Collection of Data to Measure Racial & Ethnic
Disparities in Health
Patricia Higgins, Ph.D., M.P.H.; Erin Fries Taylor,
Ph.D.
Presented by: Patricia Higgins, Ph.D., M.P.H.,
Researcher, Health, Mathematica Policy Research,
1050 Sherman Street, Apartment 409, Denver, CO
80203, Phone: (617) 953-0590; Email: thiggins@mathematica-mpr.com
Research Objective: Interest in racial and ethnic health care disparities has grown substantially in the past decade, and this topic is increasingly discussed as a key policy issue. However, a lack of valid and reliable data continues to hamper collective understanding of disparities. Data collection policies and regulations across public and private entities are largely uncoordinated and nonstandardized, though pockets of activity have begun to emerge. We profile several recent and emerging national and state policies related to racial, ethnic, and primary language (r/e/l) data collection and describe how these policies may shape the efforts of public and private organizations to reduce disparities.
Study Design: Qualitative case study including document review and interviews with over a dozen key informants, including national experts and staff from state government agencies and private organizations.
Population Studied: We interviewed staff from federal, state and private organizations.
Massachusetts, California, and New Jersey-forerunners in policy and regulatory activity related to r/e/l data collection--were a special focus of this case study.
Principal Findings: While a few new and proposed pieces of federal legislation may affect r/e/l data collection, the most significant policy and regulatory activity in this area is occurring at the state level.
Despite increasing attention to measuring disparities, few states have enacted significant legislation or regulation in this area. However,
Massachusetts, California, and New Jersey are leading strong statewide initiatives and offer valuable lessons. These states report that having a formal policy or regulation in place provides a necessary incentive for public and private entities to develop methods to collect r/e/l health data.
However, it is also crucial for states to provide leadership and technical assistance to public and private implementers, such as hospitals and health plans. State agencies, advocacy groups, and other organizations skilled in r/e/l data collection methods can facilitate implementation through standardization of r/e/l categories, staff training, and patient education about the importance of collecting these data.
Conclusion: The experiences of Massachusetts,
California, and New Jersey underscore the power of legislation and regulation in influencing health plans, hospitals, and state agencies to collect r/e/l data in a systematic, coordinated fashion. These states’ experiences provide guidance to other states interested in strengthening systems of r/e/l data collection.
Implications for Policy, Delivery or Practice:
Informants emphasized that a standard set of r/e/l categories, promoted by the federal government, would be a valuable national resource. This issue is of particular concern to national health plans, which could face a different set of regulations and standards in each state that proceeds with regulation on r/e/l data. A recently commissioned
IOM study on r/e data collection may provide important recommendations, thereby allowing for a more uniform approach that promotes best practices in r/e data collection and facilitates comparative measurement.
Funding Source(s): AHRQ
Theme: Disparities
● Mental Health Status as a Function of
Racial/Ethnic Difference & Access to Behavioral
Health
Molly Higham, B.S., O.T.R./L, M.S.; Diane Howard,
Ph.D.; Jeff Canar, Ph.D.; Robyn Golden, L.C.S.W.;
Sean Sitzes, L.C.S.W.
Presented by: Molly Higham, B.S., O.T.R./L, M.S.,
Clinical Supervisor Occupational Therapy
Psychiatry, Health Systems Management, Rush
University, 106 North Aberdeen 5G, Chicago, IL
60607, Phone: (312) 375-0946; Email:
Molly_Higham@rush.edu
Research Objective: Health care disparities for racial minorities, as well as those with mental illness have been well documented in the literature.
Health outcomes for blacks in the United States are noted to be poorer for many illnesses and diseases.
While increasing access and use of mental health services remains a national policy issue, black
Americans with mental illness have been noted to have lower rates of effective treatment. Most of the research conducted in the black community does not consider the impact of ethnic differences on the black population. Blacks of Caribbean descent comprise 7% of the black American population.
This study aims to examine the association between racial/ethnic background, access to behavioral health insurance, and mental health status.
Study Design: The study will examine the relationship of independent variables, race/ ethnicity and behavioral health insurance type to the dependent variable, mental health status. The null hypothesis states that there is no association between race/ethnicity and type of behavioral health insurance coverage, and presence of mental illness. Data analysis will include a two tailed Chi-
Square Analysis to determine the association between rac/ethnicity, type of behavioral health insurance coverage, and mental health status.
Logistic regression will be performed to examine the incidence of mental illness in the presence of racial/ethnic groups and access to behavioral health insurance.
Population Studied: The sample population is derived from the National Survey of American Life
(NSAL) database. The NSAL is one of three national studies conducted for the National Institute of Mental Health Collaborative Psychiatric
Epidemiology Surveys initiative. The NSAL used a national probability sample of African American, black of Caribbean descent, and non-Hispanic white households. It employed a geographically distributed targeted population survey approach.
Total sample includes 3,750 African Americans,
1,623 blacks of immediate Caribbean descent, and
1,006 non-Hispanic whites.
Principal Findings: Research in progress.
Findings will be available after March 6,2009.
Conclusion: Final conclusions will be available after March 6,2009.
Implications for Policy, Delivery or Practice:
Increasing access and use of mental health services remains a national policy issue because of the economic and social consequences of untreated mental illness. Improving access to and use rates of mental health services by minority groups may improve the outcome of their illness and decrease the level of functional impairment.
The wealth of research completed about disparities in the black American population group all black ethnic groups together, often missing the subtleties of the impact of these ethnic groups on results. It is important to understand the current pattern of mental health status for racial and ethnic minorities to clarify their mental health needs and barriers.
This study will assist in identifying how race and ethnicity as well as access to behavioral health insurance are associated with mental health status.
Funding Source(s): Departmental
● Public Health & Work-Based Welfare:
Potential Partnership for Reducing Health
Disparities
Eugenie Hildebrandt, Ph.D.
Presented by: Eugenie Hildebrandt, Ph.D.,
Associate Professor, College of Nursing, University of Wisconsin-Milwaukee, P.O. Box, Milwaukee, WI
53201, Phone: (414) 229-5464; Email: hbrandt@uwm.edu
Research Objective: Welfare reform in the United
States has created dramatic change in the lives of mothers living in poverty, and these changes present a community and public health challenge to the nation. Temporary Assistance for Needy
Families (TANF), the work-based welfare program that replaced Aid to Families with Dependent
Children (AFDC), was structured to move people off welfare and into the workforce with basic workrelated skills, but it has failed an alarming number of women. Health status plays a part in this failure but has been largely ignored in this reform. The effect of this major policy shift is magnified by the disparities already present within this population.
This longitudinal study was funded by the National
Institute for Child Health and Human Development
(NICHD), 1 R01 HD 054961-01-A2; data were gathered between 2007 and 2009. The purpose of this mixed method study was to gain in-depth understanding of the health, socio-cultural, and occupational experiences of women who have been unsuccessful at becoming work-ready and employed through work-based welfare.
Study Design: Mixed methods. Narrative interview design was the dominant methodology.
Population Studied: Participants were 40 women who were terminated from TANF after using up their
5-year lifetime limit of TANF cash support.
Principal Findings: Findings suggest there are barriers that deny women sustainable work and limit family self-management. These include personal and family health challenges, limited human capital, unstable lives, weak social supports, and the TANF program structure.
Conclusion: The public health system has a place in developing broad research-based evidence for more effective welfare policy and interventions to reduce health and socio-economic disparities, and improve outcomes.
Implications for Policy, Delivery or Practice: The study provides policy makers and key health care decision makers with information about the health impact of welfare reform policy, and innovative solutions that are health protective of low income women with children and support Healthy People
2010 and Healthy People 2020 Goals for the
Nation.
Funding Source(s): National Institute of Child
Health and Human Development
Theme: Quality and Efficiency: Policies and
Incentives
● Individual Insurance & Access to Care
Steven Hill, Ph.D.; Joel Cohen, Ph.D.
Presented by: Steven Hill, Ph.D., Service Fellow
Economist, Center for Financing, Access & Cost
Trends, Agency for Healthcare Research & Quality,
540 Gaither Road, Rockville, MD 20850, Phone:
(301) 427-1672; Email: shill@ahrq.gov
Research Objective: Evaluate access to care for people with non-employment-related private insurance (“individual insurance”). Do they have similar or worse access to care than those with employment-related insurance? People in poor health have greater needs for care and more experience to support their assessments of care, so we also analyze the subgroup of adults with chronic conditions.
Study Design: The sample is drawn from the nationally representative Medical Expenditure Panel
Survey (MEPS), Panels 6 through 9 (2001-2005).
The measures of access are: having a usual source of care; having any visits; unmet needs for medical care, prescription drugs, and dental care; delays getting medical care, prescription drugs, and dental care; getting needed care right away; getting routine appointments as soon as desired; problems seeing specialists; and problems getting care, tests, or treatment. These access measures refer to the prior 12 months, so we compare nonelderly adults with 12 months of individual insurance (N=785) and those with 12 months of employment-related insurance (N=18,688). For the subgroup analysis, adults with chronic conditions are those reporting at least one chronic condition (arthritis, diabetes, asthma, ischemic heart disease, stroke, obesity, emphysema, and high blood pressure), comprising
401 with individual insurance and 10,233 with employment-related insurance. We estimate regression-adjusted differences controlling for income, attitudes, health behaviors, health status, socioeconomic characteristics, and the availability of primary care providers and specialists. All statistical tests account for the complex survey design of the MEPS.
Population Studied: U.S. civilian, noninstitutionalized, nonelderly privately insured population, 2001-2005.
Principal Findings: Adults with individual and employment-related insurance report similar access to physicians. Very few privately insured adults report unmet needs and delays getting care, but those with individual insurance are more likely to report delays getting medical care and unmet needs for dental care. Adults with chronic conditions are more likely than other adults to report access problems. Compared with adults with chronic conditions and employment-related insurance, those with individual insurance are more likely to report delays getting medical care and unmet needs and delays getting dental care. In addition, among the two-fifths of adults with chronic conditions who report they or a doctor thought they needed to see a specialist, those with individual insurance are more likely to report problems seeing a specialist.
Conclusion: Adults with individual and employment-related insurance report similar, often good, access to care. Access to dental care is a bigger problem for those with individual insurance, because those with individual insurance are much less likely to report they have dental coverage.
Adults with chronic conditions are less likely than others to report good access, but differences by type of insurance are generally not large.
Implications for Policy, Delivery or Practice:
Policymakers who want to use the individual insurance market to expand coverage or change health care financing should consider including efforts to monitor access to care. They may also want to consider ways of increasing access to affordable dental coverage.
Funding Source(s): AHRQ
Theme: Coverage and Access
● Socioeconomic Disparities in Development in
Early Childhood
Marianne Hillemeier, Ph.D.; George Farkas, Ph.D.;
Paul Morgan, Ph.D.; Steven Maczuga, M.S.
Presented by: Marianne Hillemeier, Ph.D.,
Associate Professor, Health Policy &
Administration, Pennsylvania State University, 504S
Ford Building, University Park, PA 16802, Phone:
(814) 863-0873; Email: mmh18@psu.edu
Research Objective: The objective of this research is to quantify patterning of sociodemographic disparities in cognitive functioning in early childhood, and identify predictors of low cognitive functioning.
Study Design: Two-stage, multiple logistic regression analyses examine associations between sociodemographic characteristics and low cognitive functioning at 9, 24, and 48 months, and test whether gestational and birth-related factors mediate these associations.
Population Studied: Data come from 6,037 singleton and 1,248 multiple births in the Early
Childhood Longitudinal Study, Birth Cohort (ECLS-
B).
Principal Findings: Among singletons, persistent poverty and non-white race/ethnic status are not related to risk of cognitive delay at 9 months, but are statistically significant predictors at 24 and 48 months. In contrast, birth conditions—very and moderately low birthweight and very and moderately preterm birth—produce very high odds of low cognitive functioning at 9 months, but these decline by 24 and 48 months such that they are less than or equal to those of socioeconomic variables. The high risks associated with persistent poverty and race/ethnicity at 24 and 48 months occur controlling for gestational age and birthweight. Similar patterns are observed for multiple births.
Conclusion: Sociodemographic disparities in poor cognitive functioning emerge by 24 months and intensify by 48 months, but are not mediated by gestational or birth characteristics.
Implications for Policy, Delivery or Practice:
Further investigation of processes whereby social disadvantage adversely affects development in
early childhood is needed to guide interventions to eliminate disparities.
Theme: Disparities
● Development of a Chronic Disease
Prioritization Aid to Assist Policy & Decision
Makers
Michael Hillmer, Ph.D.; Jeff, Clayman, B.S.
Presented by: Michael Hillmer, Ph.D., Manager,
Health System Strategy Division, Ontario Ministry of
Health & Long-Term Care, 80 Grosvenor Avenue -
8th Floor Hepburn Block, Toronto, M7A 1R3, CA,
Phone: (416) 327-3314; Email: michael.hillmer@ontario.ca
Research Objective: Chronic disease is a major individual and system burden in virtually every healthcare system. Specific practical tools to prioritize investments and attention in chronic disease are rare. We sought to develop an easily understandable tool that could be used as an input into the decision-making process for policy makers.
The tool is based on the likelihood of chronic diseases and their risk factors to be identified, prevented, treated, and their total health system cost.
Study Design: A group of experts ranked (on a scale from 1-9) several highly prevalent chronic diseases (Alzheimer’s & Dementia, Arthritis,
Asthma, COPD, Depression, Diabetes, Heart
Disease & Stroke, Hepatitis C, HIV & AIDS,
Hypertension, Parkinson’s) and risk factors
(including physiological, external environmental, and individual behaviours) based on how likely they could be identified, prevented, and treated. These criteria scores were multiplied to yield a degree-ofcontrol coefficient (DOCC) for each disease and risk factor. The DOCC was weighted (WDOCC) using population attributable risk to estimate the relative ability to affect the trajectory of a chronic disease or risk factor. Chronic diseases with common risk factors were grouped together.
System expenditure data was obtained by defining administrative billing fee codes for each chronic disease and querying acute care, long-term and community care, physician and drug administrative databases. The expenditure results for each chronic disease (all funds in Canadian Dollars) were multiplied by the WDOCC to obtain per service and total expenditure scores to reveal the chronic diseases with the greatest system impact that can be most easily controlled.
Population Studied: Any Ontario resident (the most populous Canadian provicne with a population of aprroximately 11 million) eligible for the provincial health plan who received a service or drug for any of the chronic diseases listed above during
2006/07.
Principal Findings: The Ontario health system provided over 29 million services for the studied chronic diseases at a total health system cost of $
6,181,527,220. The WDOCC for diabetes was highest (520), followed by cardiovascular disease
(426), and HIV/AIDS (410.20). Alzheimer’s &
Dementia (47), and depression (75) had the lowest scores. HIV & AIDS had the highest Weighted Per
Service Expenditure Control Score of 646.84
(calculated by a weighted per Service cost of $1577
* WDOCC of 410). Cardiovascular disease had the highest Total Expenditure Control Score of 1441
(calculated by the Total Disease Cost of $3.4B *
WDOCC of 426), which was over 500% greater than the scores for Diabetes, Arthritis, and Asthma
& COPD (275, 120, and 128, respectively).
Conclusion: The ability to control diabetes and cardiovascular disease was found to be the highest.
The expenditures for cardiovascular disease were overwhelmingly the most controllable amongst all studied chronic diseases.
Implications for Policy, Delivery or Practice:
This method produces easily understandable and actionable information for policymakers. The results can be used to 1) target chronic diseases whose trajectories can be most readily changed, 2) develop indicators which measure the impact of current investments, 3) align investments for maximum value, and 4) map the coefficients of control to system costs allowing the organization to ensure investments are aligned with its policy goals.
Theme: Prevention and Treatment of Chronic
Illness
● Shared Decision-Making in the Primary Care
Visits of African Americans with Depression
Anika Hines, M.P.H.; Bri Ghods, B.S.; Dawna
McGlynn, M.S.W.; Gail Daumit, M.D., M.H.S.; Lisa
Cooper, M.D., M.P.H.
Presented by: Anika Hines, M.P.H., Ph.D.
Candidate, Health Policy & Management, Johns
Hopkins Bloomberg School of Public Health, 6716
Bonnie Ridge Drive, Apt 201, Baltimore, MD 21209,
Phone: (757) 343-3946; Email: ahines@jhsph.edu
Research Objective: The majority of individuals with depression in the United States who seek care receive their mental health care in primary care settings. However, treatment of depression in primary care remains suboptimal, particularly for
African Americans and other ethnic minorities.
Given several studies that suggest the role of patient involvement in improved depression outcomes,in this study we sought to examine the prevalence and extent of shared decision-making
(SDM) communication among African-American patients with depression and their primary care clinicians.
Study Design: We conducted a cross-sectional study of the visits of 76 African American patients with depressive symptoms (2 weeks or more of depressed mood and/or loss of interest within the past year and 1 week or more of depressed mood and/or loss of interest within the past month) who
were seen by 21 primary care clinicians in urban community based practices in Baltimore, Maryland and Wilmington, Delaware. Audiotapes of patient visits were examined by trained coders to assess the following elements of shared decision-making
(SDM) communication: 1) discussion of the clinical issue and nature of decision to be made; 2) discussion of the alternatives; 3) discussion of the pros and cons of the alternatives; 4) assessment of the patients’ understanding; and 5) asking the patient to express a preference or discussion of the patient’s role in decision-making. We also examined whether clinicians assessed the following patientcentered elements related to depression care for
African Americans: 1) lifestyle and coping strategies, (spirituality, social support); 2) patients’ knowledge and cultural beliefs with regard to depression; and 3) treatment concerns (side effects, addictive potential of medications, social stigma, mistrust of health care). We assessed interrater agreement on a 10% random sample of audiotapes.
Population Studied: African-Americans with depressive symptoms seen by primary care clinicians
Principal Findings: Patients' mean age is 47 years; 74% are women; 51% have at least high school education, 61% are employed, mean annual household income is $44,283, and 71% have private health insurance, while 15% are on
Medicaid. The clinicians have an average age of
44.8 years; 59% are women; 5 (24%) are African-
American, 6 (29%) are Asian, and 10 (48%) are
Caucasian, 71% are general internists, 71% are board certified, and the average number of years at their current practice is 8.5 years. Fifty-six (74%) of the visits contained at least one clinical decision about depression. There were 169 decisions related to depression, and the average number of decisions per encounter was 2.2. Of these, counseling
(32%), medication (30%), and watchful waiting
(28%) decisions each comprised approximately one third of decisions. The inter-rater agreement was
0.83. While 82% of decisions qualified as having
“minimally acceptable” SDM (including elements 1 and 5), only 9% of decisions qualified as having
“complete” SDM (including elements 1, 2 and 3), and only 6% percent of decisions addressed all 5
SDM elements. The frequency of decision-making components addressed varied with the “nature of the decision” and “patient role/preference” being present in 98% and 97% of the decisions, respectively. Alternatives were discussed in 48% of the decisions, while patient understanding was discussed in only 11%. Presence of patientcentered components was even lower with lifestyle and coping strategies being brought up in 25% of decisions, treatment concerns in 18% of decisions, and knowledge and cultural beliefs in only 6% of decisions.
Conclusion: This study shows that the prevalence and extent of shared decision-making communication about depression among African
American patients and their primary care clinicians is low and often excludes patient-centered features.
Therefore, patients’ concerns and preferences may not be adequately addressed in these encounters.
Implications for Policy, Delivery or Practice:
Inadequate shared decision-making communication may partially explain the lower observed rates of appropriate treatment and functional recovery among African American primary care patients with depression.
Funding Source(s): Foundation for Informed
Medical Decision-making, AHRQ
Theme: Behavioral Health
● Preliminary Findings from the 2008 National
Healthcare Quality & Disparities Reports
Karen Ho, M.H.S.; Ernest Moy, M.D., M.P.H.; Amy
Galifianakis, M.P.H.; William Freeman, M.P.H.
Presented by: Karen Ho, M.H.S., Health Sciences
Administrator, Center for Quality Improvement &
Patient Safety, Agency for Healthcare Research &
Quality, 540 Gaither Road, Rockville, MD 20814,
Phone: (301) 427-1342; Email: karen.ho@ahrq.hhs.gov
Research Objective: To present key themes and highlight findings from the 6th annual U.S. national reports on health care quality and disparities.
Study Design: Preliminary data come from the
2008 National Healthcare Quality Report (NHQR) and National Healthcare Disparities Report
(NHDR), which compiled national estimates on over
200 measures of health care quality from more than
35 different data sources. Most of these are nationally representative. Measures cover the quality dimensions of effectiveness, safety, timeliness, and patient centeredness and include measures representative of care for cancer, diabetes, end stage renal disease, HIV and AIDS, heart disease, mental health, maternal and child health, nursing home and home health, respiratory disease and patient safety. Trends are assessed using the geometric rate of change where the average annual rate of change is at least 1% per year and statistically significant with p<0.05.
Analyses in the NHDR focus on 22 selected core measures from the full measure set and use comparisons of whites, blacks, Asians, American
Indians and Alaska Natives and Hispanics. For each core measure, disparities are measured in both absolute and relative terms. Disparities for a baseline year and for the most current year of data are compared to assess trends in disparities.
Population Studied: Most measures cover the civilian, non-institutionalized U.S. population. Some measures are appropriate for specific groups defined by age and gender. Nursing home measures are specific to residents of these facilities.
Principal Findings: Health care quality continues to improve at a slow pace. Disparities persist in health care quality and access. In particular, data show that patient safety, colorectal cancer and mental health care are areas in health care that are in need of improvement. Overall, measures of patient safety showed a decline in improvement (-
0.9% per year). Measures include postoperative care events such as pneumonia, UTIs, and venous thromboembolic events, appropriate timing of antibiotics, adverse events associated with central venous catheters, adverse drug events, deaths following complications of care and inappropriate medications for the elderly. Screening for colorectal cancer remains low. Only about 55.5% of adults age 50 and over ever received a sigmoidoscopy, colonoscopy, or proctoscopy or an FOBT (NHIS,
2005). There are also significant racial and ethnic disparities in colorectal cancer screenings. Blacks,
Asians, American Indian/Alaska Natives and
Hispanics were less likely than Whites to ever receive screenings. Mental health care is also poor.
For example, nearly 30% of adults with mood, anxiety, or impulse control disorders had minimally adequate treatment (CPES, 2003). There were significant racial and ethnic and education disparities— among adults with mood, anxiety or impulse control disorders, only 9% of Hispanics, and 23% of Blacks had minimally adequate treatment, and individuals with less than high school education and high school graduates were less likely to have minimally adequate treatment.
Conclusion: Improvement in health care quality in the U.S. is slow. While there has been improvements noted in the reports, large gaps in preventive care especially for colorectal cancer, patient safety and mental health care remain.
Implications for Policy, Delivery or Practice:
Quality improvement and disparities reduction are achievable. Data from the National Healthcare
Quality Report and Disparities Report inform policymakers of the progress that has been made and the opportunities for improvement.
● Austria's 2007 Home Care Law: A New Pillar of
Long-Term Care Policy
Maria Magdalena Hofmarcher-Holzhacker, M.Sc.,
M.P.H.
Presented by: Maria Magdalena Hofmarcher-
Holzhacker, M.Sc., M.P.H., Senior Researcher,
Department of Economics & Finance, Institute for
Advanced Studies, Vienna, Stumpergasse 56,
Vienna, A-1060, Austria, Phone: +0043 1 59991
227; Email: maria.hofmarcher@ihs.ac.at
Research Objective: Austria was one of the first
OECD countries to acknowledge early the need for adequate provision of long-term care by appreciating freedom of choice and flexibility for recipients and families. This paper aims to give a brief review of the new home care law in Austria in the context of long-term care policy and elaborates three representative scenarios of care settings. It does not take into consideration issues of prevention of old-age disability, quality of care and issues of long-term care provision in acute care hospitals. Neither does it discuss adverse social implications and psychological issues of home care.
They may arise in spite of general belief that home care is a "gold standard" for patients and from a financial point of view.
Study Design: Descriptive cost study using various income scenarios for representative cases
Population Studied: Long-term care recipients in
Austria
Principal Findings: The Austrian long-term care system rests on four pillars. Care in the formal setting is predominantly provided in institutions owned by Federal States (“Länder”) and municipalities, or by religious and other non-profit organisations. Home-care services are largely provided by non-profit organisations. Corresponding to about 4 percent of the population, 400.000 persons are recipients of the long-term care allowance, introduced in 1993. In 2006 approximately 9 percent of these recipients received long-term care in an institutional setting, an estimated 11 percent received long-term care services given at home. Informal care provided by families or neighbours is given to about 75 percent of recipients. An estimated 5 percent receive 24h care provided by illegally practicing migrant workers who mostly come from bordering New Member
States. This had evolved as family care became insufficient to ensure comprehensive home care. In response to concerns about the economic and social impact of illegally practicing care workers a new home care law (HCL) was passed in July 2007.
Like the long-term care benefit introduced under the
1993 law legalised 24h care is financed from budgets of the Federal Republic and the Länder
(Federal States). However, whereas the 1993 longterm care allowance covers all people in need for care, regardless of the income and assets the recipients may have, the 24h home care benefit is a means tested programme. Entitlements to the universal LTC allowance are classified into seven categories of care needs, which are defined as
"average monthly time needed for regular care and assistance". Eligibility for the 24h home care programme starts for recipients who are in care category 3 corresponding to a need of 120 hours care per month. Like with the 24 h home care benefit, entitlements to social assistance are means-tested. While for the latter eligibility criteria vary across regions, i.e. “Länder” the 24h home care policy aims at harmonizing standards for access to care while ensuring freedom of choice for families and care recipients. In all pillars of care recipients pay for care delivery with their income, their LTC allowance and social assistance if entitlements are granted. In Scenario 1 social
assistance pays about 40 percent of cost for patient when treated in a nursing home. Income of the family in Scenario 2 is too high for eligibility for social assistance and is above expenses for home care. If this family needs 24h care almost all income would be used for rendering this service. The new home care law improves the situation for this family, but not much. However, where the family decides to contract with self-employed care givers, income may be twice as high as expenses for 24h care provided pay levels for the self-employed care givers can be kept to a minimum.
Conclusion: Consistent with the underlying welfare model where family responsibility is given primacy over public provision (subsidiarity principal) informal care giving mainly rendered by women is the most important care setting. Four in five long-term care recipients in Austria receive services from informal care arrangements. In regulating 24h informal care at home the 2007 home care law has recognized the need to formally introduce a further pillar for service provision. This policy aims at making
"around-the-clock" care at home more accessible by establishing a legal basis for 24h care and by increasing permitted working hours for health care workers. At the same time 24h care provided by legally qualified home help was to become more affordable through granting subsidies. The law requires contractual relations between care givers and care recipients and does – in principal – not exclude informal family care givers from entering in this kind of formal arrangement. Marketable family help in the area of long-term care would be novel in the Austrian context and may enhance formal labour market participation of women. Cost scenarios indicate that the strain on family resources to cover 24h care cost is surely alleviated when subsidies are granted, especially in selfemployment arrangements. But for this to become tangible for recipients and families the income of these care workers would need to be kept on the lowest possible level. However, this is uncertain in light of rising demand and labour shortages in this area. Also, it is rather unsustainable as income levels of cross-border health workers are likely to converge in the future.
Implications for Policy, Delivery or Practice: If the Austrian long-term care model is to continue to fund adequate care, routine adjustments for inflation of the long-term care allowance and additional funding of 24h care is needed. In addition, more attention needs to be paid to inefficiency and distributional implications arising from differences across federal states in entitlements and access to various care settings.
Theme: Long Term Care
● Conditions for Achieving Better Care
Coordination in OECD Countries
Maria Magdalena Hofmarcher-Holzhacker
Presented by: Maria Magdalena Hofmarcher-
Holzhacker, Senior Researcher, Department of
Economics & Finance, Institute for Advanced
Studies, Stumpergasse 56, Vienna, A-1080,
Austria, Phone: +43 1 59991 227; Email: maria.hofmarcher@ihs.ac.at
Research Objective: Health systems in OECD countries face major challenges to improve their performance through increasing efforts to make care delivery more responsive to patient needs. The
OECD study Improved Health System Performance through Better Care Coordination (Hofmarcher et al.
2007) aimed to assess whether -- and to what degree - better care coordination can improve health system performance in terms of quality and cost-efficiency. Coordination of care refers to policies that help create patient-centered care that is more coherent within and across care settings and over time. These issues are of particular importance for patients with chronic conditions who regularly need health care services and who may find it difficult to "navigate" fragmented health-care systems that are often found in OECD countries.
The main objective of this paper is to summarize specific problem areas likely to impede care coordination and policies that could help address these issues, largely drawing on the more detailed experiences of 26 countries responding to the questionnaire. Hofmarcher, M.M., H. Oxely, E.
Rusticelli: Improved Health System Performance through better care coordination, OECD Health
Working Paper, No 30
Study Design: The OECD Secretariat administered a questionnaire sent to 38 countries including Non-
OECD EU countries. Twenty-six countries replied.
Statistical methods have been applied to identify associations between different statements or questionnaire items. This approach was supplemented by a comprehensive literature review including an assessment of the impact of targeted programmes, e.g. disease management.
Population Studied: OECD countries and non-
OECD EU Member States
Principal Findings: The study suggests that targeted programmes, e.g. disease management programmes appear to improve quality, but evidence on cost-efficiency is inconclusive. To build on any positive impacts of these programmes system-wide approaches to improve care coordination are needed through: Making information on patient health and provider quality more readily available; Strengthening the capacity of ambulatory care providers to coordinate; Aligning payment schemes across health sectors with objectives in ambulatory care; and Reducing regulatory and administrative barriers to cooperation between health and social care. All four key areas for better care coordination appear important for achieving much-desired improvements in the performance of health-care systems, even though the precise mix of policies will necessarily
differ across countries. Enhanced information transfer across providers is key in this context. Care coordination can be improved by bridging better the administrative and other barriers that impede easy transitions from one sector or provider to another.
This issue may be particularly important for transitions into long-term care where problems appear to be the most intense. Payment models which encourage cooperation across sectors and reward multidisciplinary care need to be developed to better engage providers at all levels in reorganizing the health-care system. Pooling resources between the health and social sectors for designated care coordinators who help patients and families at these transition points may be one model to bridge administrative barriers. But there may be also scope for integrated-care models that bring specialities and services under one roof and help make fragmented and complex systems more
“userfriendly”.
Conclusion: Concern over care coordination is likely to intensify as the importance of chronic disease increases. Countries have already begun to respond. In reviewing individual country approaches by looking at initiatives in Germany, United States and England similar objectives are being adapted to the specific institutional environment in each. The review of care coordination policies in these countries suggests that:
The main objective in all countries was to reduce costly hospitalization by promoting bettercoordinated care outside hospitals, e.g. ambulatory care and long-term care. Programmes and policies in place to foster coordination are, in most cases, relatively new and clear evidence on cost savings is unavailable in all three countries. In an effort to make care delivery more patient-centred, countries have begun to restructure incentives for ambulatory care providers -- e.g. payment by result in the UK, pay-for-performance in the US -- or developed other incentives -- e.g. creating a more flexible contracting environment in Germany.
Regulatory changes have commenced in all countries to break down barriers between sectors and to stimulate cooperation across providers, e.g. incentives for sickness funds and long-term care insurers in Germany to cooperate or local development plans in England with the aim to shift services to community settings. The form of these policies has often reflected the particular difficulties and institutions specific to individual countries.
Disease or case management programmes are one approach to improving care coordination. These policies appear to provide scope for improving quality of care, although their capacity to achieve cost savings remains unclear. This may reflect the fact that, in many cases, programmes have not been followed over long enough periods. Better evaluation is needed about which policies appear to work best and in which set of circumstances.
Implications for Policy, Delivery or Practice:
More comprehensive efforts to improve care coordination require a mix of policies that go beyond disease management. Care management at transitions to long-term care or better integration of care within individual provider institutions are two of the many possible examples. These approaches need to address the continuing issue of fragmentation of health-care delivery and require payers and providers across care sectors to engage in making care more patient-centred with a focus on multi-disciplinary care. For this to happen new care delivery models and payment schemes need to be developed and evaluated. Progress in assessing the impact of policies also requires better oversight by payers backed up by information systems and regular reporting of measures of performance.
Theme: Quality and Effectivenes: Policies and
Inventives
● Factors Associated with the Coordination of
Care in VA Facilities: An Ecological Perspective
Timothy Hogan, Ph.D.; Mary Elizabeth Wilber, M.S.;
Bridget Smith, Ph.D.; Kevin Stroupe, Ph.D.
Presented by: Timothy Hogan, Ph.D., Postdoctoral
Fellow, Center For Management of Complex
Chronic Care (151H), Hines VA Hospital, 5000
South 5th Avenue, Hines, IL 60141-3030, Phone:
(708) 202-5737; Email: Timothy.Hogan@VA.gov
Research Objective: The Veterans Healthcare
System is the largest integrated healthcare system in the U.S. and is intended to meet the primary care, specialty care, and other health and social service needs of the nation’s veteran population.
Studies conducted outside of VA facilities, within the context of for-profit or nonprofit U.S. hospitals, have documented the important role that coordination plays in the delivery and receipt of quality health care. Previous studies have shown that effectively coordinated care is associated with favorable outcomes, including higher patient satisfaction and lower costs, and that both healthcare providers and consumers place considerable value on well-coordinated care.
Despite these findings, significantly less is known about the factors that shape coordination in VA settings. Using an ecological framework to highlight interaction among individual, interpersonal, organizational, and broader system levels, the objectives of this study are to understand the factors associated with coordination of care in VA facilities and those that contribute to continuity during the course of care.
Study Design: We are conducting semi-structured interviews at a large VA medical center with 30 healthcare providers recruited from the facility’s general medicine clinic and other specialty clinics.
Audio-recordings of the interviews are being transcribed into verbatim transcripts and analysis is being conducted by two investigators using iterative coding procedures and constant comparative techniques, the essence of which involves deriving
basic concepts from data and comparing them with other data to facilitate meaningful categorization.
Principal Findings: Providers interviewed have been physicians, nurse practitioners, physician assistants, and other nursing staff who have worked in VA facilities for two to over 20 years. They described continuity and coordination as critical to their daily tasks; the former being a long-term relationship between a veteran and a provider who oversees their care and the latter being a component of continuity that hinges on access to necessary resources. VA’s unified medical record has emerged from our analysis as an important facilitator of coordination efforts. At the individual level, the complex medical problems faced by veterans coupled with the pursuit of dual care through VA and Medicare-covered facilities underscore the importance of continuity, yet also make it difficult to realize. Workflow issues across teams and at the facility in question, including panel sizes, the responsiveness of different service lines, and the recent implementation of a new scheduling system were reported as pressing challenges. At the system level, mandates such as that to see new patients in 30 days, can have unintended consequences for the ability of providers to make continuous, well-coordinated care a defining feature of their clinical work.
Conclusion: Factors at the individual, interpersonal, organizational, and broader system levels coalesce to make VA facilities unique and challenging environments in which to coordinate care.
Implications for Policy, Delivery or Practice: As an integrated system, VA has great potential to provide coordinated care and to function as an effective medical home. Insights from this study can facilitate the design of contextually-appropriate, empirically-based interventions at the individual, interpersonal, facility, and system levels that address barriers to continuity and coordination.
Funding Source(s): VA
● Impact of Fungal Infections on Outcomes of
Patients Hospitalized for Bone Marrow
Transplant - Differences in Antifungal Therapy
Samuel Hohmann, Ph.D.; Jacob Gettig, Pharm.D.,
M.S.; Karl Matuszewski, Pharm.D., M.S.
Presented by: Samuel Hohmann, Ph.D., Senior
Manager, Clinical Data & Informatics, University
HealthSystem Consortium, 2001 Spring Road,
Suite 700, Oak Brook, IL 60523, Phone: (630) 954-
1740; Email: hohmann@uhc.edu
Research Objective: Compare the outcomes of antifungal therapy among bone marrow transplant patients - those with fungal infections vs those without
Study Design: Retrospective cohort study comparing antifungal therapy type and presence of fungal infection using University HealthSystem
Consortium (UHC) clinical database records of bone marrow transplant (BMT) hospitalizations
Population Studied: Patients hospitalized at academic medical centers for bone marrow transplant between October 2007 and September
2008 and reported to the UHC clincial database.
Principal Findings: More than 90% of BMT patients received antifungal therapy, of which three quarters received prophylactic therapy. Those without infection were less likely to receive therapies other than azoles. Those with infections received echinocandins, polyenes, and other antifungal agents 2 or more times as often as those without infection. Azole therapy was received by patients with or without fungal infections more than
80% of the time. Length of stay increased between
9 and 12 days (25 days to 37 days) depending on therapy for patients with infections compared to patients without infection. Mortality was also significantly higher among those patients with fungal infections (15% to 25% depending on type of therapy) compared to about 4% for patients without infections. Combinations of therapy were generally associated with longer stays and higher mortality.
Conclusion: Type of antifungal therapy leads to variation in outcomes of BMT patients who have fungal infections. Studies of therapeutic interchange should address these differences and establish the gold standard of care for managing
BMT patients with fungal infections.
Implications for Policy, Delivery or Practice:
There are additional institutional costs associated with longer stays and personal cost associated with patient deaths associated with managing BMT patients with fungal infections. Understanding best practices can lead to reducing cost of providing care and personal costs as well.
Theme: Organizational Performance &
Management
● Correlates of Ischemic Stroke Outcomes
Samuel Hohmann, Ph.D.; Matthew
Anderson;Andrew Garman, Psy.D.; Kevin Dunn,
M.S.; Ruth Ross, Ph.D.; Karen Saban, Ph.D.
Presented by: Samuel Hohmann, Ph.D., Senior
Manager, Clinical Data & Informatics, University
HealthSystem Consortium, 2001 Spring Road,
Suite 700, Oak Brook, IL 60523, Phone: (630) 954-
1740; Email: hohmann@uhc.edu
Research Objective: Identify practices that lead to better ischemic stroke patient outcomes following initial hospitalization. The Joint Commission and others have specified structural and process measures for management of stroke patients. The impact on outcomes of patients with ischemic stroke will be presented.
Study Design: Retrospective cohort study comparing ischemic stroke patient outcomes relative to management of their care using the
University HealthSystem Consortium (UHC) clinical database.
Population Studied: Ischemnic stroke patients discharged from academic medical centers between October 2007 and September 2008. The data was extracted from the University
HealthSystem Consortium clinical database.
Principal Findings: After adjusting for demographic variables, outcomes of care (length of stay, mortality, cost, ICU use, readmission within 30 days) were found to vary depending on several factors: complications of care (urinary tract infection and venous thrombosis), timely physical therapy evaluation, tPA administration, and speech pathology assessment. For example, less than
10%of patients received IV tPA, but this proportion varied widely from hospital to hospital. Similarly, about two thirds of the patients had a physical therapy evaluation during hospitalization, and the proportion by hospital also varied widely. These differences contributed to differences in patient outcomes.
Conclusion: Incorporating specific strategies for managing ischemic stroke patients can lead to better patient outcomes.
Implications for Policy, Delivery or Practice:
Adopting to practices that lead to excellence in managing ischemic stroke patients should become standard of care.
Theme: Quality and Efficiency: Measurement
● Risk Assessments of Pediatric Emergency
Transfers
Jane Holl, M.D., M.P.H.; Donna Woods, Ed.M.,
Ph.D.; Olivia Ross, M.P.H.; Anna Torricelli, B.A.
Presented by: Jane Holl, M.D., M.P.H., Associate
Professor of Pediatrics & Preventive Medicine,
Institute for Healthcare Studies, Northwestern
University, Feinberg School of Medicine, 750 North
Lake Shore Drive 10th Floor, Chicago, IL 60611,
Phone: (312) 503-4550; Email: jholl@northwestern.edu
Research Objective: To identify and address risks in the emergency transfer of pediatric patients between hospitals.
Study Design: Each of the six hospitals with comprehensive pediatric services that comprise the
Pediatric Patient Safety Consortium (Consortium) selected a specific emergency transfer process
(e.g., outside emergency department (ED) to intensive care unit) to evaluate using a prospective risk analysis method: the Failure Mode Effects and
Criticality Analysis (FMECA). Initial FMECA sessions focused on the creation of a process map of the selected emergency transfer process. Next, each FMECA team evaluated the identified steps in their emergency transfer process for potential failure modes and causes. Then, the teams determined the frequency, consequence, and safeguards of each failure mode. Finally, results were “binned” or aggregated across all six participating hospitals into generalizable risk categories. Solutions to address the major risks were designed.
Population Studied: The FMECA teams were comprised of: (1) clinicians (e.g., emergency department nurses and physicians, pediatric intensivists, transfer team members) involved in the emergency transfer of pediatric patients at the six
Consortium hospitals in Chicago, IL and (2) clinicians involved in the emergency transfer process at two referring hospitals from which each
Consortium hospital receives pediatric emergency transfers. The inclusion of clinicians from both referring and Consortium hospitals provided a comprehensive mapping of each selected transfer process.
Principal Findings: Inaccurate and delayed communication about the patient was identified as a major risk with significant consequences. Key causes of this risk included lack of pediatric-specific training at the referring hospitals and significant variation in the format and content of data communicated about the transfer patient. Within institutions, the FMECA process also revealed a general lack of knowledge and awareness by many clinicians of their colleagues’ roles and responsibilities during the pediatric emergency patient transfer process. A standardized pediatric emergency transfer form was then developed. A review of the existing transfer forms from both
Consortium and referring hospitals was conducted and a draft standardized form was developed.
Clinicians from the FMECA teams reviewed and revised the form. Additionally, three hospitals pilot tested the form while continuing to use their existing transfer process and forms. FMECA clinicians evaluated the effectiveness of the form to accurately capture the necessary transfer information. Final revisions were made based on the pilot test results.
Conclusion: The six FMECAs, involving clinicians from eighteen hospitals, identified generalizable risks during pediatric emergency transfers. A standardized pediatric emergency transfer form was developed and tested. Participation and enthusiasm for the FMECA sessions enhanced intra- as well as inter-institutional communication.
Implications for Policy, Delivery or Practice: In
2005, over 3,500 pediatric patients were transferred between the referring and Consortium hospitals. All hospitals have begun to move towards adoption of the standardized pediatric emergency transfer form.
In addition, the form is currently being integrated into a newly-funded AHRQ project (Clinical
Information Network for Safe Pediatric Emergency
Transfers: R18 HS 017912-01), for development and evaluation of a web-based version of the form that would be transmitted and used in real time by clinicians at both the referring and Consortium hospital.
Funding Source(s): AHRQ
Theme: Child Health
● Mother-Child Interactions as Mediators of the
Impact of Maternal Depression on Child
Emergency Department Visits
Margaret Holland, M.S., M.P.H.; Byung-Kwang Yoo,
M.D., Ph.D.; Harriet Kitzman, Ph.D.; Linda
Chaudron, M.D., M.S.; Peter Szilagyi, M.D., M.P.H.;
Helena Temkin-Greener, Ph.D., M.P.H.
Presented by: Margaret Holland, M.S., M.P.H.,
Ph.D. Candidate, Community & Preventive
Medicine, University of Rochester, 601 Elmwood
Avenue, Box 644, Rochester, NY 14642, Phone:
(585) 273-2549; Email: margaret_holland@urmc.rochester.edu
Research Objective: Studies have noted that children of depressed mothers have more emergency department (ED) visits than other children. Poor mother-child interactions may contribute to this association by delaying the mother’s recognition of symptoms of illness or risks of injury. In addition, when an illness or injury becomes apparent, mothers with poor interactions may interpret the problem as worse than it is. We examined whether mother-child interactions are mediators between maternal depression and child
ED visits.
Study Design: Women were interviewed 12 months after their first child's birth. Depressive symptoms were measured by the Mental Health
Inventory-5 (5-item Likert scale). Mother-child interactions were measured using the Nursing Child
Assessment Satellite Training subscale for child’s responsiveness to parent (count of 0 to 13 behaviors observed) and the Home Observation for
Measurement of the Environment subscale for mother’s emotional & verbal responsiveness (count of 0 to 11 behaviors observed). Data on child ED visits from birth to 24 months were available from medical records. We fit two ordered logistic regression models with an outcome variable of the number of child ED visits (3 categories: 0/1/2+). In the first model, the key explanatory variable was maternal depressive symptoms. In the second, a mother-child interaction measure was added. All models controlled for children's chronic conditions, birth weight, and demographic factors.
Population Studied: Control group data from the
Nurse-Family Partnership trial in Memphis, TN were used (n=432). Pregnant women were recruited at an obstetrical clinic, met two of three criteria
(unmarried, unemployed, with less than 12 years of education), and were randomly assigned to control or intervention groups. The resulting sample was primarily black (92%) and poor (47% at or below
$3000 annual income).
Principal Findings: Twenty-one percent of children had 1 ED visit and 59% had 2 or more, 23% of mothers had high depressive symptoms, 40% of mothers had higher emotional & verbal responsiveness than the sample mean, and 31% of children had high responsiveness to parent. In the first model, the adjusted odds ratio (OR) for depressive symptoms was 1.12 (95% CI: 0.95,
1.36). This indicates more maternal depressive symptoms associated with more child ED visits, as expected. When the mother’s responsiveness was additionally included, the OR for depressive symptoms decreased to 1.11 (95% CI: 0.81, 1.35) and the OR for mother’s responsiveness was 0.80
(95% CI: 0.52, 1.22). The OR for mother’s responsiveness indicates increased responsiveness was associated with fewer ED visits; the decreased depressive symptoms OR was consistent with mother’s responsiveness being a mediator (95% CI for change in coefficient: -0.0004, -0.0001; p<0.01).
When the child’s responsiveness was included, the
OR for child’s responsiveness indicated increased responsiveness associated with more ED visits
(opposite of hypothesis) and the increased depressive symptoms OR was not consistent with mediation.
Conclusion: Among this urban, mostly minority population, mother’s emotional & verbal responsiveness was found to be a mediator between maternal depressive symptoms and child
ED visits, but child’s responsiveness was not.
Implications for Policy, Delivery or Practice:
When treating a depressed mother, additional interventions specifically targeting mother‘s responsiveness to her child may decrease the child's ED visits.
Funding Source(s): AHRQ
● Self-Efficacy as a Mediator Between Child
Hospitalizations & Maternal Depression
Margaret Holland, M.S., M.P.H.; Byung-Kwang Yoo,
M.D., Ph.D.; Harriet Kitzman, Ph.D.; Linda
Chaudron, M.D., M.S.; Peter Szilagyi, M.D., M.P.H.;
Helena Temkin-Greener, Ph.D., M.P.H.
Presented by: Margaret Holland, M.S., M.P.H.,
Ph.D. Candidate, Community & Preventive
Medicine, University of Rochester, 601 Elmwood
Avenue, Box 644, Rochester, NY 14642, Phone:
(585) 273-2549; Email: margaret_holland@urmc.rochester.edu
Research Objective: Several studies have noted that children of depressed mothers have more hospitalizations than other children. Self-efficacy
(the belief that one is able to be effective at a certain task) has been suggested as a mechanism contributing to this association. Mothers with poor self-efficacy may delay seeking care for their child due to lack of confidence navigating the healthcare system or lack of certainty caring for a sick child.
Also, they may not take sufficient injury prevention precautions around the home. Each of these may result in more child hospitalizations. We examined whether self-efficacy is a mediator between maternal depression and child hospitalizations.
Study Design: Women were interviewed 12 and 24 months after their first child's birth. Depressive symptoms were measured by the Mental Health
Inventory-5 (5-item Likert scale) and were considered consistently high if above the cut-off at both 12 and 24 months. Parenting self-efficacy was measured by a 10-item Likert scale developed and validated for this study. Data on child hospitalizations from 12 to 24 months were available from medical records. We fit two logistic regression models. One or more child hospitalization was the outcome variable in each model. In the first model, the key explanatory variable was maternal depressive symptoms. In the second, self-efficacy was added. All models controlled for children's chronic conditions, birth weight, and demographic factors.
Population Studied: Control group data from the
Nurse-Family Partnership trial in Memphis, TN were used (n=432). Pregnant women were recruited at an obstetrical clinic and had to meet two of three high-risk criteria: unmarried, unemployed, with less than 12 years of education, and were randomly assigned to control or intervention groups. The resulting sample was primarily black (92%) and poor (47% at or below $3000 annual income).
Principal Findings: Twelve percent of children had
1 or more hospitalizations, 14.3% of mothers had consistently high depressive symptoms, and 48% had lower maternal self-efficacy than the sample mean. In the first model, the adjusted odds ratio
(OR) for depressive symptoms was 1.98 (95% CI:
0.86, 4.00). In the second model, the OR for depressive symptoms decreased to 1.89 (95% CI:
0.81, 4.40) and the OR for self-efficacy was 0.77
(95% CI: 0.40, 1.49). Because the addition of selfefficacy decreased the OR of depressive symptoms, self-efficacy is concluded to be a mediator as hypothesized (95% CI for change in coefficient: -0.043, -0.008; p<0.01).
Conclusion: Among this urban, mostly minority population, maternal self-efficacy is shown to be a mediator between maternal depression and child hospitalizations.
Implications for Policy, Delivery or Practice:
When treating depressed mothers, additional interventions specifically targeting self-efficacy may decrease their children's hospitalizations.
Funding Source(s): AHRQ
Theme: Child Health
● What Factors Influence the Use of
Complementary Alternative Medicine (CAM)
Treatment Options for Infertility in Women?
Laura Hollywood, R.N., C.N.M., F.N.P-B.C.,
W.H.N.P., D.N.S.; Joyce Anastasi, Ph.D., D.N.P.,
L.Ac., F.A.A.N.; Mary Wood-Byrne, Ph.D., P.N.P.,
F.A.A.N.
Presented by: Laura Hollywood, R.N., C.N.M.,
F.N.P-B.C., W.H.N.P., D.N.S., Course Coordinator,
Faculty, Frontier School of Midwifery & Family
Nursing, 23 Foley Road, Warwick, NY 10990,
Phone: (845) 987-8292; Email: laura.hollywood@midwives.org
Research Objective: Purpose: To explore factors, which influence an infertile woman's decision- making concerning Complementary Alternative
Medicine (CAM) treatment options.
Study Design: Descriptive non-experimental study using a survey design for data collection. The survey contained 24 questions addressing: demographics, infertility treatment, CAM use, and geography.
Population Studied: 415 women who belong to the on-line national organization, RESOLVE, which offers support and education for infertility. Women in the study were 25-50 years old, could speak and read English and had been diagnosed with infertility by a health care provider.
Principal Findings: this study found that 76%
Seventy six percent of the respondents used at least 1 of 16 CAM modalities for the treatment of fertility problems. CAM use was highest among women aged 35-45. Women with college degrees had the highest percentage of CAM users. In all income categories CAM was used more frequently for treatment of fertility problems than medical problems. Women with incomes over $50, 000/ year were the greatest users of CAM for fertility treatment. Women with domestic partnerships and married women used CAM more often for fertility treatments. White women used CAM more frequently for fertility problems. The largest category for who suggested you try CAM therapies was that the respondents thought it would be interesting to try, followed by friends and family.
Most were not worried about side effects of CAM therapies. Most CAM users for fertility treatment had insurance coverage, but none for CAM therapies. 56.8% of women reported that they did not choose CAM treatments because they were less expensive than conventional care. Chi-square analysis were performed to examine the relationship between Age and use or non-use of
CAM therapies. Most respondents reported using
CAM treatments for fertility, however those over 32 years did so at greater rates. Acupuncture was used less frequently for treatment of fertility problems by women between 25 and 31 years and more frequently by women over 32 years. Massage was not used by most age groups but used least frequently in women ages 25-31 years. Spirituality was not used frequently by all age groups but was used least frequently in women aged 39 and under.
Conclusion: The results of this study had consistent findings with other studies of prevalence of CAM use.
CAM for treatment of fertility problems was more common in women aged 30-39. 81% of CAM users for infertility treatment were college graduates (40% of these having graduate degrees and beyond). At least one CAM modality was used in increasing
percentages for medical and fertility problems with increasing educational preparation. Increased CAM use for fertility treatment were found in women with incomes over $50,000/ year with findings of 87% using CAM for fertility treatment and 60% using
CAM for medical conditions excluding fertility problems. In this survey 86% of CAM users were white.
Implications for Policy, Delivery or Practice:
Infertility is a life altering experience for women.
Health care providers keep women informed by providing accurate and reliable information on various aspects of care including CAM options for treatment. Health care providers must be able to discuss CAM options, known side effects, as well as adverse reactions to help women fully understand the choices available to them.
Understanding the emotional experience of the infertile women and their partners is important for nurses who often serve as emotional support for women and their partners.
● Ambulatory Computerized Provider Order
Entry: Lessons Learned from the AHRQ Health
IT Portfolio
Julie Hook, M.A., M.P.H.; Caitlin Cusack, M.D.,
M.P.H.
Presented by: Julie Hook, M.A., M.P.H., Health
Services, Health Services, John Snow, Inc., 44
Farnsworth Street, Boston, MA 02210, Phone:
(617) 385-3982; Email: jhook@jsi.com
Research Objective: Although much has been written on strategies to ease the path of implementing health information technology (health
IT) systems, such as ensuring strong leadership and clinician buy-in, these strategies are not universally adopted. Here we present lessons learned from AHRQ funded projects of ambulatory computerized provider order entry systems
(ACPOE). These lessons mirror past experiences and we recommend broader dissemination of these implementation strategies to improve the rate of successful implementations.
Study Design: We examined projects within the
AHRQ portfolio that were implementing CPOE interventions in outpatient settings. For each appropriate project, we contacted the lead investigators to schedule interviews. Questions were developed and shared with the investigators in anticipation of conducting semi-structured interviews. This format enabled us to question the investigators about core project design elements, key challenges they faced, lessons they learned along the way, and future directions for using CPOE in their organizations.
Population Studied: The five identified grantees came from diverse geographic areas, with all but one in urban areas. The majority of the implementations occurred in clinics within large integrated delivery systems and in adult primary care settings, with one in a pediatric setting and another in a long-term care facility.
Principal Findings: The interviews with the grantees provided rich detail about their successes, failures, and lessons learned on their CPOE implementation. Major themes include:
All grantees stressed the importance of leadership, both at the institutional and practice level. Adoption by clinicians was cited by all as a challenge: clinicians failing to adopt leads to project failure. A physician champion was cited by the grantees as crucial for success. Other factors important to adoption included a user friendly interface, integration of tools into current workflow, emphasizing the value-add of new tools and functions to clinicians, addressing alert fatigue, and prioritizing the speed of systems. Grantees noted three areas of continuous quality improvement important to successful implementation: conducting pilot testing, rolling out systems incrementally, and post-implementation collection of feedback.
Grantees stressed the importance of having good relationships with their IT developers, whether vendors or in-house IT departments, so that bugs and glitches are addressed in a timely fashion.
Training was noted to be another crucial component to success.
Conclusion: The lessons learned from this subset of grantees have already been noted in the literature. However, to eliminate implementation barriers, more effective ways need to be developed to disseminate this information beyond early adopters to those who are new to the field. One mechanism for doing this is via Websites, such as
AHRQ’s National Resource Center for health IT
(http://healthit.ahrq.gov). With widespread dissemination of these lessons learned, the hope is that those new to the field will implement these strategies in the initial stages of their projects, and not re-learn them as they go.
Implications for Policy, Delivery or Practice:
While research has shown that CPOE can increase patient safety and efficiency, there are many issues that affect the complexity of implementations. In light of some well known CPOE implementation failures and literature that indicates unintended adverse consequences, successful strategies and lessons from early adopters should be disseminated to a wide audience.
Funding Source(s): AHRQ
Theme: Health Information Technology
● Improving Survivorship Care for Long-Term
Colorectal Cancer Survivors: Key Findings of a
5-Year Study
Mark Hornbrook, Ph.D.; Robert Krouse, M.D.; Lisa
Herrinton, Ph.D.; Marcia Grant, Ph.D., R.N.; Carmit
McMullen, Ph.D.; Andrea Altschuler, Ph.D.
Presented by: Mark Hornbrook, Ph.D., Chief
Scientist, The Center for Health Research,
NW/HI/SE, Kaiser Permanente Northwest, 3800
North Interstate Avenue, Portland, OR 97227-1110,
P hone: (503) 335-6746; Email: mark.c.hornbrook@kpchr.org
Research Objective: To understand the determinants of health-related quality of life
(HRQOL) and the lived experiences of colorectal cancer (CRC) survivors, and to identify strategies to help maintain or enhance CRC survivors’ HRQOL.
One specific focus is on the differences between patients with and without permanent stomas.
Study Design: Mail survey and focus groups.
Focus groups for subjects with ostomy were divided by gender and high and low HRQOL. Outcome measures were the modified City of Hope Quality of
Life (mCOH-QOL)-Ostomy (abridged for anastomosis patients) and SF-36v2 questionnaires.
The SF-6D scoring algorithm was used to calculate an overall HRQOL score from SF-36v2 data. Focus groups were conducted to explore ostomy-related barriers to effective self-care and adaptation strategies.
Population Studied: Subjects were 283 ostomy and 392 anastomosis long-term CRC survivors within two regions of a large group practice HMO.
Principal Findings: CRC survivors with an ostomy experienced multiple persistent HRQOL losses that differed between men and women. Women CRC survivors with ostomies, for example, reported more sleep disruption and fatigue than men. Living with an ostomy, comorbid conditions, socioeconomic status, self-reported depression, and employment status were independent predictors of SF-6D scores. Among CRC survivors with ostomy, fistulas had important implications for HRQOL.
Psychological well-being among CRC survivors was positively associated with income. Intestinal stomas significantly influenced spiritual HRQOL. Provision or withdrawal of a partners’ support affected both short- and long-term psychosocial adjustment of female CRC ostomy patients. Focus group participants identified dietary changes to control bowel output and odor, demands of coping and adjustment, and the time it took to accept the reality of daily living with an ostomy as significant challenges, especially in persons with low HRQOL.
Conclusion: The greatest challenges reported by
CRC survivors confirmed the IOM’s findings that survivorship is a distinct, chronic phase of cancer care, and that cancer’s effects are broad and pervasive. CRC survivors could benefit from dietary and behavioral interventions even 5+ years following their cancer surgery. Women with ostomies fare more poorly than men with ostomies and report a different profile of HRQOL challenges, including sleep disruption and fatigue.
Implications for Policy, Delivery or Practice:
Depression screening and psychosocial interventions should be part of survivorship care for
CRC patients. CRC survivors should be assessed for income and financial need. Survivorship care should aim at minimizing hospitalizations, depression, and ostomy-related complications among long-term CRC survivors.
Funding Source(s): NCI
Theme: Gender and Health
● Excessive Gestational Weight Gain is
Predictive of Postpartum Weight Retention in
Obese Women
Mark Hornbrook, Ph.D.; Kimberly Vesco, M.D.,
M.P.H.; Victor Stevens, Ph.D.; Joanne Rizzo,
M.P.A.; Donald Bachman, M.S.; Patricia Dietz,
Dr.P.H.; William Callaghan, M.D., M.P.H.
Presented by: Mark Hornbrook, Ph.D., Chief
Scientist, The Center for Health Research,
NW/HI/SE, Kaiser Permanente Northwest, 3800
North Interstate Avenue, Portland, OR 97227-1110,
Phone: (503) 335-6746; Email: mark.c.hornbrook@kpchr.org
Research Objective: The prevalence of obesity among reproductive age women has increased significantly. The goal of this analysis was to determine whether gestational weight gain above the Institute of Medicine recommendation for obese women (at least 15 pounds [lbs] for term pregnancy) is predictive of postpartum weight retention at one year.
Study Design: Obesity was defined as BMI =30 kg/m-squared. Total gestational weight gain (last available predelivery weight - weight at pregnancy onset [-180 to +90 days]) was categorized as <0, 0-
15, and >15 lbs. Postpartum weight retention
(postpartum weight at one year [lowest weight measured within 243-547 days post delivery] - weight at pregnancy onset) was defined as <0, 0-
10, and >10 lbs, and by mean change. Descriptive, linear regression and multivariate logistic regression analyses were performed.
Population Studied: We conducted a retrospective cohort study of 25,789 pregnancies beginning and ending between January 2000 and December 2005 in a large health maintenance organization (Kaiser
Permanente, Northwest) and resulting in live birth.
Principal Findings: There were 1,656 obese women who met eligibility criteria and had weights available at each key time point. Compared to women who lost weight or who gained 0-15 lbs during pregnancy, women who gained >15 lbs were younger, had lower first trimester weights and BMI, had fewer live births, and were less likely to have diabetes or depression. Mean change in weight at one-year postpartum was higher for women who gained >15 lbs during pregnancy, resulting in a net gain (7.2 lbs ± 18.6), compared to a net loss among women who either gained 0-15 lbs or who lost weight during pregnancy (net loss -1.7 lbs ± 6.2 and
-2.3 lbs ± 16.3, respectively) (p<0.001). Total gestational weight gain was a significant predictor of weight gain at one-year postpartum (beta = 0.39,
R2 = 0.11, p<0.001). Women who gained >15 lbs during pregnancy were more likely to retain >10 lbs
(39.7%) at one year postpartum than women who gained 0-15 lbs (15.6%) and who lost weight
(24.1%, p<0.001). There was a 3.6 fold increased odds (Odds Ratio 3.65, 95% CI 2.69-4.95) of weighing over 10 lbs more at one-year postpartum than at pregnancy onset if pregnancy weight gain was >15 lbs compared to 0-15 lbs.
Conclusion: Weight gain above 15 lbs during pregnancy is associated with significant risk of postpartum weight retention at one year among obese women. Nutritional counseling should be provided to help prevent excessive weight gain.
Implications for Policy, Delivery or Practice:
Women of childbearing age should have their BMI measured periodically. Obese women who are planning to become pregnant or who become pregnant should be informed of the risks of additional weight gain to their babies. These women should also be referred to medically managed weight management programs.
Funding Source(s): CDC
Theme: Obesity Prevention and Treatment
● Comparative Assessment of Medical Devices
& Product Choice by Orthopedic Surgeons
Michael Housman, A.M.; Lawton Burns, Ph.D.,
M.B.A.; Robert Booth, Jr., M.D.; Aaron Koenig
Presented by: Michael Housman, A.M., Doctoral
Candidate, Health Care Management Department,
The Wharton School, University of Pennsylvania,
3641 Locust Walk, Philadelphia, PA 19104, Phone:
(215) 681-6955; Email: housman@wharton.upenn.edu
Research Objective: U.S. expenditures on medical devices comprised $80 billion in 2007 and orthopedic implants represent a substantial proportion of these expenditures. Although hip and knee implant procedures represent two of the fastest growing sources of hospital expenditures,
Medicare reimbursement rates have not kept pace with these cost increases. This has led hospitals to pursue a variety of activities designed to maximize value from their device spending and to seek out information on their comparative effectiveness.
This second half of a two-part joint study between a hospital system and academic researchers describes a methodology for conducting comparative studies of device vendors and represents the first such study for orthopedic implant vendors.
Study Design: In consultation with stakeholders, we developed and administered a survey to
Pennsylvania orthopedists. The survey asked surgeons to identify their primary vendor for hip and knee implants and to evaluate this vendor on several dimensions of product performance. We analyzed surgeon responses through the use of an analysis of variance (ANOVA) model that controlled for a variety of physician attributes, hospital characteristics, and physician-vendor ties that might influence these evaluations.
Population Studied: We identified the population of all orthopedists performing hip and knee implant procedures in the state of Pennsylvania during
2004. We excluded surgeons performing fewer than twelve procedures during that year, which yielded a final sample of 492 orthopedists. After two survey mailings, we received responses from 201 surgeons, representing a response rate of 40.9%.
Principal Findings: Vendor-level differences are consistently predicated upon a surgeon’s use of that particular vendor during his/her orthopedics training. Our independent variables indicate that hip and knee vendor evaluations are significantly associated with implant procedure volume, the physician’s number of years in practice, use of the same vendors during residency training or an arthroplasty fellowship, and receipt of consulting fees and honoraria for talks from manufacturers.
Conclusion: Our results suggest that there is not a great deal of variation across hip and knee implant vendors in terms of product performance; surgeons tend to use the same vendors that they used during their orthopedics training. We also find evidence to suggest that physicians show strong vendor preferences when they have high implant procedure volume, have been practicing for a long time, use relatively few vendors, and receive financial payments from manufacturers. These factors all have potential to create strong ties with vendors.
Implications for Policy, Delivery or Practice: The lack of variation among different vendors for hip and knee implants suggests that implant choice tends to be influenced by physician-vendor relationships rather than product differences. Any price differences between these products that may exist do not appear to be associated with better or worse performance. Hospitals appear to be justified in their efforts to implement cost-containment initiatives that are intended to standardize implant choice by using a smaller number of vendors and thereby negotiate lower implant prices.
Funding Source(s): Wharton Center for Health
Management & Economics
● Lack of Preparation & Patient Satisfaction with Obstetrical Provider: What Postpartum
Mothers are not Prepared for & Why it Matters
Elizabeth Howell, M.D., M.P.P.; Mark Chassin,
M.D., M.P.P., M.P.H.
Presented by: Elizabeth Howell, M.D., M.P.P.,
Assistant Professor, Health Policy; Obstetrics,
Gynecology, & Reproductive Science, Mount Sinai
School of Medicine, One Gustave L. Levy Place,
Box 1077, New York, NY 10029, Phone: (212) 659-
956; Email: elizabeth.howell@msnyuhealth.org
Research Objective: To determine whether lack of preparation for the postpartum experience is
associated with patient satisfaction with obstetrical provider.
Study Design: This was a prospective cohort study of women who delivered infants at an urban hospital from January through September 2002.
Patients were interviewed two weeks postpartum to assess overall preparation for the postpartum experience, provider preparation for 11 common physical and emotional postpartum symptoms, physical and emotional symptom status, selfassessed health, experience with the healthcare system, access, satisfaction with provider, and other factors. The associations of patient satisfaction and overall preparation, preparation for specific symptoms, access, demographics, and other factors were examined using bivariate and multivariable analyses.
Population Studied: 723 racially/ethnically diverse postpartum women who had uncomplicated deliveries at Mount Sinai Medical Center. Eligible subjects were women >18 years of age, English or
Spanish Speaking, who had length of stays less than 4 days for vaginal deliveries and less than 6 days for c-section deliveries, and had infants whose birthweights were > 2500 grams and 5-minute
Apgar scores >6.
Principal Findings: Mean age was 31 (range 18-
47); 26% of the sample were Latino, 15% were
African American, and 50% were white. Twentythree percent of the mothers did not feel adequately prepared by their provider for the postpartum experience. Lack of preparation for common postpartum physical and emotional symptoms was frequent: 76% of the mothers were not prepared for urinary incontinence, 62% were not prepared for breastfeeding problems, 38% were not prepared for depression, and 21% were not prepared for pain from c-section incisions or episiotomy sites repairs.
Rating your provider as excellent was more common in whites than nonwhites, patients with commercial insurance versus Medicaid insurance, women with higher educational attainment and higher SES, not having poor self assessed health, having a regular doctor, more access, and being better prepared for the postpartum experience in bivariate analyses. Physical symptom burden and depression were not associated with rating of provider in bivariate analyses. After adjusting for demographic factors, self-assessed health, and having a regular doctor, rating your provider as excellent was associated with more access
(adjusted OR of 1.33, 95% CI: 1.25, 1.42) and more preparation for common physical and emotional postpartum symptoms (OR of 1.11, 95% CI: 1.04,
1.20). In adjusted analyses, women who reported lack of overall preparation were less likely to rate their provider as excellent (OR of 0.34, 95% CI:
0.27, 0.65).
Conclusion: Lack of overall preparation for the postpartum experience, including failure to discuss common physical and emotional postpartum symptoms, and less access to health care providers are important correlates of satisfaction with obstetrical provider.
Implications for Policy, Delivery or Practice:
Understanding the natural history of maternal health is an important step to better serve women following delivery. Further research is needed to investigate whether preparing patients for expected health consequences of pregnancy improves patients’ satisfaction with providers.
Funding Source(s): AHRQ
Theme: Gender and Health
● Predictors of Timely Enrollment in the 2006
Medicare Prescription Drug Plan among
Beneficiaries with a Choice
Benjamin Howell, M.P.P.; Jennifer Wolff, Ph.D.
Presented by: Benjamin Howell, M.P.P., Ph.D.
Candidate, Health Policy & Management, Johns
Hopkins, 10 Breton Hill Road #3A, Pikesville, MD
21208, Phone: (443) 204-2818; Email: bhowell@jhsph.edu
Research Objective: To determine whether and which beneficiary and market characteristics were associated with voluntary enrollment in the
Medicare Part D Program at its inception.
Study Design: Data were drawn from the 2005
Medicare Current Beneficiary Survey (MCBS), internal Centers for Medicare and Medicaid
Services (CMS) files on retiree drug subsidy, creditable coverage, and Part D enrollment for
2006, and information on Part D contracts maintained by CMS. The primary outcome of interest was beneficiary enrollment in a Part D plan by the May 2006 deadline. The beneficiary characteristics that were examined included selfreported measures of interest in learning about
Medicare, knowledge of the Part D program, socioeconomic status, various demographic characteristics, heath status, and demand for drug coverage. The market characteristics that were examined included measures of the number, affordability, and generosity of plan options.
Frequency tabulations, stratified analyses, and multivariate logistic regression were used to assess how beneficiary and market characteristics were associated with Part D enrollment.
Population Studied: This study focused on beneficiaries for whom voluntary enrollment in a
Part D plan in 2006 was a salient decision. As such, a weighted sample of approximately 31 million community dwelling beneficiaries who were continuously enrolled in Medicare throughout calendar year 2005 and still alive at the time of the
2006 Part D enrollment deadline was subset to exclude beneficiaries who were eligible for auto enrollment into a Part D plan (estimated 3.8 million,
12.4%) or who had other sources of drug coverage
(10.2 million, 32.8%). The remaining weighted 17.1 million beneficiaries (54.8% of total) comprised our study sample.
Principal Findings: Among the estimated 17.1 million beneficiaries for whom enrollment in the prescription drug plan was a salient choice, 44.6% enrolled in a stand-alone prescription drug plan,
26.8% enrolled in a Medicare Advantage prescription drug plan, and 28.6% remained without coverage. In a multivariate model, increasing scores on a 5 item knowledge test of Part D concepts (Odds Ratio (OR) = 1.23 [1.16-1.30] and having reported skipping medications due to cost in
2005 (OR=1.58 [1.28-1.92]) were positively associated with enrollment in a prescription drug plan. Beneficiaries younger than 65 years
(OR=0.44 [0.34-0.56]) and beneficiaries with incomes greater than $25,000 (OR=0.74 [0.64-
0.87]) were less likely to have enrolled in a drug plan. The Market level characteristics that were examined did not appear to be important predictors of enrollment.
Conclusion: An estimated 4.9 million community dwelling beneficiaries who were eligible to enroll in a part D plan in 2006 elected not to do so.
Knowledge of Medicare, prior difficulties obtaining prescription medication, income, and age proved to be important factors in predicting enrollment.
Implications for Policy, Delivery or Practice:
Educating beneficiaries about the importance of obtaining drug coverage remains a priority. This is especially true for beneficiaries who may not yet recognize the need for drug coverage either through a lack of awareness about the Part D benefit or through limited experience with high prescription drug costs.
Funding Source(s): CMS
● Emergency Department Visits & Resulting
Hospitalizations by Elderly Nursing Home
Residents
Chun-Ju Hsiao, Ph.D., M.H.S.; Esther Hing, M.P.H.
Presented by: Chun-Ju Hsiao, Ph.D., M.H.S.,
Associate Service Fellow, Division of Health Care
Statistics, National Center for Health Statistics,
3311 Toledo Road, Room 3305, Hyattsville, MD
20782, Phone: (301) 458-4689; Email: hwd3@cdc.gov
Research Objective: Previous studies found that emergency department (ED) visits and hospital admissions by nursing home (NH) residents can be reduced with increased primary care within the NH.
This study examines ED visits made by elderly NH residents during 2001-06, including whether an ED visit resulted in hospital admission. The study also examines factors associated with hospitalization from the ED.
Study Design: Characteristics of ED visits made by elderly NH residents were examined using nationally representative samples of ED visits from the 2001- 06 National Hospital Ambulatory Medical
Care Surveys. Bivariate analysis and multivariate logistic regressions were performed to examine the relationship between hospitalization following ED visits and characteristics of NH residents.
Multivariate models adjusted for age, sex, race, type of insurance, metropolitan area, selected ambulatory care sensitive conditions (ACSCs), number of procedures received in the ED, number of diagnosis, and triage status.
Population Studied: ED visits made by NH residents aged 65 years and over (n=3,724).
Principal Findings: During 2001-2006, the ED visit rate by elderly NH residents was 118 visits per 100 institutionalized persons. Females comprised
65.9% of ED visits by NH residents, and the average age of NH patients in the ED was 82 years.
Nearly 15% of these ED visits had ACSC diagnoses. Nearly half (48.1%) of ED visits made by elderly NH residents had a final disposition of hospital admission, and 17.0% of these admissions had ACSC diagnoses. Selected ACSCs were associated with higher adjusted odds of being admitted to the hospital including chronic obstructive pulmonary disease (aOR= 2.22, 95%
CI= 1.08 - 4.57), congestive heart failure (aOR=
3.74, 95% CI= 2.22 - 6.29), angina (aOR= 10.03,
95% CI= 2.66 - 37.79), kidney/urinary tract infection
(aOR= 1.89, 95% CI= 1.27 - 2.81), and dehydration
(aOR= 2.80, 95% CI= 1.54 - 5.08). The adjusted odds of an ED visit resulting in a hospital admission increased with number of diagnoses, with having two or more procedures performed, and with all conditions triaged as “emergent” or “urgent”.
Conclusion: Nearly half of ED visits made by elderly NH residents resulted in hospital admission, and selected ACSCs were positively associated with hospital admission outcomes. It is possible that some of these ED visits could have been prevented if increased primary care had been provided in the NH setting.
Implications for Policy, Delivery or Practice: The role of ACSCs in hospital admissions from EDs among NH residents suggests their clinical management within NHs merits attention to improve the quality and cost of care.
Theme: Long Term Care
● The Effect of Personal Quality of Primary
Care, Continuity & Chronic Conditions on
Health Services Utilization among Older Adults
Chun-Ju Hsiao, Ph.D., M.H.S.; Jill Marsteller, Ph.D.,
M.P.P.; Bruce Leff, M.D.
Presented by: Chun-Ju Hsiao, Ph.D., M.H.S.,
Associate Service Fellow, Division of Health Care
Statistics, National Center for Health Statistics,
3311 Toledo Road, Room 3305, Hyattsville, MD
20782, Phone: (301) 458-4689; Email: hwd3@cdc.gov
Research Objective: Existing studies suggest that higher quality in the personal aspects of primary care is associated with some but not all outcomes of care. Previous studies usually included patient’s
health status as the outcome. The association between personal quality of primary care and health services utilization is not clear. This study evaluates the influence of self-reported personal quality of primary care, continuity of care, and chronic illness on health services utilization among older adults.
Study Design: This study uses data from the first three years (1998-2000) of the longitudinal study
“Choice & Quality in Senior Health Care”. Selfreported personal quality of primary care
(physician-patient communication, interpersonal treatment, knowledge of patient, integration of care, and trust in physician) and continuity of care were assessed annually using the Primary Care
Assessment Survey. Number of chronic illnesses, emergency department (ED) visits and hospital admissions in the previous 12 months were also self-reported annually. A multinomial logistic regression was employed to analyze ED visits, and a two-part model was used to analyze any hospitalization and number of hospital days.
Population Studied: Community-dwelling
Medicare beneficiaries aged 65 and older enrolled in traditional fee-for-service (FFS) or a Medicare
HMO (n= 8,717 at baseline).
Principal Findings: A positive change in quality of care from 1999 to 2000 was associated with reduced relative risk of ED visits (P=0.035). Longer relationship duration with the physician was associated with lower odds of hospitalization
(P=0.002). The number of chronic conditions was associated with higher relative risk of ED visits and higher odds of hospitalization (P<0.001). Visitbased continuity of care was related to a slightly higher incidence rate ratio of hospital days
(P=0.006). At baseline, persons with medium quality of care had an 85% increase in odds of inpatient care use in 2000 compared to those with low quality of care (P=0.023).
Conclusion: The relationships between different levels of personal quality of care and health services utilization were not consistent in the hypothesized directions. This underscores the challenges in understanding the influence of quality of care on health service utilization to provide appropriate care to older adults. Measures of continuity of care and number of chronic conditions also showed relationships with health services utilization, but not in the expected direction as well.
Additional research over a longer study period may help clarify relationships.
Implications for Policy, Delivery or Practice:
Given the rising focus on quality of care and health service reforms such as the patient-centered medical home, more information is needed to understand the impact of the personal quality of primary care upon health services utilization.
Theme: Quality and Efficiency: Measurement
● The Impact of Medicaid Disproportionate
Share Hospital Payments on Hospital
Uncompensated Care Provision
Hui-Min Hsieh, M.B.A.; Gloria Bazzoli, Ph.D.;
Michael McCue, Ph.D.
Presented by: Hui-Min Hsieh, M.B.A., Ph.D.
Student, Department of Health Administration,
Virginia Commonwealth University, 1008 East Clay
Street, P.O. Box 980203, Richmond, VA 23298,
Phone: (804) 827-1811; Email: hsiehhm@vcu.edu
Research Objective: The purpose of this study is to examine what impact changes in Medicaid
Disproportionate Share Hospital (DSH) payments has an impact on hospital uncompensated care.
Study Design: Economic theory of non-profit hospital behavior was used as a conceptual basis, and data for California hospitals for 1996-2003 were examined. In 1997, the Balanced Budget Act (BBA) was enacted and state-specific federal Medicaid
DSH allotments were reduced. California was especially affected by these changes because it is one of the states with highly concentrated Medicaid
DSH payments. Economic theory suggests that hospitals with fewer financial resources may provide less uncompensated care. However,
Medicaid DSH payments and the provision of hospital uncompensated care is endogenous, and this must be accounted for in the analysis. To eliminate the potential bias, we applied firstdifference two stage least squares with instrumental variables to analyze the research question. The primary dependent variable is hospital uncompensated care provision at time t. The key independent variable of interest is net Medicaid
DSH received by the hospital in previous year. The two-year lagged net Medicaid DSH was used as an instrumental variable for the prior year net Medicaid
DSH. This study also included control variables such as market characteristics (i.e., HMO penetration, Herfindahl Index), hospital supply factors (i.e., system affiliation, teaching hospital, payer mix), demand factors (i.e., per capita income, total population and unemployment rate within county), state health policy factors (i.e. state county appropriation for individual hospitals, Medicaid eligibility within county), and time trend variables.
Population Studied: The sample included nonfederal, short-term acute care hospitals in California that were in operation between 1996 and 2003. We excluded closed hospitals and those that experienced an ownership conversion during the study period. We also eliminated hospitals with financial data reported for a fiscal period less than 9 months. The final sample size was 2,302 hospitalyear observations, including 155 public (nondistrict) hospitals, 330 district hospitals, 540 forprofit hospitals and 1,277 of not-for-profit hospitals.
The primary data sources were the detailed hospital annual financial data from California Office of
Statewide Health Planning and Development,
American Hospital Association Annual Survey, Area
Resource File and Interstudy HMO Data.
Principal Findings: After controlling for market factors, hospital supply factors, demand factors, and other state health policy factors, there is a positive but insignificant relationship between
Medicaid DSH payments and the provision of hospital uncompensated care.
Conclusion: The result suggests that reductions in
Medicaid DSH payments after the 1997 BBA may not have had a large negative impact on hospital uncompensated care provision.
Implications for Policy, Delivery or Practice: It is widely believed that safety net hospitals rely on
DSH payments to provide charity care for uninsured and underinsured populations. Our results suggest that substantial cuts in charity care did not result after the 1997 policy changes. However, better measures and statistical techniques may be needed to assess this potential relationship.
● Perverse Incentives in the Medicare Part D
Prescription Drug Plan Market
John Hsu, M.D., M.B.A., M.S.C.E.; Jie Huang,
Ph.D.; Vicki Fung, Ph.D.; Richard Brand, Ph.D.;
William Dow, Ph.D.; Joseph Newhouse, Ph.D.
Presented by: John Hsu, M.D., M.B.A., M.S.C.E.,
Director, Kaiser Permanente Center for Health
Policy Studies, 2000 Broadway, 3rd Floor, Oakland,
CA 94612, Phone: (510) 891-3601; Email:
John.T.Hsu@kp.org
Research Objective: Medicare Part D plans receive prospective payments adjusted by a diagnosis-based risk-score (RxHCC score), and by a multiplier for low income subsidy (LIS) beneficiaries. We examined the performance of these two components of Part D prospective payments.
Study Design: We used data from Part D
Prescription Drug Plans (PDPs). To assess the LIS multiplier, we examined the ratios of plan liabilities for LIS to non-LIS beneficiaries. To assess the performance of the RxHCC score, we used ordinary least squares regressions to predict plan liability.
Population Studied: Medicare beneficiaries continuously enrolled in a stand-alone Prescription
Drug Plan in 2007, with at least some enrollment in
2006, both with and without the LIS.
Principal Findings: The mean plan liability ratio of full subsidy to non-LIS beneficiaries was 1.23, and
1.11 for partial subsidy to non-LIS beneficiaries; the current multipliers are 1.08 and 1.05, respectively.
Among LIS beneficiaries, the RxHCC score explained 24 percent of the plan liability variation; combining the RxHCC score with prior year drug costs explained up to 60 percent of the variation.
Among non-LIS beneficiaries, explained variation increased from 21 percent to 52 percent when adding prior year drug costs. Adding prior year drug information to the risk adjustment approach reduced the observed mean plan liability ratios for
LIS to non-LIS beneficiaries to levels closer to the actual multipliers.
Conclusion: The LIS multipliers are substantially lower than the observe ratios of plan liabilities for
LIS to non-LIS beneficiaries. Including information on prior year drug use improves the performance of the risk adjustment approach for all beneficiaries, and reduces the differences in plan liabilities between LIS and non-LIS beneficiaries.
Implications for Policy, Delivery or Practice: The current LIS multiplier appears to be inadequate and could create distortionary incentives for plans to avoid LIS beneficiaries by increasing Part D premiums. These distortions also could help explain the withdraw of many private plans from the LIS market, but could be ameliorated by raising the LIS multipliers, or better avoided by adding prior year drug information to the risk-adjustment formula.
Funding Source(s): Medicare Payment Advisory
Committee
Theme: Medicare
● Trends and Variation in Vaginal Birth after
Cesarean: 1996-2006
Yalin Huang, M.S.
Presented by: Yalin Huang, M.S., Doctoral
Student, Health Policy & Management, Emory
University, R623, 1518 Clifton Road, Atlanta, GA
30322, Phone: (404) 693-7546; Email: yahuang@emory.edu
Research Objective: Vaginal birth after cesarean has been a subject of debate and its rate changed dramatically last twenty years. American College of
Obstetrics and Gynecology (ACOG) published several VBAC practice guidelines as a reference for physicians and hospitals. Among those, the 1999 guidelines made a significant strict requirement regarding hospitals’ facility and staff ability when providing VBAC. This study examines the impacts of ACOG 1999 practice guidelines on VBAC rates between different types of hospitals to observe how practice guidelines affect practice pattern.
Study Design: We conducted an observational cohort study using hospital discharge data from
HCUP/NIS and reported trends and variations in
VBAC rates among different types of hospitals. We estimated multivariate logistic models with hospital random effects adjusted for maternal and hospital characteristics and reported marginal effects on the probability of VBAC. We also examined how teaching hospitals and rural hospitals respond to the 1999 guideline by including interaction terms.
Population Studied: 813,647 deliveries with a previous cesarean discharged from community hospitals in the U.S. between January 1996 and
December 2006.
Principal Findings: National VBAC rate dropped from 27.6% in 1998 to 15.6% in 2001, and was only
6% in 2006. We found that the variation of VBAC
rates between patients and between hospitals shrink as the average rates of VBAC decline. The variations were especially large in non-teaching, rural and public hospitals. The issuing of the ACOG
1999 guidelines yield a 0.74% decrease in the national VBAC rate between 1996 and 2006, while both teaching and rural hospitals yield about 1% increase in the rate of VBAC after the guidelines. In addition, we found that a a 1% increase in average malpractice payment leads to a 0.54% decrease in
VBAC rate after the issuing of the guidelines.
Conclusion: Our study showed ACOG 1999 guidelines does not have a significant effect on the decreasing of VBAC rate. That is, facility ability is not the only concern for hospitals to decide whether or not to provide VBAC. While our analysis showed negative relationship between regional malpractice pressure and VBAC rate, we believe the decreasing rate of VBAC is an example of defensive medicine in the U.S.
Implications for Policy, Delivery or Practice:
While there is no sufficient scientific evidence, a decision of VBAC falls more on mothers or physicians’ preferences than clinical factors. The guideline issued in 1999 by ACOG focusing on the facility and staff ability limit women’s options to the mode of delivery. Our evidence suggests that in addition to the guidelines, higher malpractice cost can be related to lower willingness of providing risky procedure with potentially high litigation cost. To encourage hospitals to provide VBAC, ACOG guidelines should make more specific recommendations regarding clinical conditions. In addition, tort reform with capping damage reward can also ease the malpractice pressure that obstetricians face.
● Distribution of Qualified Language Services in
Urban & Rural Hospitals
Jennifer Huang, M.S.; Christal Ramos, M.P.H.;
Karen Jones, M.S.; Marsha Regenstein, Ph.D.
Presented by: Jennifer Huang, M.S., Research
Scientist, Department of Health Policy, George
Washington University, 2021 K Street, NW, Suite
800, Washington, DC 20006, Phone: (202) 994-
8661; Email: jenny.huang@gwumc.edu
Research Objective: With the increasing linguistic diversity of patients in the U.S., there is more need for bilingual providers in the healthcare workforce.
Given financial pressures facing hospitals, there are different challenges to providing language services for urban and rural hospitals. We conducted a study to learn about who provides language services in hospitals. The purpose of this analysis is to determine if there are differences between rural and urban areas in the distribution of qualified language services.
Study Design: We developed a telephone survey for human resources (HR) directors. The datasource was completed telephone surveys, fielded in 2008. We generated descriptive statistics on hospital characteristics (ownership, teaching status, bedsize, urban or rural market) and type of language services workforce used (staff vs. volunteer). As a proxy for “qualified” language services, we examined hospital policies about using minors, friends or family, or non-trained staff as interpreters, hospital policies about using bilingual providers as interpreters, and utilization of language assessments for bilingual providers. We examined the relationship between market and type of language services provider used, use of policies, and utilization of language assessments.
Population Studied: A total of 899 HR directors from non-federal, acute care hospitals completed the survey, representing approximately 20 percent of all hospitals in the U.S. The sample included 449 hospitals from rural markets and 450 hospitals from urban markets. An oversample of California hospitals were surveyed, with 159 respondents from California. The nationwide sample was adjusted by ownership, teaching status, market, and geography (CA vs. not) to reflect the make-up of the hospital industry.
Principal Findings: Compared to urban hospitals, rural hospitals rely less on staff interpreters (21.6% vs 36.3%,p<.01) and more on volunteers for language services (55.7% vs 46.0%,p<.05), not surprising given that rural hospitals are less likely to have large volumes of encounters with limited-
English proficient patients. Rural hospitals are significantly less likely to have policies about using informal interpreters to provide language services than urban hospitals (28.2% vs 53.4%,p<.01), and less likely to have policies about using bilingual providers as interpreters (17.8% vs 38.6%,p<.01).
Rural hospitals were less likely than urban hospitals to require a language assessment for staff interpreters (66.9% vs 87.8%,p<.01) or volunteer interpreters (33.3% vs 61.5%,p<.01), and to offer an assessment to bilingual providers (7.0% vs
26.1%,p<.01).
Conclusion: Our analyses show differences in types of interpreters used between urban and rural markets, as well as differences in the qualifications of language services providers. Our findings suggest that the quality of language services provided in hospitals may vary depending on hospital location.
Implications for Policy, Delivery or Practice:
Numerous research projects have demonstrated the importance of using language services providers assessed for language fluency in order to reduce medical errors and improve communication between patients and healthcare providers.
Reliance on a language services workforce that has not been assessed for language fluency has critical implications for health care encounters, and ultimately, the quality of care for patients with language needs. Further investigation is necessary to determine how hospitals in different markets can
increase their use of qualified language services to address the needs of their patients.
Funding Source(s): The Califonia Endowment
Theme: Health Care Workforce
● Examining Differences in Performance of
Taiwan's Hospitals Across Types of Adaptive
Strategies under the Global Budget Payment
Scheme
Kuo-Cherh Huang, Dr.PH, M.B.A.; Wang-Ping Lee,
M.B.A.; Ning Lu, Ph.D., M.P.H.; Hui-Chih Chang,
M.H.A.; Jiun-Shyan Chen, M.B.A.; Hsyien-Chia
Wen, Ph.D.
Presented by: Kuo-Cherh Huang, Dr.PH, M.B.A.,
Associate Professor, Health Care Administration,
Taipei Medical University, 250 Wu-hsing Street,
Taipei, 110, TW, Phone: +886-2-27361661; Email: kchuang@tmu.edu.tw
Research Objective: Taiwan's hospitals have been under constant pressure to adapt to environmental change mainly resulting from an evolutionary payment scheme undertaken by the Bureau of
National Health Insurance. The study aimed to investigate corresponding strategic adaptations chosen by Taiwan's hospitals, and further examine differences in hospital performance across types of adaptive strategies.
Study Design: Self-administered questionnaire survey.
Population Studied: The study population included
461 district hospitals (and above) in Taiwan. Upper level managers of the hospitals such as superintendents, vice-superintendents, or other upper-level administrators who were knowledgeable of their hospital practice were explicitly asked to respond to the survey, to ensure the accuracy and reliability of the data collected. After two rounds of mailing, the final effective sample size was 182, with the effective response rate of 39.5%.
Principal Findings: 1. The results of factor analysis identified two types of adaptive strategies: market-oriented and resource-based strategies and the total explained variance by the two factors was
62.13%. Moreover, sample hospitals were more likely to execute market-oriented strategies. 2.
Logistic regression analysis revealed that predicting factors of hospitals' carrying out different types of adaptive strategies included hospital ownership, hospital level and teaching status. 3. Contradictory to the authors' expectation, the degree of local market competition had no impact on hospitals' choice of adaptive strategies. 4. With respect to hospital performance, findings revealed that market-oriented strategies fared better than resource-based strategies.
Conclusion: This investigation has revealed that
Taiwan's hospitals employ adaptive strategies of market-oriented and resource-based under the global budget payment scheme, and the former is the mainstream and performs better as well. Taken together, the study's results shed light on some interesting issues regarding hospitals' adaptive strategies in response to dramatic changes in the external environment, and provide some suggestions for hospital managers.
Implications for Policy, Delivery or Practice:
This empirical research embarks on the issues of adaptive strategies of hospitals when they are encountering significant environmental changes, such as the implementation of the global budget payment scheme. The current research also examines the differences in hospital performance across types of adaptive strategies. From a management perspective, it is imperative for hospital executives to embrace a more suitable strategy in order to survive in today's highly competitive healthcare market. Taken together, the study's results shed light on some interesting issues regarding hospitals' adaptive strategies in response to dramatic changes in the external environment, and provide some suggestions for hospital managers.
Funding Source(s): National Science Council,
Taiwan.
● A Medical Home for Children with Insulin-
Dependent Diabetes: Parental Views of Primary
& Subspecialty Physician Roles
Charles Humble, M.S.P.H., Ph.D.; Steven Wegner,
M.D., J.D.; Christine Lathren, M.D., M.S.P.H.;
Michelle Mayer, M.P.H., Ph.D.; Alan Stiles, M.D.;
Presented by: Charles Humble, M.S.P.H., Ph.D.,
Director, Analytic Services, AccessCare, 3500
Gateway Centre Boulevard, Suite 130, Morrisville,
NC 27560, Phone: (919) 380-9962; Email: chumble@ncaccesscare.org
Research Objective: To examine parental views of primary care physicians’ (PCP) and endocrinologists’ roles in the management of routine preventive and acute care, diabetes-specific care, and family education and care coordination for children with insulin-dependent diabetes (IDDM).
Understanding how parental views align with those of PCPs and subspecialists is important for building a more efficient medical home.
Study Design: Collaboration within the medical home model of care is recognized as a critical element of care by the American Academy of
Pediatrics, by Healthy People 2010 goals and by the New Freedom Initiative’s core outcomes. The benefits of collaboration between PCPs and subspecialists can include improvements in access, reduced frequency and inappropriateness of referrals, better communication, improved outcomes and satisfaction, and lower costs. This study used a mixed-mode survey of parents of children with IDDM in one Medicaid Managed Care
Network in North Carolina. Results were compared with previously-described provider attitudes regarding best sources of care. The survey
examined routine preventive and acute care, diabetes-specific care, family education and care coordination. Parents were asked which physician type they usually use for 17 specific services: the
PCP (“regular doctor”), “diabetes doctor”, or both.
The remainder of the survey focused on demographics and more general factors influencing parental decisions regarding where to seek care.
Completed surveys were returned by mail to ensure confidentiality of responses.
Population Studied: The target sample was parents of 249 pediatric patients prescribed insulin by member physicians participating in AccessCare, a provider-owned, not-for-profit medical home managed care organization serving more than
200,000 Medicaid child and adult enrollees statewide. Children with IDDM were selected because they require ongoing contact with PCPs and subspecialists to improve short- and long-term outcomes. The response rate was 43.4%.
Principal Findings: A majority of parents reported usually taking their children with IDDM to PCP for all aspects routine primary care queried and for referrals to eye doctors and mental health providers. The subspecialist was the preferred provider for most aspects of diabetes-specific care, family education, referrals to nutritionists and for talks with school or daycare personnel.
Preferences were more divided as to where to perform kidney and cholesterol checks. High proportions of parents said they would be willing to travel “a long way” to a subspecialist for serious problems. Most parents have discussed which doctor to see in different situations with both types of providers. Parent and provider attitudes regarding best sources of care generally agree.
Conclusion: An effective medical home model of care depends on establishing clear lines of responsibility between the PCP and subspecialist and sharing this information with patients and caregivers. Our findings suggest that parents and doctors have clear preferences for where to seek different types of care for diabetic children and tend to agree on these preferences for most aspects of care.
Implications for Policy, Delivery or Practice:
Knowledge of parental preferences regarding division of care between PCPs and subspecialists can facilitate co-management when it is needed and optimize care for patients with IDDM in their medical homes.
Funding Source(s): Department of Pediatrics, NC
Children's Hospital
Theme: Quality and Efficiency: Organized
Processes
● Managing Atrial Fibrillation Data Post Maze
Procedure: Registry Design, Cost &
Effectiveness
Sharon Hunt, M.B.A.; Linda Henry, Ph.D., R.N.;
Lisa Martin, Ph.D.; Niv Ad, M.D.
Presented by: Sharon Hunt, M.B.A., Senior
Database Administrator, Cardiac Surgery
Research, Inova Health Systems, 3300 Gallows
Road, Falls Church, VA 22042, Phone: (703) 776-
7070; Email: sharon.hunt@inova.org
Research Objective: Atrial fibrillation impacts over
3 million people in the United States and this rate is expected to increase to 5.6 million by 2050.
Patients with longstanding atrial fibrillation are often unresponsive to medical management resulting in the need for surgery. The maze surgery is proving to be an effective tool for restoring sinus rhythm
(SR) for those patients resistant to other treatments, but long-term evaluation of outcomes and long-term effectiveness of this surgery is not readily available.
Three years ago we designed a Maze Registry at our institute and implemented a prospective longitudinal follow-up system of information for the purpose of; 1) identifying risk factors that impact short- and long-term outcomes, 2) measuring the impact of surgery on quality of life (HRQL), 3) determining effectiveness of a clinical follow-up protocol on long term results, and 4) developing a longitudinal data system that can be used to answer research questions while maintaining operating costs.
Study Design: In 2005, we designed a maze surgery registry to capture clinical data elements associated with atrial fibrillation, both historical and during the hospital stay that were not being captured by our standard cardiac surgery registry. A prospective long-term follow-up system was then designed to capture cardiac rhythm, HRQL (SF12), events/medications at three, six, nine, twelve, eighteen, twenty-four and thirty-six months by mailed surveys with clinical data validated with documentation from physician’s office. Intermittent clinical follow-ups and events were also captured.
Dates of all treatments, events and cardiac rhythm provide a timeline for analysis and details provide the investigator with the tools to answer broad as well as very specific research questions. Domains of data are pre-hospital, hospital, discharge and follow-up for three years.
Population Studied: Includes 334 patients with
1605 clinical follow-up measurements with mean follow-up of 25.1 (13.8) months. We add 8 patients per month and solicit follow-up on 50 patients per month.
Principal Findings: Incorporating unique domain elements (pre-hospital n=168; hospital/discharge n=280; and follow-up n=308 at each time-point), an evidence based post-discharge protocol was developed to coordinate care with the outside cardiologists. Startup labor costs for designing forms and database implementation for this size of program is less than $20,000; equipment/software cost is $6500. Dedicating 55 hours per week a nurse coordinator and database administrator are able to manage data. 12 /24 month return to sinus rhythm rate is 84 and 85%. There are 94
cardioversions, 123 cardiac readmissions and 21 pacemaker insertions post discharge. Male HRQL scores significantly improve from baseline to 6 months (physical (p=0.002) and mental (p=0.04)).
136 letters were generated to physicians for treatment recommendations. Protocol effectiveness post implementation is validated at 12 months when patients on protocol return to SR at a higher rate than patients not on protocol (88% vs. 76%).
Conclusion: With increasingly expensive technological advancements, it is crucial to monitor patient outcomes beyond hospital stay. Long-term follow-up allows development of evidence based clinical protocols to improve outcomes reducing the need for later rescue treatments.
Implications for Policy, Delivery or Practice:
Small programs can effectively manage long-term follow-up without prohibitive costs.
Theme: Quality and Efficiency: Measurement
● Administrative Claims-Based Lumbar Surgery
Prediction Index (LSPI)
Constance Hwang, M.D., M.P.H.; Shantanu
Phatakwala, M.S.; Lisa Taylor, R.H.I.A.; Earl
Steinberg, M.D., M.P.P.
Presented by: Constance Hwang, M.D., M.P.H.,
Director of Clinical Research & Development,
Clinical Affairs, Resolution Health, Inc., 10490 Little
Patuxent Parkway, Suite 610, Columbia, MD
21044, Phone: (240) 295-1398; Email: chwang@resolutionhealth.com
Research Objective: To help target low back pain
(LBP) patient outreach efforts, we developed the
Lumbar Surgery Prediction Index (LSPI) to identify health plan members with uncomplicated, mechanical LBP who exhibit health care utilization patterns that suggest an increased likelihood that low back surgery (LBS) will be performed in the near future.
Study Design: Using administrative claims data, we developed a logistic regression model to identify patients with the highest odds of undergoing low back surgery, defined as laminectomy, discectomy, spinal fusion, or disc arthroplasty, within 4-26 months after an index office visit for LBP. We developed a scoring system to assign points based on the relative strength of significant variables in the final model. Both demographic (e.g. age, gender, zip code) and clinical variables (e.g. prescribed medications, diagnostic imaging procedures, physical therapy, pain management injections, presence of sciatica, depression, and comorbid burden as measured by the Charlson score) were considered. We used the split-sample method for model development and validation, by randomly allocating the members into development (67%) and validation (33%) samples.
Population Studied: We analyzed 2005-2007 administrative claims data for 1.3 million commercial health plan members, identifying those with mechanical low back problems in the absence of a neoplastic, infectious or inflammatory etiology, or pregnancy, major trauma, prior back surgery, or hospice services.
Principal Findings: Approximately 67,051 members (5.6%) were identified as having a new episode of uncomplicated low back pain between
July 1, 2005 and Aug. 31, 2006. Among these patients, 632 (1.0 %) underwent lumbar spine surgery between 4 and 26 months after the index visit. Members who underwent surgery were more likely to be male, age 51 years or older, and to have received particular diagnostic and therapeutic services. For example, patients who underwent
LBS were 10, 7.5, and 2 times more likely to utilize pain management injections, MRI imaging, and narcotic medications, respectively, relative to members who did not undergo LBS. Variables in our final model included age, gender, pain medications, pain management injections, radiologic imaging, office visits, and hospitalizations related to LBP. The optimal LSPI threshold score was associated with a sensitivity (Se) of 16.9% and a positive predictive value (PPV) of 15.1%. The performance of the model during validation was comparable to the results in the development dataset (17.4% Se, 15.7% PPV).
Conclusion: We developed and validated a
Lumbar Surgery Prediction Index (LSPI) that requires only administrative claims data to identify individuals with an increased likelihood of undergoing low back surgery 4-26 months following the first indication that a patient has a new episode of low back pain.
Implications for Policy, Delivery or Practice:
Most health plans find out about potential LBS cases when a physician or member requests prior authorization for a scheduled surgery. Because roughly 1 in 6 people identified by our LSPI as high risk will undergo LBS within 4-26 months, our LSPI enables a health plan to target LBP patients for more patient-centered, informed decision making regarding surgical vs. non-surgical therapies, and for more pro-active and cost-effective interventions.
● How Do Providers Manage their Diagnostic
Test Result Alerts? Implications for Medical
Error Reduction Through Electronic Medical
Records
Sylvia Hysong, Ph.D.; Lindsey Wilson, M.A.; Mona
Sawhney, M.B.B.S., M.T.; Donna Espadas, B.S.;
Dean Sittig, Ph.D.; Hardeep Singh, M.D., M.P.H.
Presented by: Sylvia Hysong, Ph.D., Health
Services Researcher, Houston VA Health Services
Research & Development Center of Excellence,
Michael E DeBakey VA Medical Center, 2002
Holcombe Boulevard (152), Houston, TX 77030,
Phone: (713) 794-8616; Email: hysong@bcm.edu
Research Objective: Integrated electronic medical records (EMRs) that notify providers about
abnormal diagnostic test results through “alerts” have potential to reduce errors of lack of timely patient follow-up. However, recent literature shows abnormal outpatient test results continue to be missed in systems that use computerized notifications, such as Veterans Affairs’ (VA’s)
Computerized Patient Record System (CPRS).
Some of these oversights could be avoided by using systematic strategies to manage all types of alert notifications. We determined the knowledge and use of standardized alert management strategies by primary care providers who received critical diagnostic imaging alerts on their EMR screens.
Study Design: We used techniques from cognitive task analysis to individually interview primary care providers on how and when they manage alerts received in CPRS, with specific emphasis on the strategies they use to filter alerts, reduce unnecessary alerts, and sort alerts for easier processing in their inbox. Interviews focused on four
CPRS features: sorting by alert field names for easier processing (sorting), individually customizing the reception of non-mandatory alerts according to provider preferences (notification), accelerated alert processing via appropriate use of batch processing tools (“process all”), and forwarding alerts to other providers during the ordering provider’s vacation or travel (surrogate). Participants were purposively sampled according to their high or low rates of lack of timely follow-up based upon our findings in a previous study.
Population Studied: Twenty-eight primary care providers (6 residents, 11 attending physicians, 11 allied health professionals) from a large, southern tertiary care Veterans Affairs Medical Center
(VAMC).
Principal Findings: Respondents reported being aware of each alert management feature in varying frequencies: sorting, 50%; notification, 75%; process all, 64%; surrogate, 75%. Forty-six percent of providers used none of the four techniques, and no provider used more than two techniques.
Providers with higher versus lower rates of nontimely follow-up used the four features similarly, except the notification feature; 43% of timely followup providers (n=7) reported being unaware of the notification feature, compared to none of the nontimely follow-up providers (chi-sq(1)=10.09, p=.039). Only one timely follow-up provider, compared with 42% of non-timely follow-up providers (n=5), manually scanned the alert list and processed alerts according to their judgment of clinical priority or urgency (chi-sq(1)=5.19, p=.024) rather than using systematic alert management features. Twice as many providers used multiple schedules to process alerts (e.g., both in the morning and between patients, the most commonly reported schedules) compared to the number of providers using a single schedule (e.g., only in between patients). Although seven different schedules were reported overall, the timely and non-timely groups did not significantly differ in the number of alert management schedules used (chisq(3)=1.65, p=.65). Conclusion: Considerable heterogeneity exists in provider use of alert management strategies and specific strategies may be associated with higher rates of lack of timely follow-up. Providers may lack knowledge about tools in their EMR to help them manage their critical diagnostic alerts more efficiently.
Implications for Policy, Delivery or Practice:
Standardization of alert management strategies including improving provider knowledge of appropriate tools in the EMR to manage alerts could reduce lack of timely follow-up of abnormal diagnostic test results.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Does Linking Performance Measures to
Hospital Budgets Improve Their Outpatient
Clinical Performance?
Sylvia Hysong, Ph.D.; Myrna Khan, Ph.D., M.B.A.
Presented by: Sylvia Hysong, Ph.D., Health
Services Researcher, Houston VA Health Services
Research & Development Center of Excellence,
Michael E. DeBakey VA Medical Center, 2002
Holcombe Boulevard (152), Houston, TX 77030,
Phone: (713) 794-8616; Email: hysong@bcm.edu
Research Objective: The health care industry has invested substantial resources to develop clinical quality measures that monitor healthcare facilities’ performance. More recently, hospital performance on said measures is increasingly tied to pay-forperformance initiatives, with performance measures often used to drive budgets. However, sometimes specific measures are retired when high performance levels have been sustained. At the
Veterans Administration (VA) when a performance measure is retired, it is monitored as a “support indicator” and no longer impacts facility budgets.
Whether facilities sustain high performance levels once a performance measure becomes a support indicator or vice versa is unknown. Nor is it known how long performance levels take to stabilize after a change in status. Thus, this research seeks to compare the nature and rate of change in hospital outpatient clinical performance as a function of a measure’s evaluation status (performance measure vs. support indicator). We also examined the mean stabilization time for measures after changing status.
Study Design: This research is a database review consisting of quantitative, longitudinal, retrospective analyses of outpatient clinical performance measure data from VA’s External Peer Review
Program (EPRP). We studied 17 measures covering five clinical areas: screening, immunization, chronic care following acute myocardial infarction (AMI), diabetes mellitus, and hypertension over varying quarters (minimum of
eight) between 2000 and 2006. Data were analyzed using hierarchical linear modeling and segmented regression to test for significant differences in performance before and after a change in evaluation status from performance measure to support indicator and vice versa.
Population Studied: 133 VA Medical Centers across the country.
Principal Findings: All measures either improved or remained stable over time regardless of status change. Of the six measures that changed from support indicator to performance measure, only two exhibited significant increases in slopes after the change (HbA1C <9, B_b=-0.005, p=0.004,
B_a=0.009, p<0.001; colorectal cancer screening,
B_b=-0.01, p<0.001, B_a=0.02, p=0.007*). Of the
11 measures that changed from performance measure to support indicator, four exhibited significant differences in slope; two exhibited positive slopes before the change followed by negative slopes (lipid profile every 2-years,
B_b=0.009, p<0.001, B_a=-0.007, p<0.001; and depression screening, B_b=0.01, p<0.001, B_a=-
0.01, p<0.001); two exhibited the opposite pattern
(diabetic foot inspections, B_b=-0.004, p=0.004
B_a=0.004, p=0.005; and pedal pulses, B_b=-
0.002, p=0.19 B_a=0.003, p=0.03). The remaining measures exhibited no differences in slope after the change, indicating performance levels were sustained. The amount of time required to reach stability after a change varied considerably, with some measures not reaching stability by the end of our sample period. * B_b = Regression weight before change in status; B_a = Regression weight after change in status.
Conclusion: Linking facility budgets to performance measures does not appear to directly impact clinical performance.
Implications for Policy, Delivery or Practice:
This study could indicate facilities respond to changes in VA’s performance management system with positive outcomes, irrespective of the financial impact; however, the observed effects could be explained by performance targets. Future research should determine the independent effects of performance targets and facility budget incentives, as well as the moderating effects of measure and facility characteristics.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Vaccination Decision-Making When
Interactions with Other Individuals are Allowed:
Does the Free-Riding Behavior Exist?
Yoko Ibuka, Ph.D.; Meng Li, M.S.; Jeffrey Vietri,
M.S.; Gretchen Chapman, Ph.D.; Alison Galvani,
Ph.D.
Presented by: Yoko Ibuka, Ph.D., Postdoctoral
Associate, Epidemiology & Public Health, Yale
University, 60 College Street, New Haven, CT
06520, Phone: (908) 420-3582; Email: yoko.ibuka@yale.edu
Research Objective: Because vaccination generates herd immunity, an individual’s decision regarding vaccination may be influenced by that of others in the community. Specifically, individuals may have a “free-rider incentive” to avoid the cost of vaccination and benefit from other vaccinated individuals. This study examined the existence of free-riding behavior through an experiment on vaccination decision-making that allowed interactions with others.
Study Design: Our computerized game experiment presented a fictional scenario where players decided whether they would get vaccinated against
“the flu”. 29 groups of eight to ten players played 24 rounds. After each round, players were informed of the number in their group who had chosen vaccination and of those infected. Players were initially given 2,000 points, and lost points when they were vaccinated and more points if infected.
The probability of infection depended on the vaccination rate in the group and on each individual’s vaccination status. Getting vaccinated reduced an individual’s risk of infection by half. The final points at the end of the game were translated to real monetary payouts ranging from zero to ten dollars. Our hypothesis was that the probability of vaccination acceptance would decrease as observed vaccination rate increases if free-riding behavior exists. We assessed the relationship between individual vaccination decision-making and the proportion of vaccination among other group members during the previous round. We also tested two additional hypotheses that a player’s previous flu experience, or the proportion of those infected among other group members, would positively affect vaccination decision-making. The analysis was conducted by logistic regressions incorporating the hierarchical structure of the game. Our observations consisted of 6,456 data points from
269 participants corresponding to a total of 29 groups.
Population Studied: The experiment was conducted with 269 college students at Rutgers
University.
Principal Findings: The overall vaccination rate within our game was 53 percent. Observed higher vaccination rates in the group from the previous round were associated with lower probability of vaccination acceptance (p<0.01). Individual flu experience, measured by the total number of infections the player had experienced, was positively associated with accepting vaccination
(p<0.01). The relationship between flu exposure and vaccination decision-making was tested using two alternative measures: binaries to represent if the player was infected in the previous round and if the player had ever been infected. We found that the effect was robust in both cases. Vaccination decision-making was not significantly affected by
the disease prevalence during the previous round
(p=0.47).
Conclusion: Our experiment found that participants’ choices regarding vaccination depended on others’ decisions as well as their own flu experience. Specifically, our results showed that the participants’ probability of getting vaccinated decreased as the vaccination rate in the group during the previous round increased, which suggests free-riding behavior.
Implications for Policy, Delivery or Practice: If our evidence can be generalized, an individual’s decision regarding influenza vaccination could be influenced by observed vaccination coverage in the community. It may be important to provide additional incentives for individuals to get vaccinated when high vaccination coverage has already been achieved.
Funding Source(s): National Science Foundation
Theme: Consumer Choices in Health Care
● A Community-Based Study of a Lifestyle
Intervention to Prevent Type 2 Diabetes in
People with Impaired Glucose Tolerance
Hirohisa Imai, M.D., Ph.D.; Hiroyuki Nakao, Ph.D.;
Yuchiro Yahata, Ph.D.; Fumihiro Sata, M.D., Ph.D.;
Yoshiharu Fukuda, M.D., Ph.D.
Presented by: Hirohisa Imai, M.D., Ph.D., Director,
Epidemiology, National Institute of Public Health, 2-
3-6 Minami, Wako, Saitama, 351-0197, JP, Phone:
+81-48-458-6167; Email: imaihiro@niph.go.jp
Research Objective: The incidence of type 2 diabetes mellitus is increasing worldwide.
Individuals with impaired glucose tolerance (IGT) are at high risk for developing type 2 diabetes mellitus. Type 2 diabetes mellitus results from the interaction of genetic and behavioral and environmental risk factors. Obesity and a sedentary lifestyle are the main nongenetic determinants of the disease. Prophylactic interventions that affect the lifestyles of subjects at high risk for the disease are not known. The purpose of this study was to assess whether this program would result in improved glucose levels. This intervention is a community-based self-control education program to change the lifestyles of local residents with IGT. We developed a lifestyle intervention program that included self-monitoring of blood glucose, dietary and physical activity counseling, and personal interviews with poorly controlled IGT subjects.
Study Design: This was an intervention study. A lifestyle intervention program for individuals with
IGT was conducted for six months in a town in northern Japan. Hemoglobin A1C (HbA1c), fasting blood glucose (FBG), and BMI were positioned as the main outcomes and were measured prior to conduction and at the end of the program.
Population Studied: Individuals identified in a general health check for local residents as having
IGT were invited to join the study as participants. A total of 112 individuals (61 male, an average age of
62.1 years. 51 females, an average age of 61.7 years) initially participated in baseline data collection after giving their informed consent. They were enrolled in a once-monthly lifestyle-change class in our program. The program was implemented at the health center of the town.
Principal Findings: We found the following differences in between baseline and half a year in the outcome variables for participants enrolled in the program: average hemoglobin A1C: 5.40% vs.
5.27% (p<0.0001); average FBS: 110.4 mg/dl vs.
104.8 mg/dl (p<0.05), average BMI: 26.4 vs. 25.9
(p<0.0001). All clinical outcomes exhibited statistically significant differences between baseline and the end of the study.
Conclusion: These results suggest that the lifestyle intervention program implemented was effective at encouraging better self-control in subjects with IGT in Japan.
Implications for Policy, Delivery or Practice:
Community-based, preventative lifestyle intervention programs are not a common healthcare delivery approach in Japan, even though they have been implemented in the United States of America as well as in Europe. The results of the lifestyle intervention program including self-monitoring of blood glucose, dietary and physical activity counseling, and personal interviews, demonstrated its feasibility, as well as clinical efficacy, in poorly controlled IGT subjects in Japan.
Funding Source(s): Japan Science and
Technology Agency
● The Relationship Between State Public Health
System Performance & Certain State Health
Agency Characteristics
Richard Ingram, M.Ed.; Ariel Langevin, B.A.; Martha
Riddell, Dr.P.H.; Michelyn Bhandari, Dr.P.H.;
William Mase, Dr.P.H.
Presented by: Richard Ingram, M.Ed., Research
Assistant, Center for Public Health Systems &
Services Research, University of Kentucky, 121
Washington Avenue Room 101, Lexington, KY
40536, Phone: (859) 218-2020; Email: rcingr2@uky.edu
Research Objective: This study seeks to examine the relationship between state public health system performance in assuring each of the 10 Essential
Public Health Services (EPHS) and total performance, as measured by Version 1.0 of the
National Public Health Standards Program
(NPHPSP) State Public Health System
Performance Assessment instrument, and various state public health agency characteristics contained in the 2005 Association of State and Territory
Health Officials (ASTHO) Salary and Agency
Infrastructure survey.
Study Design: Appropriate statistical analyses are used to examine the data. Pearson’s Correlation
Coefficient is used to examine the relationship between performance and numerical ASTHO survey variables. ANOVA is used to examine the relationship between performance and categorical
ASTHO survey data. Independent t- tests are used to examine the relationship between performance and dichotomous ASTHO survey variables.
Population Studied: The population studied is the sixteen states that have state health agencies that have filled out both Version 1.0 of the NPHPSP state assessment instrument and the 2005 ASTHO survey.
Principal Findings: Performance on EPHS 1 and 2 has a significant positive association with the number of full time equivalents (FTEs) in a state health agency. The state contribution to the state health agency budget and a state statute for the position of the state health officer are both significantly positively associated with performance on EPHS 3. Performance on EPHS 4 is significantly positively associated with who in the state government appoints the state health officer, who that officer reports to, having the state
Secretary for Health and Human Services involved in the budget, and agency structure. There is a significant positive association between performance on EPHS 5 and both the state statute for the position of the state health officer and the background (MD or DO) of the state health officer.
Performance on EPHS 7 is significantly positively associated with the number of years the state public health officer has been in public health. The presence of a state health officer who is board certified is significantly positively associated with performance on EPHS 8. Performance on EPHS
10 is significantly positively associated with the annual budget of the state health agency (excluding
Medicaid dollars) and the state contribution to the state health agency budget.
Conclusion: Certain state health agency characteristics are positively associated with state public health system performance in assuring some of the 10 EPHS, as measured by the NPHPSP state instrument. The results of this study are limited by the relatively small number of agencies examined, and the cross sectional nature of the data examined. More states should be encouraged to complete future versions of the NPHPSP, and further and more detailed analysis should be performed to examine the relationships uncovered in this study.
Implications for Policy, Delivery or Practice: The significant positive associations between certain state health agency characteristics and performance in assuring certain EPHS suggests that agencies that have a deficiency in these EPHS may wish to examine the agency characteristics that are positively associated with these EPHS, and seek to effect change in performance through manipulating these characteristics.
Theme: Public Health
● Composition of Medical Staff in Nursing
Homes
Orna Intrator, Ph.D.; Michelle Krukas, B.Sc.;
Vincent Mor, Ph.D.; Paul Katz, M.D.; Jurgis Karuza,
Ph.D.; Joan Teno, M.D.
Presented by: Orna Intrator, Ph.D., Associate
Professor (Research), Center for Gerontology &
Health Care Research, Brown University, Box G-
S121-6, Providence, RI 02912, Phone: (401) 863-
3579; Email: Orna_Intrator@Brown.Edu
Research Objective: Limited physician involvement precludes their full integration into the culture of the nursing facility and potentially impedes interdisciplinary communication and treatment.
Residents consequently suffer undesirable outcomes, such as avoidable hospitalizations, inappropriate prescriptions, and inadequate care.
This paper explores methods to measure the composition of medical providers billing Medicare for services of nursing home residents.
Study Design: Cohort study of nursing homes whose medical directors responded to a survey sent to a randomly selected sample of the AMDA membership. All Medicare enrollment, and part A and B claims, as well as nursing home assessments were used, and were linked to nursing home certification data.
Population Studied: Medicare beneficiaries who were nursing home residents in any of 202 cohort nursing homes in 2006.
Principal Findings: There were 61,647 fee-forservice Medicare beneficiaries in the study cohort.
Over 4.8 million part B claims were filed for them from over 162 thousand providers. Using only those claims occurring during a nursing homes stay, a study derived categorization of provider specialty showed that: a. of claims made during a continued non-SNF nursing homes stay, 46.5% of claims were for diagnostic services, 27% for primary care, 14% for specialist care, and 5% for mental health; (b) of claims made during a SNF stay, 46% were for primary care, 15% for specialist services, and 12% for diagnostics. Primary care providers billing included 47% for nursing home related CPT codes and 37% for other geriatric care, while specialists billed for 11% nursing home related CPT codes and
37% other geriatric care. Only 12,004 unique primary care providers were identified providing nursing home CPT billed care. These are potentially the resident nursing home medical staff, referred to as "Attending".
Conclusion: This paper is the first to show that it is possible to identify attending nursing home medical staff using billing data. Using this information it is possible to examine geographic variability and organizational characteristics associated with the composition of medical staff in nursing homes.
Implications for Policy, Delivery or Practice: The methods developed in this paper will allow
researchers and policy makers to examine the association between the composition of medical staff, nursing home quality, and resident outcomes, potentially leading to identification of best nursing home practices for medical staff.
Funding Source(s): NIA
Theme: Long Term Care
● Evaluating Country Health System
Performance Using Consumer Health Attitude
Segmentation
Benjamin Isgur, M.P.A.; David Chin, M.D., M.B.A.;
Sandy Lutz, B.A.; Julie Weissman, M.A.
Presented by: Benjamin Isgur, M.P.A., Director,
Health Research Institute, Pricewaterhouse
Coopers, 2001 Ross Avenue, Dallas, TX 75201-
2997, Phone: (214) 754-5091; Email: benjamin.isgur@us.pwc.com
Research Objective: Measure the degree to which consumer attitudes about health contribute to their perceptions of health system performance in 11 countries. This study used consumer segmentation analysis to construct a global attitude framework.
Study Design: PwC's Health Research Institute used a dual-method approach including interviews of multinational health experts and a quantitative survey of consumers in Australia, India, China,
South Africa, Norway, Italy, Netherlands, Germany,
United Kingdom, Canada and the United States.
The on-line consumer survey, conducted in October and November 2008, used a double opt-in screening process and all surveys were conducted in their native language. Sample size was 4,450 with approximately 400 respondents per country.
Population Studied: 4,450 total consumers ages
18-84 in Australia, India, China, South Africa,
Norway, Italy, Netherlands, Germany, United
Kingdom, Canada and the United States.
Consumers were balanced for age and gender.
Principal Findings: Consumers' perceptions of health system performance stem from personal self-interest and issues of importance. Based on the analysis, five consumer segments were identified relating to health attitudes that each consumer group values: Effectiveness -- getting their health issue taken care of, a short wait time, and a coordinated team of clinicians; Brand -- patient satisfaction ratings, education on overall wellness, and clinicians who are recognized as best in the field; Convenience -- a coordinated team of clinicians, and a short wait time and travel distance;
Price -- not having to worry about financial considerations and understanding out-of-pocket costs; and Knowledge -- knowledge of medical professionals and getting the best brand-name prescriptions. Consumers utilize technology as a tool to address health issues. More consumers seek health information from the Internet than friends and family, and many consider seeking healthcare advice by telephone or Internet.
Seventy-one percent of respondents prefer to have web-based or electronic medical records. China, the outlier, prefers to store health records in paper file format, rather than electronically. Nearly onethird of consumers surveyed did not think there was adequate medical data protection in their country.
Conclusion: Consumer attitudes about health system performance vary by country and the five segmentations. Consumers view the use of technology as a key attribute of the health system.
The Internet now rivals more traditional methods of health information and education such as friends, family and the education system.
Implications for Policy, Delivery or Practice:
Health system performance is largely based on understanding the users and potential users of the system. Consumer segmentation provides a framework for policymakers and health system leaders to best meet the unique needs of each patient. Health systems that capitalize on consumers' preference to use technology and the
Internet for health information may increase health system performance. This awareness can ultimately lead to improved health system design and patient outcomes.
Funding Source(s): PricewaterhouseCoopers, LLP
Theme: Global Health
● An Organizational Transformational Model for
Laboratory Health Systems: Application to
Personalized Medicine
Amalia Issa, Ph.D., M.P.H.; D. Joe Boone, Ph.D.;
Janet Marchibroda, B.S., M.B.A.; Linda McKibben,
M.D., Dr.P.H.
Presented by: Amalia Issa, Ph.D., M.P.H.,
Associate Professor & Director, University of
Houston, 300 Tech Building, Houston, TX 77204,
Email: aissa@uh.edu
Research Objective: Personalized medicine is increasingly moving across the "translational interface" into the clinic and the pace of implementation will continue to accelerate. Many questions remain regarding how the translation from basic science and clinical research to actual practice and health policies is going to be implemented in order to influence health care delivery and health services. A key challenge to the future implementation of personalized medicine is the present-day status of the nation’s health laboratory system and the role of laboratorians in health care quality and safety. Our objectives were to: (1) examine what is known about the real-world effectiveness of health laboratories as organizations; (2) describe a novel innovation strategy based on a health system transformation model, with four broad goals for personalized medicine that also apply to health laboratory system transformation in general (clinical utility, public health utility, personal utility and system quality); and, (3) use this model to focus policymakers on
the urgency and relevance of our proposed evidence-based, strategic solutions for transforming the health laboratory system to accommodate the integration of personalized medicine.
Study Design: We conducted semi-structured interviews with experts and opinion leaders representing a variety of perspectives from among approximately 50 participants in the CDC Division of Laboratory Systems’ 2007 strategic-planning meeting. The interviewees represented experts in laboratory medicine, health policy, and the diagnostics industry. Each was asked about the need for leadership by the health laboratory field to improve access to quality care; the role of laboratory experts in advancing patientcenteredness; and impediments to health outcomes research in laboratory medicine. A transformational model applied to health laboratory systems broadly and the integration and implementation of personalized medicine particularly was conceptualized.
Population Studied: US health laboratory systems.
Principal Findings: Interviewees named four environmental factors underlying a highly pressurized public-private health laboratory enterprise: increasingly conservative fiscal public policies, unrelenting and complex regulation, tight market competitiveness due to commoditization of laboratory tests and services and advanced workforce shortages. A fifth factor is paucity of federal funding for health outcomes research to guide integration of laboratory and clinical medicine expertise for better patient care.
Conclusion: Few health laboratorians are trained to incorporate knowledge from widely regarded social science studies about factors associated with the uptake of innovative services, practices, policies or products, including those relevant to health care.
Health system transformation involves putting innovation-diffusion concepts into practice by planning, promoting, or actively managing systems change by setting an agenda and monitoring progress towards goals. Our Transformation Model encapsulates five linked domains designed to set national goals and monitor progress toward transformation goals; including Champions &
Networks, Systems Improvement (Innovation),
Leadership, Resources & Capacity (Infrastructure) and Momentum (Measuring Progress).
Implications for Policy, Delivery or Practice: The appropriate implementation of personalized medicine using genomic information into existing health laboratory systems poses significant challenges for health services and health policy. We propose a novel Transformation Model to address these issues.
Funding Source(s): CDC
Theme: Organizational Performance and
Management
● The Nursing Shortage & Quality in Public
Health Services: Identifying Sensitive
Population Health Indicators
L. Michele Issel, Ph.D., R.N.; Betty Bekemeier,
Ph.D., R.N.; Kathleen Baldwin, Ph.D., R.N.
Presented by: L. Michele Issel, Ph.D., R.N.,
Clinical Associate Professor, School of Public
Health, University of Ilinois at Chicago, 1603 West
Taylor Street (MC 923), Chicago, IL 60612, Phone:
(312) 355-1137; Email: issel@uic.edu
Research Objective: This qualitative project sought to: (a) identify population health indicators that would be sensitive to a public health RN shortage, and (b) to hypothesize plausible pathways between a public health RN shortage and a set of selected indicators of quality of public health services.
Study Design: A purposive sample of eight
Directors of Nursing at large and small local health departments in two states and three public health nursing faculty at two major universities participated in an iterative, consensus building process. The process simulated a Dephi process, but with realtime and virtual discussions and consensus building. A paper trail was used to establish credibility, trustworthiness and replicability of the findings. After reviewing the state and the county health indicators, indicators applicable to personal health services, rather than environmental services, were selected. The participants then had a series of teleconferences, each time reducing the number of outcomes. Each shortened list was emailed to the participants, with an ensuing email discussion of the pro’s and con’s of the indicators on the list. After the third iteration, a final list of four population indicators were agreed upon by the participants.
Principal Findings: The participants established a set of four criteria for the selection of the indicators:
(a) data on the indicators must be accessible to the local health departments in a timely fashion, otherwise there is no possibility of quality improvement; (b) the population health outcome indictors must arguably be affected by a shortage of public health RNs; (c) each local health department and state must use the same metric for collecting and reporting the indicator; and, (d) the indicators had to reflect the services provided across each of the participating local health departments. These criteria forced the elimination of some indicators, primarily due to states not releasing the data on a quarterly basis, and inconsistency in the metric used by the local health department or the state.
The criteria that the indicator apply across the six local health departments posed particularly challenges given the variability across the local health departments on the types of personal health services provided. The four patient population health indicators deemed to meet the four criteria were: (1) Rate of Chlamydia per 100,000; (2)
Percent of women with first trimester prenatal care;
(3) Percent of children ages 0 to 24 months with full immunization, and (4) Percent of clients who have contacts/visits with LHD nursing staff who overall rate their satisfaction as very satisfied. The participants collectively developed hypothesized mechanisms by which each indicator could be influenced by both the adequacy and the quality of public health nursing staff.
Conclusion: All hypothesized scenarios revealed public health nursing shortages lead to worsening of the indicators.
Implications for Policy, Delivery or Practice: The challenges encountered in identifying population health indicators sensitive to the quality of public health nursing services provided suggest the need for greater uniformity in data reporting, a minimum set of personal public health services across local health departments, and greater attention to recruiting and retaining RNs in public health nursing positions.
Theme: Public Health
● Developing a Standard Occupational
Classification (SOC) Major Group for Public
Health
Tourette Jackson, M.P.H., B.S.; Robin Pendley,
M.P.H., B.S.
Presented by: Tourette Jackson, M.P.H., B.S.,
Research Assistant, College of Public Health,
University of Kentucky, 121 Washington Avenue,
Suite 101A, Lexington, KY 40536-0003, Phone:
(859) 218-2020; Email: tajack0@uky.edu
Research Objective: To evaluate the feasibility of developing a standard occupational classification
(SOC) major group for public health to assist with workforce enumeration.
Study Design: current Bureau of Labor Statistics
(BLS) standard occupational code data.
Collaborate with BLS, public health, and military manpower to develop an algorithm for public health
Standard Occupational Code (SOC). Adhere to classification principles of BLS to develop SOC major group for public health.
Principal Findings: Public health is currently not a
SOC major group under BLS. Public health professions are currently detailed under multiple
SOC major groups. The current classification leaves public health workforce professions segmented and we lack specific information related to this workforce.
Conclusion: It is feasible to develop a SOC major group for public health.
Implications for Policy, Delivery or Practice:
This can assist with enumerating the public health workforce. This will provide information to assess shortages within the public health workforce.
Theme: Health Care Workforce
● A Multivariate Approach to Identifying
Disparities in the Quality of Care within a
General Internal Medicine Practice
Muriel Jean-Jacques, M.D., M.A.; Stephen Persell,
M.D., M.P.H.; Romana Hasnain-Wynia, Ph.D.;
Jason Thompson, B.A.; David Baker, M.D., M.P.H.
Presented by: Muriel Jean-Jacques, M.D., M.A.,
Assistant Professor, Internal Medicine,
Northwestern University Feinberg School of
Medicine, 750 South Lake Shore Drive, 10th floor,
Chicago, IL 60611, Phone: (312) 503-9642; Email: mjean@nmff.org
Research Objective: To track inequities in health care, experts recommend that health care organizations collect quality data stratified by race/ethnicity. However, it is important to adjust for socioeconomic status (SES) and other covariates to assess the true magnitude of disparities. To more accurately identify disparities in the quality of care within an ambulatory care practice, we examined racial/ethnic differences in quality before and after adjusting for SES and other covariates.
Study Design: We assessed the quality of care for
18 measures of preventive and chronic disease care using data from the practice’s electronic health record system (EHRS). For each quality measure, we calculated the proportion of eligible patients with a quality deficiency as of January 1, 2008. We first used chi-square analysis to determine whether demographics (race/ethnicity, gender, and age), health insurance (commercial, Medicare, Medicaid, or self pay), or zip code-level measures of SES were univariate predictors of quality. Demographics and insurance were obtained from the EHRS. SES was imputed using 2000 U.S. Census data on the median household income and proportion of high school graduates in the patient’s zip code. We then used multivariate logistic regression models with the presence of a quality deficiency as the dependent variable and demographics, health insurance, and zip code-level SES as the independent variables to identify which factors were associated with having a quality deficiency.
Population Studied: This study was conducted in an academic general internal medicine practice with
37 attending physicians serving 43,900 patients.
Principal Findings: In univariate analyses, we found significant racial/ethnic disparities in quality for 9 of the 18 measures. The absolute magnitude of the disparity ranged from 3.2% for cervical cancer screening to 14.1% for osteoporosis screening or treatment. After adjusting for demographic, insurance, and SES variables, the racial/ethnic disparity remained significant for 3 measures: antiplatelet therapy for patients with coronary heart disease, osteoporosis screening or treatment, and screening or treatment for diabetic nephropathy (OR 1.9, 1.8, and 0.6 respectively for black patients having a quality deficiency relative to white patients, p<0.05 for all). Significant disparities
were also found by gender, insurance status, or
SES for most quality measures by univariate analyses, and these remained significant for 3 of 5 measures by gender, 5 of 10 measures by insurance, and 2 of 4 measures by SES in multivariate models.
Conclusion: The observed patterns of health care disparities in this ambulatory care setting differed substantially depending on whether we adjusted for demographic, health insurance, and SES variables.
Implications for Policy, Delivery or Practice:
Current expert recommendations to collect and report quality data stratified by race/ethnicity are inadequate. Not adjusting for covariates can give spurious findings for the presence and magnitude of racial/ethnic disparities. Moreover, solely focusing on race/ethnicity obscures other important determinants of quality. The use of zip code-level measures of SES, or even more specific geospatial measures of SES, provides a way to overcome the lack of direct data on patient income or education and allows for the routine multivariate analysis of disparities within health care organizations to help inform improvement initiatives.
Funding Source(s): AHRQ
Theme: Disparities
● Evaluating the Quality of Post-Stroke
Rehabilitation in Japan
Seungwon Jeong, Ph.D.; Katsunori Kondo, M.D.,
Ph.D.; Jeffrey Soar, Ph.D.; Nariaki Shiraishi, P.T.,
M.A.
Presented by: Seungwon Jeong, Ph.D., Research
Fellow, Research Center for Asian Social Well-
Being & Development, Nihon Fukushi University, 5-
22-35 Chiyoda, Nakaku, Nagoya, 4600012, JP,
Phone: +81(52)242-3075; Email: wonnie73@gmail.com
Research Objective: To develop and test a method for evaluating the quality of hospital-based rehabilitation after stroke.
Study Design: Controlling for disease severity (14
= numbers of days since onset = 60, 30 = length of stay in days = 180, age= 60), we evaluated the outcomes of rehabilitation, by using the Functional
Independence Measure (FIM™). For each patient, we calculated a predicted FIM™ score at the time of discharge. That calculation was based on age; modified Rankin Scale score; NIH Stroke Scale score at the time of admission, the FIM™ scores for memory, toileting, and upper-body
Dressing/Undressing at the time of admission; and the number of days from the onset of the stroke until the start of rehabilitation. We then compared that predicted discharge FIM™ score with the actual FIM™ score at the time of discharge. Each hospital's performance was categorized as bad, good, or very good based on the disparity between predicted and actual score, and the hospitals were compared.
Population Studied: As of January 2009, the
Rehabilitation Patients Databank in Japan held data on 3246 patients who were treated at 25 hospitals nationwide. We used data from the 5 hospitals that had each submitted data from at least 50 patients
(total n = 814).
Principal Findings: At the time of admission, there were no differences among hospitals in their patients' FIM™ scores, modified Rankin Scale scores, NIH Stroke Scale scores, and length of stay in days. Nonetheless, outcomes differed widely among the hospitals (p<0.05). In the bestperforming hospital, only 14% of the patients had a bad outcome, but in the worst-performing hospital
42% of the patients had a bad outcome compared with predicted score. We also found wide variation among the hospitals in the structure and process of health care. For example, regarding physical, occupational, and speech therapy, differences between good-performing and bad-performing hospitals were very large: hours of physical therapy,
73 vs 28 (0.7 vs 0.3 per day); hours of occupational therapy, 75 vs 36 (0.8 vs 0.4 per day); hours of speech therapy, 28 vs 11 (0.3 vs 0.1 per day); therapy on holidays, 99% vs 6%; therapy in hospital wards, 83% vs 0%.
Conclusion: We were able to identify goodperforming hospitals as those in which the percentage of patients with bad outcomes was lower than expected. Further studies should analyze the characteristics of good-performing hospitals.
Implications for Policy, Delivery or Practice: In
2008, Japan implemented pay-for-performance for rehabilitation. There is now a need for a comprehensive database, and for studies that account for likely confounders and use reliable, valid indicators of the quality of care.
Funding Source(s): Pfizer Health Research
Foundation Japan
● The Impact of Travel Time on Poststroke
Depression Detection among Veterans with
Acute Stroke
Huanguang Jia, Ph.D., M.P.H.; Ho-Chih Chuang,
M.S.; Xinping Wang, Ph.D.; Samuel Wu, Ph.D.; Eric
Litt, B.S.; Linda Williams, M.D.
Presented by: Huanguang Jia, Ph.D., M.P.H.,
Research Health Scientist, Department of Veterans
Affairs, Rehab Outcomes Res Center (151B), 1601
Southwest Archer Road, Gainesville, FL 32608-
1197, Phone: (352) 376-1611 Ext. 4926; Email:
Huanguang.Jia@va.gov
Research Objective: Depression after stroke is common. Poststroke depression (PSD) impairs patients’ functional recovery and increases patients’ utilization and mortality. While Veterans Health
Administration (VHA) is enhancing PSD screening among its stroke patients, little is understood about the influence of health service accessibility on
patients’ PSD detection. This study is to assess the impact of patients’ travel times to the closest VA facility on PSD detection among VHA stroke patients.
Study Design: In this retrospective, observational study, we included 5,589 veterans with acute stroke. VHA inpatient and outpatient, pharmacy, and veterans Medicare data were used. ICD-9 depression codes and antidepressant medication dispensing with guideline recommended minimum daily dosage were applied to define patients’ PSD status 12 months post index stroke. Travel times in minutes were calculated by using patients’ home zipcodes centroids and the latitude-longitude of the closest VHA facility. Logistic regression models were fitted to assess the effect of travel time on
VHA PSD diagnosis and overall VHA and non-VHA
PSD detection, respectively.
Population Studied: Veterans who were enrolled in the VHA system and diagnosed with an acute stroke in 2001.
Principal Findings: Among the study sample, 20% had PSD diagnosis within the VHA system and an additional 19% had the diagnosis outside the system. Our logistic regression analyses showed that shorter travel time to the closest VHA facility increased the odds of PSD diagnosis (Odds Ratio =
1.2, P = 0.05) within the VHA system; travel time, however, was not predictive of PSD when non-VHA
PSD diagnoses were added to the outcome, even adjusting for patients’ demographics and clinical factors. Other significant (P <0.05) predictors of
VHA PSD diagnosis included younger age, being white, and high volume of pre-stroke inpatient and outpatient utilization.
Conclusion: Our findings suggest that PSD in veterans with acute stroke is commonly detected outside the VHA system. Longer travel time to the closest VHA facility was a barrier to stroke patients’
PSD detection within the VHA system.
Nevertheless, for those stroke patients who lived further from VHA facility and possesses alternative insurance plans, their non-VHA coverage may provide a safety net for their PSD detection.
Implications for Policy, Delivery or Practice:
VHA policy makers need to consider the use of services outside the system by its enrollees with stroke when conducting program evaluation. VHA clinical providers should query patients on diagnoses and prescriptions received outside of the
VHA system and utilize the electronic medical record to reflect these findings. Future research on
PSD among veteran patients should include multiple sources of administrative data in order to more completely describe clinical care and outcomes.
Funding Source(s): VA
● Depression Diagnosis & Antidepressant
Dispensings Before & after a Stroke
Huanguang Jia, Ph.D., M.P.H.; L. Douglas Ried,
Ph.D.; Xinping Wang, Ph.D.; Hua Feng, M.D., M.S.;
Randi Cameon, M.P.H.
Presented by: Huanguang Jia, Ph.D., M.P.H.,
Research Health Scientist, Department of Veterans
Affairs, Rehab Outcomes Res Center (151B), 1601
Southwest Archer Road, Gainesville, FL 32608-
1197, Phone: (352) 376-1611 Ext. 4926; Email:
Huanguang.Jia@va.gov
Research Objective: Depression after stroke is common, ranging between 25 to 40% among stroke survivors. Post-stroke depression (PSD) impairs patients’ functional recovery and increases patients’ utilization and mortality. While the Veterans Affairs
(VA) is enhancing PSD screening among its stroke patients, little is understood about the relationship between pre-stroke depression and its impact on the onset of PSD and antidepressant treatment.
This study examined the veterans with pre-stroke depression diagnosis and compared their PSD and antidepressant treatment with those veterans with
PSD only.
Study Design: This retrospective study included veteran patients who received care for an acute stroke in a local VA network during 2001 and survived >60 days post-index stroke. To determine patients’ index stroke diagnosis, a validated stroke
ICD-9 codes algorithm was applied to VA medical and Medicare automated databases 6-months before and 12-months after the stroke. VA pharmacy database was used to obtain veterans’ antidepressant dispensing for 12-months post stroke. Relationships were explored using
Pearson’s chi-square statistics and McNemar’s test for matched pairs. The association strength was measured by Mantel-Haenszel procedure.
Population Studied: Veterans who were enrolled in a local Veteran Health Administration network and diagnosed with an acute stroke in 2001.
Principal Findings: Among the study cohort
(N=790), 9% had a depression diagnosis within 6 months pre-stroke and 29% had a PSD diagnosis within 12 months post-stroke. Compared to patients without pre-stroke depression diagnosis
(n=716), patients with pre-stroke depression (n=74) were more likely to have a PSD diagnosis 12 months post-stroke (OR=9.5; 95% CI=5.4-16.6;
P<0.0001). Further, patients with a depression diagnosis both before and after stroke (n=56) were more likely to receive SSRI antidepressants than patients with post-stroke depression only (n=177)
(OR=2.4; 95% CI=1.1-5.1; P=0.02). In addition, patients with pre-stroke depression but without PSD
(n=18) were also more likely to receive SSRI medications than those patients without a depression diagnosis both before and after stroke
(n=539) (OR=9.7; 95% CI=3.7-25.9; P<0.0001).
Conclusion: The incidence of PSD among this study cohort is similar to previous VA and non-VA reports. Our findings indicate that patients’ pre-
stroke depression status is a significant predictor of post-stroke depression diagnosis. Further, patients’ pre-stroke depression diagnosis consistently impacts the provider’s antidepressant dispensing behavior regardless of whether or not their patients had a depression diagnosis post-stroke.
Implications for Policy, Delivery or Practice:
Special attention to diagnosis and treatment of depression should be given to those patients who had a pre-existing depression diagnosis prior to their stroke. Further study is needed to explore the impact of pre-stroke depression on related health outcomes.
Funding Source(s): VA
● Impact of Smoking & Obesity on Quality of
Adjusted Life Years for U.S. States
Haomiao Jia, Ph.D.Erica Lubetkin, M.D., M.P.H.
Presented by: Haomiao Jia, Ph.D., Assistant
Professor, School of Nursing & Department of
Biostatistics, Columbia University, 617 West 168th
Street, New York, NY 10032, Phone: (212) 305-
6929; Email: hj2198@columbia.edu
Research Objective: We recently developed an estimation equation of EuroQol EQ-5D index scores from the CDC’s Healthy Days measures in order to calculate quality-adjusted life years (QALYs) at the national, state, and sub-state levels, estimate the impact of modifiable risk factors, and analyze changes over time.
Study Design: We applied a previously developed estimation equation to predict EQ-5D scores for the
Healthy Days respondents from the 2007
Behavioral Risk Factor Surveillance System
(BRFSS). With the combination of mortality data and the National Death Index linked National Health
Interview Survey data, we calculated QALYs lost contributed by smoking and obesity for each of the fifty states and the District of Columbia.
Population Studied: The BRFSS is a state-based survey of a representative sample of noninstitutionalized civilian adult residents from each of the fifty US states and the District of Columbia.
Principal Findings: The mean EQ-5D index score for U.S. adults was 0.864 and South Dakota and
West Virginia had the highest and lowest mean scores, respectively (0.888 and 0.833). Smoking and obesity contributed almost equally to QALYs lost in 2007, at 0.046 and 0.045 QALY lost per person, respectively. Smoking contributed over
0.068 QALY lost per person in Oklahoma and
Kentucky, but contributed less than 0.030 QALY lost per person in Utah and California. Obesity contributed over 0.054 and 0.056 QALY lost per person in Louisiana and Alabama, respectively, but only 0.032 in North Dakota and Hawaii. We also calculated QALYs lost due to morbidity in one year and the future QALY lost in expected life years due to excess deaths. The majority of QALYs lost to smoking was due to premature deaths, as only
24.2% QALYs lost was due to morbidity. By contrast, a much larger proportion of QALYs lost to obesity was due to morbidity (46.9%).
Conclusion: The differences in EQ-5D scores between many states would be considered to be clinically significant. Obesity and smoking contributed the most QALYs lost in Mississippi,
Louisiana, Arkansas, Alabama, and West Virginia and the least in Utah, Hawaii, Minnesota, California and Connecticut.
Implications for Policy, Delivery or Practice:
Estimates of the burden of diseases due to modifiable risk factors for each of fifty states and the District of Columbia currently are unavailable but are needed for setting targets for reducing modifiable health risks and eliminating health disparities for at-risk populations. Such data may be helpful in measuring progress with regard to the upcoming Healthy People 2020 objectives and would enable a more explicit understanding and prioritization of a set of health risks as well as the cost-effectiveness of policies or interventions used to reduce such risks. Application of our method also would permit the impact of interventions at the clinical and community level to be examined on a state and substate level as well as over time.
Funding Source(s): Columbia University School of
Nursing Dean's Translational pilot award
Theme: publichealth
● Impact of Smoking, Obesity & Diabetes on
Quality Adjusted Life Years for U.S. Adults
Between 1993-2007
Haomiao Jia, Ph.D.; Erica Lubetkin, M.D., M.P.H.
Presented by: Haomiao Jia, Ph.D., Assistant
Professor, School of Nursing & Department of
Biostatistics, Columbia University, 617 West 168th
Street, New York, NY 10032, Phone: (212) 305-
6929; Email: hj2198@columbia.edu
Research Objective: We developed an estimation equation of EuroQol EQ-5D index scores from the
CDC’s Healthy Days measures. Because the
Healthy Days measures have been included in the
Behavioral Risk Factor Surveillance System
(BRFSS) since 1993, translating Healthy Days values to a preference-based score enabled us to calculate Quality Adjusted Life Expectancy (QALE) and Quality Adjusted Life Years (QALYs) at the national, state, and sub-state levels, estimate the impact of modifiable risk factors, and analyze changes over time.
Study Design: We applied a previously developed estimation equation to predict EQ-5D scores for the
Healthy Days respondents from the 1993-2007
BRFSS. With the combination of mortality data and the National Death Index linked National Health
Interview Survey data, we calculated QALYs lost contributed by smoking, obesity, and diabetes for each of the fifty states and the District of Columbia.
Population Studied: The BRFSS is a state-based survey of a representative sample of noninstitutionalized civilian adult residents. We used
1993 to 2007 BRFSS data from each of the fifty states and the District of Columbia.
Principal Findings: Between 1993 to 2007, mean
EQ-5D scores for U.S. adults decreased from 0.883 to 0.864 while life expectancy and QALE increased slightly. The life expectancy for an 18-year-old increased 1.6 years and QALE increased 0.87 years, and gains were seen for both men and women. The contribution of smoking to the proportion of explainable QALYs lost decreased from 38.2% to 33.4% during this period. While obesity contributed only 23.0% to explainable
QALYs lost in 1993, this percentage increased consistently and, in 2007, obesity contributed
33.7% of QALYs lost. During the same time period, the contribution of diabetes to the proportion of explainable QALYs lost increased from 12.0% to
23.5%. The majority of QALYs lost to smoking was due to premature deaths and only 18.0%-26.2%
QALYs lost was due to morbidity. A much larger proportion of QALYs lost to obesity was due to morbidity (35.5%-49.0%).
Conclusion: This study estimated the impact on mortality and morbidity by smoking, obesity, and diabetes over time. In 1993, smoking contributed
66% more to QALYs lost than obesity did. But, since 1993, QALYs lost contributed by smoking have decrease while the QALYs lost contributed by both obesity and diabetes have increased. In 2007, obesity contributed slightly more to QALYs lost than smoking did.
Implications for Policy, Delivery or Practice:
Estimates of the trend of burden of diseases due to modifiable risk factors are unavailable but are needed for setting targets for reducing modifiable health risks and eliminating health disparities for atrisk populations. Such data may be helpful in measuring progress towards the upcoming Healthy
People 2020 objectives at not only the national level but also the state and local levels. Application of our method would enable a both a more explicit understanding and prioritization of a set of health risks and the cost-effectiveness of policies or interventions used to reduce such risks and permit the impact of interventions to be examined over time.
Theme: Public Health
● Measuring Health Plan Resource Cost in the
U.S.: Exploring the Burden & Complexity of
Reporting
Kiran Johal, M.P.H.; Sally Turbyville, M.A., M.S.;
David Salinas, M.H.S.; L. Gregory Pawlson, M.D.,
M.P.H.
Presented by: Kiran Johal, M.P.H., Health Care
Analyst, Performance Measurement, National
Committee for Quality Assurance, 1100 13th Street,
Suite 1000, Washington, DC 20005, Phone: (202)
955-3594; Email: johal@ncqa.org
Research Objective: In 2007, the National
Committee for Quality Assurance (NCQA) collected cost of care or Relative Resource Use (RRU) measures for diabetes and asthma. Results demonstrated substantial variation on plan performance; NCQA hypothesized that performance among plans beyond an established threshold was not a reflection of true measure variation, but in part revealed obstacles associated with reporting the measures. This study defines potential obstacles and the complexity of reporting the RRU measures.
Study Design: For 2006, NCQA calculated RRU measures across three different clinical categories from health plan claims and administrative data.
Data are risk adjusted to account for variation in health plan case and risk-mix of members. Ratios are reported as observed to expected or O/E.
Indexes greater than 1.00 indicate relatively higher cost; less than 1.00 indicates relatively lower cost.
Plans above the 95% (O/E = 2.50) or below the 5%
(O/E = 0.50) were deemed potential outliers for the diabetes and asthma RRU measures and surveyed.
These plans were asked to describe barriers in reporting. A standardized script was used, touching on known obstacles (i.e. data availability, data understanding, and data reporting).
Population Studied: Of the 462 Commercial,
Medicare, and Medicaid HMO plans from across the United States that submitted data for the RRU diabetes or asthma measures, 38 had an index below 0.50 or above 2.50 for these two measures.
These 38 plans were identified for further follow-up.
Principal Findings: Less than half (47%) of the plans contacted reported data collection errors, and approximately one fifth (20%) reported calculation errors. Particular challenges cited by the plans included human error and unfamiliarity with third party software that calculates RRU measures.
Many (75%) reported internal plan investigations to understand RRU issues and results were underway. Despite challenges, plans mentioned their continued interest in RRU measures.
Conclusion: Health plans continue to face challenges to accurate RRU measure data collection and submission. These challenges spring from a myriad of sources including human error, the robustness of data availability, and possible miscommunication with available software vendors. The findings from this study substantiate that some health plans may be encountering data collection and calculation obstacles; further plans proactively determine internal error-minimizing strategies.
Implications for Policy, Delivery or Practice:
Plan-to-plan comparisons based on resource use will allow consumers to purchase value laden health care. Further, this study allowed NCQA to improve our understanding and interpretation of the RRU
measurement set; enabling enhanced understandings of cost and resource use. Future research, should strive to gather input from all plans submitting RRU data, including those with results close to the mean.
● Occupational Class Disparities in the Use of
Routine Physical Exam among Asian Americans in the Labor Force
Dolly John, M.P.H.; Diane Martin, M.A., Ph.D.;
Bonnie Duran, Ph.D.; Butch de Castro, Ph.D.,
M.S.N./M.P.H., R.N.; David Takeuchi, Ph.D.
Presented by: Dolly John, M.P.H., Ph.D. Student,
Health Services, University of Washington, Box
357660, Seattle, WA 98195, Phone: (832) 798-
6871; Email: dajohn@u.washington.edu
Research Objective: National reports document the persistence of numerous significant inequities in the healthcare of Asian Americans, including the use of health services. European studies show that occupational class, an indicator of socioeconomic status, is a powerful, enduring determinant of health which may operate through material (e.g., income) and psychosocial (e.g., work-stress) pathways.
Because work is a significant source of income and health insurance in the United States, explicit consideration of occupational class may help explain and address health and healthcare inequities. We investigated how occupational class
(white-collar, blue-collar, service, and unemployed looking for work) is associated with use of a routine physical exam for Asian Americans in the labor force.
Study Design: We analyzed cross-sectional data from Asian respondents to the 2002-2003 National
Latino and Asian American Survey (n=2095, response rate: 66%). Routine physical exam was assessed as reporting at least one visit to a doctor, hospital, or clinic for a routine physical check-up during the past 12 months. Occupational class was categorized as white-collar, blue-collar and service following standard U.S. convention based on primary employment in one of 820 occupational groups covered by the 2000 U.S. Standard
Occupation Classification. We conducted bivariate chi-square and multivariate logistic regression analyses weighted to accommodate the complex survey design (weighted n = 5.5 million) using Stata
9.2.
Population Studied: 1530 Asian respondents who were part of the labor force (inclusion criteria: aged
18+ and being currently employed or unemployed but looking for work, exclusion criteria: student, retired, or homemaker)
Principal Findings: Overall, 26% of Asian
American workers in 2002-2003 reported no visit for a routine physical check-up during the past 12 months. Compared to 23% of white-collar workers,
34% of blue-collar, 40% of service and 21% of unemployed workers reported no visit for a routine physical check-up (p=.0061). Occupational class was strongly associated with use of a routine physical check-up. Compared to white-collar workers, blue-collar (OR = 1.69, 95% CI: 1.12-2.55) and service workers (OR=2.23, 95% CI: 1.63-3.04) had statistically significantly higher odds of reporting no routine physical check-up. In multivariate analyses, controlling for predisposing
(age, gender, ethnicity, nativity, education and
English language proficiency), enabling (income, health insurance, having a regular doctor or place of care) and need factors (health perceptions, chronic health conditions) removed the disparity for blue-collar workers (adjusted OR=1.23, 95% CI:
.71-2.11) but not for service workers (adjusted
OR=2.28, 95% CI: 1.51-3.45). We found no effect modification by gender or nativity.
Conclusion: Higher proportions of blue-collar and service workers than white-collar workers reported no routine physical check-up in the past year.
Preliminary findings support the existence of occupational class disparities in the use of routine physical exams for Asian American workers.
Controlling for predisposing, enabling and need factors removed the disparity for blue-collar workers but not for service workers.
Implications for Policy, Delivery or Practice:
Routine physical exams promote early detection of disease, use of preventive care and a good doctorpatient relationship. Eliminating disparities in care will require understanding and addressing labor related (e.g., work hours) and other factors beyond health insurance and income that also impede access to and timely use of care.
Funding Source(s): AHRQ
Theme: Disparities
● Capturing the Stories & Experiences of Teams
Caring for People with Chronic Illness
Julie Johnson, M.S.P.H., Ph.D.; Donna Woods,
Ed.D.
Presented by: Julie Johnson, M.S.P.H., Ph.D.,
Associate Professor, Centre for Clinical
Governance Research, University of New South
Wales, 10 Arthur Street, Sydney, NSW, 2031, AU,
Phone: +61 040 171 3822; Email: j.johnson@unsw.edu.au
Research Objective: We developed an Electronic
Team Survey (ETS) to better understand team member’s experience in caring for people with chronic illness and the effect of a Chronic Care
Collaborative on teams and teamwork.
Study Design: The ETS is a qualitative survey based on Electronic Event Sampling Methodology and is designed to collect information about any event that might influence team members’ daily work and subsequent outcomes. The ETS was conducted via a web-based tool. To minimize respondent burden, and to enhance response rate, each team member received an email directing
him/her to the survey for one week of each month a four-month reporting period.
Population Studied: The study included participants from the National Academic Chronic
Care Collaborative, which was funded by the
Robert Wood Johnson Foundation. Thirty-two residency programs participated and included faculty and resident physicians, nurses, and administrative staff.
Principal Findings: During 2006-2007, twentyseven of thirty-one teams participated in the ETS, resulting in an 87% response rate. Approximately
400 narrative entries immersed us into the daily work of caring for patients with chronic illness, working as part of a team, and teaching. These reflections provided a window into how
Collaborative participants think about changes in practice and changes in the institution.
An example of a narrative entry follows: “The last patient I called today had not been seen by us for about a year and a half. He lives at least an 1 ½ hour drive from the clinic. He has been seen in a clinic where he lives, but has not had blood drawn and said, ‘They don't tell me anything.’ I explained the idea of the planned visit and asked if he was willing to come the week before to have blood drawn so we could have the result at the visit. To my surprise, he agreed. I gave him instructions to come fasting for his labwork, bring insulin and food and I had him review the instructions to make sure he understood them. When I said that I looked forward to seeing him at the appointment he said with perceptible feeling, ‘I have been waiting a long time for this call.’”
Conclusion: Qualitative data collected from the
ETS provide a unique perspective into how the teams perform and the day-to-day challenges and opportunities in providing chronic illness care.
Implications for Policy, Delivery or Practice:
Specific methods for narrative analysis offer tools for teams to use their own stories to design improvements in chronic illness care. Narrative reports, such as those gathered through the ETS, can guide the improvement of the experience as well as outcomes for people with chronic illness and the work life of chronic care team members.
Funding Source(s): RWJF
Theme: Health Care Workforce
● What are the Missing Pieces? A Process
Analysis of Communication Between Hospitalbased & Primary Care Physicians During Patient
Care Transitions
Julie Johnson, M.S.P.H., Ph.D.; Jeanne Farnan,
M.D.; Vineet Arora, M.D., M.A.P.P.
Presented by: Julie Johnson, M.S.P.H., Ph.D.,
Associate Professor, Centre for Clinical
Governance Research, University of New South
Wales, 10 Arthur Street, Sydney, NSW, 2031, AU,
Phone: +61 040 171 3822; Email: j.johnson@unsw.edu.au
Research Objective: Ensuring a safe patient transition in and out of the hospital depends on communication and coordination between hospitalbased physicians and primary care physicians
(PCPs). Although current literature suggests that
PCPs are currently unsatisfied with communication at transition points between ambulatory and inpatient care and that such communication is fraught with content omissions and not performed in a timely fashion, no work to date has delineated a process and content-driven strategy for improving and standardizing communication during these transitions. We conducted a proactive risk assessment to better understand how to improve communication between hospital-based physicians and PCPs.
Study Design: We used qualitative methods – specifically critical incident interviews and focus groups – to identify barriers and facilitators to effective communication. We created process maps to define the perceptions of the current process and we conducted an artifact analysis of documents from the patient care process to demonstrate evidence of effective and poor communication between inpatient physicians and PCPs. Focus groups were conducted during December 2007-
February 2008. Focus groups were led by a moderator, lasted one hour, and were audio-taped and transcribed. Interviews of 50 discharged patients within 2 weeks after discharge from the hospital were conducted to obtain the patient perspective on the process of communication during inpatient-ambulatory transitions. Patients were asked a series of open-ended questions using critical incident analysis to identify problems that they experienced due to poor communication between their hospital-based physician and their primary care physician. Patient interviews were conducted by phone, taped and transcribed. Data from the focus groups and interview transcripts were qualitatively coded by investigators using the constant comparative method with no a priori hypothesis. Process maps were analyzed for common opportunities for improvement.
Population Studied: We conducted critical incident interviews and focus groups with key stakeholders
(PCPs, hospitalists, residents, and patients) from the hospital and community care settings.
Principal Findings: The analysis of the interviews, focus groups, process maps, and artifact analysis confirmed that communication with PCPs does not occur routinely and that certain critical decision steps in the communication process are often influenced by the presence of barriers or facilitators, such as whether the PCP is identified, PCP contact information is located, or communication preferences of the PCP are understood.
Conclusion: PCPs describe a process that is broken in which they are often not notified regarding a patient admission and have to “piece together what happened” after the fact from a variety of
sources (patient, family, home health, etc.).
Representative process maps and artifact analyses illustrate these findings.
Implications for Policy, Delivery or Practice:
Creating infrastructure to promote and support structured communication between inpatient physicians and PCPs during patient hospitalization is critical to assuring patient safety during ambulatory-inpatient care transitions. Our proactive risk assessment highlights the need for specific tools, resources, and communication strategies to address specific barriers to distinct steps in the patient transition process.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Organized
Processes
● Systematic Review of Community Health
Worker Interventions: Overview of Methods
Dan Jonas, M.D., M.P.H.; Meera Viswanathan,
Ph.D.; Jennifer Kraschnewski, M.D.; Brett
Nishikawa, M.D.; Laura Morgan, M.A.; Patricia
Thieda, M.A.
Presented by: Dan Jonas, M.D., M.P.H., Assistant
Professor, Medicine, University of North Carolina
Chapel Hill, 5039 Old Clinic Building, CB #7110,
Chapel Hill, NC 27599, Phone: (919) 966-7102;
Email: daniel_jonas@med.unc.edu
Research Objective: Community health workers
(CHWs) can serve as a bridge to the health care system for a wide range of patients and can help to minimize health disparities in low income and minority populations. We aimed to conduct a systematic review of the effectiveness of CHW interventions.
Study Design: We conducted a systematic review using the standard methods of the Agency for
Healthcare Research and Quality (AHRQ) supported Evidence-based Practice Centers. Key questions were developed in collaboration with a technical expert panel. In this panel presentation we focus primarily on one of the key questions of the review: “What is the impact of community health workers on outcomes, particularly knowledge, behavior, satisfaction, health outcomes, and health care utilization?” We searched Medline, the
Cochrane Database, and CINAHL from their inception through October 2008 using more than 10 terms for CHWs including the MESH term
“community health aides”. Two reviewers independently assessed each abstract and full text article. Disagreements were resolved by consensus. Data was extracted onto a standard form (evidence table) by one reviewer and checked for completeness and accuracy by a second reviewer. Trained reviewers abstracted data and assessed the methodologic quality (internal validity) of studies using predefined criteria based on existing international standards. We categorized interventions as being low-, moderate-, or highintensity based on the duration, nature of services, and degree of personalization.
Population Studied: We included studies comparing CHW interventions with other interventions or with usual care that were conducted in the United States, published in
English, and enrolled at least 40 participants. We excluded studies if the effect of the intervention could not be attributed to the CHW. We defined
CHWs as health workers who have a relationship with their community and do not have formal professional training.
Principal Findings: Our initial search identified 992 abstracts; we reviewed 516 full-text articles; 76 articles (representing 52 studies) met our inclusion criteria for this portion of the systematic review.
Many studies focused on low income and minority populations. Over two-thirds were randomized controlled trials and less than a third were observational. Of the 52 studies, seven were rated good quality, 25 fair, and 20 poor. Studies were generally related to one of four categories: cancer screening (13), disease prevention (11), chronic disease management (12), or maternal and child health (17). Most cancer screening interventions were low- or moderate-intensity; most disease prevention or disease management interventions were moderate- or high-intensity; most maternal and child health interventions were high-intensity.
The findings within these categories are described in separate abstracts. There was wide variation in the nature of CHW interventions.
Conclusion: CHWs often target low income, minority, and other underserved populations for a variety of health conditions and screening endeavors. A comprehensive evaluation of their effectiveness requires assessment of quality of the study and intensity of the intervention.
Implications for Policy, Delivery or Practice:
Reviewing the past success or failure of various
CHW interventions in the areas of cancer screening, disease prevention, chronic disease management, and maternal and child health can be useful for planning future CHW interventions and for resource allocation considerations.
Funding Source(s): AHRQ
Theme: Disparities
● A Multi-Layered Approach to Race Imputation using Medical Claims, Name Analyses, Inferred-
Familial Connections & a Geographic
Information System
Stephen Jones, M.S.; Soyal Momin, M.S., M.B.A.;
Allen Naidoo, Ph.D.
Presented by: Stephen Jones, M.S., Biostatistical
Research Scientist, Medical Informatics, Blue Cross
Blue Shield of Tennessee, 801 Pine Street,
Chattanooga, TN 37402, Phone: (423) 535-6915;
Email: stephen_jones@bcbst.com
Research Objective: To develop an algorithm to estimate race for members where race is unknown using a combination of medical claims, name analyses, inferred-familial connections and geographic imputation from census data
Study Design: Race information is collected for the
Medicaid population but not within the commercial line-of-business for a large southeastern managed care organization (MCO). A list of Medicaid members enrolled as of September 2008
(n=348,994) complete with first-last name and known race was created. A convenience sample approximating an 85/15 split was created for modeling purposes, where 300,000 members were used for development purposes and a hold-out
(test) sample (n=48,994) was created through simple random sampling. Race-name associations were created using a combination of member firstlast names from the Medicaid development dataset,
U.S. Census Bureau and web resources. Member residential locations were geocoded using a geographic information system (GIS). Using census block population information, a majority race was calculated for each block. For example, a block with 50 people (32 White, 18 African-
Americans) is assigned the race=“White”.
Diagnoses for members within the development dataset were extracted from medical claims for
October 2007-September 2008. A list of unique diagnoses with associated race thresholds was created. For example, of the 493 diagnoses for
ICD-9 code 282.6 (Sickle-Cell Disease), 355 (97%) were for African-Americans. Therefore any member having this diagnosis would be assigned race=“African-American”. Test dataset observations were assigned a race based on the aforementioned hierarchy: last-name then first-name, geographic census block, and race-biased medical diagnoses.
Following each imputation stage, members not assigned a race were assigned the race of a member in the same household if one was known
(i.e. inferred-familial connection). Known race was compared to imputed race for the test dataset and an accuracy frequency calculated across all races.
Population Studied: All current Medicaid members enrolled as of September 2008 in a large southeastern MCO with known race (n=348,994).
Principal Findings: We assigned race to 45,745
(93.4%) of the test dataset with an overall accuracy of 86.1%. Accuracy per race was: Asian=39%,
African-American=71%, Hispanic=85%, Native
American=9%, White=90%. Medical claims diagnoses and inferred-familial connection methodology imputed approximately 1% and 5%, respectively, of members that would have been missing otherwise. Assigning “White” to remaining missing members (n=3,249) slightly decreased accuracy in all races except white and overall accuracy was retained.
Conclusion: We were successful in developing a multi-layered approach to race imputation, yielding an overall accuracy of 86.1%. Utilizing healthcare claims data and familial inference methodology was a valuable addition to traditional methods that include only surname analysis and geoimputation.
Implications for Policy, Delivery or Practice:
Race information is commonly unavailable to commercial health insurance carriers because collection of this data may infer misuse within premium calculations and other discriminatory concerns. However, racial disparities exist relative to the quality and quantity of health care received by minority groups. The ability to impute race can help plans undertake issues of healthcare racial disparity, and address specific risk factors associated with race and mitigate them through proactive care management strategies. Member race is necessary for health plans to successfully complete a cultural needs assessment for NCQA accreditation.
Funding Source(s): BlueCross BlueShield of TN
Theme: Health Information Technology
● Child Health Insurance & Access Changes,
2004-2008
Kate Jones, M.A.; Gerry Fairbrother, Ph.D.; Lisa
Simpson, M.B., B.Ch., M.P.H., F.A.A.P.; Joseph
Schuchter, M.P.H.; Wendy Groznik, M.A.
Presented by: Kate Jones, M.A., Epidemiologist,
Child Policy Research Center, Cincinnati Children's
Hospital Medical Center, 240 Albert Sabin Way,
Cincinnati, OH 45229, Phone: (513) 803-2394;
Email: kate.jones@cchmc.org
Research Objective: To comprehensively describe child health care in Ohio and identify changes over time. The topics of interest include child health care coverage, access, and unmet needs within the context of relevant health policy changes.
Study Design: This study utilizes data from the
2004 and 2008 Ohio Family Health Surveys.
Building on previous analyses conducted nationally and in other states, child health care indicators from
2008 are examined for significant differences from comparable variables in the 2004 Ohio Family
Health Survey. Further, these indicators will be compared to national data, where applicable.
Population Studied: The population studied is children ages 17 years and under in Ohio, which is based on weighted results from the Ohio Family
Health Survey.
Principal Findings: The results of this study are a comprehensive description of child health care in
Ohio, including comparisons from previous measurements. Currently, preliminary analyses are complete using an 85 percent dataset of the 2008 survey and final results will be completed when the final dataset is available at the end of January,
2009. Results from the 85 percent dataset for 2008 appear to indicate a significant decrease from 2004 in children with no insurance in Ohio (from 5.4% in
2004 to 4.3% in 2008). The preliminary data also suggest an increase in children having at least one
preventive care visit in the past year (from 75.3% in
2004 to 78.9% in 2008), a reduction in children with one or more visits to the emergency department in the past year (from 8.4% in 2004 to 6.4% in 2008), and a decrease in children who have never been to the dentist (from 15.8% in 2004 to 13.3% in 2008).
Conclusion: Preliminary data from the 2008 Ohio
Family Health Survey show improvements in Ohio child health insurance coverage and access to care since 2004. However, there are also areas where improvements are needed, such as in quality of child health care and cost.
Implications for Policy, Delivery or Practice: The results will be included in a report that will provide relevant information for decision and policy makers.
This information can be used by these groups and other stakeholders to identify priorities and opportunities for improvement or continued improvement regarding child health and health care in the state.
Funding Source(s): Ohio Department of Jobs and
Family Services
Theme: Coverage and Access
● Testing the Sensitivity of the New York
University Emergency Department Algorithm to
Detect Changes in Emergency Department
Usage Patterns
Kari Jones, Ph.D.; Reider Hagtvedt, Ph.D.; William
Bond, M.D.; Hannah Paxton, R.N.; Jeff Etchason,
M.D.
Presented by: Kari Jones, Ph.D., Health
Economist, Community Health, Health Studies, &
Education, Lehigh Valley Health Network, P.O. Box
7017, Allentown, PA 18105, Phone: (770) 598-
4780; Email: kjones@qhr-econ.com
Research Objective: To explore the responsiveness of the Emergency Department
Algorithm (EDA) – a tool developed by New York
University Center for Health and Public Service
Research that uses administrative discharge data to categorize emergency department (ED) visits by need for emergent care – to changes in ED utilization patterns.
Study Design: The EDA maps inputs in the form of discharge ICD-9 codes for ED visits into outputs in the form of probabilities that a visit with the given
ICD-9 will be: non-emergent, emergent but care could be provided effectively and safely in a primary care setting, emergent but the condition was preventable or avoidable with better ambulatory care, in need of ED care for a condition that was not preventable or avoidable. Mathematical simulation was used to estimate EDA outcomes across all possible patient populations comprised of ICD-9s recognized by the EDA. Sensitivity analysis was then used to investigate the magnitude of changes in EDA outputs for various changes in a real-world baseline patient population.
Population Studied: Hypothetical patient populations were used for simulation. For sensitivity analysis, the baseline population consists of all ED cases from a community hospital’s administrative database that did not result in inpatient admission and for which discharge ICD-9s were available (n =
276,862).
Principal Findings: The vast majority of possible
EDA outputs do not differ significantly from one another. Thus, the vast majority of possible EDA outputs from the algorithm are so concentrated that they are not significantly impacted by significant changes to inputs to the algorithm. Starting from the mix of cases in an actual patient population, our sensitivity analyses show that the most extreme increases in high-probability non-emergent, primary care treatable, or preventable visits would produce only a 9% decrease in EDA-estimated emergent cases.
Conclusion: Our analysis explains why various
EDA users have observed EDA-generated outputs that do not change significantly with significant changes in patient populations and/or conditions in the associated healthcare system.
Implications for Policy, Delivery or Practice: The
EDA is being used nationally to assess ED usage; however, it is insufficiently sensitive to changes in
ED use patterns to be of use to policymakers or hospital administrators in addressing possibly unnecessary ED use.
Funding Source(s): Dorothy Rider Pool Health
Care Trust
● Employing Health Technology Improves
Mortality & Decreases Ventilator Use in
Critically Ill Patients
Kari Jones, Ph.D.; Matthew McCambridge, M.D.;
Elliot Sussman, M.D.; Hannah Paxton, R.N.; Kathy
Baker, R.N., M.P.H.; Jeff Etchason, M.D.
Presented by: Kari Jones, Ph.D., Health
Economist, Community Health, Health Studies, &
Education, Lehigh Valley Health Network, P.O. Box
7017, Allentown, PA 18105, Phone: (770) 598-
4780; Email: kjones@qhr-econ.com
Research Objective: To evaluate the impact on mortality, ventilator usage, and other clinical outcomes of implementing a health information technology bundle (HITB) in the intensive care unit
(ICU) of an academic community hospital.
Study Design: We conducted a single-center trial evaluation of an HITB consisting of electronic medical record (EMR), electronic medication administration system (eMAR), computer-assisted physician order entry (CPOE), bar-coded medication administration (BCMA), electronic algorithmic alert system, and a tele-intensivist program.
Population Studied: 903 control patients, receiving care in the ICU over a period of 16 months prior to the implementation of the HITB; 853 study patients
receiving care using HITB over a period of 10 months.
Principal Findings: There were no significant differences between the groups with respect to baseline characteristics. The observed mortality for the study group was significantly lower (p<0.01) than expected, at only 74% of that predicted by
APACHE Hospital Mortality. The standardized mortality ratio (SMR) was 31% lower in the study than the control group. The reduction in SMR was highest for non-stroke neurological patients, GI patients, and decreased with increasing APACHE score (above 28, where enough deaths occurred for meaningful analysis). Overall, study patients also had significantly less (p<0.05) usage of mechanical ventilation (31.1%) than the control group (36.4%).
These results were confirmed when controlling for severity of illness and diagnosis, and remained significant in sub-analyses of neurology and respiratory patients. Vasopressor use was significantly lower (p<0.05) in the study group for patients with a diagnosis of sepsis.
Conclusion: The use of our HITB significantly improves mortality and decreases ventilator use when caring for critically ill patients.
Implications for Policy, Delivery or Practice: In the United States, there is an increasing population of critically ill patients and a growing shortage of intensivists. The use of our HITB, which allows for more intensive use of scarce healthcare resources, addresses this issue while improving patient outcomes.
Funding Source(s): Dorothy Rider Pool Health
Care Trust
● Evaluating the Parma D.A.Y. Program
Katherine Jones, R.N., Ph.D.; Susan Tullai-
McGuinness, Ph.D.; Amany Farag, Ph.D.; Mary Jo
Krivanek, M.P.A.
Presented by: Katherine Jones, R.N., Ph.D.,
Professor, Bolton School of Nursing, Case Western
Reserve University, 10900 Euclid Avenue,
Cleveland, OH 44106-4904, Phone: (216) 368-
5979; Email: katherine.jones@case.edu
Research Objective: The purpose of this study is to evaluate the clinical and utilization outcomes of an innovative post-acute care hospitalization program for seniors titled Designed Around You
(D.A.Y). Individuals are identified during their acute care hospitalization as requiring post-discharge, transitional care services. The 8 week structured program provides physical, occupational, and speech therapy servces, daily nursing assessments, supervised medication administration if required, wound care if required, and mental and social stimulation. The affiliated community foundation provided funding for participants to attend.
Study Design: Retrospective chart reviews of program patient records were conducted using a structured data abstraction form; in addition, the corresponding records in the hospital electronic health record were accessed for information about physical therapy, occupational therapy, and speech therapy, and 90 day post-discharge utilization of ER and hospital services.
Population Studied: Between 2005 and 2008, 32 individuals enrolled in this 8 week day program at the hospital's Elder Center. Ages ranged from 52 to
91, with a mean age of 76. Most were female
(62.5%) and all were white. Just over half (56.3%) were referred for post-CVA therapies, with a range of other diagnoses. Medicare was the dominant source of payment (87.6%). Although the majority resided with children (37.5%), 28.1% lived with their spouse and 28.1% lived alone. The primary caregiver was most often a child (59.4%). All were discharged to home, many with continuing therapy.
Principal Findings: The level of patient/family satisfaction was high, with most indicating a willingness to pay for the program if necessary. The majority of patients showed significant progress in meeting their short-term and long-term goals of therapy. Other outcomes are currently being analyzed.
Conclusion: The 8 week, structured program appears to be a cost-effective alternative to rehabilitation services in the skilled nursing facility or at home. Multiple goals can be achieved in such a program, including safety instruction, diet monitoring, social engagement, and mental stimulation. The program also facilitates the transition of the patient from the acute care setting to the home making it easier for the primary caregivers to prepare and adjust to new safety requirements. Home safety checks are conducted by County officials as part of this program.
Implications for Policy, Delivery or Practice: The
DAY program may provide a cost-effective alternative for elders who experience CVAs and other illnesses and require long-term therapies after discharge from the hospital. Such a program gives the caregivers extra time to develop a support system, access resources, and obtain equipment/devices to ease the transition home.
Such services should be routinely covered by
Medicare reimbursement.
Funding Source(s): Community Foundation
● Advancing Child Health in Low Resource
Environments by Matching Health Care Delivery with Population Needs
Jill Joseph, M.D., Ph.D.; JoAnne, Natale, M.D.,
Ph.D.; Ben Gitterman, M.D.; Jennifer Egelseer,
D.O.; Pediatric Faculty of O.P.R.H.; First Class of
O.P.R.H. Pediatric Residents
Presented by: Jill Joseph, M.D., Ph.D., Director,
Center for Clinical & Community Research,
Children's National Medical Center, 111 Michigan
Avenue, NW, Washington, DC 20010, Phone:
(202) 476-6439; Email: jjoseph@cnmc.org
Research Objective: To reduce self-referral of low acuity pediatric patients to the outpatient facility of a nation's only pediatric referral hospital in a resource-limited environment. This objective is a priority for the Eritrean Ministry of Health as it works to build a health care system rationally distributing scarce pediatric resources.
Study Design: Longitudinal monitoring (nine months) of outpatient visits to both Orotta Pediatric
Referral Hospital (OPRH) and primary health facilities (PHFs) following implementation of a multiple-component intervention. This intervention was designed with local practitioners and residents as part of a community health rotation in a newlyestablished pediatric residency program; it included group education for parents attending OPRH as well as feedback to the PHFs regarding parental concerns. In addition, cross-sectional quantitative and qualitative data were obtained at 2 time points from both PHF providers and selected parents regarding use of health care resources.
Population Studied: We examined pediatric outpatient visits (newborn through 14 years old) in
Asmara, Eritrea March-November 2008. During this period of time, a multi-faceted intervention was implemented to discourage self-referral to OPRH and increase use of PHFs. Twelve PHFs staffed by non-physicians provide pediatric preventive and primary care services in Asmara using standard
WHO protocols for patient assessment, treatment, and referral. Staff of PHFs, as well as sampled parents seeking care for their children, participated in brief interviews to develop and refine the interventions.
Principal Findings: At baseline (March 24-29,
2008) there were 1,648 outpatient pediatric visits in
Asmara, of which 1210 (73%) were at PHFs and
438 were at OPRH. Of the children seen at PHFs,
12 (1%) were referred to OPRH and of these, 50% required admission either to emergency or inpatient unit. In comparison, of the 438 outpatient visits to
OPRH, 416 (95%) were self-referrals with parents by-passing an initial evaluation in a PHF. Among these self-referred patients, only 33 (8%) required admission. Both PHF staff and parents reported that the desire to be seen by a pediatrician and concerns about the quality of care at the PHFs motivated self-referral. We analyzed 5,639 visits between baseline and follow-up (8 months), documenting a decline in self-referral among OPRH outpatients from 95% to approximately 75%. During this same period of time, outpatient visits to PHFs increased 38% (from 1,210 to 1,667/week).
Interviews with parents demonstrated that the educational intervention increased the proportion of parents intending to use PHF for future outpatient care from 28% to 82%. PHF staff responded enthusiastically to this intervention program and requested further activities. Additional interventions are currently being implemented.
Conclusion: 1) It is possible to model and conduct research designed to understand and alter patterns of healthcare utilization in resource-limited environments. 2) Straightforward and easily designed/implemented interventions reduced selfreferral of low-acuity pediatric patients inappropriately seeking care at pediatric referral hospital.
Implications for Policy, Delivery or Practice:
Collaborative research with trainees and practitioners in a low resource environment can address important issues related to health care utilization. Such research can both train local practitioners and contribute to national efforts to link resource use to population needs
Funding Source(s): Partnership for Eritrea;
Physicians for Peace; Eritrean Ministry of Health
Theme: Global Health
● Exploring the Impact of Disease Management on Outcomes for Patients with Cardiovascular
Disease
Deborah Juarez, Sc.D., M.P.A.; James Davis,
Ph.D.; Todd Seto, M.D., M.P.H.
Presented by: Deborah Juarez, Sc.D., M.P.A.,
Research Manager, Care Management, Hawaii
Medical Service Association (Blue Cross & Blue
Shield of Hawaii), 818 Keeaumoku Street,
Honolulu, HI 96814, Phone: (808) 948-5337;
Email: deb_juarez@hmsa.com
Research Objective: To determine the impact of the disease management for patients with cardiovascular disease on clinical quality, utilization, and costs of care, after adjustment for potential confounding factors
Study Design: This was a retrospective study using administrative data from a large health plan in
Hawaii. For members with less than a full year of enrollment, their costs were multiplied by (365/days of enrollment) to get a full-year equivalent, except for members who died. Member costs were counted after they had been identified as having a condition.
Benefit reimbursement was used to measure cost.
Cost were inflated to 2006 dollars using the medical component of the consumer price index for medical costs and a GAO report for drug costs. Fixed effects models, that use members as their own controls, were used with total cost as dependent variable. Regressors included age, gender, type of coverage (HMO vs PPO), comorbid conditions
(diabetes, CAD or CHF), and morbidity level. Our main independent variable was a dummy variable indicating whether costs were post-program. We did not include the cost of the program in the estimates.
Population Studied: Algorithms from our physician pay-for-performance program were used to identify members with coronary artery disease or congestive heart failure (1999-2006).
Principal Findings: Post program implementation, cholesterol screening went from 76% to 81%, percent of patients with LDL<100 went from 40% to
50%, treatment with beta blockers after a heart attack went from 82% to 98%, and persistence of beta blocker therapy went from 69% to 77%.
Moreover, use of percutaneous coronary interventions (PCI) and coronary artery bypass surgery (CABG) decreased by more than 50 percent. Hospitalization rates dropped significantly, by 9 per 1000 per year, on average, while emergency department visits and ambulatory surgical center visits increased by approximately 2 per 1000 per year. The distribution of costs shifted substantially, with a greater share of expenditures going toward drug treatment and away from acute hospital care. The adjusted odds ratio of total cost for members after disease management implementation was 0.75 for congestive heart failure and 0.86 for coronary artery disease.
Conclusion: Following the implementation of a cardiovascular disease management program, quality improved and utilzation and costs decreased.
Implications for Policy, Delivery or Practice:
Further research is needed to determine which aspects of the disease management program might be having an impact and to identify sub-populations in need of targeted interventions.
● Determinants of Nursing Homes’ Decisions to
Terminate CNA Training Programs
Hye-Young Jung, B.A.; Denise Tyler, Ph.D.; Feng
Zhanlian, Ph.D.; Vincent Mor, Ph.D.
Presented by: Hye-Young Jung, B.A., Graduate
Student, Community Health - Health Services
Research Program, Brown University, 121 South
Main Street, Box G-S121, Providence, RI 02912,
Phone: (401) 432-7282; Email: hyeyoung_jung@brown.edu
Research Objective: The Omnibus Budget
Reconciliation Act of 1987 mandated that all certified nursing assistants (CNA) receive at least
75 hours of training. This rule went into effect in
1991. In 1997, there were 4927 facility-based CNA training programs, with a prevalence of 38%, which fell to 28% (3387 NHs) by 2006. Data were not available prior to 1997. Given their declining prevalence, and the potential influence the training they provide has on patient care, knowledge of what leads to termination of nursing home CNA training programs may be beneficial in terms of providing optimal care and oversight. The objective of this study is to identify determinants of nursing homes’ decisions to drop CNA training programs, adding insight into the decline in prevalence of these programs over the last decade.
Study Design: We used a discrete-time logistic regression analysis to identify the determinants influencing nursing homes’ decisions to drop CNA training programs. Discrete time models assume that the event of interest, dropping a CNA training program, occurs only at discrete time points. A discrete-time logit model has the benefit of accommodating both the binary outcome and repeated observations associated with the same facility over time. Odds ratios were estimated comparing the odds of terminating a program in a given year to a baseline reference group.
Population Studied: Data from the On-Line Survey
Certification and Reporting (OSCAR) system from calendar years 1997 through 2006 were used. A total of 9122 free-standing NHs with 20 or more beds were at risk of dropping a program during year intervals, with a total of 44,161 facility-years.
Principal Findings: Preliminary results show that six covariates, out of nine evaluated, are significant at the 5% level. Being part of chain (OR 1.17, p <
.001) or for profit (OR 1.19, p < .001) increased the likelihood of a NH dropping a CNA training program in any given year. Facilities located in urban areas were 55% more likely to drop a CNA training program than facilities located in rural regions. A larger number of beds (OR 0.97, p < .001) and higher occupancy rate (OR 0.17, p < .001) decreased the likelihood of dropping a CNA training program. Estimates of odds ratios associated with period effects were also obtained. These estimates were consistent with prevalence trends that show a steady decrease in facility-based CNA training programs over the study period.
Conclusion: Facilities in urban areas were much more likely to drop CNA training programs. In addition, NHs that are profit oriented or part of a chain were more prone to drop CNA training programs in any given period. Larger NHs and those with higher occupancy rates were less likely to drop training programs.
Implications for Policy, Delivery or Practice: The majority of CNA training programs are facilitybased. However, the prevalence of facility-based programs is steadily declining. Understanding both the organizational and market forces leading to
CNA training programs shifting away from nursing homes will provide insight for proper regulatory and oversight adjustments.
Funding Source(s): NIA
● Agency Characteristics & Changes in Home
Health Quality after Home Health Compare
Kyoungrae Jung, Ph.D.; Dennis Shea, Ph.D.;
Candy Warner, M.B.A.
Presented by: Kyoungrae Jung, Ph.D., Assistant
Professor, Health Policy & Administration, Penn
State University, 604 Ford Buildling, University
Park, PA 16802, Phone: (814) 863-8129; Email: kuj11@psu.edu
Research Objective: Home health care is an important source of post-acute follow-up care for the elderly. The demand for home health care is
expected to rapidly grow as the population ages and health care costs increase because home health care is viewed as preferable to costly facilitybased care. As home health services become increasingly important, assuring high quality home care is a priority. As part of quality improvement initiatives, the Center for Medicare and Medicaid
Services (CMS) recently implemented a public reporting program, Home Health Compare (HHC), for Medicare-participating home health agencies.
This study examines whether home health quality improves after HHC and whether the changes in quality differ by home health agency characteristics.
Study Design: We examine the changes in seven quality measures reported in HHC from 2003 to
2007. The measures include the improvement in bathing, transferring to bed, managing oral medications, walking, pain control, hospitalizations, and the use of emergent care. Using a linear regression model, we examine the associations between home health agency attributes and quality improvement. Agency characteristics that are examined include agency profit-status, whether an agency is hospital based or not, branch affiliation,
Medicare tenure, and the number of Registered
Nurses.
Population Studied: We use Medicare-certified agencies that submit HHC data for all years from
2003 through 2007. The number of agencies used in our analysis varies between 4,525 and 5,306, depending on the quality measure used for analysis.
Principal Findings: Significant improvements are found in quality after HHC for all indicators examined, except for the use of emergent care.
However, there was substantial variability in the level of quality change across individual agencies.
The degree of quality improvement also varies by quality measures: the largest improvement is shown in the measure of walking ability (6.3 percentage points; 19% above the baseline percentage), and the smallest improvement is in hospitalization rates
(1 percentage point; 3% improvements). There is a positive association between agencies’ not-for-profit status and quality improvement, and hospital-based agencies show greater improvement in quality than non-hospital based agencies.
Conclusion: Quality scores of the measures reported in HHC have improved over years. While the improvement is found in all types of agencies, the degree of quality improvement varies by quality indicators and by agency characteristics.
Implications for Policy, Delivery or Practice:
Despite the growing importance of home health care, little is known about home health quality. This study provides us with information about what agency characteristics are related to improvement
(or lack of improvement) in home health quality after a public reporting initiative. Large variation in quality and quality improvements across agencies suggests that there are further opportunities to improve home health care quality. It also suggests that agency characteristics and quality indicators
(e.g., utilization or functional activities) need to be considered in developing effective strategies to improve quality in home health care.
Funding Source(s): Social Science Research
Institute at Penn State University
● CDHP & Pharmaceutical Spending on Chronic
Diseases
Srikanth Kadiyala, Ph.D.; Sendhil Mullainathan,
Ph.D.; Eldon Spackman, M.A.
Presented by: Srikanth Kadiyala, Ph.D., Assistant
Professor, Pharmacy, Unviersity of Washington,
1959 Northeast Pacific. Street, H-375I, Seattle, WA
98121, Phone: (206) 543-9649; Email: harukim@u.washington.edu
Research Objective: Consumer directed health plans (CDHP) are aimed at reducing moral hazard due to insurance coverage. While preventive services such as mammograms and pap smear tests are fully covered in these plans, pharmaceuticals for chronic diseases such as diabetes, high cholesterol and hypertension are not.
How do CDHP plans affect the consumption of pharmaceuticals for chronic diseases such as hypertension, high cholesterol and diabetes?
Study Design: All enrollees in the CDHP plan in this study are initially provided an employer funded
Health Reimbursement Account (HRA) from which they can purchase medical care. The amount of funds in these accounts can range from $1000 to
$2000 depending on whether the enrollee is in an individual or a family health insurance plan. After enrollees use up the funds in the HRA account, there is a second phase in which enrollee spending is completely out of pocket (range from $500 to
$2000), after which time insurance covers medical spending at traditional cost-sharing levels. We compare average daily pharmaceutical spending for individuals with diabetes, hypertension and high cholesterol as they move through each of the different coverage thresholds (HRA, Out-of-Pocket and Full-Coverage).
Population Studied: We obtained data on all employees from a large private U.S. employer for the 2005 to 2007 time period.
Principal Findings: The population in the CDHP plan was relatively young as the average age of a claimant was 35 years of age. Using Icd-9 codes we identified approximately 28,000 individuals who had been diagnosed with diabetes, 19000 individuals who had been diagnosed with hypertension and 18,000 individuals diagnosed with high-cholesterol. For each of these individuals average daily pharmaceutical expenditure was substantially lower when individuals were in the
HRA and out-of-pocket sections of their CDHP plan as compared to when they were in the Full-
Coverage part of their insurance plan. For the three diseases combined, average daily expenditure on
pharmaceuticals was approximately $2.60 for individuals in the HRA part of the plan, $1.00 for individuals who moved into the out-of-pocket part of the insurance plan and $4.00 for individuals who were covered by insurance. These results are consistent with results from the Rand Health
Insurance Experiment which showed that increased cost-sharing reduced the consumption of medical care.
Conclusion: In this study we evaluated pharmacuetical expenditures for a set of CDHP enrollees. Preliminary findings suggest that high initial cost-sharing levels in CDHP plans affect the consumption of medications for the treatment of chronic diseases. Reductions in the use of chronic disease medications have the potential to negatively effect enrollee health outcomes.
Implications for Policy, Delivery or Practice:
Previous research on CDHP plans has shown that the CDHP enrollees do not consume less preventive services such as Mammograms when compared to traditional insurance plans. Results from this study indicate that other types of preventive care might be affected by the initial high cost-sharing levels of CDHP plans. Future research should evaluate the potential health effects from the reductions in spending identified due to higher costsharing.
Funding Source(s): Pfizer
Theme: Consumer Choices in Health Care
● Measles Vaccination Costs in Developing
Counries
Srikanth Kadiyala, Ph.D.; Jelena Zurovac, B.A.; Lou
Garrison, Ph.D.
Presented by: Srikanth Kadiyala, Ph.D., Assistant
Professor, Pharmacy, University of Washington,
1959 Northeast Pacific Street, H375I, Seattle, WA
98121, Phone: (206) 543-9649; Email: harukim@u.washington.edu
Research Objective: Measles is a leading cause of childhood death in developing countries. Recently initiatives by organizations such as the Gates
Foundation and the World Health Organization have been aimed at reducing deaths due to measles. These initiatives focus on increasing delivery of the measles vaccine via strengthening of the routine delivery system, funding supplementary immunization activities such as mass measles campaigns and funding the development of new vaccination delivery systems. Given these initiatives an important question relates to understanding the costs of delivering the current measles vaccine in developing countries.
Study Design: We use a micro-costing approach to evaluating measles vaccination costs. Routine and
Supplemental Immunizations costs are largely composed of the price of the vaccine, vaccine wastage, personnel, and supply and cold-chain costs. Estimation of current measles vaccination costs is done through both cross-sectional regression analyses of GAVI Financial
Sustainability Plan (FSP) data and analyses of Abt cost report data. An important advantage of using the GAVI FSP data is that GAVI collects information on shared health system costs between immunization programs and other health systems within a given country.
Population Studied: We estimate current measles routine and supplemental immunization vaccination costs for three developing countries, Cambodia,
Ghana and Morocco.
Principal Findings: In the countries that were evaluated, using the Abt cost report data, the average cost of measles vaccination delivery via supplementary immunization activities (SIA) is approximately 33% less than the cost of measles vaccination delivery via routine immunization (RI).
The cost savings in SIAs are achieved through both reductions in vaccine wastage and reductions in operational costs due to lower personnel costs.
Using GAVI FSP data to evaluate RI immunization costs leads to higher estimates for the RI cost of delivering measles vaccination for Cambodia and
Ghana (approximately 100% higher). The higher estimates from using the GAVI FSP data are consistent with shared health system costs accounting for an important part of total vaccination costs; suggest that not accounting for shared health system costs substantially underestimates developing country RI vaccination costs.
Conclusion: This study evaluated the costs of delivering measles vaccination in Cambodia,
Ghana, and Morocco. Results from two different data sources indicate that the cost of vaccination delivery through SIAs is substantially lower when compared to the cost of delivering a vaccination through RI. Analyses also reveal the importance of accounting for shared health system costs in developing countries. These results have implications for both developing country and donor health resource allocation as well as for understanding the potential impact of new vaccination innovations.
Implications for Policy, Delivery or Practice:
Results from this study suggest that in resource constrained countries SIAs are an effective way of delivering vaccinations. Our analyses also highlight the importance of wastage rates and personnel costs in understanding devleoping country vaccination costs. Finally, not accounting for shared healthy system costs in RI delivery can lead to underestimates of the true costs of RI vaccination delivery.
Funding Source(s): Bill and Melinda Gates
Foundation
Theme: Global Health
● Cost-Related Underuse of Medicine Due to
Medicaid Pharmacy Cost-Containment Policy
Actions
Nantana Kaisaeng, M.B.A., M.H.A.; James Zhang,
M.S., Ph.D.
Presented by: Nantana Kaisaeng, M.B.A., M.H.A.,
Student, Pharmacy, School of Pharmacy, Virginia
Commonwealth University, 1112 EastClay Street
(McGuire Hall), Room 218, Richmond, VA 23298,
Phone: (804) 551-5759; Email: kaisaengn@vcu.edu
Research Objective: We sought to evaluate the impact of cost containment policy actions on costrelated underuse of medicine, controlling for concurrent non-pharmacy cost-containment policy actions and socio-environmental differences among states.
Study Design: We used the data from the 2003
Community Tracking Study (CTS) household survey, a nationally representative survey for community-dwelling adults, and linked the CTS data with the census demographic data and the cost containment policy actions based upon 50 States
Surveys on state Medicaid spending growth and cost containment actions by the Kaiser Family
Foundation in 2003. A cross-sectional study was performed to evaluate the impact of policy actions, by comparing Medicaid beneficiaries to non-
Medicaid-pharmacy-cost-containment-actioneligible residents. The cost-related underuse of medicine measures included: patients do not receive needed medical care, patients cannot afford needed prescriptions, patients postpone needed medical care, and patients worry about the medical care cost. Each of the above four binary outcomes was analyzed using logit model, using nine prevailing pharmacy cost-containment policy actions that include: prior authorization (PA), generic drugs required, copayment method, step therapy of fail-first requirement, limit on number of prescriptions and number of refills per month, preferred drug list (PDL), over the counter (OTC) coverage, and prescription drug payment practices
(payment and purchasing policies) as predictors, controlling eight non-pharmacy cost-containment policy actions including provider payment method, benefit reduction, eligibility cut, beneficiary copayment, managed care expansion, disease and case management, fraud and abuse, and long term care reduction. We further controlled patients’ demographic and states’ socio-environmental variables including age, gender, race, education, family size, individual income, fraction of minority, fraction of women, poverty rate, homeownership rate, and electoral vote at each state in 2000.
Population Studied: 25,988 adults aged 18 or older in 2003 CTS survey linked with the Kaiser
Medicaid Policy Report and Census Data, among which 1,531 (5.9%) were Medicaid beneficiaries.
Principal Findings: Twenty-eight states have data on both CTS survey and Kaiser Medicaid Policy
Survey. On average each state has implemented
6.5 pharmacy cost containment policy actions in
2003. Among the four outcome measures, only the worrying about costs was statistically significantly associated with individual pharmacy cost-containment policy action, however, such significance disappeared after other concurrent pharmacy cost-containment policy actions were controlled for. When the total number of pharmacy cost-containment policy actions was entered in the regression instead of individual policy action, each incremental policy action was associated with 14% increase in odds ratio for unmet medical needs
(p=0.003), and 5% increase in odds ratio for postponing medical needs (p=0.02).
Conclusion: There was variable impact of pharmacy cost-containment policy actions when assessed concurrently, and collectively.
Implications for Policy, Delivery or Practice:
Further research is needed to examine the downstream costs of cost-related underuse of medicine due to Medicaid pharmacy cost containment policy actions and the collective impact of policy actions.
Theme: Quality and Efficiency: Policies and
Incentives
● Does the Complexity of General Surgeons’
Total Practice Volume Matter for In-Hospital
Complications for Gastric Bypass?
Shital Kamble, M.S., M.M.S.; Jennifer Troyer,
Ph.D.; James Studnicki, Sc.D.; Keith Gersin, M.D.
Presented by: Shital Kamble, M.S., M.M.S., Ph.D.
Candidate, Public Health Sciences, University of
North Carolina at Charlotte, College of Health &
Human Services, 9201 University City Boulevard,
Charlotte, NC 28223, Email: skamble@uncc.edu
Research Objective: The volume-outcome relationship is well-established for surgeons performing gastric bypass (GB) surgeries. However, the findings of improved outcomes after GB in high volume surgeons have been attributed only to the high volume of GB and not the volume of other nongastric bypass (non-GB) procedures. The study objective was to examine the effect of surgeons’ non-GB complex (non-GBC) and non-complex
(non-GBNC) volume on in-hospital complications among patients undergoing GB.
Study Design: Data were obtained from the 2003-
2004 Florida Hospital Discharge database. The total work Relative Value Units (RVU), available from the Physician Fee Schedule from the Centers of Medicare and Medicaid, were used to segment the surgeon’s non-GB case load into non-GBC and non-GBNC primary procedures performed by a surgeon per year. The non-GBNC procedures were those with a work RVU less than the work RVU for
GB (<16.21). The surgeon’s non-GBC and non-
GBNC volumes were dichotomized as high (vs. low) using the top tertile cutpoints of 50 for non-GBC volume and 132 for non-GBNC volume. Separate generalized estimating equation (GEE) regression
models, adjusting standard errors for the nonnested surgeon and hospital cluster effect, were constructed to examine the effect of surgeon’s non-
GBC and non-GBNC volumes on composite complications (one or more complications), technical complications (including unexpected reoperations, splenic injury, hemorrhage, anastomotic leaks, small bowel obstructions, and wound), and systemic complications (including pulmonary, cardiac, thromboembolic, genitourinary tract, and postoperative shock). Covariates included were patient characteristics (age, sex, race/ethnicity, payer, comorbidities), surgeon GB volume (high >=50 GBs/year), and hospital GB volume (high >=125 GBs/year).
Population Studied: All elective open GB patientdischarges (n = 11,398) operated on by boardcertified general surgeons (s = 108) in Florida.
Open GB discharges were identified with primary
ICD-9-CM procedure codes for High or “Mason” gastric bypass (44.31) or gastroenterostomy (44.39) confirmed by ICD-9-CM diagnosis codes for morbid obesity (278.00-278.4) or a DRG code for obesity surgery (288).
Principal Findings: In adjusted analyses, general surgeons performing 132 or more non-GBNC procedures had a higher likelihood of in-hospital composite, technical, and systemic complications compared with those performing fewer than 132 procedures (odds ratio [OR]: 1.83, p = 0.0039; OR:
1.55, p = 0.0064; OR: 1.88, p = 0.038, respectively).
In contrast, general surgeons performing 50 or more non-GBC procedures had a lower likelihood of in-hospital composite and technical complications but the effect was not statistically significant. When considering the GB volume, there was a notably lower likelihood of in-hospital composite complications for surgeons with higher GB volume
(OR: 0.77, p = 0.037).
Conclusion: The likelihood of in-hospital composite, technical, and systemic complications from GB procedures is greater for patients with general surgeons that have high non-complex surgical loads.
Implications for Policy, Delivery or Practice: The study findings suggest that while general surgeon
GB volume matters for in-hospital complications, the complexity of their overall surgical load also matters. In particular, the outcomes may improve if
GB patients avoided surgeons with a high volume of non-complex procedures. The study provides insights into the importance of procedure specificity and the composition of the total surgical practice in examining the volume-outcome relationship for general surgeons.
Theme: Quality and Efficiency: Measurement
● Designing a High-Quality Package of
Preventive Services
Douglas Kamerow, M.D., M.P.H.
Presented by: Douglas Kamerow, M.D., M.P.H.,
Chief Scientist, Health Services & Social Policy
Research, RTI International, 701 13th Street, NW,
Suite 750, Washington, DC 20005, Phone: (202)
728-1959; Email: dkamerow@rti.org
Research Objective: To define principles underlying the creation and implementation of a high-quality package of clinical preventive services
Study Design: Literature review, followed by synthesis and policy discussion
Population Studied: US adults
Principal Findings: Literally millions of years of life could be gained by Americans if we used effective preventive services, but coverage of these services by health plans, insurance programs, and government agencies varies widely. Which screening tests, immunizations, counseling, and chemoprevention services should be covered to optimize the health of Americans in the most costeffective way? Several principles should guide the creation of a high-quality package of preventive services: 1. Include the full range of clinical preventive services in the coverage package: screening tests, immunizations, counseling, and chemoprevention. 2. Use evidence-based evaluations to inform service selection. 3.
Incorporate cost-effectiveness assessments whenever possible. 4. Begin with a basic set of preventive services. 5. Update the package of services regularly. 6. Allow flexibility in the content of preventive services benefit packages, depending on local conditions and capabilities. 7. Allow for flexibility in method of delivery, depending on local conditions. 8. Provide incentives for the delivery and receipt of especially cost-effective and under delivered clinical preventive services. Just creating an appropriate package of services, however, is not sufficient to insure their appropriate and widespread delivery. Equal attention must be given to the provision of incentives—to clinicians, plans, and patients—for the delivery and receipt of the priority services. In addition, non-financial steps need to be taken to increase the uptake of preventive care, such as using standing orders and reminder systems to prompt clinicians to provide the services, as well as partnerships with community and worksite settings.
Conclusion: Assembling a package of basic clinical preventive services requires evaluating these services in an evidence-based manner and taking into consideration other issues, such as costeffectiveness and differing environments for care.
Guiding principles for the design of a package of clinical preventive services include choosing those with clear evidence of effectiveness while allowing for flexibility to expand if desired. It is also important to provide incentives for their use; simple coverage is not enough to ensure that the potential benefits of preventive medicine are fully achieved.
Implications for Policy, Delivery or Practice: A basic set of preventive services for adults can be
constructed from positive recommendations of the
US Preventive Services Task Force and the
Advisory Committee on Immunization Practices.
Just as important as the selection of services to include in such a package is the ability to modify the recommended services based on new evidence.
Finally, steps should be taken to ensure that the services are used, by providing both financial and organizational incentives for the optimal delivery of preventive care, both to patients and to clinicians and systems.
Funding Source(s): Partnership for Prevention
Theme: Prevention and Treatment of Chronic
Illness
● Awareness, Perception & Knowledge of Heart
Disease Risk & Prevention among South Asian
Immigrants
Namratha Kandula, M.D., M.P.H.; Manasi Tirodkar,
Ph.D., M.S.; Diane Lauderdale, Ph.D.; Muhamad
Paracha, M.D., M.P.H.; Gregory Makoul, Ph.D.;
David Baker, M.D., M.P.H.
Presented by: Namratha Kandula, M.D., M.P.H.,
Assistant Professor, Internal Medicine,
Northwestern University, 750 North Lakeshore
Drive, 10th Floor, Chicago, IL 60611, Phone: (312)
503-5500; Email: NKandula@nmff.org
Research Objective: South Asians in the United
States (U.S.) have a higher risk of developing coronary heart disease (CHD) compared to other racial/ethnic groups. Little is known about this community´s knowledge and attitudes about CHD risk and prevention. We studied South Asian immigrants´ knowledge and attitudes about CHD as part of an effort to design culturally-targeted CHD prevention messages.
Study Design: We conducted a survey of 270
South Asians recruited from a federally qualified health center and community centers in Chicago,
Illinois and surrounding suburbs. A standardized questionnaire was verbally administered in Hindi,
Urdu, and English. We examined associations between age, sex, education, acculturation (low, medium, and high), and CHD knowledge and attitudes.
Population Studied: Respondents ranged from 20-
75 years (mean, 49 years), and 57% were women.
While education levels were high (53% were college graduates or more), 53% were uninsured.
Two-thirds of the interviews were in Hindi or Urdu.
Participants had lived in the US on average 13 years.
Principal Findings: When asked about the leading cause of death among South Asians in the U.S.,
47% correctly identified CHD, followed by cancer
(17%), did not know (13%), and diabetes (9%).
When asked to describe in their own words, what is a heart attack, 35% of respondents used terms such as “blockage in blood flow to heart,” “coronary artery disease,” “clot in arteries.” 16% used more generic terms such as “damage to heart,” “heart does not pump,” “weak heart” or “heart stops”, another 12% said “pain in the heart.” Fifty-three percent of participants thought that most heart attacks cannot be prevented. The most frequently mentioned risk factors for a heart attack were stress
(56%), poor diet (35%), and high fat diet
(29%).When asked what things are important for preventing heart attacks, respondents most frequently cited exercise (53%), eat a better diet
(49%), reduce stress (43%), and eat a low fat diet
(39%). Very few said control of cholesterol (11%), blood pressure (12%), and diabetes (5%). In multivariate regression models adjusted for age, education, and gender, individuals with low acculturation, compared to those with high acculturation, were significantly less likely to know that CHD was the leading cause of death among
South Asians in the U.S. (OR=0.41, 95% CI= 0.20,
0.82) and were also less likely to think that heart attacks can be prevented (OR=0.28, 95% CI= 0.12,
0.68). Women were also less likely than men to think that heart attacks can be prevented (OR=0.49,
95% CI= 0.28, 0.86). Those with low education levels, independent of age, gender, and acculturation were significantly less likely to list high cholesterol, high blood pressure, and diabetes as risk factors for CHD, compared to those with higher education.
Conclusion: While South Asian immigrants are aware that CHD is the leading cause of death in their community, specific knowledge gaps and misconceptions remain. Despite high average education levels, there is low awareness about the connection between cholesterol, blood pressure, diabetes and CHD.
Implications for Policy, Delivery or Practice:
CHD education programs targeted to South Asian immigrants need to change knowledge and attitudes about the preventability of CHD and the importance of controlling cholesterol, blood pressure, and diabetes to achieve this. CHD prevention messages may need to be somewhat different for population subgroups (e.g., by gender and acculturation) to be maximally effective.
Funding Source(s): NHLBI
Theme: Disparities
● The Relationship Between Health Literacy &
Knowledge Improvement in a Spanish-Speaking
Hispanic Population after a Multimedia Type 2
Diabetes Education Program
Namratha Kandula, M.D., M.P.H.; Charles Zei, B.A.;
Quinn Stephens, B.A.; Gregory Makoul, Ph.D.; Sara
Glass, B.S.; David Baker, M.D., M.P.H.
Presented by: Namratha Kandula, M.D., M.P.H.,
Assistant Professor, Medicine/General Internal
Medicine, Northwestern University, 750 North
Lakeshore Drive, 10th Floor, Chicago, IL 60611,
Phone: (312) 503-6470; Email: nkandula@nmff.org
Research Objective: Many Hispanics in the United
States (U.S.) face the combined obstacles of limited
English proficiency, inadequate health literacy, little formal education, and cultural differences. These may make it highly problematic for some Hispanics to learn about their medical conditions and how to care for themselves. If designed well, multimedia diabetes education programs (MDEP) have the potential to overcome these obstacles and improve communication and education. However, few
MDEP have targeted low-literate, native Spanishspeaking populations, and these have not been evaluated to determine their efficacy. We evaluated a MDEP targeted to Spanish-speaking patients with low literacy and assessed knowledge gained after viewing the MDEP.
Study Design: Patients were interviewed to determine their baseline knowledge about the information in the first two modules of the diabetes
MDEP. All questions were open-ended, and patients were allowed to answer in their own words.
The interviewer then coded whether the response was correct. Patients then viewed Module 1 of the
MDEP, called "Qué es la diabetes? (What is
Diabetes?),” followed by repeat administration of the Module 1 questions. The same was done for
Module 2, "Los altos y bajos del nivel de azúcar en la sangre (The Ups and Downs of Blood Sugar).”
Health literacy was measured using the Spanish version of the Short Test of Functional Health
Literacy in Adults (S-TOFHLA), and categorized as
"inadequate" (0-16), “marginal” (17-22) and
"adequate" (23-36). Differences in knowledge gained were compared by paired t-tests.
Multivariate linear regression was used to examine differences by literacy after adjusting for baseline knowledge, age, gender, history of diabetes, and education.
Population Studied: 100 patients with and without diabetes were recruited from primary care clinics at two federally qualified health centers in Chicago that serve a Latino, Spanish-speaking population.
Principal Findings: Sixty-one percent of patients were of Mexican origin, and most (69%) were born outside the U.S. or Puerto Rico. Fifty-nine percent had less than a 10th grade education. The mean
TOFHLA score was 20.6, with 53% of individuals having adequate literacy, 14% marginal, and 33% inadequate. All individuals had significant increases in mean knowledge scores, regardless of literacy level. Individuals with inadequate health literacy gained 4.4 points, those with marginal literacy gained 4.6 points, and adequate literacy gained 5.5 points (p-value <0.001 for all groups). Although individuals with inadequate and marginal literacy had a smaller increase in post-test scores than those with adequate literacy, this difference was not significant after adjusting for baseline knowledge.
When stratified by literacy level, knowledge gains among the Spanish-speaking study participants were comparable to those seen in native Englishspeaking patients with from previous studies.
Conclusion: These findings suggest that, if properly designed, a MDEP can overcome the barriers of limited English proficiency and low health literacy and achieve similar knowledge gains regardless of literacy level or native language.
Implications for Policy, Delivery or Practice: A
MDEP developed with input from Spanish-speaking patients with low literacy can increase knowledge about diabetes.
Funding Source(s): Northwestern Memorial
Hospital Foundation
Theme: Disparities
● Evaluation of Tailored Education using
“Teach to Goal” in a Multimedia Type 2 Diabetes
Education Program
Namratha Kandula, M.D., M.P.H.; Leung, Tiffany,
M.D./M.P.H.; David Baker, M.D./M.P.H.; Sara
Glass, B.S.; Quinn Stephens, B.A.; Charlie Zei, B.A.
Presented by: Namratha Kandula, M.D., M.P.H.,
Assistant Professor, Medicine/General Internal
Medicine, Northwestern University, 750 North
Lakeshore Drive, 10th Floor, Chicago, IL 60611,
Phone: (312) 503-6470; Email: nkandula@nmff.org
Research Objective: When complicated health information is presented to patients, most cannot learn everything at once. Patients with low health literacy usually have lower baseline knowledge and, thus, more to learn. To address this, repeated assessments of knowledge gained and repetition of key learning goals that were not achieved has been recommended: the” teach-to-goal" (TTG) approach.
We evaluated the effectiveness of TTG for improving knowledge gain and knowledge retention after patients viewed a multimedia diabetes educational program (MDEP). Patients were interviewed using a pre-test, post-test design to measure diabetes knowledge before and after viewing the MDEP. Diabetes knowledge was measured using open-ended questions, administered by the interviewer. Patients were allowed to answer in their own words, and the interviewer then coded whether the response was correct. The TTG intervention was tailored to each patient; if a patient was unable to answer a posttest question, the interviewer presented the information again using a script and teaching aids, such as sections of the MDEP or print materials with pictures from the MDEP. This process was repeated until the individual exhibited learning mastery of the information or for a maximum of two times. Knowledge retention was measured by telephone 2 weeks later. Health literacy was measured using the Short Test of Functional Health
Literacy in Adults (S-TOFHLA), and categorized as
"inadequate" (0-16), “marginal” (17-22) and
"adequate" (23-36). The primary endpoints
compared changes in knowledge and knowledge retention. Differences in knowledge gained and retained were compared by paired t-tests. We compared the MDEP-TTG groups´ results with a historical comparison group (n=190), recruited form the same clinics, who received the MDEP only. 100 patients, with and without diabetes, were recruited from primary care clinics at a federally qualified health center and an academic health center. Other than the MDEP-TTG group being slightly younger than the MDEP only group (p<0.01), there were no other significant differences in demographics or
TOFHLA score between the two groups. The mean
TOFHLA score in the MDEP-TTG group was 29.2, with 10% of individuals having inadequate literacy,
10% marginal, and 80% adequate. After viewing the
MDEP, all individuals had significant increases in knowledge. In the MDEP-TTG group, individuals with inadequate health literacy gained 6.9 points
(Standard Deviation (SD) =4.2), those with marginal literacy gained 5.3 points (SD=5.7), and adequate gained 6.1 points (SD=6.1). The MDEP only group had similar knowledge gains. Although 79% of individuals in the MDEP-TTG group achieved learning mastery of the 17 key ideas within 3 passes, learning mastery differed significantly by literacy. After three attempts, 57% of individuals in the inadequate and marginal groups had achieved learning mastery compared to 84% of individuals in the adequate group (p<0.01). At the 2 week followup, there were no significant differences in knowledge retention; the MDEP-TTG group had a knowledge decrease of 4.1 points (SD = 3.7) compared to 3.1 (SD=2.5) in the MDEP only group
(p=0.09). TTG improved knowledge gain when used in conjunction with a MDEP; however,
TTG alone is inadequate to achieve long-term retention. Additional research is needed to investigate the usefulness of TTG in effecting behavior change, improving health outcomes, and overcoming health literacy barriers. Adding
Teach to Goal to a MDEP appears to be an effectibe way to improve knowledge across literacy levels in the short-term. However, TTG is not adequate to achieve long-term knowledge retention.
Other Northwestern Memorial Hospital Foundation
Prevention C 921.
● When Bureaucracy Works: Ryan White
HIV/AIDS Program Grantees’ Management of
Funding Shifts
Heather Kane, Ph.D.; Jennie Harris, M.P.H.; Amy
Roussel, Ph.D.; Megan Gray, B.A.; Julie Wise,
M.P.H.
Presented by: Heather Kane, Ph.D., Health
Services Analyst, Health, Social, & Economics
Research, RTI, International, 3040 Cornwallis
Road, P.O. Box 12194, RTP, NC 27709-2194,
Phone: (919) 541-6738; Email: hkane@rti.org
Research Objective: To examine how health care partnerships adapt to changes in the external policy and funding environments, based on the experiences of Ryan White HIV/AIDS Program grantees. The 2006 Congressional reauthorization of the Ryan White HIV/AIDS Program (formerly known as the Ryan White CARE Act) introduced policy changes to balance optimal allocation of funding with reaching the most vulnerable populations. These changes resulted in funding increases for some program grantees and decreases for others. As a result, grantees needed to respond quickly to maintain continuity of care for persons living with HIV/AIDS. This paper examines how bureaucratic structures and processes shaped grantees’ responses.
Study Design: Case studies with eight Ryan White
HIV/AIDS Program Grantees (selected to capture both increased and decreased funding from
FY2006 to FY2007), using a two-phase assessment. Phase I involved on-site, in-depth case studies of four grantees, including key informant interviews and focus groups with staff from grantees, subgrantees, providers, and community planning board members. Phase II case studies involved telephone interviews and focus groups.
Population Studied: Four Ryan White HIV/AIDS
Program Part A grantees and four Part B grantees, selected from 56 Part A grantees and 59 Part B grantees.
Principal Findings: The direction of the funding shift did not substantially change the way programs were organized and managed. Bureaucratic infrastructure and program management practices provided stability amid both losses and gains, offered structure and parameters that guided grantees’ decision-making, and served as a
“storehouse” for collective norms and values that grantees drew upon when allocating funds to subcontracted providers. When grantees experienced reduced or modestly increased funding, they strategized to buffer providers from losses by absorbing cuts at the administrative levels, leveraging external funds, and creatively managing funds (e.g., learning “to do more with less”). Hierarchical relationships supported two grantees in managing dramatic increases. These grantees lacked a process to award funds to new providers, so both drew upon and expanded existing organizational roles and relationships to resolve the problem.
Sometimes bureaucracy hindered grantees’ rapid responses. For instance, grantees needed to revise contracts in order to disperse monies to their providers, but could not do so in a compressed timeframe with their existing processes.
Conclusion: Organizational theories about bureaucracies suggest that they operate well under stable conditions, but falter under instability. That theoretical proposition was only partially supported here. Traditional bureaucratic features can benefit
organizations by maintaining the organizational priorities and values and structuring the responses of individuals consistent with such norms, particularly in more volatile policy and economic environments.
Implications for Policy, Delivery or Practice:
Healthcare organizations serving vulnerable populations operate in turbulent policy environments. Bureaucratic structural features offer potential protective benefits; comprehensive infrastructure may enable these organizations to maintain organizational values while adapting to environmental contingencies. However, differentiating those features that provide stability and support an organization’s mission from those that hinder achieving goals integrates the best features of bureaucratic and virtual organizations.
Funding Source(s): HRSA
Theme: Organizational Perfomance and
Management
● The Effect of Reason for CAM Use on
Conventional Medicine Utilization
Viji Kannan, B.B.A.; Benjamin Druss, M.D., M.P.H.;
Adam Atherly, Ph.D.; Laurie Gaydos, Ph.D.
Presented by: Viji Kannan, B.B.A., Student, Health
Policy & Management, Emory University, 1518
Clifton Road, Atlanta, GA 30322, Phone: (706)
461-3343; Email: vdkanna@sph.emory.edu
Research Objective: This study examines the effect of consumers’ reason for Complementary and
Alternative Medicine (CAM) use on their conventional medicine utilization. CAM users are characterized as high utilizers of conventional care, suggesting CAM modalities function as additions to conventional medicine instead of substitutes.
However, this characterization may be attributable to a subset of CAM users rather than a uniform tendency of all CAM consumers. We therefore distinguish between those who use CAM to treat illness versus those who use CAM to promote wellness.
Study Design: Likelihood of accessing each of eight conventional providers or services (primary care providers, medical specialists, mental health professionals, physical/occupational therapists, surgeries, emergency visits, office visits, and hospital nights) was analyzed using binary probit models. For the independent variable of interest, we segmented CAM users into two subsets: illness treatment or wellness enhancement, based on their reason for using unconventional medicine. Non-
CAM users serve as a control. Covariates affecting healthcare utilization included predisposing, enabling, and need factors.
Population Studied: We used the 2007 National
Health Interview Survey, which collected data regarding reason for CAM utilization and amount of conventional healthcare used, over a period of 12 months. Our sample consists of all adults reporting healthcare utilization, conventional or unconventional (n = 21, 729). The sample was stratified into healthy, moderate, and sick groups, allowing us to observe the interaction between reason for CAM use and health level.
Principal Findings: For each of the eight providers and services analyzed, the probability of obtaining conventional care was significantly lower for CAM use to promote wellness than CAM use to treat illness. Nonetheless, CAM generally served as an addition to conventional care; the likelihood of consuming conventional medicine for CAM users was equivalent to or higher than for non-CAM users. Exceptions suggesting CAM may serve as a substitute in some cases occurred for primary care, office visits, and emergency visits among CAM users motivated by wellness, in the moderate and healthy groups; and all CAM users in the healthy group were less likely to acquire office visits than non-CAM users.
Conclusion: CAM consumers, based on their reasons for seeking CAM, differ in their utilization of conventional medicine. Moderate to healthy consumers motivated by wellness to seek CAM are substituting these modalities for primary care and office visits. This may be partially due to decline in primary care providers or because primary care is most analogous to CAM.
Implications for Policy, Delivery or Practice: In this era of increasing chronic disease and healthcare costs, individuals’ efforts to enhance their health and wellness deserve attention. Many consumers state they engage in basic wellness efforts specifically to reduce their need for healthcare and to lower their healthcare costs.
However, insurance companies often do not offer coverage for wellness but only to treat a diagnosed disease and only if it is medically necessary. These findings indicate that with properly targeted products, insurance could cover CAM for wellness and are especially relevant for states mandating individual coverage.
● Quality of Diagnosis & Evaluation-related Care in Patients with Chronic Hepatitis C Virus
Infection
Fasiha Kanwal, M.D., M.S.H.S.; Tuyen Hoang,
Ph.D.; Jennifer Kramer, Ph.D.; Timothy Chrusciel,
M.P.H.
Presented by: Fasiha Kanwal, M.D., M.S.H.S.,
Assistant Professor of Internal Medicine,
Department of Internal Medicine, Division of
Gastroenterology & Hepatology, John Cochran VA
Medical Center, 915 North Grand Boulevard, St.
Louis, MO 63106, Phone: (314) 289-6434; Email:
Fasiha.Kanwal@va.gov
Research Objective: Chronic hepatitis C virus infection (HCV) is a prevalent, burdensome, and expensive condition within the VA. Congress has mandated a screening program for HCV among VA
users, which has been widely implemented.
However, how well the VA does in caring for patients after they are diagnosed with HCV is less clear. We sought to determine the level and determinants of the initial evaluation process of patients with HCV.
Study Design: Using a national VA HCV Registry from 1/1/2000 to 12/30/2006, we evaluated 3 modified Delphi-panel derived quality indicators
(QIs): confirmation of HCV viremia, specialty evaluation after confirmation, and testing for HCV genotype. We derived the aggregate score as the proportion of QIs satisfied among those for which patients were eligible. Using logistic regression multivariable analysis, we evaluated patient demographic, clinical, and comorbidity related factors potentially associated with the receipt of evaluation-related care in HCV.
Population Studied: National VA sample of patients with HCV
Principal Findings: Of 165,599 patients with positive HCV antibody, 65% received a confirmatory PCR test within a year follow-up after
HCV diagnosis. Of those with confirmed infection,
52% were evaluated by a specialist, and 50% received a genotype test within the following year.
Overall, patients received 52% of the recommended care, with fewer than 15% patients receiving all indicated QIs. In the multivariable analysis, African
American race (RR=0.67, p<0.0001), older age
(>65 y vs. <65 y, RR=0.63), alcohol or drug use
(RR=0.84), bipolar disorder (RR=0.83), psychosis
(RR=0.64), and end-stage renal disease (RR=0.84) were associated with poor adherence to aggregate measure of quality. In contrast, patients with a diagnosis of cirrhosis (RR=1.3) and those with HIV
(RR=1.06) were more likely to receive higher quality care. All p-values <0.001.
Conclusion: We found low overall adherence to evidence-based QIs. Several demographic, clinical, and comorbidity related factors explain part of the variation in the receipt of indicated care.
Implications for Policy, Delivery or Practice:
Although the VA has made targeted effort to improve rates of testing and specialty evaluation of
HCV patients, these efforts have not been very successful thus far. Future research will attempt to measure the role of non-patient factors in explaining the difference between expected and observed care. These data may ultimately assist researchers and policy-makers in focusing quality improvement efforts in HCV.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Quality Improvement Intervention to Reduce
Delivery Room Cold Stress in Very Low
Birthweight Infants
Heather Kaplan, M.D., M.S.C.E.; Robin Breig, R.N.,
B.S.N.; Carrie Hallett-Voss, R.N., B.S.N.; Dionni
Meek-Silvers, R.N., M.S.N.; Neka Corcoran, R.N.;
Peter Margolis, M.D., Ph.D.
Presented by: Heather Kaplan, M.D., M.S.C.E.,
Assistant Professor of Pediatrics, Neonatology;
Health Policy & Clinical Efectiveness, Cincinnati
Children's Hospital Medical Center, 3333 Burnet
Avenue, MLC 7009, Cincinnati, OH 45229, Phone:
(513) 803-0478; Email: heather.kaplan@cchmc.org
Research Objective: The National Quality Forum
(NQF) has endorsed the proportion of very low birthweight (VLBW <1500 grams) infants with an admission temperature < 36°C as a measure of quality of care. The objective of this research is to evaluate a quality improvement (QI) intervention to reduce the number of VLBW infants with low neonatal intensive care unit (NICU) admission temperatures.
Study Design: We used QI methods and reliability principles to implement a “bundle of interventions” suggested in the literature to reduce the number of infants admitted to the NICU with significant cold stress. The intervention bundle included staff education; consistent room set up prior to delivery including setting room air temperature >77°C; attention to thermoregulation during resuscitation including use of polyethylene bags; and transfer to the NICU in a warmed incubator. Staff used checklists to encourage adherence with the bundle.
Historical data on admission temperatures was obtained for the 6 months prior to the intervention
(n=97 infants) and prospectively after the intervention (n=144 infants through 10/31/2008).
Proportions of infants with temperatures < 36.5°C were compared before and after the intervention using student’s t test and statistical process control methods.
Population Studied: All VLBW infants admitted to the Good Samaritan Hospital NICU (Cincinnati,
Ohio) were included. The Good Samaritan Hospital
NICU is a 56-bed level III NICU with an annual volume of 264 VLBW admissions (2007).
Principal Findings: In the post intervention period,
64% of VLBW births had an intervention checklist completed, and of those, compliance with the complete intervention bundle reached 43% by
10/31/2008. The proportion of infants with temperatures <36.5°C was reduced from 92% to
38% (p<0.001) over a 7 month period. In the post intervention period, there was no difference in the proportion of infants with temperatures <36.5°C based on delivery in the operating room (35%) versus labor and delivery room (36%).
Conclusion: There was a significant reduction in the number of VLBW infants with a low NICU admission temperature indicating improving quality of care. Increased compliance with the intervention bundle and addition of other high reliability strategies may result in further reductions.
Implications for Policy, Delivery or Practice: We have demonstrated successful application of quality improvement methods and reliability principles that can be used by other neonatal-perinatal care
providers to improve their performance and quality of care as measured by the NQF endorsed metric.
Theme: Child Health
● Large Variation in Neonatal Intensive Care
Unit Use of Evidence Based Care: The Example of Vitamin A
Heather Kaplan, M.D., M.S.C.E.; Meredith
Tabangin, M.P.H.; Diana Henderson, M.P.H.,
M.S.W.; Peter Margolis, M.D., Ph.D.; Edward
Donovan, M.D.
Presented by: Heather Kaplan, M.D., M.S.C.E.,
Assistant Professor of Pediatrics, Neonatology &
Health Policy & Clinical Effectiveness, Cincinnati
Children's Hospital Medical Center, 3333 Burnet
Avenue, MLC 7009, Cincinnati, OH 45229, Phone:
(513) 803-0478; Email: heather.kaplan@cchmc.org
Research Objective: Vitamin A supplementation has been identified as an effective strategy for prevention of bronchopulmonary dysplasia (BPD) in premature infants. Our objective was to determine variation in the use of evidence-based Vitamin A supplementation among neonatal intensive care units (NICU).
Study Design: Retrospective cohort study of extremely low birthweight (ELBW <1000 grams) infants admitted to a children’s hospital NICU participating in the Pediatric Health Information
System (PHIS) database from January 1, 2005 to
March 31, 2008. Use of Vitamin A was defined as receipt of at least one dose because this reflects clinician intent to use Vitamin A. We calculated each center’s rate of Vitamin A use among ELBW infants along with 95% Confidence Intervals (CI).
Generalized Estimating Equation models were used to examine Vitamin A use over time, accounting for clustering by center.
Population Studied: Patients receiving care in a hospital participating in the PHIS administrative database were eligible for inclusion. The PHIS database contains inpatient data from 40 not-forprofit, tertiary care pediatric hospitals in the United
States affiliated with the Child Health Corporation of
America, a business alliance of children’s hospitals.
Patients with a BW 500-999 grams, determined using ICD-9-CM diagnosis codes, admitted within 7 days of birth were included. Infants with congenital anomalies or length of stay < 4 days were excluded. We excluded centers with <30 infants in the study period and those not submitting utilization data.
Principal Findings: Among 3,022 eligible infants cared for in 26 NICUs, 2,361 (78%) infants received no Vitamin A. Fourteen (54%) centers used
Vitamin A in some patients, while 12 (46%) centers did not use Vitamin A at all. Across NICUs using
Vitamin A, the proportion of infants receiving any
Vitamin A ranged from 0.9% [95% CI: 0.05-4%] to
92% [95% CI: 88-96%]. Infants discharged in 2007 and 2008 were more likely to receive Vitamin A compared to 2005, OR 2.3 [95%CI: 1.3-4.3] and 2.4
[95%CI: 1.2-4.5], respectively.
Conclusion: This study confirms the large variation in Vitamin A use reported by neonatologists
[Ambalavanan 2004]. Although use of Vitamin A is increasing over time, adoption has been slow, and marked variation across NICUs remains. This variation may reflect lack of awareness and agreement with published evidence. It may also be attributable to differences in organizational factors associated with reliable implementation of evidence.
Implications for Policy, Delivery or Practice: The first step in achieving high quality health care requires identifying a gap between the evidence and existing practice, as this study has done.
Based on published efficacy (risk difference 7%) and prevalence (62%) data [Tyson 1999], we estimate that as many as 165 instances of BPD or death may have been prevented if untreated infants in this study population were given Vitamin A. The results of this study will promote efforts to change physician behavior leading to improved use of
Vitamin A and improved quality of care for neonates.
Funding Source(s): Cincinnati Children's Hospital
Medical Center Outcomes Research Award
(Internal Grant Mechanism)
Theme: Child Health
● The Impacts of Break-Through Drug Classes on Total Health Expenditures: Empirical
Evidence from the 1996-2005 Medical
Expenditure Panel Surveys
Zeynal Karaca, Ph.D.
Presented by: Zeynal Karaca, Ph.D., Health
Economist / Manager, Avalere Health LLC, 1350
Connecticut Avenue, NW, Suite 900, Washington,
DC 20036, Phone: (202) 207-1335; Email: zkaraca@avalerehealth.net
Research Objective: The goal of this study is to empirically estimate the impact of break-through drug classes on total healthcare expenditures.
Study Design: This paper offers an improved way of measuring the cost impact and the identification of drug innovation. When a breakthrough occurs, varied firms introduce similar new therapies that replace numerous pre-existing therapies. These new drugs are often comparable to each other even though they may be introduced years apart.
Further, these waves do not hit all classes at the same time. A wave of major innovation in one class may occur during a period when innovation is relatively slow in another. While it may be true that usage of a superior drug will reduce contemporaneous medical expenditures, for many drugs, such as those used for chronic conditions, medical cost savings will be distributed over time.
Individual patient treatment histories vary substantially. We divided patients into three
categories of drug use: first category which consists of patients who frequently switch drugs over time; a second category that consists of patients who exclusively use break-through drugs; and a third category that consists of patients who never use these break-through drugs. Then, we estimated the impacts of break-through drug classes on total healthcare expenditure for each patient category.
Population Studied: We use the Medical
Expenditure Panel Survey (MEPS) data sets from
1996 to 2005.
Principal Findings: We studied two distinct groups of breakthrough drugs, Selective Serotonin
Reuptake Inhibitors (SSRIs) and Statins.The empirical results show that SSRIs and Statins cost more than other drugs, controlling for diagnosis, insurance, and demographics, but reduce nondrug expenditure by substantially more. The results also show that patients who regularly use SSSRIs and
Statins for their treatment experience considerable savings compared to patients who never use
SSRIs.
Conclusion: Our analysis indicates that the use of these groups of breakthrough drugs substantially reduce overall medical expenditures and have contributed to medical cost containment.
Implications for Policy, Delivery or Practice: The impact of new drugs and drug classes on overall healthcare expenditures is an important medical and economic question. Pharmaceutical firms spend billions of dollars developing new therapies, which are often sold at a substantial premium over older therapies. The use of these newer and more expensive drugs, however, is controversial. Health plans,
Medicare, and Medicaid often restrict the use of newer therapies due to cost considerations.
These policies may reduce pharmaceutical costs, but run the risk of excluding valuable therapies that can lower overall healthcare spending, reduce morbidity, lost schooling and work, and hospital stays. Others argue that newer therapies cost more and there are few differences between their therapeutic effects as compared to older, cheaper therapies. The findings of this study help Health plans, Medicare, and Medicaid to better understand to what extent these breakthrough drug classes save to U.S. healthcare system and improve the quality of life for patients.
Theme: Quality and Efficiency: Measurement
● Gender Differences in Utilization & Adherence to Cardiovascular Drugs among Medicare Part D
Enrollees
Zeynal Karaca, Ph.D.; Badi Baltagi, Ph.D.
Presented by: Zeynal Karaca, Ph.D., Health
Economist / Manager, Avalere Health LLC, 1350
Connecticut Avenue, NW, Suite 900, Washington,,
DC 20036, Phone: (202) 207-1335; Email: zkaraca@avalerehealth.net
Research Objective: The principle objective of our study is to add to this literature by determining, on an empirical basis, the extent there is a gap among male and female Medicare Part D enrollees in utilization of ACE inhibitors, beta-blockers, and statin regimens prescribed mostly for the treatment of cardiovascular diseases. The study also examines the extent drug utilization and adherence to drug treatment change between men and women when they are in the coverage gap, and assess if reaching the coverage gap has a different impact on women than men.
Study Design: All demographic, socioeconomic, and clinical characteristics are compared between men and women using Chi-Square tests for categorical variables. Demographic and socioeconomic covariates include age, gender, coverage status, payments by Part D and beneficiaries, service provider and prescriber information. All two-way interactions with gender are analyzed. We also test if there is a statistically significant two-way interaction between Medicare
Part D coverage and the effect of gender on cardiovascular drug use. Multivariate regression analyses are used to estimate the difference in cardiovascular drug use between men and women.
We measure the adherence to medications via the
Medication Possession Ratio (MPR). Following the existing literature, we calculate the MPR, we will label each in 2006 as “covered” or “not covered” by a prescription fill or refill. We assume that if a beneficiary filled a prescription for a new drug within a class or a new dose, the leftover of the previous prescription are discarded.
Population Studied: The stuyd population includes all Medicare Part D enrollees who have been prescribed at least one ACE inhibitor, beta-blocker or statin during 2006.
Principal Findings: Evidence of the benefit of short-term treatment with ACE inhibitors in women is lacking; however, long-term treatment reduces the risk of recurrent events. Although clinical trial evidence for use of beta-blockers in women is lacking, observational trials have demonstrated short- and long-term effectiveness. Statin therapy is equally effective in women and men.
Conclusion: Cardiovascular disease (CVD) is the leading cause of mortality in women. Retrospective analyses suggest that there are some clinically significant differences between women and men in terms of prevalence, presentation, management and outcomes of the disease, but little is known about why CVD affects women and men differently.
This study addresses the gap among male and female Medicare Part D enrollees in cardiovascular drug utilization and adherence to cardiovascular drugs.
Implications for Policy, Delivery or Practice:
Congress and CMS are committed to ensuring
Medicare beneficiaries, regardless of gender, receive high quality, clinically appropriate care. This research specifically enables policymakers and
clinicians who treat seniors to better understand any observed gap in cardiovascular drug utilization between men and women. These findings may also provide empirical evidence for adherence to medication for policymakers and researchers.
Funding Source(s): Avalere Health LLC
Theme: Disparities
● Does More Advertising Mean Better Quality in
Nursing Homes?
Bita Kash, Ph.D., M.B.A.; Thomas Miller, Ph.D.,
M.B.A.
Presented by: Bita Kash, Ph.D., M.B.A., Assistant
Professor, Health Policy & Management, Texas
A&M Health Science Center, School of Rural Public
Health, TAMU 1266, College Station, TX 77843,
Phone: (979) 458-0652; Email: bakash@srph.tamhsc.edu
Research Objective: The primary purpose of this study was to examine the association between advertising expenses and quality. We focused on answering two research questions: 1) Do nursing homes use advertising to signal superior quality?; and 2) Is there an association between price and quality?
Study Design: We used three alternative measures for quality to improve the robustness of this exploratory analysis. Quality measures were examined using Bonferroni correlation coefficient analysis. Associations between nursing home advertising expenses and quality were evaluated using three regression models with quality as the dependent variable. Independent variables of interest were advertising expense and price of one private bed per day while controlling for relevant facility and market level factors, such as occupancy rate, profit margin, ownership type, and level of market competition.
Population Studied: The Medicaid Cost Report for non-hospital based nursing facilities in Texas, facility level resident characteristics from the
Minimum Data Set (MDS), the Texas Quality
Reporting System (QRS), and the Area Resource
File (ARF) were merged for the year 2003. We dropped outliers based on the following decision rules: 1) facilities with advertising expenditures at more than $50,000 a year were excluded; 2) facilities with private pay prices above $500 a day;
3) facilities with $0 of advertising expenditures.
Data merging and cleaning resulted in a total sample size of 762 from the total population of
1,017 facilities.
Principal Findings: Advertising expenses were not associated with better nursing home quality as measured by three quality scales. The average price customers pay for one nursing home bed per day was associated with better quality only in one of the three quality regression models. The price of nursing home care might be a better indicator of quality and necessary to increase as advertising expenditures increase in the nursing homes sector.
Since more advertising expenditures are not necessarily associated with better quality, consumers could be mislead by advertisements and choose poor quality nursing homes.
Conclusion: We concluded that there was no significant association between quality and advertising expenditures. Therefore, advertising is not a signal of better quality in the nursing home sector. We did find a significant association between price and quality in one of three quality prediction models. Therefore, higher prices could be associated (and necessary) as advertising expenses increase in nursing homes.
Implications for Policy, Delivery or Practice:
From a policy standpoint one should be concerned about the large variation of advertising expenditures in publicly funded nursing homes. Average advertising expenses were at about $7,000 and could surprisingly amount up to $165,000 a year.
Some of this spending is surely linked to necessary marketing activities while some spending might be wasteful and not an effective marketing practice.
Therefore, consideration should be given to the definition of a level of appropriate and reasonable spending on advertising for the nursing home sector. Nursing home administrators should focus on customer relationship management tools instead of expensive adverting. Relationship management tools are proven marketing techniques, usually less expensive than advertising, and help with staff retention and quality outcomes.
Theme: Long Term Care
● Absorptive Capacity (ACAP) for
Transformation in Healthcare: A Framework for
Research
Bita Kash, Ph.D., M.B.A.; Larry Gamm, Ph.D.;
Aaron Spalding, M.H.A.; Josh Vest, M.P.H.
Presented by: Bita Kash, Ph.D., M.B.A., Assistant
Professor, Health Policy & Management, Texas
A&M Health Science Center, School of Rural Public
Health, TAMU 1266, College Station, TX 77843,
Phone: (979) 458-0652; Email: bakash@srph.tamhsc.edu
Research Objective: The purpose of this paper was to provide a research framework for the study of absorptive capacity (ACAP) for transformations specific to healthcare organizations, and to develop a measurement model that captures level of ACAP in the healthcare setting.
Study Design: We conducted a comprehensive review of literature covering various related topics including transformations in healthcare, ACAP measurement models, organizational learning, complex adaptive systems, strategy execution, organizational technologies, leadership, and culture. This comprehensive review of literature covered fields of sociology, physiology, organizational theory, management science, human
resources, and systems theory. Our analysis lead to a re-conceptualization of ACAP as it relates to transformations in healthcare. As a result we propose a measurement model for capturing level of ACAP in healthcare organizations at many levels
(individual, care team, department, and system level) before, during, and after a transformational initiative is implemented.
Population Studied: This study can be described as a meta analysis of absorptive capacity for transformation in health care. We reviewed 88 journal articles (70 from peer reviewed journals), 30 books/book sections and two websites covering theory and empirical studies related to transformational initiatives and/or absorptive capacity in healthcare and other sectors. Results from the analysis of these sources were used to develop a new framework for capturing ACAP for transformation in healthcare.
Principal Findings: A transformational initiative might prove effective in one health system and not for another. Variations in success also exist across centers and departments within a system. Some of these variations might be due to different levels of
ACAP at different levels of the organization. In order to understand the variations in the effectiveness of transformative changes within and across health systems we need to examine and measure the level of ACAP at many levels of the organization. This paper addresses these shortcomings by proposing a framework by which to examine transformational efforts that both attends to features regularly ignored in previous studies and one that utilizes an appropriate theoretical perspective for construct and measurement definition.
Conclusion: We present a framework of analysis that acknowledges healthcare organizations as complex systems, where ACAP has to be captured through multiple dimensions at many levels of the organization. ACAP profiles at many levels of the healthcare organization can be developed by capturing relevant dimensions of leadership, culture, and organizational technologies, including administrative, clinical, social, and information technologies, as defined in the proposed ACAP measurement model. By applying this measurement model to the study of transformation in healthcare issues and barriers to successful implementation of transformational initiatives can be realized, problem areas can be identified and addressed.
Implications for Policy, Delivery or Practice:
Capturing and tracking ACAP profiles can help the healthcare leaders with strategies to improve transformation implementation and success, informed decisions about the timing and selection of transformational initiatives, and decisions about the continuation or contraction of specific transformations within specific programs, teams or the entire healthcare system. The ACAP profiles can also facilitate the development of evidencebased management decision maps related to transformational efforts.
Funding Source(s): National Science Foundation,
Center for Health Organization Transformation
Theme: Organizational Performance and
Management
● Medicare Advantage Dual-Eligible Special
Needs Plans: Types & Scope of Contractual
Relationships with State Medicaid Programs
Jessica Kasten, M.A.; Paul Saucier, M.A.; Brian
Burwell
Presented by: Jessica Kasten, M.A., Senior
Research Leader, Research, Thomson Reuters,
4301 Connecticut Avenue, NW, Suite 330,
Washington, DC 20008, Phone: (202) 719-7804;
Email: jessica.kasten@thomsonreuters.com
Research Objective: To describe the types and scope of contractual relationships Medicare dualeligible Special Needs Plans (SNPs) have with state Medicaid programs. We identify states that contract with dual-eligible SNPs and provide a typology of the contractual relationships. We describe key features of the contracts such as: populations covered; Medicaid services included; models of care; and quality and performance measurement standards. In particular, the typology addresses whether Medicaid-funded long-term care
(LTC) services are covered under the contract. The
Medicare Improvements for Patients and Providers
Act of 2008 (MIPPA) requires that, as of January 1,
2010, a new dual-eligible SNP or an existing dualeligible SNP that wishes to expand its service area must have a contract with the state Medicaid agency to provide the Medicaid benefits a dualeligible enrollee is entitled to under the state’s
Medicaid program, or arrange for their provision.The MIPPA state contracting requirement highlights the need to better understand the extent to which dual-eligible SNPs currently contract with state Medicaid programs and the nature and scope of these contracts. The study, conducted for the
Office of the Assistant Secretary for Planning and
Evaluation at the U.S. Department of Health and
Human Services, is policy-relevant and timely as it provides a baseline of SNP-state contracts prior to implementation of the MIPPA requirement, and informs the 2010 Medicare Advantage contracting cycle.
Study Design: We used a combination of data sources to ascertain which states have contracts with dual-eligible SNPs: SNP proposals submitted to CMS for the 2008 contract year; research literature; and communications with state Medicaid officials. We obtained current SNP-state contracts from state Medicaid officials and Web sites.
Population Studied: Dual-eligibles enrolled in
SNPs.
Principal Findings: Roughly one-third of states
(including Puerto Rico) have contracts with dual-
eligible SNPs. We identified three major types of contracts: Medicare cost-sharing only; Medicare-
Medicaid coordination; and Medicaid only (Medicaid contracts with the same corporate entity but without any care coordination provisions). The types of dual-eligible populations covered by SNP-state contracts vary from all dual-eligibles to subgroups such as the elderly or those who require a nursing home level of care. Only a handful of contracts cover the full range of Medicare and Medicaid services. We encountered definitional issues in identifying SNP-state contracts. Many corporate entities which own and operate SNPs also have
Medicaid plans that include some dual-eligibles. It is unclear whether contractual relationships between these plans and state Medicaid programs meet the MIPPA requirement as the contracts typically apply only to Medicaid benefits and do not address coordination of Medicare and Medicaid services.
Conclusion: The landscape of SNP-state contractual relationships is more complex than many policymakers and researchers may realize.
The scope of contracts varies considerably depending on program goals and other factors.
Implications for Policy, Delivery or Practice: Our study shows policy initiatives to promote the coordination of Medicare and Medicaid-covered benefits for duals enrolled in SNPs must go beyond a simple contracting mandate. Understanding the current array of SNP-state contracts informs next steps in leveraging these relationships to increase coordination of Medicare and Medicaid services for dual-eligibles.
Funding Source(s): HHS/ASPE
Theme: Medicare
● Factors Influencing Enrollment in Medicare
Part D Enhanced Drug Coverage Plans
John Kautter, Ph.D.; Leslie Greenwald, Ph.D.;
Gregory Pope, M.S.
Presented by: John Kautter, Ph.D., Senior
Economist, Health Care Financing & Payment, RTI
International, 1440 Main Street, Suite 310,
Waltham, MA 02451, Phone: (781) 434-1723;
Email: jkautter@rti.org
Research Objective: It is well known that the
Medicare Part D prescription drug benefit has a coverage gap, commonly known as the “donut hole”. One of the objectives of the Medicare Part D program has been to increase beneficiaries’ choices of, and access to, enhanced drug coverage in the donut hole. This research focuses on factors influencing enrollment in Medicare Part D enhanced drug coverage plans.
Study Design: Using a multivariate framework, we examine the factors determining Medicare Part D enrollment in enhanced drug coverage plans. We estimate a logistic regression model of beneficiary choice between basic and enhanced coverage.
Because the vast majority of dual
Medicare/Medicaid beneficiaries are automatically enrolled in basic plans, conceptually it is not appropriate to include these beneficiaries in an enrollment choice model. Further, because the process of Medicare Advantage enrollees' choice of
Part D plans may be different from the choice process of traditional Medicare beneficiaries, we excluded Medicare Advantage beneficiaries from the analysis as well. The Part D risk adjustment score is used to control for a beneficiary’s predicted drug costs.
Population Studied: Medicare 5 percent sample of beneficiaries enrolled in Part D in July 2006.
Principal Findings: Compared to the reference group of young elderly beneficiaries age 65-74, beneficiaries eligible by disability, and older elderly beneficiaries, have a lower odds of enrolling in enhanced drug coverage. Females also have a lower odds of enrolling in enhanced drug coverage compared to males (odds ratio = 0.95).
Beneficiaries residing in the Midwest, South, and
West census regions, and also beneficiaries in more rural areas, have higher odds of enrolling in enhanced drug coverage. Sicker beneficiaries tend to enroll in enhanced drug coverage plans more than basic plans (odds ratio = 1.29), meaning that if the Part D risk adjustment score is increased by
1.00, the odds of enrolling in a enhanced drug coverage plan increases by 29 percent.
Conclusion: Although we found a statistically significant result that sicker Part D enrollees have a higher odds of enrolling in enhanced drug coverage, we also found that the mean risk adjustment score for enhanced drug plan enrollees was broadly similar to basic plan enrollees. Our multivariate finding on the risk adjustment score makes some intuitive sense as beneficiaries with greater health care needs are more likely to need more extensive drug coverage, and hence more likely to choose a richer benefit package. However, while we do find evidence that sicker beneficiaries are more likely to enroll in enhanced drug coverage plans, the substantive differences are small and arguably not large enough to be considered evidence of selection bias.
Implications for Policy, Delivery or Practice: We find the results of this study counterintuitive given the hypothesis that chronic users of prescription drugs will be more likely to enroll in enhanced drug coverage plans. Further, these results are inconsistent with our qualitative analysis findings, in which enhanced drug coverage plans claimed to have experienced adverse selection. Possibly enhanced drug coverage plans are not a better deal for beneficiaries in poorer health because they “pay for” enhanced benefits with higher premiums.
Funding Source(s): CMS
Theme: Medicare
● Comparative Effectiveness of Multiple
Treatments for Rheumatoid Arthritis using an
Observational Study Design
Aniket Kawatkar, M.S., B.P.; Michael Nichol, Ph.D.
Presented by: Aniket Kawatkar, M.S., B.P.,
Research Associate, Titus Family Department of
Clinical Pharmacy & Pharmaceutical Economics &
Policy, University of Southern California, 1540
Alcazar Street, CHP-140, Los Angeles, CA 90033,
Email: kawatkar@usc.edu
Research Objective: In observational studies, threats to internal validity stems from non-random assignment to treatment and individual heterogeneity of responses to treatment. Causality cannot be estimated unless these sources of bias are controlled. Our objective was to estimate the treatment effects of biologic disease modifying antirheumatoid drugs (DMARDs) on quarterly total health-care expenditure, while controlling nonrandom assignment to treatment (endogeneity) and allowing heterogeneity in treatment effects. We also estimated the impact of varying selection biascorrection techniques on the estimated treatment effects. The structural parameters, heterogeneous
(ATE) and homogeneous (ATE1) average treatment effects were defined as the impact of treatment on quarterly expenditure, if patients are randomly assigned to biologic DMARDs.
Study Design: We used established methods for retrospective study design using secondary data, to assemble a homogeneous cohort of rheumatoid arthritis patients with similar starting points of
DMARD utilization. Non-overlapping quarters were created from pharmacy claims for biologic
(adalimumab and etanercept) and standard
(lefluonomide, hydroxychloroquine and sulfasalazine) DMARDs. In the two-stage estimation, the treatment selection model was varied between multinomial and nested-logit, to avoid independence of irrelevant alternatives. The outcome equation was fixed-effects correlated random coefficients model, allowing heterogeneity in parameters. Endogeneity was controlled by addition of a generalized residual function, constructed using Lee’s (1983) approach for nonnormal polychotomous choice models.
Furthermore, for the multinomial-logit selection model, construction of the generalized residual function was varied based on 4 different biascorrection techniques namely, Lee’s (1983), Dubin and MacFadden’s (1984) and 2 variants of Dahl’s
(2002) approach with squared and quadratic series expansions without interactions. Hypothesis testing was based on cluster-bootstrapped errors.
Population Studied: Retrospective cohorts were constructed from California Medicaid (Medi-Cal) paid insurance claims between January 1st, 1999 to
December 31st, 2005. Final sample included 24504 episodes on 5510 patients.
Principal Findings: Controlling endogeneity significantly increased ATE1 for adalimumab, while decreasing ATE1 for etanercept, as compared to naïve fixed-effects model. ATE1 estimate from the nested-logit model based correction was lower as compared to multinomial-logit. When heterogeneity in parameters was allowed, ATE of adalimumab was not statistically significant under the nestedlogit corrected model. ATE for etanercept ($17,466, p=0.93) was not statistically significant under either of the endogeneity corrected models. Based on likelihood-ratio test for selection models, and t-test for time-varying endogeneity, the appropriate model was the nested-logit based endogeneity corrected fixed-effects model.
Conclusion: Random assignment of patients currently on standard DMARD treatment to either adalimumab or etanercept may not cause a significant difference in quarterly expenditure, if treatment effects are heterogeneous and treatment decision are based on these individual gains.
Implications for Policy, Delivery or Practice: The methodological issues addressed in this study impact our understanding of the cost effects of drug treatment choice. We may accept or reject a formulary policy to be cost-effective, by modifying the assumptions of how the treatment works and how it’s selected in real life. Models need to be realistic to mimic real life clinical decisions to inform important drug coverage decisions. We have shown that the panel data correlated random coefficients model with endogeneity correction is a practical and realistic tool to assess comparative effectiveness using observational study design.
● Understanding the Societal Im[pact of Obesity
& the Relationship of Obesity with Physical
Activity & Employment
Aniket Kawatkar, M.S., B.P.; Michael Nichol, Ph.D.
Presented by: Aniket Kawatkar, M.S., B.P.,
Research Associate, Titus Family Department of
Clinical Pharmacy & Pharmaceutical Economics &
Policy, University of Southern California, 1540
Alcazar Street, CHP-140, Los Angeles, CA 90033,
Email: kawatkar@usc.edu
Research Objective: To quantify the effect of obesity on the total direct medical expenditure and health related quality of life of US population. The second objective was to estimate the causal effect of physical activity on obesity and also understand the effect of obesity on employment.
Study Design: Obesity (OB) was defined by a binary indicator if body mass index was greater than 30. Binary indicators were also defined if an individual was employed (EMP) and if an individual spent greater than or equal to 30 minutes in moderate to vigorous physical activity (PA) at least
3-times a week. Direct medical expenditure attributable to obesity was estimated under diverse econometric models to assess sensitivity to zero
mass, non-negative values and skewed distribution.
Expenditure models compared were 1-part and 2part ordinary least-squares (OLS) on raw and logtransformed expenditure and generalized linear models (GLM) with log-link and Gamma/Poisson families. We employed Box-Cox, Modified Park’s,
LINK, RESET, Hosmer-Lemeshow, Pearson’s correlation and Copas test to determine model specification, diagnostics and cross-validation.
Incremental expenditure from the method of recycled predictions summed over population with obesity gave the total expenditure. Impact of obesity on SF12 mental and physical health component was assessed by OLS adjusting for age, gender, race, ethnicity, income, geographiclocation, and comorbidity. Causal relationships between obesity and employment and that of physical activity on obesity were studied separately.
To control for the unobservable factors affecting PA and OB, we specified separate recursive bivariate probit models. The common covariates included age, gender, race, ethnicity, geographical and metropolitan area location, smoking status, and comorbidities.
Population Studied: Data from the MEPS’s
Household Component (2006), a nationally representative survey of the U.S. civilian noninstitutionalized population, was used. Analysis accounted for the survey’s clusters, strata and sampling weights.
Principal Findings: The 2-part model of OLS on raw expenditure was the only model to pass all the specification and cross-validation tests. Based on this model, annual incremental expenditure attributable to obesity was $1188.5 (95% CI =
$1183.7 to $1193.4) while the total expenditure was
$72 billion (95% CI = $68.2 to $75.8 billion). As compared to the non-obese, the SF12 physical component score was lower by 3.7 while the mental component score was lower by 1.2 for those with obesity and these differences were statistically significant (p<0.001). Effect of selection bias was positive and significant in both the recursive probit models. Physical activity decreased the probability of obesity by 27.7% (p<0.001) in the bias corrected model. Obesity decreased the probability of being employed by 14% (p=0.009).
Conclusion: Obesity exerts an enormous economic and humanistic burden on the U.S. civilian non-institutionalized population.
Unobservable confounders may be causing an underestimate of the effect of physical activity on obesity and also underestimating the impact of obesity on employment.
Implications for Policy, Delivery or Practice:
Prevention of obesity has a significant potential to not only decrease medical expenditure but also improve humanistic costs to society. Indirect gains from increased labor force participation are also possible. Promotion of regular physical activity should be vigorously advocated.
● Compassion Satisfaction & Compassion
Fatigue among Hospice Palliative Care
Providers: The Impact of Caring
Arminee Kazanjian; Harvey Bosma, M.S.W.;
Suzanne Slocum, Ph.D.; Lars Apland, M.A.
Presented by: Arminee Kazanjian, Professor,
School of Population & Public Health, University of
British Columbia, 5804 Fairview Avenue,
Vancouver, BC, V6T 1Z3, CA, Phone: (604) 822-
4618; Email: a.kazanjian@ubc.ca
Research Objective: Little is known about experiences of compassion satisfaction (CS) and compassion fatigue (CF) among the Hospice
Palliative Care (HPC) workforce, given their constant exposure to loss, grief and bereavement.
CS and CF are useful constructs to understand workforce conditions specific to HPC but, until this study, have seldom been applied to HPC.
Study Design: The research is based on results of a cross-sectional survey of HPC personnel in
Canada in which experiences of compassion satisfaction and compassion fatigue were measured. The questionnaire also inquired about personnel background, education, employment experience, personal and organizational cultures, and other workforce-related concepts and issues.
Included in the questionnaire was a psychometric scale that was refined and validated specifically for these constructs. The findings were analyzed in relation to other survey variables such as health care discipline, level of education of provider, individual attributes, and organizational factors.
Population Studied: Administrators, managers, volunteers and HPC professionals who deliver HPC services in a range of agencies and institutions across Canada were approached to take part in the study.
Principal Findings: The research provides important information about the prevalence of CS and CF among HPC care providers and volunteers across Canada, as well as individual and organizational factors that influence them, the extent to which they are evident in different cultural contexts and information about their cross-cultural dimensions.
Conclusion: This is the first time empirical data has been collected about the prevalence of CS and
CF among HPC personnel in a wide range of settings in Canada. The findings identify sociodemographic and cultural determinants of CS and
CF and will be used to produce and test statistical models that explore in more detail their relationship to experiences of CS and CF.
Implications for Policy, Delivery or Practice: The information collected in this research can be used by managers, policy-makers, educators, front-line clinicians and volunteers to better understand the adverse outcomes of compassion fatigue and to develop strategies to mitigate them and to foster compassion satisfaction. Ultimately, the findings
can inform the development of education programs and effective workplace practices and policies that will promote positive growth for HPC providers.
Funding Source(s): N/A, Canadian Institutes of
Health Research
Theme: health care workforce
● A Pan-Canadian Study of Hospice Palliative
Care Providers: A Two-Phase Approach to
Identify the Workforce
Arminee Kazanjian, Harvey Bosma, M.S.W.;
Suzanne Slocum, Ph.D.; Lars Apland, M.A.
Presented by: Arminee Kazanjian, Professor,
School of Population & Public Health, University of
British Columbia, 5804 Fairview Avenue,
Vancouver, BC, V6T 1Z3, CA, Phone: (604) 822-
4618; Email: a.kazanjian@ubc.ca
Research Objective: Because so little is known about the numbers, distribution and mix of professionals and volunteers who provide palliative care in Canada, or the organizational structures within which such care is being delivered, our research sought to enumerate these personnel and explore their backgrounds, credentials, the wide range of services they provide, and the agencies within which they work
Study Design: The research is based on crosssectional surveys administered in two phases.
Phase I sought to enumerate the palliative care workforce across Canada, and determine the research population, by approaching administrators and managers for email contact information on hospice palliative care (HPC) personnel and volunteers in their organizations. In Phase II, a workforce questionnaire was administered to frontline HPC personnel, inquiring about personnel background, education, employment experience, personal and organizational cultures, and other workforce-related concepts and issues. Both phases were initially administered online with paper-copy mail-in questionnaires utilized to supplement response rates.
Population Studied: Administrators, managers, volunteers, and the complete range of health care professionals who provide HPC services in virtually every type of HPC agency or institution in Canada were approached to take part in the research.
Principal Findings: Currently, there is no national description or inventory of HPC professionals and volunteers in Canada. This is the first study of its kind to provide a demographic and employment profile of this sector’s workforce that includes the identification, enumeration and diversity of HPC providers across the country. Survey results also provide information about the education and training levels of professionals and volunteers in this area of health care. Finally, the results indicate the range of HPC services provided and the organizational structures within which such care is being delivered across Canada.
Conclusion: The findings of this survey provide a baseline description of the HPC workforce in
Canada and are an important backdrop of information for the recruitment and retention of an optimal number and mix of HPC personnel.
Implications for Policy, Delivery or Practice: The information gathered from this research can be used to determine an optimal confluence of personnel attributes, organization culture and system structure that will sustain best practices in
HPC and minimize workforce turnover. It can also be used for better development of HPC education and training programs
Funding Source(s): Canadian Institutes of Health
Research
Theme: Health Care Workforce
● Hospital Care on the Weekends Versus
Weekdays for Stroke Patients: Measuring the
Quality of Care
Abby Kazley, Ph.D.; Diane Gartner Hillman, D.H.A.;
Kit Simpson, D.P.H.
Presented by: Abby Kazley, Ph.D., Assistant
Professor, Health Professions, Medical University of
South Carolina, 151 Rutledge Avenue, B412,
Charleston, SC 29425, Phone: (843) 792-0012;
Email: swansoaj@musc.edu
Research Objective: This study examines the differences in hospital care on the weekend versus the weekdays by determining the aggressiveness and outcomes of care for acute ischemic stroke
(AIS) patients. Aggressive treatment for stroke is considered the administration of tissue plasminogen activator (t-PA). t-PA must be administered within three hours of the onset of symptoms. Quality of care is measured by mortality.
Study Design: To test the relationship between weekend hospital admission and quality outcomes and aggressive treatment, two logistic regressions are used. The dependent variables are t-PA and patient mortality. We control for potential selection bias two ways. First, we use a propensity score adjustment with five strata. We also test for differences in weekend admission between AIS and hemorrhagic stroke patients using logistic regression to determine if there is something unique about the AIS patients who present on the weekends that may influence the processes or outcomes of their care. We control for hospital characteristics (size, volume of AIS patients, stroke center certification, time period before or after t-PA reimbursement increase) and patient characteristics
(gender, age, race, residential distance from a stroke center, and atrial fibrillation).
Population Studied: This study includes 78,657 qualifying AIS patients in the state of Virginia from
1998 to 2006. The data come from 112 hospitals’ discharge information. We also compare these AIS patients to 20,101 hemorrhagic patients to
determine any selection bias in weekend admission.
Principal Findings: The model measuring aggressive patient treatment is statistically significant overall. AIS patients who are admitted to the hospital on the weekends are 20% more likely to be treated aggressively (p<.05). The mortality model is also significant and correctly predicts patient mortality 92.9% of the time. There is no statistically significant difference in mortality based on the day of admission. The relationships between weekend admission and aggressiveness of treatment and quality remain the same when controlled for with propensity score adjustments.
We find no differences between AIS and hemorrhagic stroke patients with regard to their day of admission.
Conclusion: We find no evidence to suggest that weekend hospital care is less aggressive or of less quality than care provided during the week. In fact, care was determined to be more aggressive on the weekends. One possible reason is that patients seen on the weekend may more quickly gain access to CT scans due to the reduced number of elective procedures. This finding is in conflict with recent research and suggests that hospitals may organize resources in a way that allows care to be better delivered. Quality is a complex construct that can be measured in many ways, and statistical control must be used to prevent selection bias threats in the process of care.
Implications for Policy, Delivery or Practice:
Conditions such as stroke require care that is provided consistently around the clock since of is of the essence with certain treatments. Hospitals must operate to allow for the rapid and high quality treatment of patients. Practitioners should determine if the process of care is designed to provide optimal access to equipment including CT scanners.
Theme: Quality and Efficiency: Measurement
● Associations Between Population
Demographics & Public Health Intraorganizational Networks
Jonathan Keeling, M.S.; Jaqueline Merrill, R.N.,
M.P.H., D.N.Sc.
Presented by: Jonathan Keeling, M.S., National
Library of Medicine Pre Doctoral Trainee in
Biomedical Informatics, Columbia University, 630
West 168th Street, VC 5, New York, NY 10032,
Phone: (212) 305-3194; Email: jonathan.keeling@dbmi.columbia.edu
Research Objective: Local health departments must use limited public resources efficiently and effectively to promote and protect population health.
Network analysis is a technique used to improve efficiency and effectiveness in private sector organizations. This non-linear empirical method offers support for management decisions by modeling complex organizational dynamics. We applied the method in local health departments
(LHDs) to determine the relationship between their organizational networks and characteristics of the local population served.
Study Design: The study is a comparative network analysis conducted in a national sample of local health departments. Health department employees were surveyed on the quality of communication with each of his/her co-workers. We computed network measurements of centralization, density, speed, administrative siloing, and characteristic path length to measure effectiveness of communication and information diffusion in the network. We computed transitivity, complexity, and task redundancy to measure the cohesiveness or coordination of the network. Correlations were performed between these measurements and demographic characteristics of the population served taken from the 2006 US census update.
Population Studied: A purposive sample of 11
LHDs was identified based on geographic location and number of employees. The sample includes departments from States with both centralized and decentralized public health systems that serve a range of populations in urban and rural locations.
1064 employees completed a network survey for a mean response rate of 84%.
Principal Findings: LHDs that are larger in size and that serve larger populations with more households, more persons per household, and higher median incomes are negatively associated with network density, speed, transitivity, and complexity and positively associated with characteristic path length and administrative silos.
LHDs that serve areas with younger populations are more centralized while those that serve more elderly populations are denser and have a higher complexity. A higher percentage of vulnerable populations such as those below the poverty line and American Indians are associated with administrative silos and increased task redundancy.
Lower population density is positively associated with increased characteristic path length.
Conclusion: Previous research has shown that larger LHDs generally perform better. This may best be explained by the increased specialization that is possible in larger LHDs which leads to increased performance and efficiency in individual tasks. The cost of this specialization is the bureaucracy required to coordinate specialization which can degrade organizational efficiency.
Although our data set was small, network measurements seen in relation to demographic attributes such as population age and vulnerability suggest that LHDs do adapt to their task environment although these adaptations may produce inefficiencies in communication and information flow.
Implications for Policy, Delivery or Practice:
These findings parallel results produced by general organizational research. This suggests that LHDs
share characteristics with other organizations such as those in the private sector and that evidencebased management strategies developed for these organizations are likely to apply in public health.
Identification of consistent organizational network features associated with population demographic characteristics may allow LHDs to optimize common elements across systems and support national standardization efforts.
Funding Source(s): RWJF
Theme: Organizational Performance and
Management
● “I don’t know much about the vaccine. I haven’t heard anything.” - Cervical Cancer, HPV
& HPV Vaccine Awareness among Rural Latino
Parents
Deanna Kepka, M.P.H., Ph.C.; Beti Thompson,
Ph.D.; Gloria Coronado, Ph.D.; Hector Rodriguez,
Ph.D., M.P.H.
Presented by: Deanna Kepka, M.P.H., Ph.C.,
Ph.D. Student, Health Services, University of
Washington, Box 357660, Seattle, WA 98195,
Phone: (206) 321-0936; Email: kepka@u.washington.edu
Research Objective: Latinas have almost a twofold higher incidence of cervical cancer compared to non-Hispanic white women. Little research has been carried out with Latino parents related to assessing awareness and acceptability of the new
HPV vaccine to prevent cervical cancer. The CDC recommends the vaccine for girls at ages 11 and 12 years old but it can also be given to girls between the ages of 9 – 26. This study aims to specifically explore rural Latino parents’ knowledge and beliefs related to HPV and the HPV vaccine in the Lower
Yakima Valley of the state of Washington.
Study Design: Purposive sampling was used to recruit Latino parents to participate in in-depth qualitative semi-structured interviews that investigated cervical cancer, HPV, and HPV vaccine awareness, knowledge, beliefs, and attitudes rooted in local socio-cultural perspectives.
Data were analyzed using Atlas.ti and grounded theory methodology.
Population Studied: 26 Latino parents (16 mothers and 10 fathers) of daughters ages 9-14 living in an agricultural region of the state of
Washington participated in this study. Half of the participants were between the ages of 41 and 50 years old (Age Range: 31-60), 89% were born in
Mexico, and an average of 8.9 years of formal education was reported. Most of the families earned less than $30,000 per year, 69% of the participants knew only or mostly Spanish, and most attended local community health centers.
Principal Findings: Overall, participants demonstrated large gaps in knowledge of cervical cancer, HPV, and the HPV vaccine. Most participants had not heard of cervical cancer or
HPV. A few had heard of the HPV vaccine.
However, most did not know the cause of cervical cancer, the number of vaccines in the series, recommended ages for vaccine uptake, vaccine costs, side effects, and where it may be administered. Most participants were supportive of the vaccine but would like more information about it.
Although most participants had a medical home for their daughters at a local community based health center, few received information from their medical providers about the HPV vaccine. Local community health centers stated that most of their patients qualified to receive the HPV vaccine at little or no cost. Most parents felt that they would be interested in speaking with a medical provider about their questions related to cervical cancer, HPV, and the
HPV vaccine.
Conclusion: Nearly all of the participants demonstrated limited exposure to and knowledge of cervical cancer, HPV, and the HPV vaccine. The majority of participants demonstrated insufficient levels of knowledge related to the logistics of vaccine administration to seek the vaccine for their daughter(s). Few received information related to the
HPV vaccine from their medical providers.
Implications for Policy, Delivery or Practice:
Culturally and linguistically appropriate cervical cancer prevention and HPV vaccine awareness programs are needed that specifically target rural
Latino parents at local community based health centers and in rural communities. Promoting parental engagement in actively learning more about cervical cancer, HPV, and the HPV vaccine may improve parental decision making ability related to supporting HPV vaccine uptake for their daughter(s). Without these targeted interventions, health disparities related to cervical cancer incidence and mortality may further widen.
Funding Source(s): NCI
Theme: Child Health
● Acculturation & HPV Infection among U.S.
Latinas
Deanna Kepka, M.P.H., Ph.C.; Beti Thompson,
Ph.D.; Gloria Coronado, Ph.D.; Hector Rodriguez,
Ph.D., M.P.H.
Presented by: Deanna Kepka, M.P.H., Ph.C.,
Ph.D. Student, Health Services, University of
Washington, Box 357660, Seattle, WA 98195,
Phone: (206) 321-0936; Email: kepka@u.washington.edu
Research Objective: Latinas have almost a twofold higher incidence of cervical cancer compared to non-Hispanic white women. In 2006, the US
Food and Drug Administration approved a new vaccine for young women that protects against the primary types of high-risk HPV (HPV-16 and HPV-
18) that cause approximately 70% of cervical cancer cases. The CDC recommends the vaccine for girls at ages 11 and 12 years old but it can also
be given to girls between the ages of 9 – 26. With increased acculturation, Latinas are more likely to engage in high-risk sexual behaviors that increase risk for HPV infection. More acculturated Latinas have shown to have a greater number of sexual partners, a greater number of partners’ sexual partners, and an earlier age of first sexual intercourse. This study aims to improve the understanding of the relationship between acculturation and HPV infection among US Latinas to inform HPV vaccine promotion efforts.
Study Design: Using multivariate logistic regression models, we investigated the relationship between four common measures of acculturation and rates of HPV-positive status for a nationally representative sample of Latinas. These four measures included one dichotomous language scale, one five-point language scale, participant country of birth, and parents’ country of birth. The relationship between acculturation and high-risk oncogenic HPV infection alone was also investigated because high-risk HPV types may differ from other HPV types in degree of persistence due to a number of biological and behavioral factors.
Population Studied: The National Center for
Health Statistics conducts the National Health and
Nutrition Examination Surveys (NHANES), the only national survey that collects extensive health information from both face-to-face interviews and medical examinations. NHANES (2003-2004) is the most recently available nationally representative dataset that measures HPV infection with a medical exam. A separate supplement of questions related to acculturation was included in NHANES (2003-
2004). 529 Latinas participated in the NHANES
(2003-2004) medical exam. Among these, 503
(95%) had adequate HPV test results (Mean Age =
28.6; Age Range: 14.0 – 60.1).
Principal Findings: In adjusted multivariate logistic regression analyses that accounted for the complex survey design, Latinas demonstrated greater levels of high-risk HPV infection with higher levels of
English language ability (1.33 OR, 95% CI: 1.02-
1.72); with birth in the US (3.29 OR, 95% CI: 1.26-
8.62); and with US born parents (1.84 OR, 95% CI:
1.19-2.85). Levels of acculturation, however, did not relate to HPV-positive status for Latinas in this study.
Conclusion: With higher levels of acculturation, increased high-risk HPV infection occurs for Latinas in the US. Further investigation is needed to explore the specific risk factors that may influence high-risk HPV susceptibility among Latinas with higher levels of acculturation.
Implications for Policy, Delivery or Practice:
Targeted interventions are needed to effectively promote the HPV vaccine to US Latinas by level of acculturation to lower health disparities related to high incidence and mortality due to cervical cancer.
Without targeted interventions to increase HPV vaccine uptake for at-risk Latinas, health disparities related to cervical cancer incidence and mortality may further widen.
Funding Source(s): AHRQ
Theme: Disparities
● Global Nurse Migration: Its Impact on
Developing Countries & Prospects for the
Future
Amir Khaliq, Ph.D.; Robert Broyles, Ph.D.; Ari
Mwachofi, Ph.D.
Presented by: Amir Khaliq, Ph.D., Associate
Professor, Health Administration & Policy,
University of Oklahoma Health Sciences Center,
College of Public Health, 801 Northeast 13th Street,
College of Health Building, Room 349, Oklahoma
City, OK 73104-5072, Phone: (405) 271-2115 x
37073; Email: amir-khaliq@ouhsc.edu
Research Objective: A large body of literature on the international migration of nurses has appeared in the last 10–15 years with a focus on source countries in Asia, Africa, and Latin America or destination countries such as the US, Canada, and the UK. Some studies have examined the patterns of migration and raised ethical questions about recruiting nurses from developing countries. Others have commented on the impact of nurse migration on developing countries and the challenges faced by source countries such as the Philippines and
India. The objective of this study is to examine the impact of nurse migration on developing countries in a comprehensive manner.
Study Design: Literature review.
Population Studied: Registered nurses trained and employed in various developing and developed countries
Principal Findings: Currently, there is a shortfall of
600, 000 nurses in Sub-Saharan Africa. In Malawi, fifty percent of available nursing posts remained unfilled in 2003. In contrast, the number of newly registered foreign-trained nurses in the US has increased every year since 1998. Approximately
4000 foreign-trained nurses were registered in the
US in 1998; the number had risen to 15000 by
2004.
Strong “pull” and “push” factors favor wealthier nations in coping with nursing shortages through overseas recruitment. The wage differential between source and destination countries is the single most important driver of nurse migration. In terms of purchasing power parity, a Canadian nurse in 2004 earned 25 times more than one in Zambia.
Between 2000 and 2003, the economic loss in the
Caribbean region through public training of nurses who left was in the vicinity of $30.2 million.
Conclusion: Developing countries are losing trained and experienced nurses at a rate much faster than the rate of production. The nurses who emigrate are often the “cream of the crop”.
Consequently, the pool of nurses who serve as educators or administrators is being rapidly
depleted. The mitigation of “push” factors requires developing countries to close the wage differential between source and destination countries.
However, given the vast difference in wages there is no hope that developing countries will be able to close the gap. Small increases in salaries will have practically no effect on the exodus of nurses in the foreseeable future.
Implications for Policy, Delivery or Practice:
Since nursing education in many developing countries is publicly financed, the migration of nurses from developing countries translates into a
“massive public subsidy” to wealthier nations. As the US nursing workforce constitutes approximately
20% of the total world stock, the impact of the US labour market on global migration of nurses is very significant.
Because of the shortage of healthcare workers, many developing countries will not be able to achieve their Millennium Development Goals. In recent years, life expectancy in 17 of the 48 sub-
Saharan African countries declined as a direct result of increased incidence of communicable diseases including AIDS. A downward trend in immunization coverage in the Caribbean region from 90% to 80% in 2003 has also been attributed to nurse migration. A similar decline in immunization rates from a high of 70% in 1993 to
60% in 2003 has been noted in the Philippines.
Theme: Health Care Workforce
● Physician Autonomy vs. Self-Regulation: You
Can't Have One Without the Other
Amir Khaliq, Ph.D.; Ari Mwachofi, Ph.D.; Robert
Broyles, Ph.D.
Presented by: Amir Khaliq, Ph.D., Health
Administration & Policy, University of Oklahoma
Health Sciences Center, College of Public Health,
801 Northeast 13th Street, College of Health
Building, Room 349, Oklahoma City, OK 73104-
5072, Phone: (405) 271-2115 x 37073; Email: amir-khaliq@ouhsc.edu
Research Objective: Physician autonomy is intricately linked with the quality of care and patient protection. It gives physicians the freedom to exercise their judgment without interference on the premise that physicians will act competently and will put the wellbeing of the patient ahead of their own personal interest. The objective of the study is to examine the dichotomous nature of physician autonomy and self-regulation in the practice of medicine and explore prospects for greater transparency and public participation in investigating and disciplining physicians.
Study Design: Literature Review
Population Studied: Physicians
Principal Findings: The medical profession has done a poor job of self-regulation and has long been accused of complicity and complacency. In recent years the profession has undertaken initiatives to promote physician competence, increased transparency and greater public participation in the regulatory process.
Recertification and revalidation requirement have helped address the issue of competence but policing physician conduct remains as a source of dissatisfaction. According to one report, 60-90% of complaints against U.S. physicians filed by the general public involve quality of care issues and
2/3rds are closed during the investigative stage because of a lack of evidence. Additional 20% are dropped as a result of clarifications and agreements with the accused physician. Thus, only 1.5% of complaints get to a formal hearing. Altogether, only
5% of complaints result in some form of disciplinary action. In 2007 in the U.S., 2.92 serious actions per
1000 physicians were taken by state medical boards. Observers suggest that historically disciplinary actions taken by regulatory authorities have been mild in relations to the seriousness of offenses reported. For example, 67% of those convicted for insurance fraud and 36% of those convicted in relation to controlled substances received only non-severe penalties.
Conclusion: The study concludes that despite the medical profession’s poor record in regulating and disciplining its members, internal control of regulation and discipline is a preferred choice than governmental oversight and control. The role of the government should be confined to monitoring the performance of the profession and to seek greater transparency and public participation. In instances where government finances the delivery of healthcare, the power of the purse can be used more effectively to demand greater transparency and public participation in the investigation and disciplining of unscrupulous or incompetent physicians.
Implications for Policy, Delivery or Practice: At the heart of regulatory procedures is the need to protect the patient (“first do no harm”) and promote professional integrity. Since individual physicians cannot always be relied to be competent and scrupulous, the social contract that gives the medical profession the privilege of autonomy goes hand in hand with the responsibility for effective self- regulation. Government delegates its regulatory and policing power to the medical profession with the expectation that the profession will fulfill its obligation toward society. To protect patients and to safeguard the autonomy of physicians, it is imperative that the medical profession effectively deals with the small number of incompetent and unscrupulous physicians. If the medical profession cannot guarantee the former, it stands to lose the latter as well.
Theme: Quality and Efficiency: Policies and
Incentives
● An Interactive Model for the Short-term
Management of Cash & the Improvement of
Profitability
Amir Khaliq, Ph.D.; Robert Broyles, Ph.D.; Steven
Mattachione, J.D.
Presented by: Amir Khaliq, Ph.D., Associate
Professor, Health Administration & Policy,
University of Oklahoma Health Sciences Center,
College of Public Health, 801 Northeast 13th Street,
Room 349, College of Health Building, Oklahoma
City, OK 73104-5072, Phone: (405) 271-2115 X 5;
Email: amir-khaliq@ouhsc.edu
Research Objective: Failure to maintain an adequate cash balance or the decision to retain excessive funds in idle balances lowers the organization’s net surplus and can increase the organization’s dependence on external sources of credit. Accordingly, the first objective of the paper is to develop a model that estimates the costs of adopting various strategies concerning the amount of cash balance that the organization maintains during the planning period. The second objective is to develop an interactive spreadsheet that allows the analyst to carry out sensitivity analysis regarding the cash balance and changes in the values of the parameters of the model.
Study Design: Based on hypothetical data, the proposed model for the short-term management of cash is illustrated in an Excel spread sheet. The model is based on the proposition that the healthcare organization anticipates random and nonrandom cash flows during the planning period.
In particular, the model assumes that the timing and amount of cash disbursements to labor, vendors and creditors are known with certainty.
The spreadsheet illustration focuses on the expected total costs, given by the sum of the expected long and short costs. The short costs consist of fixed and variable components whereas long costs are represented by investment income that is foregone when surplus funds are allowed to remain idle. The expected long cost function in the illustration assumes that the opportunity rate of return is 5 percent and that the cash balance and the excess of the cash balance relative to cash disbursements may be invested for the entire planning period. The illustration also assumes that the organization expects to invest fifty percent of any net cash receipt during the planning period.
Principal Findings: As the value assigned to the cash balance increases, expected short costs decline and expected long costs grow. Accordingly, the optimal cash balance corresponds to the strategy that minimizes expected total costs. The model and sensitivity analysis indicate that an increase in the fixed cost, the interest rate, or the cash deficits that must be financed during the planning period result in higher expected short costs, an outcome that motivates the organization to maintain a larger cash balance during the planning period. The interactive spreadsheet enables management to assess the sensitivity of the cash balance decision to changes in one or more parameter. The illustration indicated that, if fixed costs increase by $100, the value of the desired cash balance doubles and increases from
$2,000 to $4,000.
Conclusion: The results of the spreadsheet illustration merely represent an approximate value of the cash balance that minimizes expected total costs and are influenced by the reliability and validity of the values assigned to each of the parameters. Since parameter estimates are projections of conditions that will prevail during the planning period, specific values must be based on recent data or the subjective judgments of management. Accordingly, the model and related outcomes should be regarded as guidelines that assist the organization in reaching decisions concerning cash holdings.
● Record Review of the Management of Patients
Admitted for Diabetic Foot Complications
Rukhsana Khan, M.B.B.S., M.P.H., M.S.
Presented by: Rukhsana Khan, M.B.B.S., M.P.H.,
M.S., Consultant, Ministry of Health, Health
Services Academy, Health Services Academy,
Chak Shehzad, Islamabad, PK, Phone: 0092 51
9255590 ext 126; Email: rukhsana_khan@embanet.com
Research Objective: To determine clinical features, foot-related and cardiovascular risk factors, management, and clinical outcomes of patients admitted with diabetic foot in general surgery wards of the Pretoria Academic Hospital,
Pretoria -South Africa
Study Design: Cross Sectional, descriptive
Population Studied: A retrospective audit of the surgical ward register and patient files was conducted for a period of two years – 2005 to 2006.
Patients admitted for the first time with diabetic foot complication were included in the analysis.
Principal Findings: A total of 81 records were identified, reviewed and analysed. The mean duration of diabetes was 12 years. Glucose on admission was measured in 95 % , it was
12.6mmol/l . HbA1c was measured in 31% cases with mean of 10%.Peripheral pulses were examined in 85% and in 18% of cases; both pulses were absent in the affected foot. Semmes Weinstein 5.07 monofilament was not used. Ulcers were assessed in 44% of cases . Sixty seven % had HT and were taking treatment. Lipid values measured in less than 10% of subjects. Renal function was measured in 94%. Outcome was 48% below knee and 18.5 % above knee amputations. Readmissions were done in33% of patients of these 23% for problems in the same foot. On discharge 43% were prescribed ACE inhibitors, 33% aspirin and only 12% statins.
Conclusion: There was no standardised protocol for ulcer grading, detection of neuropathy or evaluation of cardiovascular risk.
Implications for Policy, Delivery or Practice:
Patient should be treated as a whole ,the current complication should not only be focussed. Longterm care of these patients could possibly be improved by involving other disciplines during the admission of patients in the surgical ward with
“diabetic foot”. A guideline is required to assist health care providers in surgical wards. Ideally, these guidelines should be used in conjunction with a multidisciplinary diabetic foot team.
● Impact of Prescription Drug Coverage on Use of Prescription Drugs among Elderly Patients with Arthritis
Nasreen Khan, Ph.D.
Presented by: Nasreen Khan, Ph.D., Assistant
Professor, Pharmacy Practice, College of
Pharmacy, University of New Mexico, 1 University of New Mexico, Albuquerque, NM 87110, Phone:
(505) 272-5294; Email: nkhan@salud.unm.edu
Research Objective: Arthritis is one of the most common chronic diseases among elderly.
Approximately, 50% of all elderly in the US suffer from the disease. Individuals suffering from the disease report functional impairment, depression, and lower quality of life. Drug therapy plays an important role in decreasing morbidity and mortality associated with the disease. The objectives of this study are to evaluate the effect of prescription drug insurance coverage on use of prescription drugs among elderly patients with arthritis. Specifically, I analyze the effect of prescription drug coverage on use of any prescription drugs, and use of drugs for specific co-morbid condition (cardiovascular, diabetes, and cough and cold) besides arthritis.
Study Design: A fundamental problem in estimating the impact of drug coverage on utilization is that the same person is never simultaneously observed with and without prescription drug coverage. Further, it is unlikely that self-selection into coverage is random and thus it is not possible to use those without coverage as controls. To address this issue, I use longitudinal data and a multivariate regression model with controls for person-specific fixed effects. Person fixed effect accounts for unmeasured time invariant individual characteristics. Analyses were based on the Medicare Current Beneficiary Survey (MCBS) for years 1992 - 2004. MCBS collects self-reported information on drug coverage and prescription drug use over a 3-4 year period among Medicare population.
Population Studied: Sample was limited to elderly patients with arthritis.
Principal Findings: Estimates show that prescription drug coverage increased use of any prescription drugs by 2% - 15%, among patient with arthritis, depending on the generosity of the coverage. Gaining public coverage increased use of arthritis drugs by 8%, cardiovascular drugs by 9%, diabetic drugs by 27%, and cold medications by
3%. Gaining managed care coverage had relatively low impact on use of arthritis and cardiovascular drugs but increased use of diabetic drugs by 23% and decreased use of cold and cough medication by 12%.
Conclusion: Drug coverage significantly increased use of medications among elderly patients with arthritis. The results of the analysis also suggest the importance of controlling for selection bias.
Estimates on drug coverage were reduced markedly using a within person design.
Implications for Policy, Delivery or Practice:
Given the choice and voluntary nature of Medicare part D plans. Any, future analysis of part D should account for selection bias.
Funding Source(s): RWJ Health Policy Center at
UNM
● Breast Cancer Screening Practices of Primary
Care Providers
Amal Khoury, Ph.D., M.P.H.; Nedra Lisovicz, Ph.D.;
Kiran Yadav, M.D., M.P.H.
Presented by: Amal Khoury, Ph.D., M.P.H.,
Associate Professor, Health Services
Administration, East Tennessee State University,
42 Lamb Hall, Johnson City, TN 37660, Email: khoury@etsu.edu
Research Objective: Breast cancer is the most frequently diagnosed cancer and the second leading cause of cancer-death for women. Breast cancer screening rates have steadily increased during the last decade. However, recent reports revealed significant declines in mammography rates for women overall and among specific groups that have prompted public health concern. Studies have demonstrated that a provider’s recommendation is a key driver of screening rates. The purpose of this study was to examine: 1) patterns in the provision of breast cancer screening services among primary care providers; 2) facilitators and barriers to screening; and 3) impact of physician and practice characteristics on the provision of screening services.
Study Design: Mail survey of physicians, with follow-up using computer-assisted telephone interviewing. The sampling frame was constructed using the database of the American Academy of
Family Physicians. The survey questionnaire was informed by phase I of the project -- a qualitative study involving in-depth interviews with 34 primary care providers, as well as by a systematic review of the literature. The questionnaire collected data about screening practices, follow-up, facilitators and barriers to screening, and physician and practice characteristics. Data were entered and analyzed using SPSS v. 15.0.
Population Studied: Family physicians practicing in four southeastern states in 2007.
Principal Findings: A total of 489 physicians responded to the survey -- 65% males and 10% members of minority groups. The majority of physicians were in group practices. Practices were located in urban/suburban (56%) and rural (43%) areas. Although the vast majority of physicians believed in the importance of regular mammograms, only 2 out of 3 physicians regularly referred 80 – 100% of their eligible female patients for a screening mammogram. Almost two-thirds of respondents indicated that their clinics did not have a patient reminder system for scheduling mammograms. In addition, one-third of clinics did not follow-up with patients who missed their scheduled mammograms. With regard to the clinical breast exam, and although the majority of physicians believed in its importance, only 35% regularly provided the exam to 80 - 100% of their eligible patients. Major barriers to both mammography referral and clinical breast examinations were lack of annual physical exams, patient refusal/preferences, and patient having another provider. An additional barrier to mammography referral was patient limited resources. An additional barrier to clinical breast examination was provider time constraints. The majority of physicians believed in the importance of the breast self exam (BSE), but only 42% educated the majority of patients (80 – 100%) about it. Most common educational methods were explaining the procedure to the patient and distributing educational materials. Only 15% of providers asked patients to demonstrate the procedure to them. Lack of annual physical exams and provider time constraints and competing demands emerged as barriers to BSE education. Analyses are underway to examine the associations between physician characteristics, practice features, provision of screening services, and follow-up.
Conclusion: The findings identify major gaps in providing mammography referrals and clinical breast exams, as well as in education about the breast self exam. Significant proportions of practices lacked mammography reminder and follow-up systems. Barriers to the provision of services included variables at the woman-level (e.g. refusal), provider-level (e.g. time constraints, lack of reminder and follow-up systems), and health care system-level (e.g. lack of health insurance, lack of coordination among providers).
Implications for Policy, Delivery or Practice:
Reversing the recent decline in screening rates may require multiple interventions to 1) increase community awareness of the value of early detection; 2) institutionalize clinic systems that facilitate screening; and 3) expand health insurance and service coordination.
Funding Source(s): The Susan G. Komen Breast
Cancer Foundation
Theme: Gender and Health
● Cost Differential Between Michigan Medicaid
& the U.S. Following Michigan's Multiple Cost
Containment Policies
Jennifer Kibicho
Presented by: Jennifer Kibicho, 2071 South
Summit Avenue, Milwaukee, WI 53202, Phone:
(414) 955-7754; Email: jkibicho@mcw.edu
Research Objective: Confronted with a 100 percent increase in Medicaid prescription drug expenditures between 2000 and 2002, Michigan implemented the following 4 cost containment policies: (1) a preferred drug list with supplemental manufacturer rebates in February 2002; (2) a joint purchasing arrangement with Vermont in February
2003; (3) a maximum allowable cost pricing system for pharmacy reimbursement in November 2003; and (4) a multi-state pooling supplemental rebate arrangement in May 2004. The need to reduce cost without negatively impacting access to care motivated Michigan’s cost containment efforts.
These policy efforts were aimed at encouraging physicians to prescribe products that were costeffective for taxpayers and safe and clinicallyeffective for beneficiaries. This study compared the daily cost of therapy (expenditure per day of therapy) for select cardiovascular prescription drugs in Michigan and the United States between 2000 and 2004. Specifically, Michigan’s and the United
State’s daily costs of therapy were compared in the pre-policy (2000-2001) and post-policy (2002-2004) periods. The comparative study was conducted to evaluate the impact of Michigan’s policies on prescription drug costs.
Study Design: The study employed a pre-post research design. Data for Michigan were obtained from Michigan’s Medicaid outpatient prescription drug program. Data for the U.S. were obtained from the Medical Expenditure Panel Survey (MEPS) database. The Michigan Medicaid database is an administrative database that contains prescription drug utilization and expenditure data for Medicaid beneficiaries. The MEPS data is gathered through nationally-representative population surveys, and is used to estimate prescription drug utilization and expenditures for the non-institutionalized U.S. civilian population. Seventeen of the most frequently prescribed cardiovascular drugs were included in the study. In 2000, these drugs accounted for 72% of Michigan Medicaid expenditures on cardiovascular prescription drugs.
These drugs can be classified under the following therapeutic sub-classes: angiotensin converting enzyme inhibitors, angiotensin receptor antagonists, beta blockers, calcium channel blockers and lipotropics. The analyses utilized
31,175 Michigan Medicaid and 186,567 MEPS data claims. The main outcome of interest was the daily cost of therapy, which is the ratio of prescription drug expenditures to days supply of prescription drugs aggregated on an annual basis. Michigan
prescription drug expenditures included reimbursements to pharmacists by Medicaid and private insurance, and out-of-pocket spending by beneficiaries, but did not include dispensing fees.
U.S. prescription drug expenditures included amounts paid by federal, state, local and private insurance. All expenditures were inflated to 2004 constant dollars using the Consumer Price Index for all urban consumers (CPI-U). The CPI-U is the index used by Medicaid to analyze manufacturer price increments. The average price and market share by product type (generic versus brand) as well as the daily cost of therapy was calculated for each calendar year for both Michigan and the U.S.
Trends in the average price by product type depicted the rate of inflation/deflation by product type. Changes in generic market share reflected shifts in prescription drug utilization patterns from brand to generic products. Analyses were conducted using Statistical Package for Social
Sciences (SPSS) software.
Population Studied: Michigan’s Medicaid data were restricted to dual-eligible beneficiaries
(individuals who qualify for both Medicaid and
Medicare). Dual-eligibles have a greater demand for prescription drugs when compared to the general Medicare population as a result of poorer health status and a higher prevalence of chronic conditions. MEPS data were derived from a sample of the nationally-representative survey of the U.S. non-institutionalized civilian population.
Principal Findings: Several trends in daily cost for
Michigan and the U.S. between 2000 and 2004 were found. First, between 2000 and 2004, the daily cost of therapy fell by 20% in Michigan while it increased by 23% in the U.S. Specifically, in
Michigan the daily cost increased during the prepolicy period (2000 and 2001), and then fell in subsequent years. In contrast, the U.S. daily cost increased steadily from 2000 to 2004. Second, even though brand prices increased in both
Michigan and the U.S., the rate of increase was nearly twice as large in the U.S. when compared to
Michigan (41% in the U.S. versus 23% in Michigan).
Third, generic prices decreased by 45% in Michigan but increased by 14% in the U.S. Fourth, even though a shift away from brand products was observed in both Michigan and the U.S., generic market share went up more than twice as much in
Michigan as the U.S. (75% in Michigan versus 35% in the U.S.). Fifth, the daily cost differential between the U.S. and Michigan increased substantially over the study period. The daily cost of therapy for the
U.S. was 9% higher in 2000 and 66% higher in
2004, compared to Michigan.
Limitations of the study: First, expenditures were not adjusted for the supplemental manufacturer rebates received by Michigan. Manufacturer rebate information for specific drugs is proprietary information and therefore detailed individual drug rebate information could not be obtained. Thus the reported amount of cost savings generated by
Michigan’s cost containment policies may be understated. Second, whereas Michigan data was restricted to individuals 65 years and older, the
MEPS data contained information on individuals 18 years and older. This is not considered a major concern for the following two reasons: First, the main unit of analysis was expenditure and not utilization. Second, cardiovascular disease is highly correlated with age.
Conclusion: Michigan experienced price deflation starting in 2002 when the state implemented the first of 4 cost containment policies, whereas the
U.S. experienced price inflation from 2000 to 2004.
This finding is consistent with the 2004 Office of the
Inspector General report, which noted that Michigan paid consistently lower prices for most drugs compared to other states. The daily cost differential between the U.S. and Michigan has the following financial implications: Michigan saved $0.29 per day of therapy between 2000 and 2004. If the U.S. had paid the same prices as Michigan, it could have saved more than $3 million over this time period. In summary, it appears that Michigan’s cost containment efforts had a significant effect on both utilization patterns and the daily cost of therapy.
Michigan saved health care dollars by implementing a combination of strategies that shifted utilization patterns towards generic products, reduced the amounts paid for generic drugs, and moderated the rate of brand product inflation.
Implications for Policy, Delivery or Practice:
Two policy implications can be derived from the study. First, although Michigan's generic market share increased substantially and generic prices fell significantly, the daily cost of therapy did not decrease proportionately. This suggests that cost containment efforts that concentrate on generic products have their limitations. Study findings suggest that in order to significantly curb escalating costs, policies must target brand products, which constitute a significant share of the total prescription drug expenditure. Second, study findings suggest that well-coordinated multiple cost containment policies can have a positive and significant impact on the daily cost of therapy. As noted above,
Michigan’s implementation of various policies increased the daily cost differential between the
U.S. and Michigan. If the U.S. had paid for the 17 cardiovascular prescription drugs at similar prices to
Michigan, the nation would have generated huge cost savings. If these findings are extrapolated to all prescription drugs, it would suggest that multiple cost containment policies have the potential to substantially curb escalating prescription drug costs, and save scarce health care dollars for both the private and public sectors.
Funding Source(s): National Institute of Mental
Health (NIMH) grant T32-MH19985, Wayne State
University Graduate School Dissertation Award,
Blue Cross Blue Shield of Michigan Dissertation
Award
Theme: Medicaid, SCHIP and State Health Reform
● Is Private Long-Term Care Insurance
Affordable for Older Adults?
Nayoung Kim, M.A.; Gail Jensen, Ph.D.
Presented by: Nayoung Kim, M.A., Economics,
Wayne State University, 656 West Kirby, 2074 FAB,
Detroit, MI 48202, Phone: (313) 577-3345; Email: nayoung@wayne.edu
Research Objective: Why do so few Americans buy private long-term care (LTC) insurance?
Several theories have been offered as possible explanations, including the availability of Medicaid, misperceptions that Medicare or other policies cover LTC, beliefs that one’s own risk of needing
LTC services is small, or desires to simply rely on children and spouses for LTC. This paper examines another possible explanation – that private LTC insurance is simply “unaffordable” for most older Americans, which may be why they don’t buy it.
Study Design: We begin by investigating the meaning of affordability in the context of private
LTC insurance. We propose several definitions for affordability, drawing on concepts recently developed to gauge the affordability of acute-care health insurance and housing. We consider definitions for affordability based on simple normative standards, such as whether remaining household income after paying for LTC insurance is above some (arbitrarily-set) threshold, as well as more behavioral definitions of affordability, such as whether other adults with similar economic, demographic, and family circumstances are seen to purchase LTC insurance.
Population Studied: Then using nationally representative data from the ongoing Health and
Retirement Study (HRS) and the Rand HRS data we examine the incidence of “unaffordability” of
LTC insurance premiums among Americans over age 50, given each of our alternative definitions for it.
Principal Findings: Using 2 and 3 times the poverty level as a gauge for “other spending” after buying LTC insurance, we find even the leastgenerous policy is unaffordable for 25-40% of adults in their 50’s, and this same policy is unaffordable for 35-53% of adults in their 60’s. In addition, affordability is highly sensitive to the gauge used for reasonable other spending. Using 3 times the poverty level, the affordability of all policies dramatically decreases.
Conclusion: Lack of affordability can explain why many older adults forgo private LTC insurance. Age at first purchase has a large effect on affordability.
The outcomes from different definitions can be compared and contrasted, so that more general conclusions can be reached regarding the role of affordability in explaining non-purchase.
Implications for Policy, Delivery or Practice:
This analysis offers researchers and policymakers an operational framework for defining affordability, and for evaluating its relative importance as an explanation for non-purchase.
● Content Analysis of On-line Site for
Complementary & Alternative (Medical) Therapy on the Domestic Internet in Korea
Chun-Bae Kim, M.D., Ph.D.; Ki Bong Um, M.P.I.; Ji
Eun Lee, M.P.H.; Ki-Kyong Kim, Ph.D.; Jun-Sang
Yoo, O.M.D., Ph.D.; In-Sik Kong, M.D.
Presented by: Chun-Bae Kim, M.D., Ph.D.,
Professor, Department of Preventive Medicine,
Yonsei University Wonju College of Medicine, 162
Ilsan-dong, Wonju City Kangwon-do 220-701, KR,
Phone: +82-33-741-0344; Email: kimcb@yonsei.ac.kr
Research Objective: Recently the cancer patients are utilizing more the complementary and alternative medical (CAM) therapies in the world because of the advancement of access of information in the internet on-line sites. However, the quality of information is not verified. The purpose of this study is verifying the current status of internet information for CAM therapy in Korean on-line sites.
Study Design: This study design was crosssectional study.
Population Studied: The study sites were the domestic famous search sites - Daum
(http://www.daum.net/), Yahoo
(http://kr.yahoo.com/), and Naver
(http://www.naver.com/) in Korea. We searched the on-line sites using key words including
`complementary and alternative medical (CAM) therapies', et al from January to December, 2007.
After all, we finally selected 823 on-line sites. 2 analytical teams evaluated independently the sites according to the evaluation protocol. If different results may be gotten, we decided through consensus process.
Principal Findings: The average integrated score on the availability of access was high in the manual
/physical acting services and group/organization type among CAM therapies. E-mail (63.8%) and opinion collecting method (75.3%) were high on the mutual interaction parts.
The score of the private information policy was higher in the (CAM) goods sale site than other site on the privacy of correspondence. Also, 475 on-line sites (57.7%) reveal their aims on the object parts.
Conclusion: The evaluation score of many CAM on-line sites in Korea was inferior on all domain compared general health related on-line sites.
Implications for Policy, Delivery or Practice:
Therefore, it is necessary that continuous management and evaluation about on-line site of
CAM therapy for consumer' safety.
Funding Source(s): Korean Cancer Institute
Theme: Health Information Technology
● The Relationship of Financial & Mission
Factors to the Level of Uncompensated Care
Provided in California Hospitals
Tae Hyun Kim, Ph.D.; Michael McCue, D.B.A.; Jon
Thompson, Ph.D.
Presented by: Tae Hyun Kim, Ph.D., Assistant
Professor, Department of Health Administration,
Governors State University, 1 University Parkway,
University Park, IL 60466, Phone: (708) 534-4033;
Email: t-kim@govst.edu
Research Objective: Community hospitals in the
US have experienced a substantial rise in the burden of uncompensated care over the past few years. However, there is an ongoing debate about whether hospitals, especially private not-for-profits, are providing sufficient levels of uncompensated care. This study assesses how hospital financial performance, such as free cash flow, liquidity, and debt level, influences the provision of uncompensated care, whether different mission characteristics reflected in hospital ownership as well as level of community benefit are related to the level of uncompensated care, and if competitive pressures are associated with the provision of uncompensated care.
Study Design: With secondary datasets from the
California Office of Statewide Health Planning and
Development, a pooled cross-sectional time-series design was used to investigate the study questions retrospectively. we used three different measures of uncompensated care (i.e., charity care, charity care plus bad debt, and combined bad debt, charity care and county indigent program) available in the data that account for differences in hospital reporting and which are more sensitive to the broader concept of uncompensated care provided by hospitals.
Population Studied: We studied private short-term acute-care hospitals in California during the fiscal year 2005 to 2007. To control for any accounting variations within a given year and to reflect consistency of cash flow performance, we selected hospitals that reported three consecutive fiscal years of operating, financial data and accounting periods of at least 365 days. The final sample included 193 private acute-care hospitals, with 80 percent of the sample facilities being not-for-profit and 20 percent being for-profit.
Principal Findings: Among the financial performance measures, free cash flow was positively associated with the provision of uncompensated care in not-for-profit hospitals, while debt level was negatively related to uncompensated care. Neither ownership type (forprofit vs. private not-for-profit) nor the number of community benefit services was statistically significant. Overall levels of uncompensated care in the local market were positively associated with a hospital’s level of uncompensated care.
Conclusion: At least in not-for-profit hospitals, free cash flow appears to be used for the community-atlarge, the implied stockholders/owners, through funding uncompensated care. Obligations of debt may force a hospital to reduce the distribution of its community dividend in the form of the provision of uncompensated care. Hospital ownership may not make a significant difference in the level of uncompensated care. The level of a hospital’s uncompensated care may be related to competing hospitals’ provision of it.
Implications for Policy, Delivery or Practice: In the future, non-profit hospitals with high free cash flow in California may face an array of obstacles in serving their community needs in the form of uncompensated care. A host of factors—including a rising uninsured population, slowing growth in
Medicare and Medicaid reimbursement due to a declining economy, and increasing healthcare supply and capital costs—may erode away any excess cash and reduce not-for-profit hospitals’ ability to maintain current levels of uncompensated care. Additional research should examine changes in uncompensated care levels over time as
California hospitals address these increasing and significant influences.
Theme: Health Care Markets and Competition
● Medicare Expenditure among Racial/Ethnic
Groups for Elderly Medicare Beneficiaries with
Diabetes
Jee-Ae Kim, M.P.P.; Xinhua Yu, Ph.D.; Marshall
McBean, M.D., M.Sc.
Presented by: Jee-Ae Kim, M.P.P., Ph.D. Student,
Health Policy & Management, University of
Minnesota, 420 Delaware Street, Southeast,
Minneapolis, MN 55108, Phone: (612) 986-4925;
Email: kimxx607@umn.edu
Research Objective: The growing financial burden of Medicare has been an ongoing concern.
Understanding Medicare expenditures and their determinants are an integral part for the optimal allocation of Medicare resources. The study objectives are 1) to describe Medicare expenditures and examine the racial/ethnic differences in projected expenditures among elderly beneficiaries with diabetes; 2) to examine the association of a pattern of healthcare service with Medicare expenditures among racial/ethnic groups.
Study Design: Claims for a 5% random sample of
Medicare beneficiaries for the year 2003 was used.
The racial/ethnic groups of main interest were whites, blacks, and Hispanics. Information from the
Medicare Inpatient, Outpatient, Carrier, DME and hospice files was used, and predicted mean costs were estimated using a two part model. To examine the pattern of health service use and costs, inpatient costs and outpatient costs were separately examined, in addition to total cost. The model predicted costs were obtained from generalized linear models adjusted for covariates such as sociodemographic characteristics, comorbidity and
healthcare services use. Socio-demographic characteristics: age, sex, rural residence, zip code for median income, and state’s subsidy program status were obtained from the Beneficiary Eligibility
Summary file. Physician office visits and emergency visits were identified from specialty codes and
BETOS codes in the Medicare Carrier claims. A
Charlson score for comorbidity was calculated from
ICD-9 diagnosis codes in the 2003 Medicare
Carrier, Inpatient and Outpatient claims.
Population Studied: Elderly Medicare beneficiaries with diabetes who were older than 66 years old were studied. Application of exclusion criteria resulted in 230,764 persons: whites
(83.8%), blacks (11.0%), and Hispanics (2.3%)
Principal Findings: In 2003, predicted mean outpatient, inpatient and total costs for blacks and
Hispanics were greater than for whites. Outpatient costs were 7 percent and 16 percent higher, for blacks ($9,323) and Hispanics ($10,149), respectively, than for white ($8,722). The difference was greater in inpatient costs. Blacks ($16,526) had
16 percent, and Hispanics ($17, 280) had 21 percent higher cost than whites ($14,271). After adjusting for covariates, cost differences among racial/ethnic groups decreased, however, they were all statistically significant (p<0.0001). Increasing comorbidity scores and emergency department visits were significantly related to an increase in costs (p <0.0001). For physician office visits, predicted costs among those with 5-10 and 10-15 visits were lower than among those with less than 5 visits and those with more than 15 visits (p
<0.0001).
Conclusion: Elderly blacks and Hispanics with diabetes consistently showed higher predicted costs than whites. The predicted costs are strongly related to comorbidity scores which were higher in
Blacks and Hispanics. Higher comorbidity for black and Hispanics may be associated with a different healthcare service seeking behavior; for example, fewer regular physician office visits and greater use of emergency department.
Implications for Policy, Delivery or Practice: The higher Medicare expenditures for black and
Hispanic beneficiaries with diabetes may indicate that Medicare dollars can be spent on these minority populations in a more cost effective way.
Encouraging minority groups to use the most costeffective healthcare services such as more regular physician office visits for the prevention and early detection of disease progression can be one strategy.
Funding Source(s): CMS
Theme: Medicare
● The Comparative Evaluation of Expanded
National Immunization Policies in Korea Using an Analytic Hierarchy Process
Chun-Bae Kim, M.D., Ph.D.; Taeksoo, Shin, Ph.D.;
Yang-Heui Ahn, Ph.D.; Hyo-Youl Kim, M.D., Ph.D.;
Dong-Han Lee, M.D.; Un-Yeong Go, M.D., Ph.D.
Presented by: Chun-Bae Kim, M.D., Ph.D.,
Professor, Department of Preventive Medicine,
Yonsei University Wonju College of Medicine, 162
Ilsan-dong, Wonju City Kangwon-do 220-701, KR,
Phone: +82-33-741-0344; Email: kimcb@yonsei.ac.kr
Research Objective: The purpose of this paper is to propose new evaluation criteria and an analytic hierarchy process (AHP) model to assess the expanded national immunization programs (ENIPs) and to evaluate two alternative health care policies.
One of the alternative policies is that private clinics and hospitals would offer free vaccination services to children and the other of them is that public health centers would offer these free vaccination services.
Study Design: Our model to evaluate the ENIPs was developed using brainstorming, Delphi techniques, and the AHP model.
Population Studied: We first used the brainstorming and Delphi techniques, as well as literature reviews, to determine 25 criteria with which to evaluate the national immunization policy; we then proposed a hierarchical structure of the
AHP model to assess ENIPs.
Principal Findings: By applying the proposed AHP model to the assessment of ENIPs for Korean immunization policies, we show that free vaccination services should be provided by private clinics and hospitals rather than public health centers.
Conclusion: Application of the proposed AHP model to Korean immunization policies showed that private clinics and hospitals are a better alternative to public health centers in terms of providing free vaccination services.
Implications for Policy, Delivery or Practice: We hope that other countries would be implemented this AHP model for evaluating their health policy.
Funding Source(s): Korea Centers for Disease
Control and Prevention, Republic of Korea
Theme: Quality and Efficiency: Policies and
Incentives
● Inter-Hospital Collaboration & the
Performance of Hospital Network in Korea
Tae Hyun Kim, Ph.D.; Jae-San, Park, M.P.H.
Presented by: Tae Hyun Kim, Ph.D., Assistant
Professor, Department of Health Administration,
Governors State University, 1 University Parkway,
University Park, IL 60466, Phone: (708) 534-4033;
Email: t-kim@govst.edu
Research Objective: Facing recent health care environmental pressures, such as change in hospital payment systems, introduction of a hospital evaluation system, and increased market competition, Korean hospitals have formed alliances and networks. Particularly, smaller
regional community hospitals have been striving to build collaborative networks with large tertiary hospitals to improve their operations by boosting patient referral base. However, there are relatively few studies assessing the performance of hospital networks in Korea. The objective of this study was to examine the relationship of inter-hospital collaboration to the performance of hospital networks.
Study Design: We focused on dyadic relationships between tertiary hospitals and community hospitals.
Each tertiary-community hospital collaboration is independent of the others; there are no relationships among the community hospitals. Data were collected via a cross-sectional survey using self-administered questionnaires. The key persons in the community hospitals responsible for the collaboration were asked to respond to the 5-point
Likert scale questions regarding the collaboration activities and hospital network performance. The dependent variables include a quantitative measure
(i.e., the number of inpatients and outpatients referred between collaborating hospitals) as well as a subjective measure of network performance, such as perceived organizational performance and management efficiency. Independent variables were various inter-hospital collaboration activities such as information sharing, inter-hospital communication, and quality of relationships, such as flexibility of relationship and expectations of relationship continuity.
Population Studied: A sample of 175 community hospitals that participated in the collaborative network with tertiary hospitals in 2005. Most of the facilities are mid- or small-size (400 beds or smaller), private not-for-profits. There were no statistical difference in terms of rural vs. urban location and teaching status.
Principal Findings: The results of multivariate regression analysis on the hospital network performance revealed that three independent variables, information sharing (beta=0.214), flexibility of relationship (beta=0.248), expectation of relationship continuity (beta=0.234), and teaching status (beta=0.116), were statistically significant and positively associated with the dependent variables, the subjective measures of network performance.
Conclusion: From the perspective of the community hospitals participating the networks, information sharing, flexibility of relationship, and expectation of relationship continuity with the tertiary hospitals are positively associated with perceived organizational performance and management efficiency.
Implications for Policy, Delivery or Practice: The findings of this study provide implications for hospital management. Both the community hospitals and the partner hospitals should effectively communicate about the patient’s medical information and bed utilization. The dyadic relationship between participating institutions needs to be flexible so that the mutual relationship and the process of patient referrals can be adjusted as the situation changes. It is also important to create an atmosphere that the collaboration will continue in the future.
● Heterogeneity in Behavior Problem
Trajectories among Maltreated Children &
Related Ecological Factors
Jiyoung Kim, Ph.D.; Richard Thompson, Ph.D.;
Alan Litrownik, Ph.D.
Presented by: Jiyoung Kim, Ph.D., Postdoctoral
Fellow, Institute for Healthcare Studies,
Northwestern University, 750 North Lake Shore
Drive, 10th Floor, Chicago, IL 60611, Phone: (312)
503-5552; Email: jiyoung-kim@northwestern.edu
Research Objective: Decades of research have consistently demonstrated a strong association between the experience of abuse or neglect and the later development of behavior problems. However, the extent to which maltreated children develop behavior problems may vary widely according to their ability to manage their experiences of maltreatment, and the extent to which support is available through family, community, and health and social services. This study is a longitudinal investigation of unobserved heterogeneity in the developmental trajectories of maltreated children.
The goal of the study is to inform effective treatment and prevention plans with respect to behavior problems of maltreated children by examining the different trajectories of behavior problems (Aim1) and by assessing ecological risk factors related to each trajectory (Aim2).
Study Design: This study utilized data from
LONGSCAN (Longitudinal Studies of Child Abuse and Neglect). LONGSCAN is designed to measure the long-term effects of maltreatment on the development of children at five sites - East,
Midwest, Northwest, Southern, and Southwest. All five studies share measures, definitions, interviewer training, data collection strategies, data entry, and data management, and the study sites are linked through a governance agreement and a coordinating center at the University of North
Carolina at Chapel Hill. Behavior problems were repeatedly measured using Achenbach’s Child
Behavior Checklist from age 4 to age 10, and ecological factors predicting each trajectory were measured at the baseline. Missing data were replaced using multiple imputation. A group-based trajectory model was used to investigate distinctive sub-group trajectories (Ami1), and multinomial logistic regression was used to examine the ecological factors related to each trajectory (Aim2).
Population Studied: The population studied includes 827 maltreated children who have been followed from age 4 to age 10.
Principal Findings: This study identified five distinctive developmental trajectories in maltreated
children: the lowest, low-medium, decreasing, increasing, and high-chronic trajectory groups.
Child developmental disability was significantly related to membership in the increasing and highchronic groups, and caregiver alcohol problems and low community quality were significantly related to membership in the increasing group compared with the lowest group. Child mental health services were more likely to be utilized by the low-medium, decreasing, increasing, and high-chronic groups than by the lowest group. Finally, caregiver depression was found to significantly predict the divergence of membership between the decreasing and increasing groups.
Conclusion: This study found distinctive sub-group trajectories and different ecological factors predicting each trajectory. The study provides significant empirical evidence for heterogeneity among maltreated children and risk factors related to diverging behavior problem trajectories in maltreated children’s ecology.
Implications for Policy, Delivery or Practice:
Many studies have demonstrated that early behavior problems are highly predictive of later juvenile and adulthood behavior and psychological problems. Examining developmental trajectories in early childhood, before they are entrenched in adolescence and adulthood, will be crucial for the formulation of early treatment plans, and will help us target interventions toward those most likely to benefit.
Funding Source(s): Fahs-Beck Dissertation Grant
Award, NIDRR, DHHS
● A Randomized Control Trial of Telephone
Care Management for Medicaid Recipients with
Depression: 18-Months Follow-Ip
Sue Kim, Ph.D., M.P.H.; Allen LeBlanc, Ph.D.;
Charles Michalopoulos, Ph.D.; Francisca Azocar,
Ph.D.; Evette Ludman, Ph.D.; Greg Simon, M.D.
Presented by: Sue Kim, Ph.D., M.P.H., Research
Scientist, MDRC, 475 14th Street, Suite 750,
Oakland, CA 94612-1900, Phone: (510) 844-2243;
Email: sue.kim@mdrc.org
Research Objective: Telephone care management has been shown to improve outcomes for depressed patients. However, there is a lack of studies on its effectiveness for individuals receiving
Medicaid. This study examines a one-year telephone care management intervention designed to help low-income individuals who are experiencing major depression to enter and remain in in-person treatment.
Study Design: This study used random assignment. The intervention group received intensive monitoring from care managers to facilitate and support clinical treatment. The control group received usual care, which included referrals to mental health treatment providers without followup. The Quick Inventory of Depressive
Symptomalogy (QIDS) was used to measure depression severity at baseline, 6 months, and 18 months. Administrative and survey data collected information on utilization of mental health services.
Population Studied: This study included 499
Medicaid participants who screened positive for depression (QIDS>5) and were eligible to receive care from a managed behavioral health plan in
Rhode Island at baseline. 245 were assigned to the intervention group and 254 were assigned to the control group. Age ranged from 18 to 64 years
(average age of 35), 90% were women, 98% had children, 45% non-Hispanic White, 33% Hispanic,
13% African American, and 6% other (Asian, Pacific
Islander, American Indian, Alaskan Native and other).
Principal Findings: At 18 months follow-up, a smaller percentage of the participants in the intervention group (4%) had “very severe” depressive symptoms (QIDS >21) than the control group (10%) (p<.05). Overall, the mean depression scores were lower than at baseline for both groups
(15.2 to 11.5 for the intervention group; 15.6 to 12.1 for control group), but there was no significant difference in mean depression at 18 months between the two groups. Significantly more participants in the intervention group received inperson professional treatment specifically for depression in the first 6 months than the control group (32% vs. 22%; p<.01). Although a higher percentage of the participants in the intervention group continued in-person treatment during the last
6 months (months 7-12) of the intervention, this trend did not continue beyond the intervention period. There was no difference in the use of inperson treatment between the intervention and control group at 6 months following the intervention
(months 13-18). Similarly, compared to the control group, a higher percentage of the intervention group filled prescription medications at an adequate therapeutic dosage in months 7-12 (27% vs. 18%; p<.05), but there was no difference between the groups for months 13-18. Finally, although there appeared to be an improvement in depression among the Hispanic subgroup midway through the intervention, there was no difference in depression at 18 months follow-up between the intervention and control group for this subgroup.
Conclusion: Telephone care management initially had positive effects on increasing utilization of inperson treatment, but the impacts diminished with time –particularly once the one-year intervention ceased operation. The intervention had minimal effects on the overall depression for this Medicaid population, but may have helped people from becoming very severely depressed.
Implications for Policy, Delivery or Practice:
Care management with an active telephone outreach program may be an effective strategy in helping individuals receiving Medicaid enter and remain in in-person treatment.
Theme: Behavioral Health
● Health-Related Unemployment among
Survivors & Siblings in the Childhood Cancer
Survivor Study (CCSS)
Anne Kirchhoff, M.P.H.; Wendy Leisenring, S.D.;
Kirsten Ness, P.T., Ph.D.; Elyse Park, Ph.D.; Debra
Friedman, M.D.; Kevin Oeffinger, M.D.
Presented by: Anne Kirchhoff, M.P.H., Graduate
Student, Health Services, University of Washington,
Box 357230, Seattle, WA 98195-7230, Phone:
(314) 477-4042; Email: akirchh@u.washington.edu
Research Objective: Mortality for childhood cancer has decreased substantially due to improvements in therapy and supportive care; however, long-term effects from surgery, chemotherapy and radiation may continue throughout adulthood. Although adult survivors generally report lower levels of employment in contrast to healthy comparison populations, it is unknown whether unemployment is due to ongoing illness or disability. Female survivors, who often have poorer health-related outcomes, may be at higher risk. Therefore, we examined current employment status in a cohort of adult childhood cancer survivors compared to their age and gender matched sibling controls who were participating in the Childhood Cancer Survivor
Study (CCSS).
Study Design: The CCSS represents 14,370 subjects diagnosed under the age of 21 years with cancer between 1970-1986 and randomly selected nearest age siblings (N=3,418) as comparison.
Employment status was obtained via questionnaire; categories included currently employed (full-time
=30 hours/week; part-time <30 hours/week), unable to work due to illness/disability, unemployed by choice (not seeking paid work, student, retired), and unemployed but seeking work. Multivariate logistic regression calculated the odds ratio (OR) and 95% confidence interval (95% CI) of being unable to work due to illness/disability for survivors compared to siblings using generalized estimating equations to account for the case-sibling pairs. Within survivors, multivariate logistic regression estimated the risk of being unemployed due to illness/disability by sex. Models were adjusted for current age, race, years of education, and marital status, and the survivor-only model also included cancer type at diagnosis.
Population Studied: To assess participants most likely to be in the labor force or wanting employment, we included the 6,373 survivors and
1,968 siblings aged =25 years as of 2002 and not unemployed by choice.
Principal Findings: Survivors’ average age was
34.0 (range: 25-54) years compared to 35.9 (range:
25-57) for siblings. Survivors were less likely to be female (45% vs. 50%), college educated (48% vs.
57%) and married (48% vs. 63%) than siblings;
90% of survivors and siblings were White, non-
Hispanic. Over 66% of survivors and 74% of siblings reported current full-time employment with no differences in the proportion employed part-time
(7% each). For the main outcome of interest, significantly more survivors (9%) compared to siblings (1.5%) reported unemployment due to illness/disability (multivariate OR 5.25; 95% CI 3.66,
7.54). Within survivors, females reported unemployment due to illness/disability more often than males (12% vs. 8%, respectively; multivariate
OR 1.84, 95% CI 1.53, 2.21), but this same effect was not apparent in siblings (1% males and 1% females). Younger age, being non-Hispanic white, higher education and being married were all significantly associated with a lower probability of health-related unemployment within survivors.
Conclusion: Childhood cancer survivors are more likely to report health-related unemployment compared to their siblings. Female survivors are at significant additional risk for unemployment.
Implications for Policy, Delivery or Practice:
Health-related unemployment may disproportionately affect adult survivors of childhood cancer. As survival continues to increase for childhood cancers, efforts to mitigate health-related work limitations are necessary, especially for female survivors, who often have poorer health outcomes.
Funding Source(s): NCI
● Dimensions of Physician Quality
Barrett Kitch, M.D., M.P.H.; Elizabeth Mort, M.D,
M.P.H.; David Blumenthal, M.D., M.P.P.; James
Mountford, M.D., M.P.H.; Sandra Feibelmann,
M.P.H.; Elyse Park, Ph.D., M.P.H.
Presented by: Barrett Kitch, M.D., M.P.H., Senior
Scientist, Institute for Health Policy, Massachusetts
General Hospital, 50 Staniford Street, 9th Floor,
Boston, MA 02114, Phone: (617) 726-4201; Email: bkitch@partners.org
Research Objective: Performance measures and public reporting are believed to be powerful levers for improving the quality of healthcare, yet there is ongoing debate about the accuracy and usefulness of these mechanisms for examining physician quality. Debate is often most forceful among physicians themselves and their concerns about the ability of current performance measures to reflect physician quality is a major barrier to the physician community’s embracing quality measurement. One understudied area is which dimensions physicians themselves regard as being the most important markers of physician quality. The primary aim of this study was to examine the views of expert clinicians and clinical leadership on dimensions of physician quality.
Study Design: In order to develop a qualitative instrument for focus group data collection, we reviewed the literature, including regulatory body
(Accreditation Council of Graduate Medical
Education) conceptions of physician quality and
existing physician peer assessment instruments
(e.g. American Board of Internal Medicine peer assessment survey). We then conducted three focus groups that consisted of clinicians from
Massachusetts General Hospital (Boston, MA) in
May 2007.
Population Studied: The focus groups included a total of 18 physicians and 3 nurses from a variety of disciplines, including primary care and both procedural and non-procedural specialties.
Principal Findings: We identified 11 dimensions of physician quality that could be subsumed under three domains: system, self, and communication.
System consisted of use of the system, quality of service to patients, relationships with staff and colleagues, and supervision and teaching of trainees (of note, supervision and teaching of trainees may have limited relevance outside of academic health centers). Communication consisted of communication with patients and their families and psychosocial aspects of care. Self consisted of professional behavior, clinical knowledge, diagnostic skills, and procedural/technical skills. Patient management was a dimension that was categorized across all 3 domains.
Conclusion: Based on our focus groups, we developed a physician-generated model of physician quality that consisted of three broad domains of quality that encompassed a number of dimensions. The content and interactions among the three domains determined quality of care.
Implications for Policy, Delivery or Practice: Our findings, consistent with other work in this area, suggest that the tools in current use to measure physician quality, including peer assessment instruments, do not address a number of dimensions that physicians themselves identified as important elements of physician quality. Next steps include the development and piloting of a physician peer assessment instrument that addresses these dimensions of quality.
Funding Source(s): Massachusetts General
Hospital institutional grant
● Public Reporting of Nursing Sensitive Quality
Measures in Massachusetts & Maine: Views from Hospital Leaders
Barrett Kitch, M.D., M.P.H.; Patricia Noga, R.N.,
Ph.D., M.B.A., N.E.A.-B.C.; Joyce Clifford, Ph.D.,
R.N.; Sharon Gale, M.S., R.N.; Sandra Feibelmann,
M.P.H.; Joel Weissman, Ph.D.
Presented by: Barrett Kitch, M.D., M.P.H.,
Instructor in Medicine, Institute for Health Policy,
Massachusetts General Hospital, 50 Staniford
Street, 9th Floor, Boston, MA 02114, Phone: (617)
726-4201; Email: bkitch@partners.org
Research Objective: We conducted a study to analyze the views of hospital chief executive officers (CEOs) and chief nursing officers (CNOs) in
Massachusetts and Maine on the development and impact of statewide programs for the public reporting of nurse sensitive quality measures
(NSMs) endorsed by the National Quality Forum
(NQF).
Study Design: In 2008 the research team conducted a survey of all acute care hospital CEOs and CNOs in Massachusetts and Maine. The survey items were developed in part based on issues identified through a series of interviews and hospital case studies conducted in these 2 states in
2006-2007. Interviewees were asked questions that assessed opinions about newly developed statewide programs for public reporting of NSMs, burden of the NSM program, impact on quality improvement activities, specifications and accuracy of the measures, and the public’s interest in the data.
Population Studied: Two hundred and five hospital leaders in Maine and Massachusetts, consisting of the CEOs and CNOs of all acute care hospitals in these states, were mailed a survey and cover letter describing the study. Names of hospital leaders were provided by the Massachusetts
Hospital Association (MHA) and the Maine Health
Data Organization (MHDO); titles and mailing addresses were confirmed by directly contacting each acute care hospital in the two states.
Principal Findings: Sixty CEOs and seventy-five
CNOs completed the survey (66%). Most hospital leaders believe that public reporting of NSMs is likely to have a positive impact both on quality of nursing care and patient outcomes. In both states, most believed that a public mandatory program would work best to improve quality of nursing care.
Programs were viewed as at least somewhat burdensome by most of the leadership, and as very burdensome by nearly a quarter of CEOs. Most of the respondents were concerned about the accuracy and consistency, across hospitals, of how data were collected and reported. A large proportion did not believe that the public was aware of their hospital’s performance on NSMs.
Conclusion: Statewide programs for public reporting of NSMs are possible and are viewed by most CEOs and CNOs as likely to have a positive impact both on the quality of nursing care and patient outcomes.
Implications for Policy, Delivery or Practice:
Mandatory programs for public reporting may be viewed as more likely to improve quality of nursing care than confidential or voluntary reporting programs. Despite their belief in the value of the programs, public reporting of NSMs is perceived as burdensome to hospital leadership. Leaders see limited capacity to accommodate expanded numbers of measures, whether nursing-sensitive or other types of quality measures.
Funding Source(s): RWJF, RWJF Interdisciplinary
Nursing Quality Research Initiative
● Coronary Artery Disease Quality Measurement from Electronic Health Record Systems: An
Analysis of Physician Exception Reporting
Karen Kmetik, Ph.D.; Michael O’Toole, M.D.;
Carmen Brutico, M.D.; Stasia Kahn, M.D.; Gary
Fischer, M.D.; John Schaeffer, M.D.
Presented by: Karen Kmetik, Ph.D., Director,
Clinical Performance Evaluation, Clinical
Performance Evaluation, American Medical
Association, 515 South State Street, Chicago, IL
60654, Phone: (312) 464-4221; Email: karen.kmetik@ama-assn.org
Research Objective: Many physician performance measures developed by the Physician Consortium for Performance Improvement (PCPI) allow physicians to identify patients with exceptions (eg, drug allergy) to recommended processes of care.
Those patients are removed from the denominator when calculating performance. Exception data are collected, categorized as medical, patient, or system reasons, and reported separately with performance rates. The objectives of this study are to assess the prevalence and patterns of exception reporting, document specific reasons for exceptions, and evaluate the accuracy of reported exceptions.
Study Design: The five physician practices in the
Cardio-HIT quality measurement/EHR collaborative project submitted to a data warehouse patient-level data elements from their EHRS for four coronary artery disease (CAD) measures (antiplatelet therapy, lipid-lowering therapy, beta-blocker therapy for prior MI, ACEI/ARB therapy) developed by the
PCPI with the American College of Cardiology and the American Heart Association. Retrospective manual reviews of the EHRS were performed at the practice sites to validate the exceptions reported to the warehouse and to identify specific reasons behind the reported exceptions. Records thought to be “apparent quality failures” were also reviewed to determine whether a valid exception existed in the
EHRS which was not submitted to the data warehouse.
Population Studied: Samples of patients in the data warehouse 18 years of age and older, with two visits within a 12 month period, and a diagnosis of
CAD, seen during the first or second quarters of
2007, with reported exceptions (545 records), or an
“apparent quality failure” (585 records).
Principal Findings: The overall performance rate for all sites for all CAD measures was 77%. The majority of records with reported exceptions also met the numerator (drug prescribed); records with exceptions and no drug prescribed are called “true exceptions.” The true exception rate across all sites and all measures was 3.6%, with variation across measures, sites and site characteristics.
Most exceptions were for medical reasons, with
96% agreement between reported exception and documentation in the EHRS, and specific reasons noted as clinical contraindication (78.3%), drug allergy (14.1%), drug interaction (0.7%), and drug intolerance (7.0%). Only 35% of “apparent quality failures” were found to be opportunities for improvement. The remaining 65% met the measures, mostly due to an exception being located.
Conclusion: Exception reporting was generally low with high rates of agreement. The specific reasons for a medical exception suggest standard categories of medical exceptions (eg, clinical contraindication). Because many exceptions are not absolute, physicians may decide to “over-ride” an exception and provide the relevant aspect of care.
Implications for Policy, Delivery or Practice: The ability to collect and analyze exception data may prove valuable in understanding variations in care.
Physician access to exception data from the EHRS at the point of care is critical for clinical decision making and improving patient outcomes. If widely available, payers and policy makers could reliably use physician performance measures reported from an EHRS for quality reporting, pay for performance, and identifying outlier rates in larger, populationbased measurement programs.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Measurement
● Development of Internationally Compatible
Instruments to Assess the Obesogenicity of
Urban Environments
Cecile Knai, Ph.D., M.P.H.; Joceline Pomerleau,
Ph.D. M.S.; Martin McKee, C.B.E., M.D., D.Sc.
Presented by: Cecile Knai, Ph.D., M.P.H.,
Lecturer, Public Health & Policy, London School of
Hygiene & Tropical Medicine, Keppel Street,
London,WC1E7HT, UK, Email: cecile.knai@lshtm.ac.uk
Research Objective: To develop a portfolio of instruments that can capture the consequences of policies relevant to obesity in international studies.
There is an extensive body of research that has sought to measure various aspects of the
“obesogenicity” of environments, capturing both conduciveness to physical activity (e.g. walkability scales and measures of neighbourhood attractiveness) and the food environment (such as density of types of food outlets). However these measures have, almost entirely, been developed and used within single countries, even though there is evidence that findings in some countries do not apply in others (for example, the US findings associating density of fast food outlets with obesity are not replicated in the UK or Australia). There is a need to develop and validate instruments that can capture important contextual differences among countries.
Study Design: Development and validation of a
Community Questionnaire, comprising a Food
Environment Component and a Built Environment
Component. The instruments were developed following a review of the determinants of obesity and of existing methods used to assess the environment related to obesity. Following detailed discussions by an international team of researchers, two draft instruments were prepared and piloted, in areas of different socioeconomic status in cities in five countries. The Food environment component measures the numbers and types of shops selling foods in given neighbourhoods, the cost of indicative food items, marketing in and outside stores, the nature and extent of television advertising on children’s television, and the cost and marketing of selected fast food items. The Built environment component examines indicators of the ‘walkability’ and
‘bikability’ of the neighbourhoods including cycle lanes, parks, public transport stops, road crossings, pavements, and evidence of incivilities. Field manuals were developed to standardise the methods used. Evaluation drew on methods used to assess reliability and validity of psychometric instruments (e.g. inter-rater reliability, content validity).
Population Studied: Five European cities (Ankara,
Turkey; Brno, Czech Republic; Marseille, France;
Riga, Latvia, Sarajevo, Bosnia and Herzegovina).
Principal Findings: The development of such instruments faces many practical and methodological challenges. The former include access to high quality maps, including variable definition coverage by Google Earth; permissions for researchers to collect information on, for example, food prices in some countries. The latter include difficulties in defining and categorising (by socioeconomic status) areas for sampling; differences in the meaning of contextual variables, such as the presence of graffiti; difficulties in defining comparable food baskets and foodstuffs, and choosing appropriate quality criteria for aspects of the built environment. However, many of these can be addressed.
Conclusion: There is an urgent need for a better understanding of what an obesogenic environment is in different countries and to develop methods to measure it. This paper describes how some of the challenges are being overcome.
Implications for Policy, Delivery or Practice:
International studies of obesogenic environment must take account of the different contexts within which they are undertaken.
Funding Source(s): European Commission, 6th
Framework Program
Theme: Obesity Prevention and Treatment
● Expanding HIV Rapid Testing Via Point-of-
Care Paraprofessionals
Herschel Knapp, Ph.D., M.S.S.W.; Henry Anaya,
Ph.D.; Jamie Feld, B.A.; Matthew Goetz, M.D.
Presented by: Herschel Knapp, Ph.D., M.S.S.W.,
Health Science Officer; National Project Director,
Quality Enhancement Research Initiative-HIV /
Hepatitis, Center for the Study of Healthcare
Provider Behavior, Health Services Research &
Development Center of Excellence, Quality
Enhancement Research Initiative HIV/HEP, U.S.
Department of Veterans Affairs Greater Los
Angeles Healthcare System, 11301 Wilshire
Boulevard, Mail Code 111G, Los Angeles, CA
90073, Phone: (310) 478-3711 ext. 48568; Email:
Herschel.Knapp@va.gov
Research Objective: HIV counseling and testing has traditionally been performed by highly trained professionals in clinical settings. With HIV rapid testing, a reliable and easy to use diagnostic tool, paraprofessionals can be trained to administer onsite HIV testing in a variety of non-traditional settings, broadening HIV detection rates. Our objective was to create and implement a robust and sustainable paraprofessional training module to facilitate off-site HIV rapid testing in non-clinical settings.
Study Design: Trainees attended four training sessions detailing the use of the instrument, HIV education, rapid test instructions, and communication techniques.
Principal Findings: After these sessions, trainees competently carried out HIV rapid testing in homeless shelters throughout Los Angeles County.
Conclusion: Agencies motivated to expand HIV screening programs may use trained paraprofessionals to administer a full range of services (recruitment, pretest counseling, administration, interpretation of results, posttest counseling, documentation) via this training model, enabling more highly trained healthcare providers to focus efforts on tending to patients identified as HIV positive.
Implications for Policy, Delivery or Practice:
This transportable training module is particularly relevant for use in resource-limited settings worldwide where health professionals are scarce or unavailable.
Funding Source(s): VA
Theme: Prevention and Treatment of Chronic
Illness
● Community Socioeconomic Factors
Associated with Public Hospital Closure &
Privatization, 1983-2003
Michelle Ko, M.D.; Ninez Ponce, Ph.D.; Jack
Needleman, Ph.D.;
Presented by: Michelle Ko, M.D., Graduate
Student, Health Services, University of California,
Los Angeles School of Public Health, P.O. Box
951772, Los Angeles, CA 90095, Phone: (310)
8925-8242; Email: jassmine@ucla.edu
Research Objective: Relative to private hospitals, public hospitals face a unique set of financial, social and political stresses, which may affect their survival over time. Because public hospitals are key providers in the health care safety net, it is critical to understand the characteristics of communities that close or convert public hospitals. Prior studies examining factors associated with hospital closure and privatization have generally combined public and private hospitals in model estimation.
Importantly, although recent research has shown that the hospital-level characteristics associated with closure and privatization differ dramatically between public and private hospitals, no known study has examined how key contextual factors affect public hospital closure and privatization. The objective of this study is to examine how community sociodemographic factors associated with closure and privatization of U.S. public hospitals from 1983 to 2003.
Study Design: Multivariate competing hazards models were tested to examine associations between predictors and time to closure versus privatization. Predictor variables consisted of yearly county-level sociodemographic characteristics, such as racial and ethnic composition and measures of community socioeconomic status. The models were stratified by rural and non-rural location and controlled for hospital size, provision of psychiatric services, teaching affiliation, nurse staffing ratios, expenditures per inpatient day, share of revenues paid by Medicare and Medicaid, presence of other public hospitals and hospital competition in the county.
Population Studied: Retrospective cohort study of general, acute care, nonfederal hospitals under public ownership in 1983.
Principal Findings: Among rural hospitals, the likelihood of closure was higher for hospitals in areas with higher poverty rates (HR 10.88, p=0.008). The likelihood of privatization was lower for hospitals in areas with a higher proportion
Hispanic population (HR 0.09, p=0.006). Among urban hospitals, the likelihood of closure was higher for hospitals in areas with a higher proportion non-
Hispanic white population (HR 2.61, p=0.032). The likelihood of privatization was lower in areas with higher poverty rates (HR 0.012, p=0.017).
Conclusion: The relationships between community context and public hospital survival differ for rural versus non-rural hospitals, as well as for those that close versus those that privatize. In rural areas, loss of the public hospital via closure or conversion appears to be associated with indicators of lower socioeconomic status. In urban areas, public hospital closure is associated with the proportion of residents who are white, which may proxy for a preference for privately provided health care, leading to reduced community support for safety net services.
Implications for Policy, Delivery or Practice:
Given the current economic downturn and continued rise in the number of uninsured, policymakers need a greater understanding of factors that impact the stability of the health care safety net. In rural areas, public hospital status changes have occurred in areas with a relatively greater need for safety net services, suggesting that these communities require funding and support to maintain access to care. In urban areas, policymakers should address whether community support to maintain public hospitals is concomitant with local need for services.
Funding Source(s): RWJF
Theme: Coverage and Access
● A Qualitative Assessment of Pap Test
Provider’s Knowledge of HPV Testing
Guidelines
Kim Kobus, Ph.D.; Teri Larkins, Ph.D.; Katherine
Roland, M.P.H., C.H.E.S.; Vicki Benard, Ph.D.;
Presented by: Kim Kobus, Ph.D., Senior
Researcher, Institute for Health Research & Policy,
University of Illinois at Chicago, 1747 South
Roosevelt Road, Room 558, M/C 275, Chicago, IL
60608, Phone: (708) 275-8714; Email: kkobus@uic.edu
Research Objective: Research evidence and consensus recommendations among professional medical organizations support decreasing the frequency of cervical screening in women 30 and older who present with either three or more consecutive normal Pap tests, or with a concurrent normal Pap and HPV test result. Benefits for these recommendations include decreasing health care costs and minimizing unnecessary invasive procedures. The CDC is implementing a pilot study to examine the adoption of HPV testing by Pap test providers participating in a national screening program. In order to inform the development of an appropriate intervention for the pilot study, a sample of providers from a Midwest urban area were interviewed to assess level of knowledge regarding cervical screening practices. The findings from those interviews are presented here. Our research objective was to determine provider-identified barriers to adoption of cervical cancer screening guidelines and technology.
Study Design: Six physicians with no relationship to the pilot study were contacted and interviewed in
June to August, 2008. Information was collected in a one hour in person meeting. Compensation was provided. Information collected included practice specialty, years in practice, patient demographics, screening procedures and tests used, knowledge of guidelines and recommendations, perceived barriers to screening for patients, and barriers for the provider to increase screening intervals for women who qualify (concurrent negative Pap and
HPV test results, and 30 years of age and older), ideas for office-based interventions to increase
knowledge of guidelines, and patient education needs.
Population Studied: Six physicians practicing in the Chicago metropolitan area.
Principal Findings: Physicians interviewed were in practice an average of 9 years. Identified barriers to
HPV test utility/screening intervals include: comfort level of providers, knowledge of guidelines and resistance to change, fear of missing cervical cancer in women skipping yearly examination, discouragement from colleagues, financial concerns, burdens on staff, and tradition. Updates to guidelines were most commonly obtained at conferences, in light of reported disagreement among providers. Half of physicians reported understanding and willingness to implement guidelines for decreasing frequency of Pap test for women 30 and older with normal Pap and negative
HPV test, or 3 consecutive normal Pap tests. Of these, only 2 reported implementing this into their practice. For all physicians, monogamy status and sexual history weighed heavily into their decision to recommend increased length of time between screenings.
Conclusion: In addition to informing the development of a provider intervention, these findings concur with previous work stating some practitioners are either unaware of changes to cervical screening recommendations, or are aware but resistant to implementing them into clinical practice.
Implications for Policy, Delivery or Practice:
Based on these findings, our targeted provider intervention will include reviews of critical issues such as HPV epidemiology and disease progression, patient risk profiles, medico-legal issues with increasing intervals, and the perceived impact of patient sexual behaviors on screening protocols.
Funding Source(s): CDC
Theme: Public Health
● Utilization of Clinical Services for Populations with Long-Term Health Conditions at Louisiana
Community Supported Emergency Shelters
Following the 2005 Gulf Hurricanes
Yordanka Koleva, M.P.H., Ph.D.
Presented by: Yordanka Koleva, M.P.H., Ph.D.,
Research Associate, Health Systems Management,
Tulane University, 1440 Canal Street, Suite 1900,
New Orleans, LA 70112, Phone: (504) 756-3802;
Email: ykoleva@tulane.edu
Research Objective: The 2005 hurricanes created an opportunity to analyze utilization of disaster relief services and identify areas for improvement in public health emergency preparedness. Clinical services and needs for populations with long-term health conditions at general emergency shelters were assessed based on a survey conducted by
Louisiana Department of Health and Hospitals and
Department of Social Services.
Study Design: A needs assessment survey was performed at 168 general shelters in Louisiana in early September 2005. Data collected through assessment forms contained contact information, summary findings, and data about shelter demographics, capacity, disease surveillance, medical drugs/equipment and environmental health.
Data analyses were based on the assumption that long-term health conditions are predictors of social vulnerability in emergency situations. Multivariate analyses on data determined the effect of sheltering operations on the health status of residents and integration of emergency shelters in community health preparedness system.
Population Studied: Populations with long-term conditions and mental health needs living in the community who are not eligible for special needs shelters (SNS) and utilize general community supported emergency shelters. Emergency management coordinators from faith-based,
American Red Cross, community, city, parish and privately operated shelters.
Principal Findings: Preliminary analyses demonstrated dependency of vulnerable populations on community support in emergency situation. Individuals seeking for community support were predominantly medically dependent, elderly, children, people with long-term health conditions and low socioeconomic status. Findings confirmed hypothesized vulnerability of individuals with longterm conditions who were not hospitalized or eligible for special needs shelters. Analyses of various aspects of shelter services, including disease surveillance, health care referral systems, access to medical care, transportation and communications demonstrated the need for better integration of community organizations in the state level disaster relief activities. Significant challenges were associated with the lack of comprehensive data collection and management system, inadequate medication supplies, insufficient trained staff and volunteers, and difficulties predicting the duration of shelter operations.
Conclusion: Results suggested that pre-event planning with community partners should include programs for recruiting and training of staff and volunteers, improving sanitation conditions and implementing information management system for health-related data and drug inventories. Referral system for individuals with long-term conditions needs to be incorporated in pre-disaster mitigation programs on national and state levels.
Implications for Policy, Delivery or Practice: The survey provided data for stakeholders in the community disaster response and supported the importance of an interdisciplinary approach in public health preparedness analysis and planning. State agencies and community relief organizations can formulate local and multijurisdictional strategies for recruitment and training of volunteers and develop
need assessment instruments, based on identified best practices.
● Issues in Designing a Health Insurance
Program for Bermuda's Seniors: Evidence from
The FutureCare Program
Gerald Kominski, Ph.D.; Jay Ripps, F.S.A.,
M.A.A.A.; Andrea Polischuk, F.S.A., M.A.A.A.;
Jeffrey Burkhardt, Ph.D.; Miriam Laugesen, Ph.D.;
Andrew Cameron, Ph.D.
Presented by: Gerald Kominski, Ph.D., Professor,
Health Services & Center for Health Policy
Research, University of California, Los Angeles
School of Public Health, 10960 Whilshire
Boulevard, Suite 1550, Los Angeles, CA 90024,
Phone: (310) 794-2681; Email: kominski@ucla.edu
Research Objective: To develop an enhanced publicly subsidized health insurance program for
Bermudian seniors, ages 65 and older, that improves their access to primary care, outpatient services, and prescription drugs that is also affordable from the perspective of both the government and seniors.
Study Design: Policy analysis of alternative design features for a publicly subsidized health insurance program, called FutureCare, based on international experience, including U.S. experience under the
Medicare program, and pre/post comparison of health care expenditures among seniors under alternative policy scenarios.
Population Studied: Bermuda citizens and longterm residents ages 65 and older. Baseline health expenditures obtained from the Ministry of Health, which provides coverage for about 45% of seniors through a basic Health Insurance Plan (35%) or to retired government employees through the
Government Employees Health Insurance plan
(10%), and from 3 private insurance companies that provide insurance to about 50% of seniors. Data on out-of-pocket health expenditures obtained from a
2007 household survey of seniors.
Principal Findings: A substantial portion of
Bermuda's seniors -about 40% - currently have inadequate insurance; 35% are enrolled in the government subsidized HIP plan, which primarily provides coverage for inpatient hospitalization, with limited coverage for prescription drugs and dental care; another 5% are completely uninsured. The remaining 60% of Bermuda's seniors have fairly comprehensive insurance, but pay substantial premiums out-of-pocket for that coverage, ranging from $700-1200 per month. Based on our analyses of 3 optional benefit designs, we estimated that government spending for health care would increase from $33.8 to $43.5 million in FY 2011 (or by 54-70%) if the government subsidized a new standard benefits package for all Bermudian seniors, with 75% of the additional costs paid by the government and 25% by seniors, except those below poverty, who would be fully subsidized. We estimate that these additional government revenues could be raised through a total payroll tax of 1.2-
1.8%, with a 50/50 contribution from employers and employees. Furthermore, we estimate that seniors above poverty would experience sufficient savings in out-of-pocket expenses to be able to afford increased premiums under each of these options.
The increase in government expenditures for health care under these scenarios would represent about a 3-4% increase in Bermuda's total government expenditures.
Conclusion: Government subsidies for health insurance could substantially increase benefits for about 40% of Bermuda's seniors, and increase the affordability of health insurance for the remaining
60% who are relatively well insured but who pay high out-of-pocket premiums. These improvements in access and affordability can be achieved at a relatively modest increase (i.e., 3-4%) in the government's total annual budget and can be financed through a modest payroll tax (i.e., 1.2-
1.8%). FutureCare would provide stability and equity for Bermudian seniors, and would extend the principal of shared employer/employee responsibility that currently exists through a mandate on employers to the entire retired, aged population.
Implications for Policy, Delivery or Practice:
Providing additional government subsidies within an existing mixed public/private system of financing for seniors is an effective strategy for improve access, equity, and affordability.
Funding Source(s): Bermuda Ministry of Health
Theme: Coverage and Access
● A Systematic Review of Health-Related
Outcomes from Community Health Worker
Interventions in Disease Prevention & Chronic
Disease Management
Jennifer Kraschnewski, M.D.; Dan Jonas, M.D.,
M.P.H.; Meera Viswanathan, Ph.D.; Brett
Nishikawa, M.D.; Laura Morgan, M.A.; Patricia
Thieda, M.A.
Presented by: Jennifer Kraschnewski, M.D.,
National Research Service Award Primary Care
Research Fellow, Medicine, University of North
Carolina at Chapel Hill, 5039 Old Clinic Building,
CB 7110, Chapel Hill, NC 27599, Phone: (919)
966-2276; Email: krasch@med.unc.edu
Research Objective: Community health workers
(CHWs) can serve as a bridge to the health care system for both disease prevention and management. We aimed to conduct a systematic review of the effectiveness of CHW interventions in the areas of disease prevention and chronic disease management.
Study Design: We conducted a systematic review using standard Evidence-based Practice Center methods. We searched Medline, the Cochrane
Database, and CINAHL from their inception through
October 2008 using a variety of terms for CHWs.
We included studies with a comparison group that were conducted in the United States, published in
English, with at least 40 participants. Two reviewers independently assessed each abstract and full-text article for inclusion, resolving disagreements by consensus. Two reviewers abstracted data and checked for accuracy.
Trained reviewers abstracted data and assessed the methodologic quality of studies using predefined criteria from international standards.
Population Studied: Our initial search identified
992 articles. Of these, 23 studies met our inclusion criteria. All of the studies focused on low income and minority populations.
Principal Findings: Trial duration ranged from three months to four years. Of 23 studies, 11 focused on disease prevention and 12 studies addressed disease management. Studies that addressed disease prevention focused on a wide variety of conditions, including pediatric immunizations, cardiovascular disease, diabetes prevention, HIV prevention, second-hand smoke exposure, colorectal cancer prevention, and general preventive care. Within disease prevention, nine studies were randomized controlled trials and two studies were observational. Of the 11 studies, one was rated good quality, six fair, and four poor. Five studies used high-intensity CHW models, four used moderate-intensity interventions, and two used lowintensity interventions. Studies evaluated a wide variety of outcomes, including measures of knowledge, behavior, health, and health care utilization. Preliminary results suggest variations in the consistency and volume of evidence across outcomes, with more evidence for outcomes involving changing knowledge and health care utilization, and inconsistent evidence for outcomes of behavior and health. Twelve studies addressed disease management, including diabetes mellitus, hypertension, asthma , back pain and tuberculosis.
Within disease management, ten studies were randomized controlled trials and two studies were observational. Of the 12 studies, one was rated good quality, eight fair, and three poor. Seven studies used high-intensity CHW models and five studies used moderate intensity interventions.
Studies evaluated a wide variety of outcomes, with most reporting on health outcomes specific to the condition of interest. Preliminary results suggest variations in the consistency and volume of evidence across outcomes.
Conclusion: CHWs have been used in many different health conditions, largely targeting low income and minority populations. Preliminary analysis suggests variations in consistency, volume, and quality of evidence for disease prevention and management. Future research of
CHW interventions should take into account evidence of past effectiveness in specific conditions.
Implications for Policy, Delivery or Practice:
CHW interventions may serve as an effective means of improving certain outcomes related to preventing disease in underserved, minority populations. Further research is necessary to determine the role of CHWs in chronic disease management.
Funding Source(s): AHRQ, NRSA Primary Care
Research Fellowship
Theme: Disparities
● Using Real-time Pharmacy Data & Telephonic
Management to Improve Medication Adherence in a Bipolar Population
Neil Krohn, Ph.D.
Presented by: Neil Krohn, Ph.D., Senior Health
Information Consultant, Analytics, Health
Management Corporation, 700 Broadway, Denver,
CO 80273, Phone: (720) 225-6748; Email: neil.krohn@anthem.com
Research Objective: Adherence to a medication regimen is one of the most important factors in the effective treatment of Bipolar Disorder. Nonadherence to Bipolar medications presents an important quality of care concern that negatively impacts short-term recovery, long-term stability, and the cost of care associated with treatment. Bipolar patients have high rates of refilling medications late or not refilling medications at all. As patients fail to sustain continuous medication supplies, they become increasingly at risk for longer and more severe recurrent episodes of the illness. This study sought to determine whether using real-time pharmacy data to outreach Bipolar patients would be an effective means of increasing medication adherence and improving clinical stability.
Study Design: Medical claims were queried for
Bipolar diagnosis codes to determine program eligibility. Daily pharmacy refill data was linked to eligible program members who were five days late refilling Bipolar medications. The program then provided telephonic medication-related assessment and education via a behavioral health nurse for these late-refilling members, as well as case management services for higher-acuity members.
Analyses compared one-year pre-post changes on medication adherence and utilization variables for the program cohort, compared with a non-selected, intact group of individuals with a diagnosis of
Bipolar Disorder.
Population Studied: The study included individuals 18 years of age and older in a California
Commercial health plan who had a primary diagnosis of Bipolar Disorder, and who filled a medication used to treat Bipolar Disorder (N=838).
Principal Findings: For a one-year evaluation period, medication adherence, defined as the percentage of on-time medication refills, improved
22 percent relative to baseline levels. Preliminary
cost analyses showed decreases in both medical and behavioral health utilization.
Conclusion: The Bipolar Condition Care program in California demonstrated the effectiveness of realtime pharmacy-based intervention by a marked improvement in medication adherence rates. Realtime monitoring of medication refill patterns, coupled with personalized patient interventions, represents an effective means of addressing medication adherence treatment issues. Program members experienced greater clinical stability with fewer psychiatric rehospitalizations and decreased use of other intensive services while participating in the program. The program demonstrated cost effectiveness by reducing behavioral health and medical utilization, and by contributing to the stabilization of the most at-risk members in this population.
Implications for Policy, Delivery or Practice:
Real-time monitoring of medication refill patterns, coupled with personalized educational and support intervention, represents an effective means of addressing medication treatment issues. Programs of this nature can be a cost effective means of providing improved patient care and stability.
● Immigrants' Contribution to Medical
Expenditures
Leighton Ku, Ph.D., M.P.H.
Presented by: Leighton Ku, Ph.D., M.P.H.,
Professor, Health Policy, George Washington
School of Public Health, 2021 K Street, NW, Suite
800, Washington, DC 20006, Phone: (202) 416-
0479; Email: leighton.ku@gwumc.edu
Research Objective: To compare insurance coverage and medical expenditures for immigrant and U.S.-born adults.
Study Design: Analyses of the 2003 Medical
Expenditure Panel Survey, comparing recent immigrant, established immigrant and U.S.-born adults. Includes two-part multivariate analyses of medical expenditures, controlling for health status, insurance coverage, race/ethnicity and other factors.
Population Studied: Nationally representative sample of U.S. adults, immigrant and native-born.
Principal Findings: About 44% of recent immigrants and 63% of established immigrants are fully insured over 12 months. Immigrants’ per person unadjusted medical expenditures are about half to two-thirds as high as for U.S.-born adults, even when they are fully insured. Recent immigrant adults are responsible for only about 1% of public medical expenditures, although they constitute 5% of the population. After controlling for other factors, immigrants’ medical expenditures are about 14% to
20% (p < .001) less than those of U.S.-born adults.
Conclusion: Insured immigrants have much lower medical expenses than insured U.S.-born adults, even after controlling for insurance coverage.
Implications for Policy, Delivery or Practice: It is plausible that insurance premiums paid for immigrants may be cross-subsidizing care for the
U.S.-born. If so, benefits could be rechanneled back to immigrants to improve their care.
Theme: Disparities
● Factors Associated with Utilization of Urban
Emergency Departments by the Homeless
Population in the United States
Bon Ku, M.D.; Stephen Pitts, M.D., M.P.H.; Kevin
Scott, M.D.
Presented by: Bon Ku, M.D., Assistant Professor,
Emergency Medicine, Thomas Jefferson University,
1020 Samson, 239 Thompson Building,
Philadelphia, PA 19107, Phone: (718) 986-0475;
Email: bon.ku@jefferson.edu
Research Objective: Homeless individuals are frequent users of many emergency departments
(EDs) but there have been few studies looking at the unique needs of this highly vulnerable population. This study characterizes the homeless who visit urban EDs across the entire U.S. To our knowledge, this is the first attempt to describe homeless ED visits from a national perspective.
We sought to characterize and find demographic and clinical variables that were associated with visits made by homeless individuals to urban EDs in the U.S. during a two-year period.
Study Design: We performed a secondary analysis of the ED components of the 2005 and 2006
National Hospital Ambulatory Care Surveys
(NHAMCS-ED). NHAMCS-ED is nationally representative survey of U.S. hospitals. We examined bivariate and multivariable associations with a limited group of variables chosen on the basis of a priori clinical relevance. We used Stata version 10 for all statistical analyses. Significance tests and confidence intervals were obtained using
Stata’s “svy” programs (tabulation and logistic regression), which adjust standard errors for correlation within primary sampling units and allowed the use of probability weights to obtain national estimates.
Population Studied: Homeless individuals who visited urban EDs in the U.S. based upon 2005-
2006 NHAMCS-ED data.
Principal Findings: Homeless individuals made
550,000 (95% CI 0.4, 0.7 million) ED visits annually, or 72 visits per 100 homeless persons in the U.S. population. Homeless patients were older than other users with a mean age of 44 (95% CI 42 to
46) years compared to 36 (95% CI 35, 37).
Demographic variables independently associated
(p-value <0.01) with homelessness included male sex (AOR 3.34, CI 2.33, 4.78), Medicaid status
(AOR 1.96, 95% CI 1.34, 2.86), lack of insurance
(AOR 3.17, 95% CI 2.04, 4.90), and Western geographic region (AOR 3.12, 95% CI 1.98, 4.97).
Clinical variables independently associated (p-value
<0.01) with homelessness included arrival to the ED by ambulance (AOR 1.76, 95% CI 1.17, 2.64), seen by a resident or intern (AOR 1.81, 95% CI 1.10,
2.98), having a psychiatric diagnosis (AOR 5.83,
95% CI 3.60, 9.42) or having a substance abuse diagnosis (AOR 9.05, 95% CI 5.03, 16.29).
However, there was no association with race, triage status, time of arrival, or rate of ED admission to hospital.
Conclusion: We found that homeless patients who seek care in urban EDs often utilize ambulance services, lack medical insurance, and have psychiatric and substance abuse problems. This population represents a heavy cost burden to the health care system and poses a challenge for hospitals to treat and discharge them safely from the ED.
Implications for Policy, Delivery or Practice: Our findings can provide policymakers and health providers with important information on ED utilization by the homeless. Unfortunately, in light of a challenging economy and ongoing losses of health coverage, we do not anticipate that the issue of homelessness or increased utilization of overburdened EDs by the homeless will disappear within the foreseeable future. New strategies are needed to meet this growing challenge.
Theme: Disparities
● Surprises & Strategies among Families with
High-Deductible Health Plans
Jeffrey Kullgren, M.D., M.P.H.; Tracy Lieu, M.D.,
M.P.H.; Jeff Solomon, Ph.D.; Jim Sabin, M.D.;
Virginia Hinrichsen, M.P.H.; Alison Galbraith, M.D.,
M.P.H.
Presented by: Jeffrey Kullgren, M.D., M.P.H.,
Resident in Internal Medicine, Department of
Medicine, Brigham & Women's Hospital, 75 Francis
Street, Boston, MA 02115, Phone: (517) 214-7366;
Email: jkullgren@partners.org
Research Objective: High-deductible health plans
(HDHPs) are a new and controversial approach to encouraging patients to become more “discerning consumers” of health services by placing them at greater risk for the cost of their care. Little is known about the health care experiences of families enrolled in HDHPs. Our aims were to (1) describe surprises encountered, strategies for limiting health care costs, and perceptions of control about health care use among families enrolled in these plans, and (2) identify specific areas where HDHPs could be improved, to better support effective health care use by families.
Study Design: We conducted 4 focus groups with parents of families who had HDHPs and had experienced high out-of-pocket health care costs during the past 12 months. Focus groups were conducted using a semi-structured interview guide.
All groups were audio recorded and transcribed verbatim. Initial themes were identified independently by investigators using a grounded thematic analysis approach. Consensus regarding the final themes was achieved among investigators through ongoing discussion of transcripts.
Population Studied: Parents in families enrolled in a New England-based health plan’s HDHP with a deductible of at least $1,000 or more per individual and $2,000 or more per family before selected health benefits are covered.
Principal Findings: Parents identified specific situations where they were surprised by unexpected health care costs, including tests ordered during preventive visits and emergency department visits.
They reported often making health care decisions without knowledge of the financial consequences.
Strategies for limiting costs involved deferring medical visits, including watchful waiting, and seeking advice by telephone or on websites. In contrast, if they were to make a medical visit, parents did not expect doctors to know about or help them avoid out-of-pocket costs. Parents expressed a sense of lack of control over health care costs, in part due to the complexity of the
HDHPs. They also observed that lack of medical knowledge limited their control over health care spending decisions.
Conclusion: Families with high-deductible health plans experienced unexpected costs in specific areas, and their strategies to limit costs involved deferring medical visits rather than discussing costs with clinicians during visits. Few parents felt their
HDHP enabled them to have a sense of control over health care decisions and costs.
Implications for Policy, Delivery or Practice:
Families in HDHPs tended to defer making medical visits as a strategy to limit costs, but did not report other behaviors consistent with being the
"discerning consumers" that HDHPs are intended to encourage. HDHPs could offer better anticipatory guidance for families and might be more usable if their structure and information flow were simplified.
Clinicians should be aware that families with highdeductible health plans may seek more telephone advice in order to reduce medical visits, and might have concerns about costs but not initiate such discussions during medical interactions.
Funding Source(s): National Institute of Child
Health and Human Development
Theme: Consumer Choices in Health Care
● Code Blue: Shifts in Payer Sources of Patients
Visiting & Admitted Through Los Angeles
County Emergency Departments puts Financial
Stress on Hospitals
Heather Kun, M.S., Ph.D.; Andrew Barnes, M.P.H.;
J.Eugene Grigsby, Ph.D.
Presented by: Heather Kun, M.S., Ph.D., VP,
Research & Evaluation, Research & Evaluation,
National Health Foundation, 515 South Figueroa
Street, Suite 1300, Los Angeles, CA 90071, Phone:
(213) 538-0753; Email: hkun@nhfca.org
Research Objective: Hospital reimbursement is critical to the solvency of business. Due to recent hospital closures in Los Angeles County(LAC) and increasing number of the uninsured, hospitals find it difficult to remain financially secure when uncompensated for care. This study examines the change in payer mix of Los Angeles County hospitals between 2000 and 2006.
Study Design: Data from the 2000 and 2006 Office of Statewide Health Planning and Development
(OSHPD) Patient Discharge and Emergency
Department databases were used to conduct the analysis. Patients who were admitted through the
ED were broken down by payment source: self pay/county indigent program (SP/CIP), Medi-Cal,
Medicare, private insurance and other sources.
Hospitals studied were grouped by Service
Planning Areas (SPAs), disproportionate share status (DSH), and whether they were county owned
(public). Payer mix for hospital groups were calculated using the frequencies of hospital admissions via the ED by payer type. Descriptive statistics are used to compare changes in payer mix within and between hospital groups over time.
Population Studied: Hospital groupings (e.g.,
DSH, public/private, SPA) are the unit of analysis.
All admissions via the ED from the 74 general acute care hospitals in continuous operation from 2000 to
2006 were included in the analysis.
Principal Findings: Between 2000 and 2006, LAC had 27 private DSH, 43 private Non-DSH and 4 public hospitals. Key findings show that during this time, Non-DSH and DSH private hospitals experienced an overall increase in proportion of admits covered by inadequate payment sources
(Medi-Cal, SP/CIPand other) while public hospitals experienced little change or a slight decrease in proportion of admits covered by inadequate payment sources (Medi-Cal, Self Pay,
Indigent/other). Specifically, DSH private hospitals saw a 4% decrease in proportion of admits covered by Medicare and Private payer sources (60% versus 56%) and a 4% increase in proportion of admits covered by Medi-Cal and SP/CIP payers combined (39% versus 43%). Non-DSH private hospitals saw a 3% decrease in proportion of admits covered by Medicare and Private payers combined (82% versus 79%) and a 3% increase in proportion of admits covered by Medi-Cal and
SP/CIP payers combined (18% versus 21%). Public hospitals experienced virtually no change in the proportion of admits covered by Medicare and
Private payment combined (11% versus 12%) or
Medi-Cal and SP/CIP combined (90% versus 89%).
Regional analyses (e.g. by SPA) indicated most private hospitals have experienced unfavorable changes in payer mix.
Conclusion: Previously it was thought that the uninsured depended on public hospitals for care, but this study illustrates that private hospitals are experiencing significant increases in caring for the uninsured. As such, overall reimbursement for private hospital services continues to deteriorate threatening the financial viability of many private hospitals.
Implications for Policy, Delivery or Practice:
Private hospitals, regardless of DSH status, are bearing an increasing majority of the risk of serving the underinsured. As the economy weakens and unemployment rises, numbers of underinsured will increase placing further stress on an already distressed hospital system. Without adequate reimbursement for hospital services, additional emergency department and hospital closures should be expected.
Funding Source(s): Hospital Association of
Southern California
Theme: Coverage and Access
● Factors Related to Timing of Receiving
Colposcopy Procedure Following Cytological
Abnormalities
Tzy-Mey Kuo, Ph.D., M.P.H.; Nancy Berkman,
Ph.D.; Vicki Benard, Ph.D.; Lisa Richardson, M.D.,
M.P.H.; Carolyn Martin, M.S.W.
Presented by: Tzy-Mey Kuo, Ph.D., M.P.H.,
Research Psychologist, RTI International, 3040
Cornwallis Road, RTP, NC 27709, Phone: (919)
541-6622; Email: mkuo@rti.org
Research Objective: American Society for
Colposcopy and Cervical Pathology (ASCCP) guidelines for management of abnormal cytology recommend a colposcopy procedure following a
Pap test result of high grade squamous intraepithelial lesion (HSIL), low grade squamous intraepithelial lesion (LSIL), and atypical squamous cells that cannot exclude high-grade squamous intraepithelial lesion (ASC-H). Using a large administrative database, we examined the length of time it took for a woman to receive a colposcopy procedure after she was diagnosed with HSIL,
LSIL, or ASC-H. This study seeks to identify factors associated with differences in the time from diagnosis to the recommended colposcopy procedure.
Study Design: This study was a retrospective review of women with an index Pap test in 2004 who were followed for 12 months. The time interval from the diagnosis of HSIL, LSIL, and ASC-H to the colposcopy procedure was computed. Data include member enrollment files combined with physician and facility claims. We also included contextual variables obtained from census and area resource file data describing socio-demographic characteristics of women’s neighborhoods.
Descriptive and multivariate analyses were conducted to examine the relationship between the time interval and individual level factors, provider specialty type on the index Pap, and area level factors.
Population Studied: The study sample included women between the ages of 18 and 64 years who were members of a large geographically diverse major U.S. health care organization. Women were considered eligible if they were enrolled for 9 months prior to an index Pap test and 12 months after. We excluded women who were pregnant or had a procedure for diagnosing or treating a cervical cytological abnormality in the 9 months prior to the index Pap test. Therefore, the Index Pap test was not a follow-up to previous treatment but was the initial screening preceding a new diagnostic period. We then selected the study sample among those who received a colposcopy diagnostic procedure following HSIL, LSIL, and ASC-H diagnoses. All women were commercially insured.
Women enrolled in Medicare or Medicaid were excluded.
Principal Findings: Among the total eligible sample of 486,782 women who received an index
Pap in 2004, 9,152 (1.9%) had a HSIL, LSIL, or
ASC-H cytological result. Among these women,
4,456 (48.7%) received follow-up diagnostic colposcopy. Between the diagnoses of interest and colposcopy, the median time interval was 30 days and the mean interval was 49 days (with a standard deviation of 61 days). Our regression model shows that women who were younger, had less severe cytological abnormalities, received an index Pap from doctors other than gynecologists, and lived in areas with higher proportions of African American residents received colposcopy procedures after a longer time interval following their abnormal cytological results.
Conclusion: Using claims data, this study identified risk factors that are associated with a longer time interval to the recommended diagnostic procedure following the diagnoses of HSIL, LSIL, and ASC-H.
Implications for Policy, Delivery or Practice:
These findings are important for identifying populations that may need to be targeted by health care providers to reduce disparities related to cervical cancer early detection and treatment.
Funding Source(s): CDC
Theme: Gender and Health
● Gender Based Differences in Management &
Outcomes of Cholecystitis
SreyRam Kuy, M.D.
Presented by: SreyRam Kuy, M.D., Robert Wood
Johnson Clinical Scholar, Robert Wood Johnson
Clinical Scholars Program, Yale School of
Medicine, 333 Cedar Street, New Haven, CT
06520-8088, Phone: (210) 535-2877; Email: sreyram.kuy@yale.edu
Research Objective: Biliary tract disease is common, with women predisposed for its occurrence. Among patients hospitalized for cholecystitis, we identified gender based differences in demographics, type of management, and economic and clinical outcomes.
Study Design: This is a cross-sectional study using the Health Care Utilization Project-Nationwide
Inpatient Sample database. We identified all adult men and women hospitalized with cholecystitis,
1999-2006. Characteristics included interventions, number of days from admission to intervention, race, admission urgency, insurance, comorbidity, income, hospital size, location, teaching status, and region. Outcome measures were in-hospital mortality, mean length of stay, and mean hospital cost (inflation adjusted to 2005 dollars). Among patients who underwent cholecystectomy, surgical complications were measured. Statistical analyses were performed with chi-square analysis, Student’s t-test, ANOVA, and multivariate linear and logistic regression.
Population Studied: The study population was all inpatient hospitalizations with ICD-9-CM diagnosis codes for cholecystitis. We excluded discharges where data was missing on gender or age, where there was a neonatal or maternal code indicating pregnancy, and all patients younger than 18 years.
Principal Findings: Adult non-pregnant women
(n=518,826) compared with adult men (n=276,404), accounted for the majority of all admissions with the diagnosis of cholecystitis, 65.2 %. Compared with men, women admitted with cholecystitis were younger, (mean age 53.9 vs. 60.9 years, p<.0001) and more likely to be black (10.7 vs. 6.8%, p<.0001) or hispanic (15.5 vs. 12.1, p<.0001).
Women had lower rates of percutaneous cholecystostomy (0.1 vs. 0.2%, p<.0001) or ERCP
(4.4 vs. 4.9%, p<.0001), but more likely to undergo cholecystectomy (84.4 vs. 79.6%, p<.0001).
Overall, women had lower mortality (1.4 vs. 2.5%), shorter length of stay (5 vs. 6.5 days), and lower inflation adjusted cost ($11,898 vs. $15,606), all p<.0001. Among the patients who underwent surgical management (cholecystectomy), women had their surgery performed earlier then men (1.6 vs. 1.9 days after admission) and were less likely to have cholecystectomy complications (6.3 vs. 9.3%) or general complications (22.1 vs. 30%), all p<.0001. On multivariate analysis, gender, procedure and time to procedure remained independent predictors of outcome.
Conclusion: Contrary to many diseases where women have worse outcomes, in cholecystitis women have better clinical and economic outcomes
(lower mortality, shorter length of stay, lower cost, and lower rates of surgical complications). In addition, women were more likely then men to undergo surgical management and to have their cholecystectomy performed earlier during the hospitalization.
Implications for Policy, Delivery or Practice:
This study revealed a gender based difference in both management of cholecystitis and in the clinical and economic outcomes of the disease. Unlike gender, surgical management and timing of surgery
are modifiable variables. Identifying that these modifiable characteristics are associated with better clinical outcomes will help both providers and patients in treatment decision making. The disparity in economic outcome is significant to policy makers in examining comparative effectiveness and in identifying ways to improve the health of patients with cholecystitis.
Funding Source(s): RWJF
Theme: Gender and Health
● An Examination of Strategic Group
Membership & Technology of Nursing Homes
Alexandre Laberge, M.B.A., P.T.; Robert Weech-
Maldonado, Ph.D.; Jeff Harman, Ph.D.; Ann
Horgas, Ph.D.; Kathy Hyers, Ph.D.; Amy
Yarbrough, Ph.D.
Presented by: Alexandre Laberge, M.B.A., P.T.,
Student, Health Services Research, Management &
Policy, University of Florida, 3725 Northwest 110th
Terrace, Gainesville, FL 32606, Phone: (352) 871-
0589; Email: laberger5@comcast.net
Research Objective: Nursing homes need to identify strategies to be able to survive in the current healthcare environment. Some of the key strategic decisions include which processes should be provided and how to allocate resources in order for those processes to be provided effectively. This study examines whether business level decisions of technology provision (i.e. processes of care) and resources utilization can be utilized to identify the different strategies used in nursing homes. The study will then determine if there is any quality, cost and financial strategic-performance differences between the strategies.
Study Design: This study is a retrospective cohort study of the nursing home industry using the
Minimum Data Set (MDS), the Online Survey
Certification of Automated Records (OSCAR), and the Medicare cost reports. Study Design: The study used 57 process variables (including restorative, skin care, nutritional, bowel and bladder) and 11 resource variables (including staffing levels, staffing mixes and wages in a factor analysis). Strategic groups were identified using K-means cluster analysis based on the 9 process variables and 10 resource variables derived from the factor analysis.
Strategic-performance relationships of the strategic groups were determined by using generalized linear model with gamma distribution and log link for the quality and costs variables and ordered logit for the financial variables. Dependent Variables: Long term care measures included bladder, bowel, activities of daily living 4 point decline, as well as pressure sore prevalence. Post acute quality measures included walking improvement and pressure sore incidence.
Other measures included quality of care deficiencies, patient cost, operating margin and total margin with county level fixed effects and
White’s correction for heteroskedascity.
Independent Variables: Dichotomous variables for each strategic group, size, payer-mix, and case-mix variables (including acuindex, adlindex, tubefeeds, injections, IVs, suction, tracheotomies and respiratory care).
Population Studied: All free standing, nongovernment nursing homes that had Medicare beds in the United States(n=11701).
Principal Findings: Nursing homes that provide a high amount of long term processes of care had better long term care quality outcomes than the other groups. Nursing homes that provided a high level of care processes had the same costs ($136) as the nursing homes whose strategy was to keep costs low ($135) (both of which were lower than the nursing rehab group ($146) and specialty group
($144). The high care providers (1.20), the low cost providers (1.25) and the rehab focus providers
(1.46) had greater odds of being in the highest tier than those facilities that lack strategy.
Conclusion: Nursing homes whose strategy involves providing a high amount of nursing home processes appear to have better nursing home quality outcomes. The study supports the notion that facilities can provide quality care and still control costs. Finally, there are some strategies that are better than others with respect to financial performance.
Implications for Policy, Delivery or Practice:
Efforts by policy makers to improve quality in nursing homes should pay special attention to encouraging good care processes. Managers can also benefit with better financial performance by attending to their care processes.
Theme: Long Term Care
● Minding the Gap: Decomposing Differences in Utilization of Health Services Between
Depressed & Non-Depressed Elders in Europe
Keren Ladin, M.Sc.
Presented by: Keren Ladin, M.Sc., Graduate
Student, Health Policy, Harvard University, 478
Shawmut Avenue, Apartment 6, Boston, MA 02118,
Phone: (617) 960-6465; Email: kladin@fas.harvard.edu
Research Objective: This paper explores differences in utilization of health services between depressed and non-depressed persons aged 50+ in eleven countries across the European Union.
Study Design: After documenting statistically significant differences in health services utilization between depressed and non-depressed respondents in all eleven countries, we explore the extent to which these differentials can be explained by prevailing differences in the prevalence of covariates predicting utilization (e.g. demographic, socioeconomic, geographic, chronic disease, selfreported health, etc.) versus varying impact of these variables in depressed and non-depressed persons.
Population Studied: The sample was taken from
The Survey of Health, Ageing and Retirement in
Europe (SHARE, Wave 1 Release 2), a prospective observational study of 31,115 randomly selected men and women aged 50 years and above living in
Austria, Germany, Sweden, the Netherlands, Spain,
Italy, France, Denmark, Greece, Switzerland, and
Belgium in 2004.
Principal Findings: Using Blinder-Oaxaca decomposition methods for multivariate linear regression models, we find robust evidence that the gap in utilization between depressed and nondepressed can be accounted for by both prevalence
(explained) and impact (unexplained) differences between depressed and non-depressed. The prevalence effect accounted for 57.7% of the gap in utilization rates whereas the impact effect explained
42.3% of differences in utilization rates between depressed and non-depressed respondents.
Despite cross national differences in quality and coverage of health services, in all countries the prevalence effect was explained entirely by need based factors. The impact effect varied cross nationally, but was entirely explained by enabling factors, particularly socioeconomic status and urbanity.
Conclusion: While social policy has struggled to determine how to best intervene to reduce the uneven burden of disease and subsequent utilization of persons suffering from depression, gaps in understanding whether targeting depression treatment through primary care services or income redistribution would improve observed disparities in utilization.
Implications for Policy, Delivery or Practice: Our findings suggest that increasing attention, screening, and treatment of depression in most countries could reduce utilization differences.
Theme: Behavioral Health
● Sustainability of a Quality Improvement
Initiative to Improve Obesity Management in the
Outpatient Setting: Lessons Learned from the
BMI Collaborative
Neda Laiteerapong, M.D.; Keith Naylor, M.D.; Lisa
Vinci, M.D.; Julie Oyler, M.D.; Vineet Arora, M.D.,
M.P.H.
Presented by: Neda Laiteerapong, M.D., Chief
Resident, Medical Education, MacNeal Hospital,
3249 South Oak Park Avenue, Berwyn, IL 60402,
Phone: (312) 613-2354; Email: nedalai@gmail.com
Research Objective: Obesity is epidemic and accounts for $100 billion annually due to obesityrelated diseases. Body mass index (BMI) is the obesity-screening tool recommended by the United
States Preventative Services Task Force. We report on the short and long-term success of a quality improvement (QI) initiative in an internal medicine residency clinic to increase rates of BMI documentation.
Study Design: The initial QI initiative in January
2007 entailed BMI education for residents and nurses, distribution of patient and resident education materials, and accessibility to height/weight charts that simplified BMI determination (BMI charts). A process map of clinic workflow was created to assess the current process of BMI documentation. Clinical nurses were instructed to document patient weight and height on modified data intake forms. Plan-do-study-act cycles of QI were used for sustainability. At 6months post-intervention, educational handouts were provided to residents and patients on the availability of local exercise and nutrition programs.
At 9-months post-intervention, the most cited barrier to documenting BMI was missing BMI charts, which were replaced. A retrospective review of the
University of Chicago dictation system (EMDAT,
Inc.) was performed on PGY-2 dictations based on appointment date for 2 weeks of encounters at 2weeks pre-intervention, and 2-weeks, 2-months, and 6-months post-intervention. Also data was collected on all resident dictations for a 2-week period 12-months post-intervention. For each period, all encounters with a dictation were included. Documentation of height, weight, diagnosis of weight status, BMI, weight loss counseling, and referral to dietician was recorded.
Population Studied: Internal medicine resident dictations of patient encounters seen in at an urban academic primary care clinic.
Principal Findings: For PGY-2 residents, 206,
162, 175, 119 and 188 patient encounters were audited in the 2-week pre-intervention and 2-week,
2-month, 6-month, and 1-year post-intervention periods, respectively. The percentage of patient encounters with height dictated significantly increased from 11% to 88% post-intervention with sustained results at 2-months (87%), 6-months
(84%) and 12-months (75%). The percentage of encounters with weight documented increased significantly from pre-intervention, 90%, to 2-weeks post-intervention, 97%, and also remained high at
95%, 94%, and 93% at 2, 6, and 12-months postintervention. The percentage of encounters with
BMI dictated increased significantly from 4% to 79%
2-weeks post-intervention. The percentage of charts with BMI dictated decreased to 69% and
41% at 2 and 6-months post-intervention, respectively, remaining at 43% at 12-months postintervention. Interestingly, at 12-months post, PGY-
1 residents were least likely to document BMI compared to PGY-2 and PGY-3 residents (19% vs.
39% and 43%). BMI documentation was significantly associated with a higher likelihood of counseling obese patients (p<0.01) but not referral to a dietician (p=0.61).
Conclusion: Sustaining a QI intervention in a resident primary care clinic to document BMI is challenging, especially due incoming PGY-1
housestaff who may not be familiar with current QI initiatives. Documenting BMI, however, is an important proxy for counseling obese patients.
Further cycles of change should include educating nursing staff and residents on the importance of obesity and BMI and evaluating barriers to documentation.
Implications for Policy, Delivery or Practice: The care of the obese patient begins at the identification of the problem. This study shows that an increase in the frequency of this necessary step is possible and relatively easy.
● Academic Primacy Care Clinician
Participation in the CMS Physician Quality
Reporting Initiative (PQRI): Turn Up the
Neda Laiteerapong, M.D.; Vineet Arora, M.D.,
M.P.P.; Andrew Davis, M.D., M.P.H.
Presented by: Neda Laiteerapong, M.D., Chief
Resident, Medical Education, MacNeal Hospital,
3249 South Oak Park Avenue, Berwyn, IL 60402,
Phone: (312) 613-2354; Email: nedalai@gmail.com
Research Objective: To describe a systems-based intervention to implement PQRI in an academic primary care practice and explore factors associated with clinician participation. The PQRI is a pay-for-reporting program that provides clinicians with a bonus if data is reported to CMS on at least three quality indicators for at least 80% of qualified visits. We selected three indicators centered on diabetes because of its importance in our population. To minimize workflow disruption, diabetes billing codes on standard charge tickets were supplemented with CPT Category II codes for hemoglobin A1c (HbA1c), low density lipoprotein
(LDL), and blood pressure (BP). Each code denotes a range within that indicator, e.g., 3044F represents HbA1c less than 7.0. The BP indicator requires systolic and diastolic blood pressures be reported. Quality codes were collected on all adult diabetics to improve sample size and provide clinicians information on their practice. Email feedback was provided and reinforced at monthly meetings.
Study Design: Data on clinicians, volume of diabetics, and quality codes was collected from
7/1/07 to 1/25/08. Primary care clinicians (PCCs) were identified as “high” or “low” volume using median number of diabetic encounters as the cutpoint. Data on gender and years since training were collected from the departmental website.
Correlations between reporting more than one indicator were calculated. One-sided t-tests were used to determine the association between volume and reporting for selected PQRI measures.
Population Studied: Academic primary care clinicians.
Principal Findings: All PCCs participated, including 2 nurse practitioners (41/41). 46.3% were male. The median years since training were 15 (4 to
41, standard deviation (SD)=8.3). The median volume was 117 (2 to 387, SD=107.2). 73.1% of clinicians qualified for bonuses. Reporting on one quality indicator was highly correlated to reporting others (HbA1c and LDL and BP, r=0.98 and 0.97, respectively; LDL and BP, r=0.98). High volume
PCCs were significantly more likely to report HbA1c
(79.7 vs. 90.4%, p=0.04) and LDL (76.5 vs. 87.6, p=0.04). A similar trend existed for high volume
PCCs and BP (77.4 vs. 87.0%, p=0.07). No significant differences existed in by gender or years since training. There was a nonsignificant trend towards an increased proportion of higher volume clinicians qualifying for bonuses than lower volume clinicians (85% high volume vs. 62%,low volume p=0.095).
Conclusion: We found that the CMS PQRI program was feasible to implement and well accepted in an academic primary care practice with over 70% of clinicians reporting successfully. The average bonus for 2008 is anticipated to be $450
(range $50-1100) per provider. Our experience highlights the value of reaching clinician consensus on measures, enlisting front-line support staff, and integrating data collection into existing workflow.
The workflow design likely led to high correlations in reporting of quality indicators.
Implications for Policy, Delivery or Practice:
Higher volume clinicians were more likely to qualify for bonuses, raising questions of whether they are better providers, more engaged in patient care activities, or were motivated by higher financial incentive to participate. Further work is needed to better understand this relationship in the context of future incentive programs.
● Nursing Unit Organization & Very Low Birth
Weight Infant Outcomes in Neonatal Intensive
Care Units
Eileen Lake, Ph.D., R.N., F.A.A.N.; Jeannette
Rogowski, Ph.D.; Jeffrey Horbar, M.D.; Douglas
Staiger, Ph.D.; Thelma Patrick, R.N., Ph.D.; Robyn
Cheung, R.N., Ph.D.
Presented by: Eileen Lake, Ph.D., R.N., F.A.A.N.,
Associate Professor, Center for Health Outcomes &
Policy Research, University of Pennsylvania School of Nursing, 418 Curie Boulevard, Philadelphia, PA
19104, Phone: (215) 898-2557; Email: elake@nursing.upenn.edu
Research Objective: Very low birth weight infants are among the highest-risk patient populations.
These infants are treated in NICUs where they are closely monitored and provided with an array of life support measures and intensive interventions.
Large variations in outcomes across NICUs currently exist that cannot be explained by differences in patient or NICU characteristics, such as volume or NICU level. Since NICU cases are among the most nurse-intensive hospitalizations,
nursing care may explain some of this variation and provide an opportunity for improving outcomes for these high-risk infants. In this first large-scale U.S. study of nursing effects on NICU outcomes, we examine whether variation in acuity-adjusted nurse staffing and environments contribute to the variation in NICU patient outcomes.
Study Design: In this observational study, we collected data in March 2008 on patient acuity and the practice environment via web survey of 6400 nurses working in 104 NICUs from the Vermont
Oxford Network (VON), an international quality improvement collaborative. The independent variables were the ratio of the observed to expected nurse-to-infant ratio for each unit, compositional features of the nursing staff (education, certification, experience), the professional practice environment and the hospital’s ANCC certified magnet status.
Infant outcomes were measured from the Vermont
Oxford Network Database for calendar year 2007
(N=8490). Outcomes were analyzed in randomeffects logit models which correct for the clustering within hospitals of patient outcomes. Patient characteristics, including gestational age, 1-minute
APGAR score, race, gender, and prenatal care were included as control variables.
Population Studied: Very low birth weight infants in neonatal intensive care units
Principal Findings: The average nurse cared for two infants, but this varied considerably by infant acuity. For the highest and lowest acuity infants, the average nurse-to-infant ratio was 0.95 and 0.34, respectively. The overall proportions of staff with bachelor’s degrees, neonatal specialty certification, and five or more years of nursing experience were
56%, 19% and 74% respectively. The sample of hospitals is exceptional in the large fraction (42%) with certified nursing magnet status based on achieving exemplary nursing standards, compared to hospitals nationally (6%). There was significant variation across units in both infant survival and infection, with average mortality rates of 11% and nosocomial infection rates of 17%. Preliminary analyses indicate that nurse staffing and staff composition variables are not able to explain this variation. A more professional practice environment appears to be associated with a lower likelihood of infection.
Conclusion: Nurse staffing and environment explains a small amount of the variation in VLBW infant outcomes across NICUs. After adjusting for multiple nursing factors, significant variation in outcomes across units remains. Future work should investigate the reasons for this outcome variation.
Implications for Policy, Delivery or Practice: Our research findings will support evidence-based management decisions to improve the quality of
NICU care by informing managers about staffing ratios and staff qualifications and identifying the features of practice environments that are most influential for infant outcomes.
Funding Source(s): RWJF
Theme: Child Health
● Common Characteristics of High-Impact
Investigator Initiated Research
Tim Lake, Ph.D., Kate Stewart, Ph.D., Tara Krissik,
M.P.P., Marsha Gold, Sc.D.
Presented by : Tim Lake, Ph.D., Senior
Researcher, Mathematica Policy Research, 600
Maryland Avenue, SW, Suite 550 Washington, DC
20024, Phone: (202) 484-4522 Email: tlake@mathematica-mpr.com
Research Objective : Health services research
(HSR) is a well-established and growing field.
However, the ways in which this research is used by policymakers are not well understood. AHRQ has historically been a key funder of investigatorinitiated HSR. As part of a larger evaluation of
AHRQ-funded investigator-initiated research on health care costs, productivity, organization and market forces, we developed case studies that described 7 AHRQ grant-funded studies that were used by policymakers and other end-users in various decision-making processes. The goal of was to understand the pathways through which research is disseminated and used to inform policy or managerial decisions.
Study Design : We reviewed grant applications, final project reports, selected journal articles and other relevant materials. We interviewed the principal investigators and key users of the findings.
We conducted approximately 5 telephone interviews for each case. Draft write-ups of each case study were shared with the respective principal investigators, with a request for feedback on the accuracy of stated facts.
Population Studied : Seven studies that appeared successful in terms of publications, other dissemination, and use by decision-makers. The studies were selected from a total of 149 AHRQ funded studies, and were identified through document review and discussions with AHRQ project officers. We selected the studies to achieve diversity in terms of grant size and topical focus; illustrative pathways for dissemination; and uses for private and public policy decision-making.
Principal Findings : The seven studies were funded between 1998 and 2006. The grants ranged in size from $203,000 to $5.3 million. Topics studied included managed care, patient cost-sharing, and physician payment methods. The grants illustrated multiple pathways for translating research into use for decision-makers, including (1) collaborating with intermediaries or end-users; (2) publishing in highly visible journals; and (3) targeting research and results to specific users in advance. Factors that helped PIs to disseminate their research included having established relationships with potential users, understanding how results might be used in
decision-making, and placing results in the relevant context.
Conclusions : There are multiple pathways through which research may reach policy audiences and inform policy discussions, and a variety of factors may contribute to successful dissemination and use. Studies published in highly visible journals often reach broader, non-research audiences through, in part, general media attention, although publication in these highly selective journals is not always possible. Producing alternative publications such as issue briefs, and developing relationships with potential end-users of research to help understand relevant issues and developing expertise in a particular area can greatly facilitate use by decision-makers.
Implications for Policy, Delivery or Practice :
Investigators and funding agencies interested in increasing the visibility of their research may want to consider some of the strategies of successful case studies outlined above. Potential additional strategies for funders such as AHRQ include establishing formal mechanisms to support dissemination of investigator-initiated research, such as facilitating contacts between policymakers and expert researchers, and developing broader syntheses of results from individual research projects.
Funding Source(s) : AHRQ
● Evaluation of Aging & Disability Resource
Centers: A Mixed Methods Approach
Glenn Landers, M.B.A., M.H.A.; Amanda Phillips
Martinez, M.P.H.
Presented by: Glenn Landers, M.B.A., M.H.A.,
Senior Research Associate, Georgia Health Policy
Center, Georgia State University, 14 Marietta
Street, Suite 221, Atlanta, GA 30303, Phone: (404)
413-0294; Email: glanders@gsu.edu
Research Objective: Aging and Disability
Resources Centers (ADRC) are promoted by the
Centers for Medicare and Medicaid Services and the Administration on Aging as models that make the long term service system more person-centered and consumer-directed, make it easier for people with disabilities of all ages to access information about home and community-based alternatives to institutional services, and support people of all income levels to live independently in their communities. ADRCs currently exist in 43 states.
The study had three main objectives: 1. Evaluate the implementation of three ADRC beta sites against the goals of the national ADRC program 2.
Evaluate the knowledge outcomes of ADRC stakeholders participating in a one-day workshop designed to increase interaction with regional
ADRCs. 3. Evaluate the fiscal impact of ADRCs on the provision of institutional and community based long-term care.
Study Design: For the beta site process evaluation, qualitative analyses including: 1. Review of program documents, including beta site proposals 2. Observation of stakeholder meetings
3. Site visits and semi-structured interviews with
ADRC stakeholders For the evaluation of knowledge outcomes: 1. Pre and post test of workshop objectives and outcomes at five sites using 16 Likert scale and true/false questions and two open-ended questions. For the fiscal impact analysis: 1. Retrospective analysis of 2003, 2004, and 2005 Medicaid claims data 2. Costs of institutional and HCBS long-term care were analyzed for counties with and without ADRC beta sites pre and post beta site implementation.
Population Studied: 1. 45 ADRC coalition members at three sites. 2. 305 workshop participants at five sites. 3. 233,356 consumers of
Medicaid long-term care services in calendar years
2003, 2004, and 2005
Principal Findings: For the beta site process evaluation: ADRC beta sites demonstrated high degrees of fidelity to the national ADRC goals of awareness, assistance, and access. Challenges included lack of common information systems across partner sites and integration of long-term care processes at the highest levels of state government. For the evaluation of knowledge outcomes: Knowledge outcomes related to ADRC,
Medicaid waivers, and the enhanced services resource database increased from an average of
24% excellent/good to 87% excellent/good. On a scale of zero to ten, 72% responded between eight and ten that they were likely to change their interaction with the local ADRC as a result of the their participation in the workshop. For the evaluation of fiscal impact: By 2005, 47% of
Medicaid long-term care consumers in ADRC counties received services through HCBS versus
45% of consumers in non-ADRC counties. Total long-term care costs in the 24 counties served by
ADRCs increased by two percent between 2003 and 2005 versus an increase of six percent in the counties without an ADRC. In 2005, had the percent of Medicaid consumers residing in institutions in ADRC counties mirrored the percent of consumers residing in institutions in non-ADRC counties, 319 additional consumers might have resided in institutional care in the ADRC counties in
2005. The savings from these consumers being in
HCBS totaled $411,469 per month or $4,937,628 in
2005.
Conclusion: Aging and Disability Resource
Centers can be successfully replicated or adapted in states, and the transition to their expanded use may be facilitated by educational programs targeted to public and private long-term care providers and consumers. Over time, ADRCs may facilitate transitions to HCBS, providing the opportunity for lower total Medicaid long-term care costs.
Implications for Policy, Delivery or Practice:
Aging and Disability Resource Centers may be a
viable solution for the delivery of long-term care information, referral, and assistance across all ages and disability types; however, their implementation is not without challenges, particularly the integration of aging and disability cultures. Implementation support that increases awareness of and knowledge about the purpose of ADRCs may improve the implementation process. Fiscal impact analyses are preliminary, and more rigorous fiscal studies should be undertaken to determine potential savings resulting from widespread adoption of the
ADRC model.
Funding Source(s): CMS, Administration on Aging
Theme: Long Term Care
● Evaluation of Medicaid Disease Management
Programs for the Aged, Blind & Disabled: A
Mixed Methods Approach
Glenn Landers, M.B.A., M.H.A.; Mei Zhou, M.S.,
M.A.
Presented by: Glenn Landers, M.B.A., M.H.A.,
Senior Research Associate, Georgia Health Policy
Center, Georgia State University, 14 Marietta
Street, Suite 221, Atlanta, GA 30303, Phone: (404)
413-0294; Email: glanders@gsu.edu
Research Objective: Medicaid disease management programs for the aged, blind, and disabled have historically achieved mixed results.
This study's first objective was to evaluate the clinical outcomes of aged, blind, and disabled
Medicaid consumers enrolled in two disease management programs in a large southern state.
The study's second objective was to evaluate consumers' self-reported health status and knowledge of chronic disease self-management and providers' knowledge and use of chronic disease management programs.
Study Design: Retrospective cohort analysis of
2006 and 2007 Medicaid claims data for aged, blind, and disabled consumers experiencing one or more of the following conditions or events: coronary artery disease (CAD), congestive heart failure
(CHF), diabetes, asthma, chronic obstructive pulmonary disease (COPD), schizophrenia, inpatient days, emergency department visits, and inpatient admissions. Consumers were divided into two groups as determined by whether they resided in the northern or southern region of the state and received disease management services by one of two private disease management firms assigned to the north or south. Clinical indicators were established for each condition, and baseline data were calculated for 2003/2004 (preimplementation). Additionally, 800 consumers (400 in each region) out of a population of 69,535 were surveyed at baseline (fall 2005) and at the end of performance year one (PY1: 2006/207) to determine overall self-reported health status and knowledge of chronic disease self-management.
444 providers out of a population of 3,055 participating in the disease management programs were surveyed at baseline (fall 2005) and at the end of performance year one (PY1: 2006/207) to determine knowledge and use of chronic disease management programs.
Population Studied: 69,535 consumers and 3,055 providers enrolled or participating in two Medicaid disease management programs 2004/2005 through
2006/2007.
Principal Findings: Clinical outcomes:
Both disease management programs achieved statistically significant differences across the following diseases: CAD, CHF, diabetes, and schizophrenia. One program also achieved a statistically significant difference in asthma and
COPD. Example indicators included: percent of
CAD members who received an influenza vaccine within the past 12 months, percent of members with heart failure who are taking spironolactone, and percent of members with diabetes who had at least two A1C tests in the measurement year. Both programs achieved statistically significant reductions in their respective populations' inpatient admissions and inpatient days; however, neither program achieved reductions in emergency department visits. Survey outcomes: Neither program's patient populations achieved statistically significant change in any of the questions related to self-reported health status or knowledge of chronic disease self-management except that one program's patient population rated their ability to get the services and support they need to manage their health lower in PY1. Only one program's participating providers achieved statistically significant positive change for two indicators: knowledge and understanding of the difference between disease management and case management and willingness to establish standing orders consistent with evidence based guidelines.
Conclusion: Patient clinical improvement along selected chronic disease indicators exceeded improvements in self-reported health status and management of chronic conditions. Programs achieved minimal improvement in provider knowledge and use of chronic disease management programs.
Implications for Policy, Delivery or Practice:
States often turn to disease management programs to reign in Medicaid expense. While lower Medicaid program cost may be the ultimate objective, the degree to which clinical outcomes are improved and disease managed are equally important. This study demonstrates just such clinical improvement; however, patients may not perceive the same degree of improvement or confidence in chronic disease self-management. Future disease management program design might consider how to improve patients' perception of program benefit.
Additionally, future research might explore the relationship among clinical outcomes and improvements in self-reported functioning.
Theme: Prevention and Treatment of Chronic
Illness
● Innovation in Simulating Lifetime Diabetes
Risk among Mexican-Americans Living along the US-Mexico: An Agent-Based Modeling
Approach
Holly Lanham, M.B.A.; Angela Winegar, M.S.; H.
Shelton Brown, III, Ph.D.; Jean-Paul Watson, Ph.D.;
Reuben McDaniel, Jr., Ed.D.; William Sage, M.D.,
J.D.
Presented by: Holly Lanham, M.B.A., Doctoral
Candidate, Information, Risk, & Operations
Management, The University of Texas at Austin,
3305 Glenview, Austin, TX 78703, Phone: (512)
471-3322; Email: holly.lanham@phd.mccombs.utexas.edu
Research Objective: The purpose of this research is to develop an agent-based model capable of describing the effects of alternative health policy strategies on the incidence and progression of diabetes mellitus (DM) among Mexican Americans
(MA) living along the US-Mexico border. DM incidence and its related complications are highest among MA living in low-income communities.
Existing computational models in health care focus largely on simulating physiological and pharmaceutical interventions. Our approach focuses on building models that are both openaccess and scaleable and that have the capacity to generate new insights about how behavioral, social and cultural factors shape population-level health outcomes.
Study Design: A comprehensive dataset with demographic and health information about residents living in south Texas was used to instantiate the agents. Model development also included developing rules for agent interactions.
Previously published studies, a large, comprehensive dataset collected in south Texas and a diverse and distinguished panel of subject matter experts informed model development. Three simple but plausible versions of a patient-centered medical home (PCMH) were compared for outcomes related to DM incidence and progression in our target population. We tracked DM incidence and progression under each of the PCMH policy scenarios over a 20 year time period.
Population Studied: A simulated community of MA adults, age 35-60, living along the US-Mexico border.
Principal Findings: The model generated discernable and interesting differences among the three policy scenarios. For example, under
Alternative 1 (high reimbursement scenario), diagnosed DM rose initially because lower cost medical visits gave more people access to medical care and increased the likelihood of diagnosis.
Under this same scenario, more individuals were diagnosed with complications but the number of people who died from DM was lower compared to baseline (14% vs. 9%).
Conclusion: We developed an agent-based model that evaluates three versions of a PCMH for their influence on the incidence and progression of DM among MA living along the US-Mexico border.
Well-designed agent-based models are powerful decision support tools that can rapidly and inexpensively generate useful insights. Using this approach, policy scenarios aimed at improving health outcomes can be tested and evaluated against each other. Interesting new policy and research questions also emerged from the analysis.
Implications for Policy, Delivery or Practice:
This research seeks to improve the policy decision making affecting the health of individuals living along the US-Mexico border by simulating pathways to improving the quality and costeffectiveness of health care delivery in these communities. This method is applicable in a variety of health care service delivery contexts. Agentbased modeling allows one to evaluate how changes in health care system design can affect decisions available to individuals and institutional community members (e.g., health care providers), the consequences of choices that are made as decision makers interact with each other, and the health outcomes that emerge from these interactions for both individuals and populations over extended periods of time.
Funding Source(s): The University of Texas and
Sandia National Laboratories
Theme: Prevention and Treatment of Chronic
Illness
● Age Differences in Met & Unmet Long Term
Care Needs
Mitchell LaPlante, Ph.D.
Presented by: Mitchell LaPlante, Ph.D., Professor,
Institution for Health & Aging, University of
California, 3333 California Street, San Francisco,
CA 94118, Phone: (415) 502-5214; Email: mitch.laplante@ucsf.edu
Research Objective: To compare access of working age persons and older persons to personal assistance with I/ADL, including informal and formal help and personal attendants, hours of help received, met and unmet needs, adverse consequences of unmet needs, and satisfaction with help.
Study Design: Statistical comparison of persons ages 18-64 and 65 and older using data from a large national survey and multivariate models
(multinomial and logistic) models controlling for need variables including severity of disability and diagnoses, and other predisposing and enabling variables.
Population Studied: All adults in the US living in households needing help with IADL or ADL, of whom half are 18-64 years old and half are ages 65
and older. Data are from the 1994-1997 National
Health Interview Survey on Disability.
Principal Findings: Of all adults who need help with I/ADL, controlling for need variables, older persons are twice as likely as working age persons to pay for help out-of-pocket and to receive help paid by private or public programs (OR=3.2, p<.001). About 18 percent of older persons have unmet needs, however working age persons have a higher rate (OR=1.4, p<.001), and for most measures analyzed, greater incidence of adverse consequences and higher levels of dissatisfaction with the amount and quality of help received. The major barrier appears to be that working age persons are less likely to be able to afford purchasing paid personal assistance.
Conclusion: Younger and older adults living in households who need help in I/ADL express substantial problems getting their needs met, but working age persons have more problems.
Implications for Policy, Delivery or Practice:
Advocates for younger adults with disabilities maintain that access to personal assistance services is essential for self-determination, which has become codified in the Americans with
Disabilities Act and the United Nations Convention on the Rights of Persons with Disabilities. However, these goals apply to persons of all ages. National policy should continue to identify and implement cost-effective ways to provide greater access to paid help for adults with disabilities of all ages in order to reduce unmet need, adverse consequences, and dissatisfaction and to improve self-determination. The results especially call attention to the problems working age persons have in getting their long term care needs met. Key national policy proposals include the Community
Choice Act to expand Medicaid access to paid personal assistance and the Class Act to establish a social insurance program for long term care, both of which would apply to adults of all ages. The results support the need for alternative service delivery and practice models, such as Medicaid
Cash and Counseling, than can better address unmet needs of working age persons. Future research should address the extent that preferences for paid personal assistance differ by age and identify specific barriers to financing paid personal assistance.
Funding Source(s): National Institute on Disability and Rehabilitation Research
Theme: Long Term Care
● The Primary Language Spoken at Home &
Disparities in Medical & Oral Health, Access to
Care, & Use of Health Services in U.S.
Adolescents
May Lau, M.D., M.P.H.; Hua Lin, Ph.D.; Glenn
Flores, M.D.
Presented by: May Lau, M.D., M.P.H., Assistant
Professor, Pediatrics, University of Texas
Southwestern Medical Center, 5323 Harry Hines
Boulevard, Dallas, TX 75390-9063, Phone: (214)
648-3717; Email: may.lau@utsouthwestern.edu
Research Objective: 7.5 million US adolescents speak a non-English primary language (NEPL) at home. Little is known about whether healthcare disparities exist for NEPL adolescents. The goal of this study was to examine disparities in medical and oral health, access to care, and use of health services in adolescents from non-English primary language households (NEPL).
Study Design: The 2003-2004 National Survey of
Children’s Health was a phone survey of a nationally representative sample of parents of children 0-17 years old. Bivariate and multivariable analyses were performed for all adolescents
(n=48,742) 10-17 years old, to identify disparities in
40 measures of medical and oral health, access to care, and use of services. Among NEPL adolescents, sub-analyses compared Latinos (Ls),
Asian/Pacific Islanders (APIs), and whites (Ws).
Population Studied: A nationally representative sample of US NEPL adolescents.
Principal Findings: Significantly (P<.05) higher proportions of NEPL (vs. EPL [English as the primary language spoken at home]) adolescents had suboptimal health status (47% vs. 14%), no health insurance (32% vs. 7%), no usual source of medical care (42% vs. 15%), no preventive medical care in the past year (44% vs. 26%), no dental insurance (56% vs. 80%), never seen a dentist (5% vs. 1%), no routine preventive dental visit in the past year, and teeth in fair/poor condition (29% vs.
8%). Among NEPL adolescents, compared with
Ws, significantly higher proportions of Ls were uninsured (35% vs. 20%), had no usual source of medical care (46% vs. 12%), were in poor/fair health (15% vs. 0.2%), had poor/fair oral health
(33% vs. 3%), and had never visited a dentist (5% vs. 2%). In multivariable analyses adjusting for 8 covariates, NEPL adolescents had significantly greater adjusted odds vs. EPL adolescents of suboptimal medical and oral health status, no medical and dental insurance, no usual source of care, and no medical, mental health care, or dental visit in the last year. Compared with NEPL Ws,
NEPL Ls had greater odds of suboptimal medical health status (odds ratio [OR], 4.6; 95% confidence interval [CI], 2.0–10.6), suboptimal oral health status (OR, 2.6; 95% CI, 1.4–4.9), and bone/joint/muscle problems (OR, 7.4; 95% CI, 1.5–
35.3), whereas NEPL APIs had greater odds of no medical visit in the last year (OR, 2.9; 95% CI,
1.02–8.0), and NEPL Latinos (OR, 4.8; 95% CI,
2.3–10.1) and APIs (OR, 3.3; 95% CI, 1.9–10.1) had greater odds of no usual source of care.
Conclusion: Compared with EPL adolescents,
NEPL adolescents experience multiple disparities in medical and oral health, access to care, and use of health services. Among NEPL adolescents, Latino
and API adolescents experience several specific disparities.
Implications for Policy, Delivery or Practice:
Elimination of health and healthcare disparities for
NEPL adolescents will require addressing financial, access, and language barriers to care.
● Racial/Ethnic Disparities among U.S.
Adolescents in Medical & Oral Health, Access to
Care, & Use of Health Services
May Lau, M.D., M.P.H.; Hua Lin, Ph.D.; Glenn
Flores, M.D.
Presented by: May Lau, M.D., M.P.H., Assistant
Professor, Pediatrics, University of Texas
Southwestern Medical Center, 5323 Harry Hines
Boulevard, Dallas, TX 75390-9063, Phone: (214)
648-3717; Email: may.lau@utsouthwestern.edu
Research Objective: 14 million US adolescents are non-white, but not enough is known about racial/ethnic (R/E) adolescent health disparities.
The objective of this study was to assess R/E disparities in medical and oral health, access, and use of services in US adolescents.
Study Design: The National Survey of Children’s
Health surveyed a nationally representative sample in 2003-2004 by phone of parents of children 0-17 years old. We examined disparities in 40 measures of medical and oral health, access, and use of services for all adolescents 10-17 years old
(n=48,742). Bivariate and multivariable analyses were conducted for white (W), African-American
(AA), Latino (L), Asian/Pacific Islander (API), Native
American (NA), and multiracial (MR) adolescents.
Population Studied: A nationally representative sample of R/E minority adolescents in the US.
Principal Findings: Marked, significant (P<.05) disparities were noted for many measures: uninsurance rates were 6% for Ws, 24% for Ls, 9% for AAs, 8% for APIs, and 13% for NAs, and the proportions with a usual source of care were 88% for Ws, 65% for Ls, 75% for AAs, 85% for APIs, and
68% for NAs. After adjusting for 8 covariates in multivariable analyses, many disparities persisted for greater than one R/E minority group, including increased odds of suboptimal medical and oral health status, overweight, emotional difficulties, asthma, skin allergies, medical uninsurance, no usual source of care, not receiving all needed medical or dental care, unmet medical and oral health needs, problems getting specialty care, no medical or dental visit in the past year, no mental health visit in the past year, and difficulties getting prescriptions. Certain disparities were especially marked for specific R/E groups: for Ls, suboptimal health status; for AAs, suboptimal oral health status, skin allergies, and not receiving all needed prescriptions; for APIs, unmet medical needs due to health plan problems, unmet dental needs due to transportation and scheduling barriers, difficulties getting prescriptions, and no physician or mental health visit in the past year; and for NAs, overweight, having limited abilities, asthma, ADHD, hearing/vision problems, no medical or dental insurance, no usual source of care, not receiving all needed medical care, problems getting specialty care, and no preventive dental visit in the last year.
MR youth also had many disparities.
Conclusion: US adolescents experience many R/E disparities in medical and oral health, access, and use of services. Certain disparities are particularly striking for specific R/E groups and MR youth.
Implications for Policy, Delivery or Practice: The elimination of R/E disparities in US adolescents may require better disparities monitoring, and innovative, community-based interventions targeting specific disparities in medical and oral health, access to care, and use of services.
● Updating Medicare Payments: Comparing
Specialty & Primary Care
Miriam Laugesen, Ph.D.; Roy Wada, Ph.D.
Presented by: Miriam Laugesen, Ph.D., Assistant
Professor, Health Services, University of California,
Los Angeles School of Public Health, CHS 31-293A
Box 951772, Los Angeles, CA 90095-1772, Phone:
(310) 825-6958; Email: laugesen@ucla.edu
Research Objective: To determine whether primary care physicians are disadvantaged by the advice provided by the Relative Value Update
Committee RUC to the Centers for Medicaid and
Medicare Services CMS during the update process for the Resource-Based Relative Value Scale
RBRVS.
Study Design: We used evaluation and management services as a proxy for services provided by primary care physicians. Using a generalized ordered logistic regression model we estimate the effect of physician categories on the
CMS relative value unit decisions while controlling for confounding factors.
Population Studied: We assembled an original dataset of all payment decisions made by CMS on
Medicare RBRVS relative values between 1994 and
2006.
Principal Findings: Between 1994 and 2006, CMS has decreased relative work valuations 154 times, increased them 112 times, and made no changes
1744 times. After controlling for confounding factors, CMS was found to be more likely to reduce
AMA-RUC recommended work values for all seventeen major physician specialty categories than for evaluation and management services. This suggests AMA-RUC is more likely to recommended increases for physician specialties. Compared to evaluation and management, which is a proxy for primary care, CMS is more likely (P<0.01) to decrease anesthesia, audiology, cardiovascular, dialysis, endocrinology, indictable, neurology and neuro-muscular, physical therapy, and psychiatry codes. Pathology and laboratory (P<0.05) and
radiology (P<0.10) were also more likely to be decreased. The existing codes are more likely to be kept the same or be increased by the CMS than for newly introduced codes. There are also differences over time: the CMS was more likely to accept a recommended fee increase during the earlier years.
Conclusion: RUC is more likely to recommend lower relative work values for primary care physicians than for physician specialty services, when compared to the final decisions made by
CMS. CMS may be attempting to compensate for the tendency of RUC to make lower recommendations for primary care services.
Implications for Policy, Delivery or Practice:
Even small changes in the values of work RVUs have large implications for health expenditure. The incentives provided by the RBRVS affect the kinds of services physicians want to provide to patients, as well as specialty choices among the medical workforce in the United States. In 1989 the RBRVS was created to try to reduce income differences between primary and specialty providers. Although our research suggests that CMS is attempting to balance specialty and primary care payments,
Medicare should revisit the process used by CMS to determine payments under the RBRVS and consider bundled or other payment methods for paying primary care physicians.
Funding Source(s): AHRQ
Theme: Medicare
● Continuity of Public Mental Health Care for
Vulnerable Groups in the City of Amsterdam,
The Netherlands.
Steve Lauriks, M.Sc.; Onyebuchi Arah, M.D., Ph.D.
Presented by: Steve Lauriks, M.Sc., Researcher/
Ph.D. Student, Department of Social Medicine,
Academic Medical Center, University of
Amsterdam, Meibergdreef 9, P.O. Box 22700,
Amsterdam, 1000 DE, NL, Phone: +31 20 555-
5495; Email: slauriks@ggd.amsterdam.nl
Research Objective: Care coordination, continuity and client tracking are among the most challenging aspects of public mental health care (PMHC). They are challenging but essential because the target groups suffer a plethora of vulnerabilities such as homelessness, lack of income, psychiatric disorders and addictions. Healthcare providers involved in
PMHC often use registration systems that are incompatible with those of other providers and are sometimes hesitant to supply client-information to a third party given strict privacy laws and client confidentiality. The Municipal Health Departments are now responsible for PMHC coordination. To fulfill this task, the Municipal Health Department of
Amsterdam appointed a ‘field-director’ who inventorizes the status, goals and progress of
PMHC clients on multiple domains, and links with providers to obtain client-information. So far, the information gathered has only been used to coordinate care on an individual-level and little is known about the determinants and effects of care continuity/coordination within the PMHC system.
This study investigates the determinants of continuity of care for the vulnerable groups within the
PMHC system.
Study Design: We compiled the data gathered by the field-director on housing, income, debt, vocational activities, care-provider, psychiatric disorders, addiction, and demographics of vulnerable groups. Continuity of care was expressed as the quarterly period between two consecutive PMHC assessments. We used multivariate analysis of variance to quantify the adjusted influence of care-status, achievement of care goals set by the case-manager, psychiatric disorders and drug addiction on continuity of care.
Population Studied: 6437 clients of PMHC in the city of Amsterdam whose status on at least one care-domain was as-sessed at least once between the first quarter of 2006 and the first quarter of
2008.
Principal Findings: On average 715 (range 105 –
2146) new clients were registered per quarter.
About 20% of the clients was assessed only once within the observed period. Over 71% of the assessments were done one quarter after the initial assessment. The variance in continuity of care
(time between an assessment and the next) could be explained by housing status (F = 2.903, P =
0.000), income status (F = 2.267, P = 0.045), care status (F = 2.185, P = 0.026) and having psychiatric disorders (5.575 = F = 13.017, 0.000 = P = 0.018).
Conclusion: Once clients enter the PMHC-system, tracking them and keeping them in care seems successful. Continuity of care depends on having housing, income and care-status. Drug addiction and psychiatric disorders are associated with positive outcomes, perhaps because of more intensive care offered to those vulnerable clients.
Implications for Policy, Delivery or Practice:
Providing secure housing, income and a main care provider should be priorities in securing continuity of care for vulnerable clients within the public mental health care system. Addiction and psychiatric care could be useful starting points for the development.
Theme: Behavioral Health
● Discontinuous Public Health Insurance among
Low-Income Children & the Causal Link with
Access to Physician Services: An Instrumental
Variables Analysis
Shana Lavarreda, M.P.P., Ph.Dc.
Presented by: Shana Lavarreda, M.P.P., Ph.Dc.,
Senior Research Associate, University of California,
Los Angeles Center for Health Policy Research,
10960 Wilshire Boulevard, Suite 1550, Los
Angeles, CA 90024, Phone: (310) 794-2261;
Email: shana@ucla.edu
Research Objective: This project examines the causal link between discontinuous enrollment in
California’s Medicaid program or State Children’s
Health Insurance Program (MC/HF) on access to physician services, as measured by three indicators: 1) having a usual source of care, 2) having seen a doctor in the past year, and 3) having a delay in needed care.
Study Design: This study will use the 2007
California Health Interview Survey (CHIS 2007), which includes the relevant health outcomes, insurance, individual- and family-level variables.
Discontinuous insurance status is conceptualized as: 1)MC/HF coverage all year, 2) had MC/HF and became uninsured, 3) was uninsured and gained
MC/HF coverage, and 4) uninsured all year.
Contextual variables at the county level will augment the dataset from a variety of sources, and will be used as policy-level instruments in the regression models. Demographic, employment, and health status variables will be used as a set of constant independent predictors in each of three different models. Multivariate probit instrumental variable regression models will be used to estimate the causal effect of the four categories of insurance status on the three dichotomous outcome measures, including calculations of odds ratios and relative risks.
Population Studied: 4,249 children, ages 0-18, who either had current MC/HF coverage or were currently uninsured at the time of their CHIS 2007 interview.
Principal Findings: Preliminary results from an examination of reduced logit models using CHIS
2003 data show that, compared to having MC/HF all year, children who lost their public health insurance were two and half times as likely to report a delay in getting care (RR = 2.25; OR = 2.53, p =
0.000) and slightly less likely to have seen a doctor in the past year (RR = 0.93; OR = 0.59, p = 0.044).
Children who were uninsured all year fared the worst on all measures. The multivariate probit models that include contextual-level variables will likely support these findings, but will contribute the additional dimension of accounting for the endogeneity of the regressor of interest (insurance status) with the access and utilization outcome measures.
Conclusion: It is hypothesized that this study will indicate that losing MC/HF does indeed worsen access to physician services. Children who gained coverage, however, will likely have better outcomes than did children who were uninsured all year, indicating MC/HF enrollment has beneficial effects on access to care.
Implications for Policy, Delivery or Practice:
These findings will support claims that increased enrollment and retention of children in MC/HF will improve their access to physician services, indicating MC/HF expansion would positively impact children’s health. This study will demonstrate the important linkage between continuous enrollment in
Medi-Cal and Healthy Families and positive health outcomes for children.
Funding Source(s): California Program on Access to Care
Theme: Cverage and Access
● Enhancing Respiratory Infection Control in
High Risk Health Care Settings
Sherri LaVela, Ph.D., M.P.H., M.B.A.; Lauren
Rapacki, M.S.; Frances Weaver, Ph.D.
Presented by: Sherri LaVela, Ph.D., M.P.H.,
M.B.A., Research Scientist, Department of
Veterans Affairs, Center for Management of
Complex Chronic Care, 5th Avenue and Roosevelt
Road, Hines, IL 60141, Phone: (708) 202-5895;
Email: sherri.lavela@va.gov
Research Objective: To encourage good infection control practices and prevent/reduce the spread of influenza in veterans at high risk for respiratory complications and their health care workers
(HCWs).
Study Design: Health professional education program/intervention. Educational materials were distributed to HCWs and follow-up questionnaires were distributed in 2008 to assess outcomes.
Statistical analyses were primarily descriptive; bivariate analyses were used to compare outcomes for HCWs who received the educational intervention and those who did not.
Population Studied: HCWs who had direct contact with high risk patients (complex physical disabilities) in health care settings.
Principal Findings: Across 21 VA facilities nationwide, 511 HCWs received the educational intervention and 264 did not. The questionnaire response rate was 63%. The sample was comprised of 67% nurses, 20% therapists, and 13% practitioners working in inpatient (87%), outpatient
(10%), and home care (3%) settings. More HCWs who received the educational intervention (vs. did not) were vaccinated during the influenza season
(70% vs. 64%, p=0.09) and utilized antiviral medications for influenza (17% vs. 13%, p=0.17).
Among antiviral medication users, HCWs who received the intervention (vs. those who did not) indicated the following as appropriate conditions for use: to prevent HCWs from getting influenza (34% vs. 63%, p=0.004); during a community outbreak
(25% vs. 6%, p=0.01); and for treatment of influenza (48% vs. 29%, p=0.05). Overall, 41% of
HCWs worked with symptoms of a respiratory infection during influenza season. HCWs who received vs. those who did not receive the intervention worked fewer average days while symptomatic (7 vs. 14 days), p=0.05.
Conclusion: Tailored educational materials made readily accessible to HCWs positively affected infection control practices such as vaccination uptake and fewer days working with symptoms.
Other areas were uninfluenced, suggesting that education alone is not enough. Areas of limited infection control knowledge/behavior highlight room for improvement. For example, 41% of HCWs worked while symptomatic and the understanding of appropriate conditions for antiviral use is limited.
Implications for Policy, Delivery or Practice:
Educational efforts encouraging recommended respiratory infection control practices are helpful for
HCWs in working in high risk areas, but may not be sufficient alone. The findings provide useful information for clinicians/administrators to guide future strategies and suggest the need for enhanced efforts to implement infection control best practices in high risk areas.
Funding Source(s): VA
Theme: Health Care Workforce
● Chronic Health Conditions in Male Veterans with MS
Sherri LaVela, Ph.D., M.P.H., M.B.A.; Thomas
Prohaska, Ph.D.; Sylvia Furner, Ph.D.; Frances
Weaver, Ph.D.
Presented by: Sherri LaVela, Ph.D., M.P.H.,
M.B.A., Research Scientist, Center for Management of Complex Chronic Care, 5th Avenue and
Roosevelt Road, Hines, IL 60141, Phone: (708)
202-5895; Email: sherri.lavela@va.gov
Research Objective: Disability-related concerns may displace other health concerns which can increase the risk for physiological conditions. The objective of this study was to compare the prevalence of physiological conditions in male veterans with MS to that of non-MS male cohorts
(overall and 50+) and to identify variables associated with disease occurrence in male veterans with MS.
Study Design: Cross-sectional survey data was collected and compared with CDC Behavioral Risk
Factor Surveillance System national data. Bivariate comparisons were made by group and multivariate logistic regression was used to identify covariates associated with chronic diseases.
Population Studied: Male participants included: veterans with MS (n=1,142), general veterans
(n=31,500), and the general population (n=68,357).
Principal Findings: Nearly half of male veterans with MS had high cholesterol (HC) (48.50%) and hypertension (HBP) (47%), followed by diabetes
(16%), CHD (11%), asthma (8%), stroke (7%), and prostate cancer (6%). HBP and stroke were more prevalent in male veterans with MS than in general veterans. Diabetes, HBP, HC, stroke, and prostate cancer were more prevalent in veterans with MS than the general population. For those aged 50 years and older, higher prevalence of diabetes,
HBP, and HC remained significant compared with the older general population (and stroke remained higher compared with both older non-MS groups).
Conclusion: With the exception of asthma, condition prevalence was higher in male veterans with MS than in general veterans (HBP and stroke) and the general population (diabetes, HBP, HC, stroke, and prostate cancer). Variables associated with occurrence are similar to what literature reports, e.g., dynamic associations between cardiometabolic and cerebrovascular conditions and the impact of age, education, and race for select conditions. Nearly half of male veterans with
MS had HC and HBP; these conditions occurring alone or concomitantly are implicated in numerous other diseases, making them important to address.
Although a lower proportion of males in general are affected by stroke; it is a concern due to consistent disparities across age groups independent association with the highest odds of having other conditions (asthma, diabetes, and HBP).
Implications for Policy, Delivery or Practice:
Research in this area is relatively scarce and these findings help bridge a gap in the MS literature. The study results identified chronic disease priorities and highlighted areas of unmet need in male veterans with MS that may be targeted for interventions to improve health and reduce disparities.
Funding Source(s): UIC, Paul Q. Peterson Public
Health Doctoral Scholarship
Theme: Prevention and Treatment of Chronic
Illness
● Immunization Information Systems as a
Resource to Improve Health Plan’s Efficiency &
Quality Measures
Christine Layton, Ph.D., M.P.H.; Alan O'Connor,
M.B.A.; Bobby Rasulina, Ph.D.; Todd Osbeck,
M.M.; Therese Hoyle
Presented by: Christine Layton, Ph.D., M.P.H.,
Public Health Researcher, Health, Social &
Economics Research, RTI International, 234
Homewood Terrace, Baltimore, MD 21218, Phone:
(410) 467-2907; Email: layton@rti.org
Research Objective: To assess the use of
Immunization Information Systems (IIS) as a single point of data entry for use in managed care organization’s (MCO): (1) physician incentive program, (2) internal quality tracking program, and;
(3) external quality reporting program (HEDIS).
Study Design: A business case study of a managed care organization (MCO). Data were collected through key informant interviews and from the MCO’s data systems. Immunization coverage rates and MCO quality measures from 2004 through 2007 were simulated with and without the
IIS as a data source for performance measurement and attainment towards incentive benchmarks. The study included a cost-benefit-analysis.
Population Studied: The study focused on a
Michigan-based MCO with 480,000 members and
Michigan’s IIS known as the Michigan Care
Improvement Registry (MCIR).
Principal Findings: IIS data increased MCOobserved immunization rates from 6.49 to 54.13 percentage points for childhood immunizations and
57.63 to 77.97 percentage points for adolescent immunizations. The HEDIS administrative rate for childhood immunizations doubled from 43.38% to
80.89% in the year following IIS data acceptance.
While the number of charts reviewed for HEDIS declined from 582 in 2003 to 51 in 2007, the most significant source of savings was in administration of the MCO’s physician incentive program which saw 18,881 fewer chart reviews over a 4 year period. Total costs were estimated to be $14,318 and net benefits were $107,854, corresponding to a benefit-to-cost ratio of $8.06.
Conclusion: The IIS provides the MCO three benefits: (1) immunization data which are more reliable, timely, and comprehensive, (2) a singe point of entry for immunization data resulting in a need for fewer chart reviews for HEDIS and the provider incentive program, and; (3) enhanced transparency in the MCO’s physician incentive and internal quality ratings programs. In addition, the
MCO’s use of IIS data leverages a state mandate for entry of all immunizations into the MCIR. We found that the MCO accrued cost saving benefits which amounted to a 310% annual return on investment. Benefits in the form of cost savings and improved immunization rates were seen with the
MCO’s application of IIS’ data. IIS data also increased assurance that MCO members were up to date on immunizations and permitted providers to quickly determine the immunization status of their individual patients.
Implications for Policy, Delivery or Practice: IIS and MCOs are a natural partner in which each can strengthen the other’s ability to serve as a data resource for quality improvement. Health plans which take advantage of IIS may accrue cost savings as well as improvements in quality measures.
Funding Source(s): CDC
Theme: Quality and Efficiency: Organized Policy
● Quality Measurement in Diabetes Care
Brian Leas, M.S., M.A.; Bettina Berman, R.N., B.S.;
Kathryn Kash, Ph.D.; Albert Crawford, Ph.D.;
Richard Toner, M.S.; Neil Goldfarb, B.A.
Presented by: Brian Leas, M.S., M.A., Project
Manager, Jefferson School of Population Health,
Thomas Jefferson University, 1015 Walnut Street,
Suite 115, Philadelphia, PA 19107, Phone: (215)
955-4689; Email: brian.leas@jefferson.edu
Research Objective: The quality of diabetes care is assessed in multiple ways by a broad variety of stakeholders. These efforts may have become too extensive, hindering their use and impact while overlooking important measurement gaps. The objectives of this study were to delineate and categorize current quality measures addressing diabetes; identify gaps that are not adequately addressed by these measures; examine barriers and challenges to new ways of assessing performance; and explore future directions in quality measurement that are likely to impact physicians and diabetes care.
Study Design: An environmental scan of diabetes quality measures was conducted. Measures were selected from the National Quality Measures
Clearinghouse, and from quality indicators promulgated by leading measurement organizations. Key informant interviews were completed with thought leaders who develop, promote, and use quality measures.
Population Studied: Leaders of prominent organizations were selected for the interviews.
Interviewees included representatives from programs that develop and endorse measures, and other stakeholders in performance assessment.
Physician specialty societies, insurers, and employers were included.
Principal Findings: One hundred forty six distinct measures spanning 31 clinical processes or outcomes were identified. Many of these are redundant and inconsistent, including 21 measures of hemoglobin A1c and 22 measures of lipid profiles. Interviewees believe that current diabetes measurement efforts are excessively broad and complex. They call for better harmonization of these measures. Several gaps were also found, including measures focusing on population health, healthcare disparities, the uninsured, access to care, structural elements of healthcare, and prevention of disease.
Conclusion: Diabetes quality measurement is characterized by a vast array of measures that can lead towards better care and outcomes. Use of these indicators is hindered in part by their breadth and variety. Important gaps in measurement also require the attention of physicians, researchers and policymakers.
Implications for Policy, Delivery or Practice:
Streamlining and harmonizing measures is an important challenge that must be addressed to encourage wider use of performance assessment and lead towards improved patient care and outcomes. Developing approaches to remedy major gaps in measurement can also drive improvements in care.
Funding Source(s): National Changing Diabetes
Program of Novo Nordisk, Inc.
Theme: Quality and Efficiency: Measurement
● The Hospital Information Technology
Adoption with the New Investment Index
Jinhyung Lee, M.S.; Yoonho Choi, M.S.
Presented by: Jinhyung Lee, M.S., Research
Assistant, Health Policy & Management, University of Minnesota, 1029 29th Avenue, Southeast #A,
Minneapolis, MN 55414, Phone: (612) 298-9397;
Email: leex2474@umn.edu
Research Objective: The factors which influence the adoption of Health Information Technology
(HIT) have been examined by few studies. But, even these few have not considered that there is interdependence among HIT adoption decision, which could lead to biased results because of possible correlation between concurrently installed
HITs. Thus, to take into account this interdependence among HIT adoption decisions, this study develops a new variable, Clinical
Information Technology Index, which is utilized to investigate the factors that influence adoption of
HIT in hospitals. Also, this results are compared to the outcomes of utilizing only one HIT,
Computerized Physician Order Entrey (CPOE).
Study Design: Clinical Information Technology
Index (CITI), including nine HITs, consists of four levels based on the diffusion of HIT. To examine the factors that influence the adoption of HIT in hospitals, ordered logit model is applied.
Population Studied: This is a cross sectional study. Healthcare Information and Management
Systems Society (HIMSS) and Office of Statewide
Health Planning and Development (OSHPD) are collected in 2004 and applied in this analysis. A total of 303 hospitals were matched from both data sets. While HIMSS provides data about HIT software applications in each hospital, OSHPD provides data about the characteristics of hospitals and the market.
Principal Findings: More independent variables reach significance with CITI as opposed to with only one HIT (CPOE). When CITI is utilized to analyze the data, hospitals with more beds are less likely to adopt clinical information technologies and those with higher managed care penetration and more
Medicaid patient are more likely to adopt clinical information technology. Also, hospitals in the more competitive market are more likely to adopt clinical information technologies. On the other hand, the average length of stay is only associated with the
CPOE adoption.
Conclusion: One HIT adoption at a hospital can influence subsequent adoptions and therefore adoption decisions within hospitals are not independent.
Using CITI with its nine HITs in the adoption decision model explains more factors influencing the adoption of HIT compared to using only one
HIT.
Implications for Policy, Delivery or Practice: HIT adoption decisions are interdependent each other so that researchers should avoid to use only one specific HIT.
● Developing Strategies to Promote Getting
Cervical Cancer Screening Through Focus
Group Interview with Middle Aged Women in
Korea
HeyJean Lee, M.D.; Heui Sug Jo, M.D., Ph.D.; Bo
Young Lee, R.N.; Yeon Hee Won; Jean Lee
Presented by: HeyJean Lee, M.D., Clinical
Assistant Professor, Preventive Medicine, Kangwon
National University Hospital, Hyoja3dong 17-1,
Chuncheon, Gangwon, 200-947, Phone: (823)
258-2178; Email: vishue@ewhain.net
Research Objective: To find out facilitators and barriers for getting cervical cancer screening
To develop practical alternatives to promote getting cervical cancer screening
Study Design: For the Focus Group Interview, 6 interview groups are organized according to age group(40’s, 50’s, 60’s) and the type of residence area(city, county). For each interview group, about
10 persons were recruited via telephone call through RDD with consideration about experience and intention for getting cancer screening.
Population Studied: Middle aged(40-65years) women who dwells in the certain city or counties of
Gangwon Province in Korea.
Principal Findings: 1) People have undergone cancer screening before but only a few have regular visit to physician for cancer screening. Most of participants have got cancer screening right after childbirth but it has not linked to regular cancer screening. 2) Position to get Pap smear make women feel embarrassed and reject screening.
Women favor female gynecologist rather than male and request slacks type clothes rather than gowns.
3) In Korea, all the women aged 30 years or older can get free cervical cancer screening every 2 years at the several assigned facilities. But this guideline and governmental support for cervical cancer patients are not well recognized to women.
4) Women are not satisfied to public cancer screening because it’s neither confidential nor comfortable. Some people go to general hospitals at the expense of themselves and some people ask for the refund of the cost for cancer screening. 5)
People are affirmative to remind call, postcard, diary to check up screening. People also ask for customized information and notice for cervical cancer screening.
Conclusion: It’s necessary to raise confidence and amenity for promoting public cervical cancer screening. Environments and conditions should be considered detailed at the point of recipient.
Physician’s regular check up and recommendation seems to facilitate people getting cervical cancer screening. It can be helpful to let female gynecologists do Pap smear or educate physicians not to touch thigh while doing Pap smear.
Implications for Policy, Delivery or Practice:
Guidelines and information about cervical cancer screening should be written in easy words and delivered repetitively through various materials.
It is necessary to consider refund for the cost of cancer screening rather than ambulatory screening facility to promote public cervical cancer screening.
Funding Source(s): Korean Ministry of Health,
Welfare, and Family
Theme: Consumer Choices in Health Care
● The Association Between Hospital Outcomes
& Diagnostic Imaging: Could More Be Better?
David Lee, Ph.D.; David Foster, Ph.D.
Presented by: David Lee, Ph.D., Senior Director,
Health Economics & Outcome Research, GE
Healthcare, 3000 South Grandview Boulevard, w-
460, Waukesha, WI 53005, Email: david.w.lee@ge.com
Research Objective: This study tests whether the
“more is better” hypothesis holds for diagnostic imaging by examining the association between the utilization of inpatient diagnostic imaging services and three key hospital outcome measures: mortality, length of stay and costs.
Study Design: We used patient-level data to construct two hospital-specific, risk-adjusted imaging utilization measures for CT, MR, ultrasound and x-ray—a binary indicator of whether the patient received the service, and an estimate of the mean number of services received (both adjusted for the patients’ demographic characteristics, severity level, discharge disposition of death, and for hospital characteristics). We also constructed three hospital-specific, risk-adjusted outcome measures from patient-level data: inpatient mortality; length of stay (LOS); and total inpatient hospital costs. We then assessed correlations between the imaging utilization measures and the outcome measures using Pearson correlation coefficients to test for statistical significance (p<0.05). The correlation analyses were weighted by hospital patient volume.
Population Studied: We examined 2007 data from the Thomson Reuters Hospital Drug Database for
1.1 million patients treated at 102 hospitals. The majority (53.7%) of patients were female; 32.8% were between 45 and 64 years old, with another
29.9% between 70 and 84 years old. The hospitals ranged from fewer than 200 beds (26.5%) to more than 500 beds (12.8%), and 10.8% were teaching institutions; the majority (70.6%) were in the
Southern region of the U.S.
Principal Findings: The risk-adjusted probability of receiving an imaging service had an inverse and statistically significant correlation with risk-adjusted mortality for all four imaging modalities, ranging from -0.20 (MR) to -0.24 (ultrasound). The correlation between the risk-adjusted mean number of imaging services received by patients at a hospital and that hospital’s risk-adjusted mortality score ranged from -0.07 for MR to -0.20 for ultrasound (the only statistically significant result).
Utilization of imaging services (both measures) was positively correlated with costs and length of stay.
None of the correlations between imaging and costs achieved statistical significance; however, associations between imaging utilization and LOS were statistically significant for the likelihood of receiving any ultrasound, the number of ultrasounds received and the number of x-ray services received.
Conclusion: These findings suggest that the increased use of imaging services is associated with lower mortality while at the same time having little or no impact on LOS or cost. Substantially more research is needed to examine whether the observed association is causal.
Implications for Policy, Delivery or Practice:
Because providers are only compensated for the professional component of inpatient diagnostic services, use of these services likely reflects a belief in their clinical benefit. This study highlights the need for further assessment of whether and how diagnostic imaging shapes key clinical, operational, and economic outcomes in the inpatient setting. This study may inspire others to explore this question with other data sources, imaging intensity measures that include the number and type of studies performed, and alternative statistical methods.
Funding Source(s): GE Healthcare
Theme: Quality and Efficiency: Policies and
Incentives
● Disparities in Forgoing Seeing a Doctor Due to Cost & Having a Consistent Health Care
Provider for Older Adults with Disabilities
Jae Chul Lee, Ph.D.; Denys Lau, Ph.D.; Rowland
Chang, M.D., M.P.H.
Presented by: Jae Chul Lee, Ph.D., Postdoctoral
Fellow, Institute for Healthcare Studies,
Northwestern University, 750 North Lake Shore
Drive, Chicago, IL 60611-3152, Phone: (312) 503-
5547; Email: jaelee2@northwestern.edu
Research Objective: To examine disparities in forgoing seeing a doctor due to cost and having a consistent health care provider for older adults with disabilities, and the extent to which the disparities may be accounted for by Andersen’s Socio-
Behavioral Model (predisposing, enabling, and health need factors).
Study Design: This study used data from the 2006
Behavioral Risk Factor Surveillance System
(BRFSS), a cross-sectional national telephone survey. Chi-square test was used to examine whether the disparities between older adults with and without disabilities existed, controlling for age and sex. A series of multivariate logistic regression analyses were performed step-wise by including predisposing (sex, age, race, education), enabling
(income, social and emotional support, health care coverage), perceived health need (general health status, physical health, mental health), and evaluated health need (five chronic health conditions) to determine their association with the outcome variables. All analyses accounted for the complex survey design of BRFSS.
Population Studied: The study sample included unweighted 91,005 individuals aged 65 and older in
50 states and the District of Columbia who had any activity limitation due to physical, mental or emotional problems, or used any special equipment due to any health problem.
Principal Findings: Even though a smaller proportion of older adults with disabilities had no consistent health care provider compared to their counterparts without disabilities (4.07%, vs. 6.58%), a higher percentage of older adults with disabilities forwent seeing a doctor due to cost (6.40% vs.
2.98%), after controlling for age and sex. Logistic regression analyses showed that older adults with disabilities were 2.23 times more likely (95% CI =
1.95, 2.56) to forgo seeing a doctor due to cost without adjusting for any predisposing, enabling, and health need factors. The unadjusted odds ratio decreased from 2.23 to 2.21 (95% CI = 1.92, 2.54) while adjusting for predisposing factors; to 2.10
(95% CI = 1.81, 2.43) while adjusting for enabling and predisposing factors; to 1.52 (95% CI = 1.30,
1.77) while adjusting for perceived health need, predisposing, and enabling factors; and to 1.47
(95% CI = 1.25, 1.72) while adjusting for evaluated health need and all other factors. Perceived health need factors decreased the odds ratio in forgoing seeing a doctor due to cost the most while adjusting for predisposing and enabling factors.
Conclusion: This study provided empirical evidence that although older adults with disabilities were more likely to have a consistent health care provider than those without disabilities, they were more likely to forgo seeing a doctor due to cost.
Perceived health need factors accounted the most for the risk of forgoing seeing a doctor.
Implications for Policy, Delivery or Practice: To address disparities in access to health services, our findings may inform the development of future policies to reduce financial hardships that may hinder older adults with disabilities from seeing their physicians. This information also suggests that health care reform must be cautiously designed to avoid any unintended consequences of making the disadvantaged population further vulnerable to difficulty accessing health care because of cost.
Funding Source(s): The contents of this presentation were developed under a grant from the
Department of Education, NIDRR grant number
H133P080006 (PI: Allen Heinemann, Ph.D.).
However, those contents do not necessarily represent the policy of the Department of
Education, and you should not assume endorsement by the Federal Government.
● Decomposition of Income-Related Health
Inequality by Socio-Economic Determinants &
Health Dimensions: A Case Study of South
Korea
Jun Hyup Lee, Ph.D.; Kook Hwan Rhim, Ph.D.;
Kyusuk Chung, Ph.D.
Presented by: Jun Hyup Lee, Ph.D., Department
Chair, Department of Healthcare Management,
Korea University, San 1, Jeongneung-Dong,
Seongbuk-Gu, Seoul, 136-703, KR, Phone: +82-2-
940-2875; Email: junhlee@korea.ac.kr
Research Objective: The concentration index has been frequently used as a measure of incomerelated inequality in health. The objective of this study is to examine the major sources behind health inequality among South Koreans. For this purpose, this study decomposes inequality in health, which is measured as the concentration index, into relative contributions from socioeconomic determinants.
Study Design: Data came from the 2005 Korea
National Health & Nutrition Examination Survey, based on a national representative cross-sectional sample of the total non-institutionalized population.
Income-related inequality in health was measured by the concentration index computed from the EQ-
5D score. The EQ-5D Index includes 5 health dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. As the
EQ-5D score is a sum of 5 dimensions, its concentration index can be decomposed into 5 components. The contributions of each of the 5 dimensions to income-related inequality as well as the overall concentration index for the EQ-5D score were estimated by multivariate regression analysis.
All explanatory variables in the regressions were categorical variables except income (log household income per capita): age (19-30,31-45,46-60,61-
70,71+), gender, marital status (unmarried, divorced/separated, married), educational level (9 school years, 10-12, 13+), activity status ( unemployed, house worker, student, employed), region (rural, urban).
Population Studied: 24,169 South Koreans age 19 or older were interviewed using the health status instrument, EQ-5D, as part of the Korea National
Health & Nutrition Examination Survey in 2005.
Principal Findings: The contributions to the overall predicted concentration index varied by determinant, in terms of both magnitude and direction (positive and negative signs). Larger contributing factors included younger age (aged 19-
30, 31-45), low income, and low education.
Meanwhile, even though older male and female
(aged 61-70,71+) were large contributors, they had negative sign.
Conclusion: When using the EQ-5D as a health status measure, the decompositions allowed for determinants to explain the sources of inequality in health. It was shown that the impact of sociodemographic and socio-economic variables varied considerably, but negative signs make difficult to interpret the contributions.
Implications for Policy, Delivery or Practice:
Public health policies aiming at reducing health inequality should be implemented to specific
groups, such as low income group or low educated group, which had larger effects.
● Comparison of Characteristics & Follow up of
Female Victims Between Adult and Underage at the Center for Female Victims of Sexual
Assaults in a Korean City
HeyJean Lee, M.D.; Hyang Ah, Lee, M.D.; Eun
Jung Soh, M.D.; Jong Yun Hwang, M.D.; Jun Sik
Cho, M.D.; Dong Heon Lee, M.D.; Heui Sug Jo,
M.D., M.P.H., Ph.D.
Presented by: HeyJean Lee, M.D., Clinical
Assistant Professor, Preventive Medicine, Kangwon
National University Hospital, Hyoja3dong 17-1,
Chuncheon, Gangwon, 200-947, KR, Phone:
+8233-258-2178; Email: vishue@ewhain.net
Research Objective: The purpose of this study is to compare characteristics and patterns of follow up between adult and underage victims at the Center for female victims of sexual assaults.
Study Design: We collected medical records and counseling sheets of female victims visited the
Center for female victims of sexual assaults in C city between February 2006 and July 2007. Based on the legal support people under 18 years old were grouped as underage and people as same as or older then 18 years old were grouped as adult.
We compared socio-demographic characteristics, type of assaults, and follow up on medical recommendations of female victims between adult and underage female victims of sexual assaults.
Population Studied: Female victims visited the
Center for female victims of sexual assaults in C city between February 2006 and July 2007.
Principal Findings: There were 39 underage female among 81 female victims visited the Center.
Underage female less visited facilities in comparison to adult female within 24 hours after sexual assaults. More of Underage female washed out vagina, bodies and lost physical evidence to investigate.
Underage female victims showed higher tendency to dwell in rural area. There were more recurrent cases among underage victims. More of assaulters were drunken, did sexual acts, companied threats or violence among adult victims. More of underage victims were decided to get psychiatric counsels.
Conclusion: Alternatives to prevent sexual assault should be considered based on the characteristics of victims and the physical and cultural environments of assaults. It’s necessary to develop guidelines to prevent sexual assault considering age.
Implications for Policy, Delivery or Practice:
Nowadays patrol for kids is newly organized to prevent sexual assault to underage at community level. It would be better to place the patrol for kids mainly in rural areas.
Funding Source(s): Gangwon Provincial Police
Agency
● Assessment of a Hepatitis Education Group
Serving Veterans in a Substance Use Disorders
Clinic
Tabitha Leighton, M.P.H.; Hildi Hagedorn, Ph.D.;
LeeAnn Heim, B.A.
Presented by: Tabitha Leighton, M.P.H., Research
Project Coordinator, Minneapolis VA Medical
Center, Department of Veterans Affairs, 1 Veterans
Drive (2E/152), Minneapolis, MN 55417, Phone:
(612) 629-7421; Email: tabitha.leighton@va.gov
Research Objective: Three years after the
Substance Use Disorders (SUD) clinic at the
Minneapolis Veterans Affairs Medical Center
(VAMC) launched the Healthy Liver Group (HLG), a survey took place to measure change in patient knowledge, participant satisfaction, and vaccination attitudes in those attending the educational group.
The HLG is an hour long intervention led by a SUD clinic nurse (RN). The first half of the intervention provides materials to convey basic information about liver health and hepatitis. The second half is devoted to individualized review of veterans’ lab results with the RN. Depending on the nurse’s recommendations, veterans have the immediate opportunity to receive the hepatitis A and/or B vaccines or be immediately scheduled for an intake into the hepatitis clinic.
Study Design: From August, 2007 to May, 2008
102 veterans who attended a HLG meeting filled out a survey regarding basic hepatitis facts and overall satisfaction with the educational group. The patients were asked to complete the first 11 questions before the group started and then answer the remaining questions after the group finished.
Population Studied: The Minneapolis VAMC houses an SUD clinic serving over two thousand veterans each year. The veterans seeking treatment are predominantly male, Caucasian, nearing fifty years of age, and most commonly have a diagnosis of alcohol dependence.
Principal Findings: A portion of the survey was designed to capture the participants’ hepatitis knowledge before and after the group. Questions asked basic information such as the organ hepatitis predominantly affects, risk factors, and prevention.
Before attending the HLG participants were on average able to answer 56.7% of multiple-choice hepatitis questions correctly. After attending the group, participants were able to answer on average
80.6% of questions correctly. This indicates the participants’ knowledge was considerably higher after attending the HLG. Before the group, participant’s rated their likelihood of accepting vaccination for hepatitis A and/or B if recommended. Afterwards a question addressed whether or not the vaccinations actually took place.
Eighty-five participants gave valid responses to both questions. 56% reported that they were somewhat or very likely to accept vaccination.
83.5% reported receiving a vaccination. Of participants that reported that they were somewhat or very unlikely to accept vaccination, 95.0% received a vaccination. These data suggest that knowledge gained during the group may have convinced these patients of the value of vaccination. The majority of attendees of the HLG
(79%) stated they were either “somewhat satisfied” or “very satisfied” with the group. The majority of attendees (74%) also rated the HLG as either
“somewhat helpful” or “very helpful.” The aspect rated as the most helpful was having the nurse available to explain personal lab results.
Conclusion: Overall, the results of this survey show the Healthy Liver Group yields a large increase in basic knowledge of hepatitis, high levels of patient satisfaction, and a strong acceptance of vaccinations, even among those patients that were unlikely to accept vaccination prior to attending the group.
Implications for Policy, Delivery or Practice:
Group educational meetings can be an effective way to increase hepatitis knowledge, screening, prevention and treatment referrals in clinics serving high risk patients.
Funding Source(s): VA
● The Length of a Genetic Counseling Session
Improves Patient Satisfaction but not
Information Recall or Psychosocial Factors
Barbara Lerner, M.S.
Presented by: Barbara Lerner, M.S., Pre-Doctoral
Fellow, Health Policy & Management, Boston
University School of Public Health, 9 Minola Road,
Lexington, MA 02421, Phone: (781) 771-8892;
Email: lerner@bu.edu
Research Objective: Genetic counseling involves exchanges of complicated biomedical and sensitive psychosocial information. In disclosing test results clinicians provide patient-specific susceptibility risk estimates, address individual and family related psychosocial concerns, and discuss follow-up for medical management options and additional services. A considerable amount of time may be required to accomplish these objectives. However, little is understood about the relationship between the amount of time devoted to a genetic counseling session and the quality of counseling. We examined the relationship between session length and selected patient outcomes in genetic counseling sessions in which genetic test results for
Alzheimer’s disease (AD) were disclosed. Although susceptibility testing for AD is not currently standard genetic counseling practice, it serves as a model for common diseases, such as diabetes and heart disease, with modest genetic risk factors for which testing will soon be available.
Study Design: Data are from a randomized controlled trial to assess the impact of providing genetic risk information to individuals with a family history of AD; 262 test result disclosure sessions were observed. We used multiple regression modeling to estimate the effects of session length on patient satisfaction immediately after the session, test result recall, and psychosocial factors
(i.e. anxiety, depression, distress and impact of test result disclosure) six weeks post disclosure.
Covariates include test result, provider type (genetic counselors vs. physicians), gender, and race.
Population Studied: Unaffected adult children of
AD patients were recruited at four medical centers across the United States: 21% were African-
American, 71% were female, and the mean age was 58.
Principal Findings: The mean session length was
15.5 minutes (range 2.1–55.12 min). Physicians took less time than genetic counselors (12 vs. 17 min, p<.001). Session length was not associated with psychosocial factors, test result recall, patients’ perceptions of their clinicians’ abilities to explain the test and its meaning, or their expectations that they could explain the test and its meaning. However, the longer the session, the more likely patients liked their clinicians (ß=0.139, p=0.03) and felt the clinician liked them (ß=0.148, p=0.02). Interestingly, the longer the session, the more likely patient felt that clinicians said or did something to help them feel both better and worse emotionally (ß=0.144, p=0.03 and ß=0.166, p=0.01, respectively).
Conclusion: Although quality of care factors such as patient psychosocial status and test result recall are not affected by the length of a genetic counseling test result disclosure session, some aspects of patient satisfaction are. These mixed results highlight the trade-offs in patient care quality when determining the appropriate length of time to allocate to these encounters.
Implications for Policy, Delivery or Practice:
Genetic susceptibility testing for common diseases is likely to increase in the coming years, exceeding the supply of professionally qualified counselors.
While traditional approaches have involved long counseling sessions, new models emphasizing brevity may be more efficient. These findings suggest that instituting shorter sessions may not result in decreased quality, albeit with diminished patient satisfaction.
● Quality of Care Results from the Medicare
Physician Group Practice Demonstration
Musetta Leung, Ph.D.; Michael Trisolini, Ph.D.,
M.B.A.; Sherry Grund, R.N.; Gregory Pope, M.S.
Presented by: Musetta Leung, Ph.D., Health
Services Researcher, Program on Health Care
Quality & Outcomes, RTI International, 1440 Main
Street, Suite 310, Waltham, MA 02451-1623,
Phone: (781) 434-1730; Email: mleung@rti.org
Research Objective: To measure quality performance among participating physician groups in their care of Medicare beneficiaries.
Study Design: This Medicare value-based purchasing initiative measures the quality and efficiency of care for ten large physician group practices (PGPs). The project uses both Medicare claims and clinical record data to measure 32 quality indicators in four disease topics (diabetes, heart failure, coronary artery disease and hypertension) and in preventive care. Quality indicators were developed by the Centers for
Medicare & Medicaid Services in conjunction with the National Committee for Quality Assurance and the American Medical Association. To date, baseline and second performance year results are available for the 10 diabetes (DM) measures, the 10 heart failure (HF) measures, and the 7 coronary artery disease (CAD) measures. Six of the 27 quality measures were calculated from claims with the option for the sites to improve their scores using clinical record data. Performance targets in the second performance year were based on 75% compliance, the 2004 mean Medicare HEDIS values reported by Medicare Advantage plans, the
70th percentile of the 2004 Medicare HEDIS level, or quality improvement targets set from baseline results.
Population Studied: Medicare fee-for-service beneficiaries assigned to the ten PGPs were included if they met quality measurement inclusion criteria of (1) having 2 or more evaluation and management visits to a physician’s office, and (2) meeting criteria for each of the three disease topics.
All beneficiaries with the relevant conditions were evaluated for the claims-based quality measures. A random sample of 615 beneficiaries was included for the clinical record based measures in each topic.
Principal Findings: The number of beneficiaries eligible for evaluation ranged from 527 to 5,868 depending on the PGP and the disease topic. Five of the PGPs achieved benchmark quality performance on all 27 quality measures, and all
PGPs met all of their quality targets in the HF and
CAD topics. Several DM quality measures targets, such as in blood pressure management, remained difficult to attain for a few sites, but the practices improved their quality scores in general. Compared to baseline results, the PGPs increased their quality scores an average of 9 percentage points across the DM measures, 11 percentage points across the
HF measures, and 5 percentage points across the
CAD measures. The PGPs’ performance on claims based measures also remained favorable when compared with secular trends in their comparison groups.
Conclusion: Second performance year results for the PGP Demonstration showed marked improvement in the comprehensive approach to care for patients with chronic conditions over baseline. Improvement from baseline was attributed to various interventions, including practice redesign, improved care management, and enhanced information technology. However, several diabetes quality measures remain challenging for some sites.
Implications for Policy, Delivery or Practice:
Quality of care measurement and reporting allows physician practices to evaluate their patterns of care for patients with chronic conditions, improve care delivery, and track improvements over time.
Paying for increased quality of care may provide additional resources or incentives for physician groups to implement interventions to improve care.
Funding Source(s): CMS
Theme: Quality and Efficiency: Measurement
● Are Patients who Receive CT Scans Young
Enough to Face a Significant Increase in
Lifetime Cancer Risks from Radiation
Exposure?
Rebecca Lewis, M.P.H., Mythreyi Bhargavan,
Ph.D.,
Presented by : Rebecca Lewis, M.P.H.,
Researcher, Research Department, American
College of Radiology, 1891 Preston White Drive,
Reston, VA 20191, Phone: (703) 295-6771 Email: rlewis@acr.org
Research Objective : Recent increases in the use of CT have led to concern regarding the risks to patients due to the cumulative radiation exposure during a patient’s life time. If patients have CT scans when they are relatively young, particularly scans with high radiation doses, they may have a significant increase in their lifetime risk of cancer from the radiation exposure. The objective of this study is to identify patterns in distribution of CT scans by age, gender, and type of CT.
Study Design : We used a large national employer plan (LNEP) of physician claims for 2003, which covered approximately 4.7 million lives. We analyze claims for CT utilization, by body part
(chest, abdominal, pelvic, spine (lumbar, cervical and thoracic), head/neck/brain, CT angiography and other), and calculate the number of scans and number of patient receiving scans, overall and by age and gender. Likewise, we calculate the number of scans and number of patients receiving scans, overall and by age and gender, for those patients who received CT within 6 months, 12 months and 36 months of date of death. Data were weighted to make the age distribution representative of the U.S. population, using estimates of population size for 2003 from the U.S.
Census.
Population Studied : Large national employer plan population (LNEP) for 2003.
Principal Findings : Age distribution of CT utilization is similar across body parts. Overall persons age 75-84 receive the highest number of
CTs, but for age 45-74, the proportion of CT scans exceeds the percent of U.S. population of that age group. Overall, women receive higher numbers of
CTs than men, but at ages 55-84, men receive
higher numbers of CTA (computed tomographic angiography) than women. Analysis of end-of-life data reflect that a relatively small percentage of patients have scans within 6 months of date of death, both over all ages (2.7%), and for patients age 65 and older (1.9%.) For those patients who had scans within 12 months of date of death, the percentages over all ages (3.6%) and for those age
65 and older (2.5%) were slightly higher but still under 5%. For those patients who had scans within
3 years of date of death, the percentage over all ages was 5.9% and for those age 65 and older was
4.1%.
Conclusions : Older patients receive a disproportionately large portion of CT scans, but the disproportionality begins at age 45. Patients at age
45-54 have life expectancies of 20 years or more during which cancers may develop. For these patients, risks from radiation exposure must be carefully evaluated against the benefits of the scan before the test is ordered. Only a small fraction of
CTs are performed at the end of patients’ lives when the benefits from the CT may vastly exceed the risks.
● Technology Adoption for Diagnosis of
Pulmonary Embolism in a Medicare Population
Rebecca Lewis, M.P.H., Mythreyi Bhargavan,
Ph.D., Jonathan Sunshine, Ph.D., Saurabh Jha,
M.D., Jean Owen, Ph.D.
Presented by : Rebecca Lewis, M.P.H.,
Researcher, Research Department, American
College of Radiology, 1891 Preston White Drive,
Reston, VA 20191, Phone: (703) 295-6771 Email: rlewis@acr.org
Research Objective : Pulmonary embolism (PE) is a common pulmonary vascular emergency. When the diagnosis of acute pulmonary embolism is confirmed and treated, mortality and chances of recurrence are low. Results from a large multicenter trial (PIOPED II) published in 2006 found CT angiography of the chest to be the definitive diagnostic tool in the diagnosis of PE, and this replaced the previous guideline of using ventilationperfusion scans for this purpose. The objective of this study is to identify whether clinical practice among Medicare patients is consistent with recommendations in the academic literature, and to explore whether there are variations based on place of service, physician specialty, or patient characteristics.
Study Design : Using the 2005 Medicare 5%
Research Identifiable Files (RIF) claims data, we identify the tests most commonly used to diagnose or rule out PE, or rule out alternative diagnoses for similar symptoms. For two groups of patients: those with a diagnosis of PE or PE-related symptoms, and those identified with PE on an inpatient stay/admission, we calculate the distribution of tests used overall and as a first line of investigation, by place of service and by physician specialty. Logistic regression analyses are used to measure the effect of patient characteristics, physician specialty, and place of service on receipt of the five most common diagnostic tests for PE.
Population Studied : Medicare enrollees in 2005
Principal Findings : Just over 2% of patients who might potentially be suspected of having PE are actually found to have PE. For patients for whom
PE may be suspected , the most common tests were CT or CTA of the chest, echocardiograms, perfusion only scans, and duplex scans. For all groups of claims studied, the vast majority of patients had a chest X-ray. For approximately a third of the patients with PE, there is no evidence that they had any of the tests recommended for diagnosis of PE. These patients were likely diagnosed based on clinical findings. For patients with symptoms that may be related to PE, logistic regression results show that patients seen first by an emergency physician are more likely to have an echocardiogram or perfusion-only study than patients first seen by a generalist; rural patients are less likely than urban patients to have any of the tests. For patients with PE, there are fewer significant findings: patients seen in ER and inpatient settings have higher odds of having tests than patients seen only in the inpatient setting for
PE.
Conclusions : Physicians appear to be adopting findings from the literature in their practice even before they were formally released. Even before the release of findings from PIOPED II, CT was in extensive use and VQ was used relatively little.
Implications for Policy, Delivery or Practice :
Because clinical practice often fails to adopt widelypublicized guidelines it would be valuable to learn how and why common practice was so progressive in this instance.
● American College of Chest Physicians
Evidence-Based Guideline Methodology
Sandra Zelman Lewis, Ph.D.
Presented by: Sandra Zelman Lewis, Ph.D.,
Assistant Vice President, Health & Science Policy &
Quality Improvement, American College of Chest
Physicians, 3300 Dundee Road, Northbrook, IL
60062, Phone: (847) 498-8322; Email: slewis@chestnet.org
Research Objective: It is optimal, and will soon be required, for performance measures to be based on rigorously developed guidelines that meet the criteria of the Physician Consortium for
Performance Improvement (PCPI). The recommendations must be graded utilizing a simple, user-friendly system that rates evidence and recommendations based on both the balance of benefits and harms and the quality of the underlying research. Resource allocation and
patient values and preferences should be incorporated to produce patient-centered, costsensitive recommendations. Conflicts of interest policies must be strictly enforced. The guideline methodology should be replicable.
Study Design: The ACCP guideline methodology will be reviewed for feasibility, usability, patientfocus, and assessment relative to the AGREE instrument. Additionally, Web hits and literature citations will be reported for the three methodology articles.1-3
Principal Findings: The American College of
Chest Physicians (ACCP) has developed a precise, evidence-based, methodology1 for the development, dissemination, and implementation of clinical practice guidelines. Two working groups were convened to develop (1) a grading system2 that would be user-friendly and consistent across all
ACCP guidelines; and (2) methods of incorporating resource allocation3 and patient values and preferences into guideline development. Nominated panelists are screened for conflicts of interest and not seated or removed if they have conflicts that cannot be managed. ACCP’s standardized methodology has been cited in the literature and other guideline developers have adapted it for their purposes. The Institute of Medicine and Agency for
Healthcare Research and Quality have expressed interest in the process. The methodology and grading scheme are quite versatile and have been successfully used for a variety of clinical diagnostic, preventive, and therapeutic topics, eg, lung cancer, antithrombotic therapy, and pulmonary rehabilitation. However, the same methodology and grading system have also been implemented in the development of a non-clinical guideline assessing the effectiveness of continuing medical education, including simulation education. Several implementation projects are underway using these guidelines.
Conclusion: The ACCP protocol is a rigorous and systematic process for development of graded evidence-based guidelines that facilitates patientcentered and resource-considerate care.
Implications for Policy, Delivery or Practice:
This replicable methodology can support evidencebased guideline development efforts of other developers and meets or exceeds the PCPI standards for methodological rigor.
Theme: Organizational Performance and
Management
● Coordinating Care for Patients with Complex
Chronic Diseases in Primary Care
Rachel Lewis, R.N., B.A., M.A.; Dr. Martin Charns,
Ph.D.
Presented by: Rachel Lewis, R.N., B.A., M.A.,
Community Matron (UK) Harkness Fellow, Center for Organization, Leadership, & Management
Research, VA Boston, 150 South Huntington
Avenue, Boston, MA 02130, Phone: (857) 364-
5686; Email: rachel.lewis5@comcast.net
Research Objective: The purpose of this study was to evaluate the role of a community matron in moving services closer to home for older patients with advanced, co-morbid conditions. This was done through fostering a unique partnership between specialist/tertiary services and primary care, and through enhanced coordination. Aims of the intervention were to: reduce the disease burden on patients by coordinating services to meet individual needs closer to the patient’s home; coordinate primary and secondary services to reduce duplication and omissions of care; provide ongoing monitoring with appropriate and timely intervention in primary care reducing hospital admissions and/or prolonged inpatient stays; preemptive prioritization of needs and timely referral to appropriate clinician/service e.g. community geriatrician, dietician, palliative care; liaise closely with specialist medical staff, maintaining confidence that their patients were appropriately managed in primary care.
Study Design: This is a pilot study of patients with advanced chronic kidney disease referred to the active case management team under the care of a community matron with experience in managing patients with complex chronic conditions.
Community matrons are senior nurses working autonomously in primary care with a caseload of clinically unstable, older patients. Their role is to optimize the wellbeing of patients, reducing hospital admissions and unscheduled care.
Population Studied: Twenty patients aged 46-94,
12 men, all had multiple chronic diseases, were frail and often housebound making access to services difficult. Many had recent unscheduled stays in hospital (13/20), stage IV/V kidney disease who were clinically unsuitable for, or who had refused dialysis (14/20); or older patients whose health and function were deteriorating despite haemodialysis (6/20). Patients were referred from secondary care (18/20) or their general practitioner.
Principal Findings: At 10 months: Routine outpatient visits were reduced for most patients
(15/20); 123 acute bed days were avoided through treatment in community facilities; 46 acute bed days were avoided through supported discharge; 8 incidents of inappropriate prescribing were identified; 104 referrals were made to other services including social services, housing, equipment and transport.
Conclusion: A community matron, with experience in chronic disease management, can effectively coordinate care for vulnerable older people with multiple co-morbid conditions in primary care. Early identification of potential problems and timely interventions optimize patient well-being and reduce dependency on secondary care services, including episodes of unscheduled care. Critical to the success of the pilot was the unique relationship
between the matron and the hospital based specialist teams. An element of trust was needed to reassure the specialists that the care of their patients would be enhanced by the intervention.
Over the course of the pilot we were able to effectively illustrate this.
Implications for Policy, Delivery or Practice:
Given the current and predicted pressures on healthcare services, using community matrons to coordinate the primary/secondary care interface can optimize clinical care and lead to more costeffective use of resources.
● Years of Potential Life Lost Due to Male
Genital Cancers in the United States —
Projection through 2050
Chunyu Li, M.D., Ph.D.; Donatus Ekwueme, Ph.D.;
Sun Rim, M.P.H.; Florence Tangka, Ph.D.
Presented by: Chunyu Li, M.D., Ph.D., Associate
Service Fellow, Center for Disease Control &
Prevention, 4770 Buford Highway, NE, MailStop:
K55, Atlanta, GA 30341, Phone: (770) 488-4866;
Email: hsf6@cdc.gov
Research Objective: Although years of potential life lost (YPLL) due to cancers have been widely studied, specific information on such burden from male genital cancers (prostate, testicular, penile, and other male genital cancers) is currently lacking.
With unprecedented growth in population aging, increasing racial diversity, and higher risk of male genital cancers among minorities (e.g. prostate cancer), YPLL from male genital cancers is expected to increase. The primary objective of this study is to project YPLL due to male genital cancers through 2050 accounting for changes in demographics and death rates using nationally representative data.
Study Design: Using death rates – predicted from
1999-2004 trends in national mortality data based on log-linear regression – and 2008 census population demographic projections, we projected age- and race/ethnicity-specific number of deaths for male genital cancers and all malignant cancers among men through 2050. All projections were based on the 2000 U.S. standard population. We extracted data by 5-year age groups and created four mutually exclusive race/ethnicity categories
(i.e., non-Hispanic white, non-Hispanic black,
Hispanic, and non-Hispanic other). We applied the life expectancy method to estimate YPLL for male genital cancers and all malignant cancers by multiplying the projected number of deaths and remaining life years for each age- and race/ethnicity-specific group from 2004 U.S. life tables. Sensitivity analyses were performed by varying both death rate and population projections.
Population Studied: The study population included all males in the U.S. who died of cancer at age =
15.
Principal Findings: YPLL due to male genital cancers (1.0 million) is projected to represent 9.8% of the estimated YPLL attributable to all malignant cancers (10.5 million) among American men in
2050, increasing from 6.8% (303,989 out of 4.5 million) in 2004. Of the projected 729,043 increase in YPLL due to male genital cancers in 2050, 45.5% are due to changes in demographic composition,
25.3% to population growth, and 29.2% to increasing death rates. 98.5% of the projected increase in YPLL due to male genital cancers in
2050 is from prostate cancer. The anticipated fastest growing racial/ethnic group with YPLL due to male genital cancers are Hispanics (+843% from
2004-2050) with non-Hispanic blacks (+534%), non-Hispanic other (+248%), and non-Hispanic whites (+130%) following. The largest percent increase is expected among men who died at age =
85 years.
Conclusion: This study provides the first estimation of YPLL due to male genital cancers and projection through 2050 in the United States. YPLL associated with all malignant cancers among men is projected to increase by 2.4 times in 2050 compared with 2004. Male genital cancers are projected to account for a much higher proportion of this burden.
Implications for Policy, Delivery or Practice:
Male genital cancers may be a greater burden in the 21st century due to demographic changes and the aging population. More efforts are needed to reduce the burden of these cancers.
Theme: Gender and Health
● Effect of Diabetes Complication Prevention on
Medicare Expenditure in the Elderly Medicare
Population: An Instrumental Variable Analysis
Suying Li, Ph.D.; Jiannong Liu, Ph.D.; David
Gilbertson, Ph.D.; Allan Collins, M.D., F.A.C.P.
Presented by: Suying Li, Ph.D., 914 South 8th
Street, Suite S-406, Minneapolis, MN 55404,
Phone: (612) 347-5226; Email: sli@usrds.org
Research Objective: To examine the effect of diabetes complication prevention on Medicare expenditure.
Study Design: This is a retrospective follow-up study. We identified 12,635 incident diabetes patients from 1998 and 1999 Medicare 5% claims data, and calculated Medicare payments from
2000-2003. We applied a 2-stage least square model with instrumental variable (IV) methodology to estimate the effect of prevention on Medicare expenditure. Diabetes complication prevention at baseline was measured basing the American
Diabetes Association (ADA) guideline for caring people with diabetes – receiving at least 2 HbA1c tests annually aimed at controlling glycemia.
Medicare expenditure was payment per patientyear in the follow-up period and adjusted for inflation. When calculating payment per patient-
year, we censored the follow-up time at death, end of Medicare entitlement, or December 31, 2003. We considered an IV related to the distance between a patient home and a Diabetes Education Program
(DEP).
Population Studied: The study population included
1998 and 1999 incident diabetes patients who were
Medicare Part A and Part B eligible, aged at least
67 years, and without known retinopathy, neuropathy, renal disease, or cardiovascular disease at baseline.
Principal Findings: In our study sample, only
27.7% (3503/12,635) of patients received at least two HbA1c tests annually. Univariate analysis showed that receiving at least two HbA1c tests was significantly associated with lower Medicare payment in the follow-up period, with a difference of
$1311 per patient-year ($5453 vs. $6764; p<0.0001). Results from the ordinary least square estimation showed that receiving at least two tests annually was associated with a $831 decrease in
Medicare payment per patient-year (p=0.0002).
Results from the 2-stage least square IV estimation showed that receiving at least two tests annually was associated with a $953 decrease in Medicare payment per patient-year (p=0.0498).
Conclusion: Preventing diabetes complications could save Medicare considerable costs. Given that the estimated effect of at least two HbA1c tests on
Medicare payments was a $953 decrease per patient-year, about $8.7 million ($953 x 9132) of the
Medicare budget would have been saved each year if those 9132 patients had received the tests.
Because our study sample was from the Medicare
5% random sample database, our estimations could be multiplied by 20 for the entire Medicare population. Each year during 2000-2003, about
$174 million in Medicare expenditures could have been saved through the Medicare patients aged 67 years old and older who developed diabetes in
1998 and 1999, had no diabetes complications at baseline, and did not receive at least two HbA1c tests annually after diabetes onset.
Implications for Policy, Delivery or Practice:
This study demonstrates the role of ADA DEPs in delivering better quality of care for diabetes patients. Currently, DEPs are not available in all areas of the US, especially in rural areas. Studies should be initiated to explain the lower availability of
DEPs in rural and in populous areas, and to explore ways to extend programs nationally. In addition, the rate of patients receiving at least two HbA1c tests was very low in the study period; policy should encourage health providers to prescribe this service.
Funding Source(s): NIDDK
Theme: Prevention and Treatment of Chronic
Illness
● The Influence of Health Behaviors on Survival in Lung Cancer Patients in Taiwan
Ya-Hsin Li, Dr.P.H.; Shwn-Huey Shieh, Ph.D.; Chi-
Yi Chen, M.D.
Presented by: Ya-Hsin Li, Dr.P.H., Assistant
Professor, Health Policy and Management, Chung
Shan Medical University, 15, Lane268, Sec2,
ChungDe Road, Taichung, TW, Phone: +886-963-
438065; Email: yli.tulane@gmail.com
Research Objective: Many studies reported health behaviors (such as tobacco consumption, alcohol consumption, regular physical check-up, etc.) play an important role in increasing the chance of acquiring lung cancer, but only a few studies focused on the relationships between health behaviors and patients survival. The study is going to discuss the effects of health behaviors (such as smoke, alcohol consumption, betel-nut consumption, regular physical examination) on lung cancer patient survival outcome.
Study Design: The sample consists of 1,139 newly diagnosed, histologically confirmed lung cancer patients from a medical center hospital in central
Taiwan. The patient medical records were collected from January 1, 1998 to March, 2005. Besides descriptive statistical analyses, t-test and ANOVA were used to analyze the relationship between patient characteristics, patient health behavior and survival. Chi-square tests were used to analyze the relationship between patient characteristics, patient health behaviors, and stage of disease. Cox’s proportional hazards regression was computed for the risk of survival by patient characteristics, and patient health behaviors.
Population Studied: The sample consists of 1,139 newly diagnosed, histologically confirmed lung cancer patients from a medical center hospital in central Taiwan. The patient medical records were collected from January 1, 1998 to March, 2005.
Principal Findings: Patients with no smoking and no alcohol consumption behavior had significantly longer survival time than patients with smoking and alcohol consumption behavior. Patients without betel-nut consumption behavior had tendency of longer survival time than patients with betel-nut consumption behavior although it is not significantly different. Patients with regular physical check-up behavior had significantly longer survival time than patients without regular physical check-up behavior.
When holding constant with all of the variables, smoking behavior, physical check-up, gender, and lung cancer stage at diagnosis will effect lung cancer patients’ survival.
Conclusion: From the study, patients’ gender, smoking behavior, physical check-up, and lung cancer stage at diagnosis will effect patients’ survival.
Implications for Policy, Delivery or Practice:
Patients’ education should be intervened to improve patients’ health behaviors. Otherwise, public education about regular physical check-up is
needed to increase the early diction rate and improve the outcome of lung cancer.
● Development of an Evidence-Based National
Healthcare Performance Measurement
Framework for Singapore
Eng Kok Lim, M.B.B.S., M.S.; Jian Ming Kung,
M.B.B.S.; Matthew Niti, M.B.B.S., M.P.H., Ph.D.;
Swee Bee Hong, B.S.; Serena Low, M.B.B.S.,
M.P.H.; Mabel Deurenberg-Yap, M.B.B.S., M.S.,
Ph.D.
Presented by: Eng Kok Lim, M.B.B.S., M.S.,
Deputy Director (Clinical Benchmarking), Clinical
Quality Improvement Division, Ministry of Health,
Singapore, College of Medicine Building, 16
College Road, Singapore, 169854, SG, Phone:
+632-51282; Email: lim_eng_kok@moh.gov.sg
Research Objective: Many countries have developed performance measurement frameworks for measuring and managing their health system performance to ensure effectiveness, quality, efficiency and equity, to identify gaps for improvement and to provide a mechanism for monitoring changes over time. This paper describes the development of a locally relevant healthcare performance framework for evidence based health system performance assessment, to guide policy decisions while enabling national and international benchmarking.
Study Design: In developing a National Healthcare
Performance Framework, an extensive literature review of similar frameworks available in other countries was conducted. This process also involved consultations with various international organizations, key stakeholders and international partners where relevant. The locally relevant
Healthcare Performance Framework attempts to link the tiers of ‘Determinants of health’ and ‘Health system performance’ to ‘Health status and outcomes’, reflecting different domains of quality of care (effective, appropriate, patient-centred, accessible and safe care) at the national, institutional and specialty levels. From this framework, priority areas for indicator development were selected based on national goals, disease burden/costs, improvability, and known variations in quality. Potential measures for priority areas which met well-accepted criteria for scientific reliability, comparability, acceptability and feasibility were then established as initial performance measures.
Principal Findings: This framework is currently being implemented nationally, beginning with the public sector. Ten initial priority areas have been identified, with measures being operationalised for the different tiers and domains of care. For example, for ischemic heart disease (IHD), a initial priority area, initial measures cascaded from a national level (e.g incidence and premature mortality from IHD for tier: ‘Health status and outcomes’, prevalence of smoking, obesity, diabetes mellitus for tier: ‘Determinants of health’,) to institutional and specialty-specific levels (e.g. for tier: ‘Health system performance’, measures include percentage of acute myocardial infarction patients who were given aspirin and beta blockers on discharge, hospital standardized mortality ratio), to examine performance across the various tiers of healthcare.
Conclusion: Our Healthcare Performance
Measurement Framework and indicators are a
‘living set’ in terms of future development and refinement to ensure relevance.
Implications for Policy, Delivery or Practice: The development of this evidence-based Healthcare
Performance Measurement Framework and key performance measures for Singapore allows the alignment of performance measurement across the health care continuum, helping to create a regulated performance environment and stimulating a culture of continuous improvement.
● Nursing Home Organizational Characteristics
& Utilization of Cancer-Related Medical Services
Chunchieh Lin, M.B.A., Ph.D.
Presented by: Chunchieh Lin, M.B.A., Ph.D.,
Graduate Research, Department of Health
Administration, Virginia Commonwealth University,
1008 East Clay Street, Richmond, VA 23298-0203,
Phone: (804) 267-0662; Email: linc5@vcu.edu
Research Objective: Cancer, which is the second leading cause of death in the U.S (National Center for Health Statistics, 2005 ), is more common among the elderly with about 55.8% of all cancer cases diagnosed among those aged 65 and older
(SEER, 1992-1998 ). Since frailty and other age related conditions put the elderly at risk for nursing home care, nursing homes may be the site of care and death for many elderly with cancer. However, relatively little is known about the variation of cancer diagnosis and care in these facilities. After controlling for resident and nursing home market characteristics, this study will examine whether nursing home organizational characteristics are related to the use of cancer-related medical services for treatment (oncologist visits, cancerdirected surgery, chemotherapy and radiation therapy), and palliative care (pain medication and hospice services) among Medicaid and Medicare insured residents of nursing homes in Michigan from 1996-2000.
Study Design: Using data from the Medicare claim file, Michigan tumor registry, Area Source File,
Michigan Medicaid Nursing Home Cost Report, and
Online Survey, Certification and Reporting
(OSCAR) during the period of 1996 through 2000, the study used logistic regression to predict the logit of utilization of cancer-related medical services from different nursing home organizational characteristics.
Population Studied: The study group is comprised of Michigan Medicaid and Medicare insured nursing home residents, also known as “dually-eligible,” diagnosed with cancer during the period 1996 through 2000. To examine the role nursing homes play in resident cancer-related health service utilization, this study selected residents diagnosed with cancer after entering a nursing home. To be included in the study, a resident had to have stayed in the same nursing home for at least 30 days before diagnosis. Residents who were diagnosed through death certificate or autopsy are discussed in another study and are excluded. Since half of the study subjects died within 3 months, this study will focus on cancer-related medical services within
6 months after the diagnosis date.
Principal Findings: Residents who stayed in nursing homes with higher nurse staffing level and higher skill mix are more likely to utilize more cancer-related medical service.
Conclusion: Nursing home with higher nurse staffing level and higher skill mix are more likely to help residents with earlier cancer diagnosis, provide better support, and motivate residents to accept cancer-related medical services.
Implications for Policy, Delivery or Practice:
Nearly 1 in 10 nursing home residents had a diagnosis of cancer (Johnson et al., 2005) and the rate is likely to increase because of the aging population. The government should put effort to monitor nursing homes in providing necessary attention in helping residents to access needed medical services and therefore, improve their quality of life.
Funding Source(s): NCI
Theme: Long Term Care
● Nursing Home Quality: The Link Between
Ownership & Quality Measures
Michael Lin, Ph.D., M.S.P.H.; Judith Lave, Ph.D.;
Nicholas Castle, Ph.D. M.H.A.
Presented by: Michael Lin, Ph.D., M.S.P.H.,
Assistant Professor, Health Policy & Management,
University of Pittsburgh, A649 Crabtree Hall,
Pittsburgh, PA 15261, Phone: (412) 624-2743;
Email: linm@pitt.edu
Research Objective: The primary aim of this study is to expand the knowledge base regarding the quality of US nursing home facilities using the publicly-reported quality measures from the Nursing
Home Compare website. A secondary aim is to compare the quality of facilities based on their ownership along the dimensions of tax-exemption, government control, and chain affiliation.
Study Design: This study utilizes data from the
Center for Medicare and Medicaid Services (CMS) to create a profile of nursing homes on licensure and certification characteristics, and publiclyreported quality measures. We use descriptive statistics and multivariable regression models to compare the average profiles for size, staffing, and quality measures.
Population Studied: All free-standing Medicare and/or Medicaid-Certified nursing homes that were in operation during 2007 are included in this analysis.
Principal Findings: Comparisons based on profit status indicate that nursing homes quality is different based on ownership: for-profit nursing homes have worse quality, on average, than nonprofit nursing homes on 8 of the 19 quality measures, while they have better quality on 3 of the
19 measures. In terms of chain affiliation, independent facilities have better quality than chainaffiliated nursing homes, on average, for 5 of 19 quality measures, and worse quality on 3 of 19 quality measures. Including chain-level measures in the regression models, such as the size of the nursing home chain, the number of states in which the chain operates, and the number of beds in the nursing home chain, we do find that these measures are independently-associated with the quality of care. This suggest that including more fine-grained detail about chain-level characteristics may improve risk-adjusted models of nursing home quality. However, excluding state fixed effects from the regression models, all of the remaining independent measures account for only a modest amount of observed variance, ranging from one to seventeen percent.
Conclusion: The relationship between ownership and quality of care among nursing homes is confounded by the relationship between profit status, chain affiliation, and operational characteristics such as staffing, payor mix, and competition. While more nursing homes are chainaffiliated than independent, and more nursing homes are for-profit than either non-profit or government owned, it is important to recognize the extent of variability among facilities based on these ownership classes. One important issue is how to profile quality accurately and parsimoniously across the various measures.
Implications for Policy, Delivery or Practice:
Although there are differences in quality based on ownership, it remains to be seen whether these are more important than operating characteristics such as nurse staffing levels and resident mix. Recent legislation regarding increased transparency of nursing home ownership should include additional reporting requirements regarding residents and operating characteristics over the course of a year so that researchers can more accurately quantify and characterize the linkage between ownership and quality.
Theme: Long Term Care
● The Quality of US Nursing Home Chains
Michael Lin, Ph.D., M.S.P.H.; Judith Lave, Ph.D.;
Nicholas Castle, Ph.D. M.H.A.
Presented by: Michael Lin, Ph.D., M.S.P.H.,
Assistant Professor, Health Policy & Management,
University of Pittsburgh, A649 Crabtree Hall,
Pittsburgh, PA 15261, Phone: (412) 624-2743;
Email: linm@pitt.edu
Research Objective: The primary aim of this study is to describe the quality of nursing home chain facilities operating across the US. A secondary aim is to compare the quality of these chain-affiliated nursing homes based on their ownership by a forprofit company.
Study Design: This study utilizes a chain-linked database to compare chain-affiliated nursing homes on structure, process, and outcome characteristics.
To compare non-profit to for-profit nursing home chains, we use t-test statistics to compare the average profiles for size, geographic dispersion, staffing, and quality measures.
Population Studied: All free-standing Medicare and/or Medicaid-Certified nursing homes that were in operation during 2003 and 2007 are included in this analysis as long as they are owned by a company that owns another nursing home.
Principal Findings: The number of nursing home chains operating in the US during 2003 and/or 2007 are roughly the same (852 vs. 835). Across most structure, process, and outcome measures, quality has not changed significantly. In 2007, the majority of nursing home chains are owned by for-profit companies (69.0%), are affiliated with only one or two other nursing homes (51.0%), and are located in only state (65.4%). On average, for-profit chains have more facilities than non-profit chains, and are more geographically dispersed. Also, using 2007 data, chain-affiliated for-profit nursing homes have a lower average percentage of residents (both short-stay and long-stay) who receive the recommended influenza and pneumococcal vaccinations, and have higher average percentages of residents with pressure ulcers, residents who are bedfast / confined to a chair, and residents in physical restraints. In contrast, chain-affiliated nonprofit nursing homes have higher average percentages of residents who are depressed or anxious, residents who lose bowel or bladder continence, and residents in moderate to severe pain. However, it is important to recognize that several of these differences between for-profit and non-profit chains mirror the differences observed in a comparison of independently-owned for-profit and non-profit nursing homes.
Conclusion: The relationship between for-profit status and quality of care among nursing home chains is mediated by the relationship between operating characteristics (e.g., staffing levels, bed size) and quality of care. While more chain-affiliated nursing homes are for-profit than non-profit, it is important to recognize the extent of variability among chains within an ownership class.
Geographic dispersion across multiple states appears to be one of the most important chain characteristics affecting quality. One important conceptual and empirical issue yet to be resolved is how to profile appropriately the quality of care across a nursing home chain’s portfolio of facilities.
Implications for Policy, Delivery or Practice:
Proposed Congressional legislation (S2641 and
HR5799) regarding the transparency of nursing home ownership and quality reporting must contend with important “unit of analysis” concerns in order to provide a parsimonious yet meaningful profile of quality. One particular dimension that may be of great relevance but has yet to be included in regulatory data is centralization. We suggest that policymakers consider collecting information about the locus of strategic decision-making which affects both the types of residents who are admitted and the staff that are hired.
Theme: Health Care Markets and Competition
● Risk Adjustment for Medicare Beneficiaries with Alzheimer’s Disease & Related Dementias
Pei-Jung Lin, Ph.D.; Matthew Maciejewski, Ph.D.;
John Paul, Ph.D.; Andrea Biddle, Ph.D.
Presented by: Pei-Jung Lin, Ph.D., Post-Doctoral
Fellow, Department of Pharmaceutical Health
Services Research, School of Pharmacy, University of Maryland, 220 Arch Street, 12th Floor, Baltimore,
MD 21201, Phone: (410) 706-0871; Email: plin@rx.umaryland.edu
Research Objective: Accurate prediction of health expenditures and risk adjustment are essential to balance the health care needs of beneficiaries with greater disease burden, and the financial viability and stability of participating plans, especially health plans serving exclusively or disproportionately frail populations. The objective of this study was to compare the performance of various prospective risk adjustment measures in predicting overall and drug expenditures among Medicare beneficiaries with Alzheimer’s disease and related dementias
(ADRD).
Study Design: This retrospective cohort study analyzed the 1999-2004 Medicare Current
Beneficiary Survey linked with Medicare claims.
Base-year demographic characteristics, diagnoses, self-reported health, and Medicare expenditures were used to construct prior expenditure, risk adjustment measures, comorbidity measures, and functional status measures that were used to predict subsequent year’s total health care and drug expenditures. Adjusted R-squared, predictive ratios, and receiver operating characteristic curves were used to compare overall predictive power, accuracy of subgroup prediction, and accuracy in identifying beneficiaries with the top 10% of expenditures, respectively.
Population Studied: A two-year panel of 1,452 person-year observations representing 726 elderly, community-dwelling beneficiaries with ADRD was analyzed.
Principal Findings: Prior expenditures predicted subsequent year’s expenditures best by far in the overall sample (adjusted R-squared=20.4% and
38.7% for total and drug expenditures), the lowest cost quartile subgroup, and the highest cost quartile subgroup. Two diagnosis-based risk adjustment measures, the Centers for Medicare and Medicaid
Services-Hierarchical Condition Category (CMS-
HCC) and the Chronic Illness and Disability
Payment System-Medicare (CDPSM), had higher predictive power for total expenditures for all groups. The Prescription Drug Hierarchical
Condition Category (RxHCC) exhibited greater predictive power for drug expenditures than other measures, and outperformed all other measures in identifying ADRD beneficiaries with extremely high drug expenditures (c-statistics=0.69) and extremely high overall expenditures (c-statistics=0.73).
Adding functional status to single-measure models generally improved predictive power of overall health expenditures by 2.3-3.6% but not for drug expenditures.
Conclusion: The CMS-HCC and the RxHCC measures currently in use are more accurate than other risk adjustment measures for overall and drug expenditure prediction for beneficiaries with substantial disabilities and comorbidities.
Prediction of overall expenditures may be modestly improved for these beneficiaries by using a combined model of age, gender, CDPSM, and functional status.
Implications for Policy, Delivery or Practice: Our findings are particularly relevant for payers, insurers, and health plans, such as Medicare
Special Needs Plans, that enroll a high proportion of chronically ill beneficiaries with persistently high expenditures. Additional research is needed to evaluate the performance of pharmacy-based measures and to improve the Part D payment methodology. As new, expensive drugs become available over time, it will be important to update risk adjustment models for reimbursement purposes and to identify beneficiaries with high-cost risks for disease management, medication therapy management, and care coordination.
Theme: Medicare
● Determinations of Parternership
Disintegration for the Primary Community-
Based Care Networks in Taiwan BNHI
Demonstration Project
Blossom Yen-Ju Lin, Ph.D.; Chia-Yi, Liau, M.H.A.;
Blossom Yen-Ju Lin, Ph.D.
Presented by: Blossom Yen-Ju Lin, Ph.D.,
Associate Professor, Health Services
Administration, China Medical University, 91 Hsueh
Shih Road, Taichung, 404, TW, Phone: +886-922-
709359; Email: yenju1115@hotmail.com
Research Objective: Primary Community-based
Care Network (PCCN) Demonstration Project was launched by the Bureau of National Health
Insurance (BNHI) in 2003 and it is still on the way.
Partnership structures in the PCCNs represent the virtual vertical (i.e., between the clinic and hospital members) and virtual horizontal (i.e., among the clinic members) aspects of organizing, which designate the formal relationships between individuals and the total network and include organizational design to ensure effective communication, coordination, and integration across the total network. Previous studies focused largely on how this virtual integration among the partnerships and its effects. However, little studies were tried to understand the partnership disintegration in PCCNs. This study was aimed to explore why some partnerships in PCCNs were disintegrating.
Study Design: A qualitative methodology with semi-structured interview questions was conducted.
Semi-structure questions were pre-designed for study interviews to understand the factors driving partnership disintegration.
Population Studied: Thirty-seven clinic members and five hospital members, who have been withdrawn from PCCNs, identified from the 2003-
2005 Taiwan Primary Community-based Care
Network List, were included in this study.
Principal Findings: It found the reasons for partnership disintegration were: 1) PCCN members' limited involvement in personal times and limited personal abilities for family medicine practicing; 2)
PCCNs members' limited facility capability such as human resources and required medical equipment involvements; 3) lack of consensus, communication and trusts among PCCN members; and 4) the complicated administrative processes, such as paper work burdens.
Conclusion: Partner disintegration in PCCN demonstration projects could be attributed to providers' personal involvements, facility capabilities, and administrative burdens.
Implications for Policy, Delivery or Practice: To tighten partnership relationships, several suggestions could be made in this study, including the establishment of clinic and hospital member relationships, and the reductions of administrative works. In addition, educating the public for the concepts of family doctors and well-organized the national health policies could support health care providers to smooth the integration processes.
Funding Source(s): Taiwan National Science
Council
● Effect of Culture Differentiation Between a
Hospital & its Emergency Department on
Emergency Department Medical Staff
Blossom Yen-Ju Lin, Ph.D.; Wei-Chen, Huang,
M.H.A.
Presented by: Blossom Yen-Ju Lin, Ph.D.,
Associate Professor, Health Services
Administration, China Medical University, No.91
Hsueh-Shih Road, Taichung, 404, TW, Phone:
+886-922-709359; Email: yenju1115@hotmail.com
Research Objective: This study was aimed to explore how the culture differentiation between a hospital and its emergency department (ED) might influence its ED medical staff’s job satisfaction and leaving intention.
Study Design: The study was a mailed crosssectional survey design. Data, including hospital culture, ED culture, and ED medical staff's job satisfactions and leaving intention, were collected.
Pearson, t-test, one-way ANOVA, and multiple regressions were used to examine the effects of the culture differentiation between a hospital and its hospital-based ED on ED medical staff’s job satisfaction and leaving intention.
Population Studied: With individual ED medical staff as unit of analysis, one hundred and ninetythree ED medical staff among 52 hospitals responding the survey was analyzed in this study.
Principal Findings: The results showed that the culture differentiation between a hospital and its ED was not related to the ED medical staff’s job satisfaction and leaving intention. Hospital culture was also not related to the ED medical staff’s job satisfaction; however, the ED culture (i.e. subculture) was related to the ED medical staff’s job satisfaction. Future analysis showed that the medical staff in the ED characterized as flexible and decentralized culture have higher job satisfaction than those in the ED characterized as dominative and centralized culture. And the medical staff in the
ED characterized as internal focus and stable culture have higher job satisfaction than those in the ED characterized as external focus and competitive culture. No evidences showed the hospital culture and its ED culture would influence the leaving intention of the medical staff.
Conclusion: In this study, it was found that the culture differentiation between the organization and its department was not related to department employees' job satisfaction. However, department subculture is a key for influencing department employees' job satisfaction.
Implications for Policy, Delivery or Practice:
From the managerial perspective, it suggests that hospital managers should develop a working environment full of conformity and human nature and empower the medical staff in autonomy and decision-making.
Funding Source(s): Taiwan National Health
Research Institute
● Effects of Director Professionalism on
Management Styles in Community Health
Centers
Blossom Yen-Ju Lin, Ph.D.; Cheng-Chieh, Lin,
M.D., Ph.D.; Yung Kai Lin, M.D., M.B.A.
Presented by: Blossom Yen-Ju Lin, Ph.D.,
Associate Professor, Health Services
Administration, China Medical University, No.91
Hsueh-Shih Road, Taichung, 404, TW, Phone:
+886-922-709359; Email: yenju1115@hotmail.com
Research Objective: Based on the national regulation on qualification of director professionalism in the public health centers in 2000, medical physicians have not been the only medical professionalism to qualify for the position of director of public health centers in Taiwan. The potential director candidates also include those in the professionalisms of nursing, pharmacy, and medical technology. However, few studies have been explored the leadership effectiveness of various director professionalisms in public health centers.
Therefore, this study was aimed to explore how directors in the public health centers with different medical professionalisms, named as physicians, nurses, pharmacists, and medical technologists were related to their leader styles and effectiveness.
Study Design: This was a mailed survey study with individual public health centers as the unit of analysis. The directors and staff in the individual public health centers were surveyed by the structured questionnaires in terms of director personal and professional backgrounds, organizational strategy, communication, and coordination, organizational congruent with national health policy, employee empowerments, and employee satisfaction. In addition, public health centers’ service population and service location
(rural vs. urban) were also collected. Descriptive analyses and structural equation modeling was performed. Data were collected through multiple stages to get all the needed information in 2005 and all the data were aggregated into the level of individual public health centers.
Population Studied: There were 230 health centers responded our study with 66% (230/347) response rate.
Principal Findings: The results revealed that there were not statistical significance in leadership styles and effectiveness including organizational strategy, communication, and coordination, organizational congruent with national health policy, employee empowerments, and employee satisfaction among the various medical professionalisms of directors in the public health centers. However, it was found that the directors ever trained and having research experiences in the discipline of public health would empower their employees with less extent and dominate more in the business activities in the public health centers, be more active in service creations, and be more sensitive to external environments. And the directors with longer position duration would empower their employees with less extent. Moreover, elder directors involved less in communication and coordination in the management of public health centers. The overall model was shown as a moderate fit in this study.
Conclusion: The different medical professionalisms in the directors of the public health centers were not related to their leader styles and effectiveness. However, the directors’ personal backgrounds such as ages, position length, and trained and research experiences in public health played important roles in their leader styles and effectiveness.
Implications for Policy, Delivery or Practice:
Medical professionalisms were not related to the leader styles and effectiveness in managing public health centers. However, we did find the role of education, that is, directors’ trained and research experiences in public health did make the directors to be different in their leader styles and effectiveness. The value of education of training program in public health discipline might be better criteria for recruiting, selecting, or training the potential director candidates in the public health centers than the criteria of just focusing on the specified medical professionalism.
Funding Source(s): Taiwan Bureau of Health
Promotion, Department of Health, ROC
● Cost & Effectiveness Analyses of Clinical
Pathway in Coronary Artery Bypass –
Experience of Cardiovascular Department in
One Medical Center
Yung Kai Lin, M.D., M.B.A.; Chia-Pin Chen, Ph.D.;
Wen-Chen Tsai, Dr.P.H.; Yu-Ching Chiao, Ph.D.;
Blossom Yen-Ju Lin, Ph.D.
Presented by: Yung Kai Lin, M.D., M.B.A.,
Attending Physician, Division of Cardiovascular
Surgery, Taichung Veterans General Hospital,
Taichung, Taiwan, R.O.C., 160, Sec. 3, Chung-
Kang Road, Taichung, 407, TW, Phone: +886-
922-709359; Email: floyd330@ms2.hinet.net; yenju1115@hotmail.com
Research Objective: Clinical pathway contains multidisciplinary plans outlining the sequence and timing of actions necessary to achieve expected patient outcomes with optimal efficiency. However, few studies have been devoted to the exploration of the effect of clinical pathways on coronary artery diseases treated with coronary artery bypass surgery. This study was aimed to investigate the cost-effectiveness of the clinical pathway on coronary artery bypass surgery in one medical center of the central Taiwan.
Study Design: This was a retrospective study with secondary datasets 2003-2007. The costs, postoperative complication occurrence, the number of graft bypass, and length of postoperative stays, were traced. Patient demographics such as age and gender, and patient surgical risk indicator,
EuroSCORE, were also collected.
Population Studied: Two hundred and twelve CAB surgery patients, met to the third-payer case payment criteria, were included.
Principal Findings: Overall, it revealed that there was no statistically differences of total costs and complication occurrence in heart arrest CAB surgery patients with vs. without enrolled clinical pathways. However, there exist different patterns in the specific cost items in CAB surgery patients with and without enrolled clinical pathways. It was found that length of ordinary ward stays were positively related to the costs of heart beating CAB surgery patients in most cost items.
Conclusion: The implemented clinical pathways in the studied medical center seem not to be shown cost-effectiveness in CAB surgeries.
Implications for Policy, Delivery or Practice: The implemented clinical pathways in the studied medical center seem not to be shown costeffectiveness in CAB surgeries and it provides an opportunity to re-evaluate the existing clinical pathways. Several possible issues could be discussed for these findings.
Funding Source(s): Taichung Veterans General
Hospital in Taiwan
● Health Insurance Claim Compliance
Management: The Experience of Taiwan’s
Universal Health Insurance
Chun-Mei Lin, Ph.D; Chinho Lin, Ph.D.
Presented by: Chun-Mei Lin, Ph.D, Manager,
Health Expenditure Management, Bureau of
National Health Insurance Southern Region Branch,
96, Kung-Yuan Road, Tainan, 700, TW, Phone:
+886-6-2244396; Email: melin@mail.nhisb.gov.tw
Research Objective: Medical claim fraud is a problem in health care systems around the world, and non-compliance with regard to payment will lead to the waste of health care spending and divert investment from other sectors of the economy.
Consequently, claim compliance has become an issue of rising political concern in many countries.
Claim compliance works to ensure that the appropriate level of resources is devoted to health care and that those resources are used optimally.
To help meet these aims, more transnational studies should be undertaken so that experiences of dealing with this problem can be compared and shared.
Study Design: An in-depth case study is conducted. The author analyzes the strategies employed by administrators in Taiwan to identify anomalous practices in the universal health insurance program, verify their causes, and then prevent them and the related fraudulent claims.
Population Studied: The detailed medical claim compliance strategies and management policies in
Taiwan’s universal health insurance were investigated in order to draw practical lessons to enable insurance payers to develop appropriate claim compliance designs and instruments.
Principal Findings: The claims auditing process used in Taiwan is based on knowledge-driven
intelligent systems to help insurance payers to recognize and describe novel or unusual practice patterns in order to control and recover health insurance payments. Intelligence analysis and sophisticated compliance strategies can thus manage the formal relations between noncompliance indicators and potentially inappropriate claims. With this system the thirdparty payers can monitor a GP’s adherence to prescription guidelines and deal with this information under individual physician profiles. We find that providers’ compliance is highly reliant on the perception of their monetary payoff and the cognitive experience of clinical care. Specifically, when third-party payers increase the coverage rate of their auditing activities, the providers are likely to pay more attention to billing integrity and will have an incentive to coordinate changes in inappropriate practice. Therefore, the insurance payers have to make concerted efforts at educating and communicating with providers in order to achieve voluntary compliance with regard to claim integrity and benefit quality assurance.
Conclusion: This study helps to obtain valuable experience from Taiwan’s universal health insurance claims auditing practices with regard to compliance strategies and procedures, and consequently to show the relevance and applicability of the instrument developed in this work in translating claim-derived knowledge into policy and action.
Implications for Policy, Delivery or Practice:
This study provides an opportunity to understand how Taiwan's health care authorities make use of claim-derived knowledge in their claim auditing and compliance strategies, in order to achieve a better and more financially effective governmentsponsored national health insurance program.
Theme: Medicare
● Chronic Conditions, Transitions & Care
Coordination for the Perspective of Patients &
Caregivers
Keith Lind, J.D., M.S.
Presented by: Keith Lind, J.D., M.S., Senior Policy
Advisor, Public Policy Insitutute, American
Association of Retired Persons, 601 E Street, NW,
Washington, DC 20049, Phone: (202) 434-3845;
Email: klind@aarp.org
Research Objective: Describe the experience of chronic conditions from a consumer perspective, highlight problem experiences with the health care delivery system for these patients and for caregivers, and identify populations most at risk for experiencing problems.
Study Design: During the fall 2008, we conducted two large national opinion surveys. We asked respondents about the quality of care, their transitional care experiences, personal care coordination activities, level of assistance, and level of activation or engagement in managing their care.
Population Studied: Populations surveyed included (1) one of people over 50 with chronic conditions (n=2,453) and (2) one of informal caregivers over 45 (n=978). Patients and care recipients had all experienced at least one transition from a health care facility to home within the last three years.
Principal Findings: Among respondents with chronic conditions: 23% had experienced medical errors;
15% were readmitted to a hospital within 30 days of discharge; 16% received unnecessary medical tests;
11% did not receive preventive services; 18% felt their transitional care was not well coordinated;
14% had their first follow-up appointment more than four weeks after discharge or never had one; 15% were readmitted to a health care facility within 30 days of discharge. Poor transitions were significantly associated with (1) greater number of chronic conditions, (2) poorer health status, and (3) less education.
Respondents with chronic conditions who reported poor transitions were more likely to be readmitted within 30 days of discharge. Readmission was also associated with (1) needing assistance with patient care coordination activities, (2) having poorer health status and (3) being male. Respondents with more frequent transitions reported fewer transitional care problems suggesting that they had learned from their experiences. Among caregiver respondents:
25% reported transitional care was poor for their care recipients; 15% reported their care recipient never had a follow-up visit after discharge; 32% reported their care recipient was readmitted within
30 days of discharge; 84% believed they understood transitional care instructions whereas
70% or less of care recipients understood. People spent significant amounts of time coordinating and arranging activities related to direct provision of health care and related services including: respondents with chronic conditions spent an average of 10 hours per month; caregivers for patient respondents who received assistance with these activities spent an average of 20 hours per month; and caregiver respondents spent an average of 80 hours per months. People with chronic conditions sought health information from multiple sources – 82% from their usual care provider; 53% from the internet; 15% from Medicare or other government agency.
Conclusion: People with multiple chronic conditions and low activation are more likely than others to experience quality of care problems, poor transitions following discharge from a health care facility, readmission and to require assistance with patient care coordination activities.
Implications for Policy, Delivery or Practice:
Many people with chronic conditions and their
caregivers could benefit from improved transitional care and other care coordination services.
Transitional care services might be most effectively targeted at patients at high risk of poor transitions and readmission during the period immediately following discharge from a health care facility.
Greater attention should be focused on caregivers when patients are discharged from health care facilities.
Funding Source(s): AARP Public Policy Institute
Theme: Prevention and Treatment of Chronic
Illness
● Physician Performance Assessment: Setting a
Credible Standard
Rebecca Lipner, Ph.D.; Brian Hess, Ph.D.; Gerald
Arnold, Ph.D.; Weifeng Weng, Ph.D.
Presented by: Rebecca Lipner, Ph.D., Vice
President for Psychometrics & Research Analysis,
Psychometrics & Research Analysis, American
Board of Internal Medicine, 510 Walnut Street,
Suite 1700, Philadelphia, PA 19106-3699, Phone:
(215) 446-3541; Email: rlipner@abim.org
Research Objective: Challenges in assessing physician practice performance, specifically in achieving measures that are evidence-based and statistically sound, have received a lot of attention over the last decade. Less research is focused on standard-setting techniques used to determine standards by which physicians are being held accountable for public reporting or pay-forperformance. This study sought to create a structured process that would lead to a credible and defensible absolute passing standard.
Study Design: The ABIM Diabetes Practice
Improvement Module (PIM<sup>SM</sup>) was used to collect medical record and patient experience data for ten clinical (intermediate outcomes and processes) and two patient experience measures. The Angoff standard-setting method, frequently used for multiple-choice examinations, was used to judge how minimallycompetent physicians would perform on individual measures. This required an expert panel to review the twelve measures and establish performance criteria for delivering minimally competent diabetic care. For example, each expert answered the question “what percent of patients seen by a borderline physician would receive an ophthalmologic exam?” Experts then rated each measure to determine relative importance and these were scaled using the Dunn-Rankin method.
An overall pass/fail standard was derived using a continuous scoring approach.
Population Studied: Performance data were from
957 physicians who completed the Diabetes
PIM<sup>SM</sup> with at least 10 patients between the ages of 18-75. This yielded 20,131 chart data and 18,706 patient surveys. The expert panel had four general internists, two nephrologists, one endocrinologist, and one geriatrician.
Principal Findings: The expert panel successfully used the Angoff method and importance rating scale to establish reasonable criteria and weightings. The scoring approach was in contrast to those used by recognition programs where physicians who meet minimum performance criteria for a measure are awarded all points allocated while physicians who do not receive zero points.
Instead, we multiplied physicians’ actual performance rate for each measure by the derived importance value and summed these products.
Continuous scoring uses more information from a physician’s performance making it more sensitive to identifying differences among physicians. The approach yielded a normal distribution and a passing standard of 48.002 (almost half of the possible 100 points) which classified 4.3% of physicians in the sample as incompetent.
Performance on clinical outcomes were weighted highest for importance (54%) followed by process measures (30%) and patient experience (16%). The decision consistency was quite high (.95). Validity was substantiated because those classified as incompetent were more likely to be of lower cognitive ability (as measured by standardized internal medicine certification scores and overall clinical competence program director ratings) and more likely to be in solo practice.
Conclusion: The standard-setting approach yielded a credible and defensible absolute passing standard whose outcome was reasonable and reproducible, was based on informed judgment, due diligence, performance data, and research.
Implications for Policy, Delivery or Practice: In making high-stakes decisions about physician performance that affect not only physicians but the patients they treat one must not only ensure that the measures used in the assessment are evidence-based and statistically sound but also that the method used to determine a standard of competence is rigorous and credible.
Theme: Quality and Efficiency: Measurement
● Impact of Environmental Interventions on
Lead Poisoning among Children
Tianli Liu, M.S., Ph.D.; Stephen Buka, Sc.D.
Presented by: Tianli Liu, M.S., Ph.D., Community
Health, Brown University, 121 South Main Street,
2nd Floor, Providence, RI 02912, Phone: (401)
996-4664; Email: tianli_liu@brown.edu
Research Objective: This prospective longitudinal study evaluates the impact of environmental interventions including lead inspection, abatement and enforcement on reducing lead poisoning among children under 6 years old living in the
Rhode Island State (RI).
Study Design: Prospective longitudinal study
Population Studied: Data from the RI Childhood
Lead Poisoning Prevention Program (RI CLPPP)
Lead Elimination Surveillance System (LESS) were utilized in the study. Lead levels were screened each year for all children 0 to 6 years old.
Properties were eligible for lead inspection if children living in the properties were identified as suffering significant lead poisoning (one venous blood lead level 20ug/dl or higher, or two consecutive venous blood levels 15ug/dl to19ug/dl).
There were 4 intervention processes/outcomes for these eligible properties: (1) eligible for inspection, but not performed; (2) inspection performed, no hazards found; (3) inspection offered, performed, and hazards were not abated; (4) inspection offered, performed, and hazards were abated. From
1997 to 2007, 2115 properties in nine core cities of
RI including Central Falls, Cranston, East
Providence, Newport, Pawtucket, Providence,
Warwick, West Warwick, and Woonsocket had completed one of the above four processes, hence were identified as closed cases. There were 4425 children had been living in these properties, and
10665 lead level measures were collected and recorded for these children.
Principal Findings: We were interested in whether lead levels of children living in abated properties decreased more and faster in comparison with children living in unabated properties from the time when environmental interventions completed until
12 months follow up. Generalized linear regression was employed in the analysis. The results showed there was no significant difference in lead levels of the two groups of children living in abated and unabated properties at the time when the environmental interventions were completed.
During the 12 month follow up period, the lead level of children living in the abated properties on average decreased 0.67ug/dl per month, which is significantly greater (p = 0.037) than the lead level decrease of 0.41ug/dl per month for children living in the unabated properties.
Conclusion: These preliminary results suggest that household lead abatement effectively reduced lead poisoning among children. We are currently conducting more analyses to test the hypothesis.
Implications for Policy, Delivery or Practice:
This ongoing research will provide useful and important information in making and improving environmental prevention strategies to reduce lead poisoning among children.
● CMS Hospital Acquired Conditions: Accuracy
& Completeness of POA Coding
Danielle Lloyd, M.P.H.; Kathy, Belk, B.A.
Presented by: Danielle Lloyd, M.P.H., Senior
Director, Reimbursement Policy, Advocacy, Premier
Inc., 444 North Capitol Street, Northwest, Suite 625,
Washington, DC 20001, Phone: (202) 879-8002;
Email: Danielle_Lloyd@PremierInc.com
Research Objective: To review coding completeness and accuracy for the Present on
Admission (POA) flag which is vital in determining hospital-acquired conditions (HACs)
Study Design: In the Inpatient Prospective
Payment System fiscal year 2008 and 2009 rules the Centers for Medicare & Medicaid Services finalized policy whereby 10 categories of HACs will no longer qualify cases for higher payment for discharges occurring on or after October 1, 2008.
The policy is contingent upon the accurate coding of whether the conditions are POA for all diagnoses
- a requirement for discharges on or after October
1, 2007. For completeness of coding the number of diagnosis codes with a valid POA flag was compared to the total number of diagnosis codes present in the population excluding diagnoses not requiring POA coding per the ICD-9-CM Official
Guidelines for Coding and Reporting. A review of the trending of each code across time was utilized to assess the uptake in POA coding.
Analyses were conducted for four separate populations (All patients, Medicare patients,
Medicaid patients, and Other payor patients) and for states with early implementation of POA requirements.
Population Studied: Analyses utilized Premier’s
Perspective™ data for more than 12 million acute inpatients for the July 2006 through December 2008 timeframe.
Principal Findings: Trending revealed that hospitals are not yet fully coding all principle and secondary diagnoses. Evaluating the first 9 codes that CMS accepts versus the 25 allowed on an electronic claim did not substantially improve the coding completeness. Hospitals are coding POA flags for appreciably more diagnoses associated with Medicare cases than for private payor cases, and substantially more for private payor cases than for Medicaid. Some early adopter states (CA, FL,
IL and NY) started higher in terms of completeness, however were not among the top performing states
(MA and WI) after national implementation. The number of missing/invalid POA flags has dropped precipitously since September 2007. At the same time, coding indicating a condition was POA has increased dramatically. It appears that the majority of formerly missing codes are now being coded as
POA. The flags that indicate insufficient documentation (“U”) or clinical efforts could not determine POA (“W”) are used infrequently.
Conclusion: Hospitals are not coding completely for all payors, individual payors, or for all diagnoses on an electronic claim. By state, hospitals have a wide range of completeness and early adopter status does not necessarily relate to high-coding compliance, suggesting that additional time to educate and train providers will not be enough to ensure hospitals code all diagnoses. Given the low use “W” and “U,” providers may benefit from additional education in this area.
Implications for Policy, Delivery or Practice:
Understanding POA coding is vital to assessing the appropriateness and effectiveness of HAC-related payment reductions. In addition to the current
Medicare regulations, there are numerous other payors considering implementation of similar policies. This analysis demonstrates that improvements must be made in POA coding before the payment reductions under HAC policies can be implemented accurately. Such information could be useful to Congress as it deliberates expansions of value-based purchasing initiatives and health reform.
● The Business Case for Fighting Nosocomial
Infections
Danielle Lloyd, M.P.H.; Kathy, Belk, B.A.
Presented by: Danielle Lloyd, M.P.H., Senior
Director, Reimbursement Policy, Advocacy, Premier
Inc., 444 North Capitol Street, Northwest, Suite 625,
Washington, DC 20001, Phone: (202) 879-8002;
Email: danielle_lloyd@premierinc.com
Research Objective: To look at the business case for hospitals reducing the number of hospital acquired conditions (HACs) both in terms of additional cost and reduced payment as well as the possible implications for the broader healthcare system in terms of reduced costs for avoided HACs.
Study Design: In the Inpatient Prospective
Payment System fiscal year 2008 rule CMS finalized a policy that for discharges occurring on or after October 1, 2008, hospitals will not receive additional payment for cases in which one of the selected conditions was not present on admission.
That is, the case would be paid as though the secondary diagnosis were not present. This study focuses on central line associated blood stream infections (CA-BSI). Patients were categorized into three groups: 1) patients with central line where the blood stream infection was present on admission,
2) patients with central line without blood stream infections, and patients with central line with blood stream infections acquired during hospitalization.
These groups were examined for differences in mortality rates, lengths of stay (LOS), readmissions, charges, costs as well as use of post-acute care.
Analyses were conducted for four separate populations: 1) All patients (all payors), 2) Medicare patients, 3) Medicaid patients, and 4) Other payor patients.
Population Studied: This analysis utilized
Premier’s Perspective™ data for more than 4 million acute inpatients from inpatient facilities in the
January 2008 through December 2008 timeframe.
Principal Findings: While the impact of CA-BSIs on individual patients can be catastrophic, we anticipate that the prevalence across all payors to be quite low. We anticipate that the reduced payment under the HAC policy will be a small portion of the total amount hospitals lose on these cases due to higher than average LOS and resource utilization. In addition, we expect a higher use of post-acute care (PAC), which has strong payment implications for payors and the system as a whole.
Conclusion: We believe the analysis will show that a strong business case exists for hospitals to invest in education, training and technology that will reduce the number of CA-BSIs. We also believe that the analysis will show that the broader cost implications for the system through increased length of stay, readmissions, and post-acute care will emphasize that this is an important area for further payment policy developments such as payfor performance to increase the value of healthcare.
Implications for Policy, Delivery or Practice: The incremental costs data will be useful to a hospital as a benchmark against which it can compare.
Finance departments at hospitals can use this benchmark information to justify procurement or increases in human resources. CMS can also use this data to determine if CA-BSI truly meets the high-cost criteria for inclusion in the policy.
Heretofore, CMS has only been able to look at total case costs. If higher incidence of PAC use is found, CMS should consider ways to tie this policy to these other settings.
● Systematic Evaluation of Teaching Qualities of Medical Faculty: Development & Validation of a System Combining Self- & Residents’
Assessment
Kiki Lombarts, Ph.D.; Onyebuchi Arah, M.D., Ph.D.
Presented by: Kiki Lombarts, Ph.D.,
Researcher/Consultant, Quality Process Innovation,
Academic Medical Center, Meibergdreef 9,
Amsterdam, 1100 DE, NL, Phone: +31205668635;
Email: m.j.lombarts@amc.uva.nl
Research Objective: Providing clinician-educators with feedback on their teaching skills and selfreflection are known to be helpful in improving teaching qualities. The aim of our study was to examine if a valid, reliable and feasible system for the specialty-specific evaluation of teaching qualities of medical faculty could be developed for use in academic medical centers.
Study Design: We developed a system (SetQ) for the evaluation of teaching qualities of faculty in an academic medical center in The Netherlands. The
SetQ system consisted of specialty-specific (i) webbased self-evaluation questionnaire filled in by faculty, (ii) web-based questionnaire for residents’ assessment of faculty, (iii) individualized faculty feedback reporting, and (iv) individualized faculty follow-up support. Depending on specialty, both self and resident questionnaires shared at least 24 core items based on the validated 26-item Stanford
Faculty Development Program (SFDP26) instrument. All questionnaires contained two items of global ratings: faculty being a role model and
overall training skills. Each specialty could have between two and eleven additional specialtyspecific items. The resident questionnaire included two open questions for narrative feedback on faculty. Analyses were conducted using SPSS.
Population Studied: In total 157 faculty and 152 residents from four specialties, namely anaesthesiology, internal medicine, obstetrics and gynaecology, and surgery were invited to participate. Residents chose which faculty to assess. Participation was anonymous. The evaluation period lasted the month of September
2008.
Principal Findings: The response rates were 82% and 78% for faculty and residents respectively.
Residents completed 1389 assessments for 152 faculty. There were 9.1 assessments per resident
(4.5 – 18.6 across specialties), and 8.9 ratings per faculty (5.5 – 15.7 across specialties). Explorative factor analysis identified 5 teaching domains: teaching climate, professional attitude towards residents, communications of goals, evaluation, and feedback. The internal consistency was high with
Cronbach’s alpha ranging from 0.88 to 0.94. Faculty feedback was positively received. For reliable feedback to faculty, assuming a minimum
Cronbach’s alpha of 0.70 for the instruments, at least 4 to 9 assessments per faculty are required.
Residents were generally positive about the teaching qualities of faculty in our academic institution. ‘Professional attitude towards resident’ was the highest scoring domain (mean 4.2) and
‘communications of goals’ the lowest (mean 3.5).
Residents’ and faculty self-assessment showed no substantial correlations. Using cross-classified multilevel models, we found that between 23% and
46% of the total variations in the teaching domains and two global ratings were attributable to betweenfaculty differences in residents’ evaluations. 10
Days after closure of the evaluation period faculty received their individualised feedback reports, presenting the results of the self and residents’ assessment and including a peer group ranking.
Reports concluded with an overview of the narrative feedback as formulated by residents.
Conclusion: The SetQ system appears reliable and valid for the evaluation of teaching qualities of faculty in an academic medical center. The needed numbers of residents’ assessments per faculty are achievable. Faculty feedback reports seem wellreceived, sometimes leading to improvement activities and are useful in designing individual development tracks. Future re-assessments need to measure actual improvements.
Implications for Policy, Delivery or Practice:
Residency program directors should formatively discuss the results with faculty and support them in initiating change.
Funding Source(s): AMC, Amsterdam
Theme: Health Care Workforce
● Racial & Gender Differences in Emergency
Room Triage Assessment & Test Ordering for
Chest Pain
Lenny Lopez, M.D.; Andrew Wilper, M.D.;
Alexander Green, M.D.; Joseph Betancourt, M.D.
Presented by: Lenny Lopez, M.D., Assistant in
Health Policy, Institute for Health Policy,
Massachusetts General Hospital, 50 Staniford
Street, 9th Floor, Boston, MA 02114, Phone: (617)
270-6600; Email: llopez1@partners.org
Research Objective: Racial and ethnic disparities have been documented in the clinical care of myocardial infarction (MI). A patient’s presenting symptoms and initial emergency department (ED) triage assignment likely affect the assessment and treatment of MI. The objective of this study is to assess whether presenting symptom and triage assignment could contribute to reported differences in care.
Study Design: Nationally representative ED data for all adult visits (> 18 years) were obtained from the National Hospital Ambulatory Health Care
Survey of Emergency Departments (NHAMCS-ED) for 1997-2006. Patient demographic characteristics, reason for the visit, triage category assignment, tests ordered, insurance status, final
ED diagnosis, and hospital characteristics were analyzed. We used weighted logistic regression in
SUDAAN to examine the multivariable-adjusted associations between race and presenting symptom, triage assignment, test ordering and final
MI diagnosis. Survey year was included to assess changes in test ordering practices over time.
Population Studied: The total sample consisted of
235,117 visits. Of these, 21,629 presented with a complaint of chest pain, representing 78 million such visits nationally over the 10 year period.
Principal Findings: Compared to whites, Blacks
(OR 0.93 [CI 0.88-0.95]) and Hispanics (OR 0.82
[CI 0.76 – 0.89]) are less likely to present with an initial complaint of chest pain. Men and older patients are more likely to present with an initial complaint of chest pain. Of those presenting with chest pain, Blacks (0.62 [CI 0.52-0.73]) and
Hispanics (OR 0.69 [CI 0.53 – 0.90] are less likely to be categorized in emergent triage categories compared to whites. The uninsured and those with
Medicaid were less likely to be categorized in emergent triage categories compared to those with private insurance. Of those presenting with a complaint of chest pain, Blacks (OR 0.84, [CI 0.74-
0.90]) and those insured with Medicaid and the uninsured were less likely to have an EKG ordered.
Blacks (OR 0.69, [CI 0.61 – 0.80]), Hispanics (OR
0.65 [CI 0.54 – 0.78]), and those insured with
Medicaid and the uninsured were less likely to have a cardiac monitor ordered. Blacks (OR 0.79, [CI
0.68 – 0.91]) and Hispanics (OR 0.74 [CI 0.61 –
0.89]) were less likely to have pulse oximetry ordered. Men and older patients were more likely
to have all three tests ordered. These findings persist after adjusting for year of survey. Of those with a final diagnosis of myocardial infarction, there were no significant differences in triage category or test ordering by patient or hospital characteristics.
Conclusion: Minority patients are less likely to present with chest pain to an ED. Of those presenting with chest pain, minority patients are less likely to be triaged to emergent triage categories and are less likely to receive essential evaluative tests. However, there were no significant differences in triage category or test ordering for those with a final diagnosis of MI.
Implications for Policy, Delivery or Practice:
These findings suggest gender and racial/ethnic differences in triage which may result in differences in initial evaluation and urgency of evaluation but not in final diagnosis of MI.
● Acculturation & Self-Reported Hypertension &
Diabetes in Latino Subgroups: The California
Health Interview Survey 2001-2005
Lenny Lopez, M.D.; Lina Pabon-Nau, M.D.; Leroi
Hicks, M.D., M.P.H.
Presented by: Lenny Lopez, M.D., Assistant in
Health Policy, Institute for Health Policy,
Massachusetts General Hospital, 50 Staniford
Street, 9th Floor, Boston, MA 02114, Phone: (617)
270-6600; Email: llopez1@partners.org
Research Objective: Little is known whether acculturation factors impact hypertension and diabetes prevalence among Latino adult subgroups.
We tested the hypothesis that diabetes and hypertension prevalence among U.S. Hispanics vary by country of origin and by degree of acculturation.
Study Design: We stratified the analysis by selfidentified Latino subtypes categories: Mexico
(n=19336), Central America (n=2370), South
America (n=666), Puerto Rico (n=370), Latinos having two or more countries of origin (n=1901), and other Latino (n=925). We also analyzed non-
Latinos by self-identified race: Afircan-American
(n=7409), White (n=92,214). We used self-reported
English language ability and years in the US (<=1 yr, 2-4 yrs, 5-9 yrs, 10-14 yrs, 15+ yrs) as measures of acculturation and assessed presence of participants’ self-reported hypertension/diabetes by race/ethnicity, Latino subtype, and acculturation.
Using multivariable logistic regression, we assessed the relationship of these characteristics to self-reported hypertension/diabetes adjusting for patients’ comorbid disease, health care utilization, insurance status and socio-economic status.
Population Studied: We examined the adult participants in the 2001, 2003, and 2005 California
Health Interview Survey (CHIS).
Principal Findings: We found that over 40% of all participants lived in the US for greater than 15 years. Insurance rates ranged from 62% for
Central Americans to 91% for non-Latino Whites.
Hypertension prevalence was lowest in South
Americans (16%) and highest in African-Americans
(34%). Diabetes prevalence was lowest in South
Americans (4%) and highest in other Latino (12%).
After multivariable adjustment, hypertension was less likely to be reported by Mexicans (OR 0.49
[95% CI: 0.37, 0.65]) and Central Americans (OR
0.61 [CI: 0.45, 0.83]), relative to non-Latino Whites.
Diabetes prevalence was not different among
Latino subtypes compared to non-Latino Whites.
Those participants with less than 15 years in the US were less likely to report hypertension and diabetes
(OR range 0.33-0.64) compared to those residing in the US for greater than 15 years. English language ability did not influence self-reported prevalence of either disease. There was a significant interaction between Latino subtype, acculturation and hypertension and diabetes prevalence. Living in the US for less than 15 years was associated with a disproportionately lower prevalence of hypertension and diabetes among Mexican (p<0.003) and
Central American (p=0.04) immigrants compared to other Latino subtypes.
Conclusion: We found significant heterogeneity in hypertension but not diabetes prevalence among
Latino subtypes in California compared to non-
Latino Whites after adjustment for differences in health care utilization and access. Those participants with less than 15 years in the US were less likely to report hypertension and diabetes.
Further we found that the degree of acculturation may be an important predictor of disease prevalence among Latino subgroups.
Implications for Policy, Delivery or Practice:
Future work examining ethnic differences in hypertension prevalence and outcomes should account for the heterogeneity among Latino participants based on their country of origin and degree of acculturation to the US.
● The Impact of Obstetric Unit Closures on
Maternal & Neonatal Outcomes of Pregnancy
Scott Lorch, M.D., M.S.C.E.; Sindhu Srinivas, M.D.,
M.S.C.E.; Jeffrey Silber, M.D., Ph.D.; Corinne
Fager, M.A.
Presented by: Scott Lorch, M.D., M.S.C.E.,
Assistant Professor of Pediatrics, Neonatology, The
Children's Hospital of Philadelphia, 3535 Market
Street, Suite 1029, Philadelphia, PA 19104, Phone:
(215) 590-1714; Email: lorch@email.chop.edu
Research Objective: Between 1997-2005, 9 of 19 obstetric units closed in Philadelphia, reducing the supply of obstetric beds by 40% and increasing the average number of deliveries in the remaining hospitals by 31%. There are no studies to suggest whether this degree of obstetric unit closures would result in improved care, through the closure of lesser quality obstetric units, or worse care, through increased crowding at the remaining hospitals. The
objective of this study, then, was to determine the association between obstetric unit closures and pregnancy outcomes in Philadelphia, controlling for changes in these outcomes in six cities (Pittsburgh,
Kansas City, St. Louis, Los Angeles, San
Francisco, San Diego) that did not experience a similar reduction in obstetric units over the same time period.
Study Design: Birth and death certificates linked to maternal and neonatal hospital discharge records for all births between 1995-2005 were used to identify neonatal deaths, deliveries by Cesarean section, births < 32 weeks gestation, and various maternal (hemorrhage, blood transfusion, wound infection) and neonatal (asphyxia, birth injury, seizures) birth complications. Using multivariable logistic regression models to control for maternal comorbidities, congenital anomalies, and sociodemographic factors, we constructed a difference-in-differences model to compare the difference in outcomes between Philadelphia and six cities during the pre-closure period of 1995-1996 with differences during the 3 post-closure periods of
1997-1999, 2000-2002, and 2003-2005.
Population Studied: All mothers and infants delivered in Philadelphia and six control cities between 1995 and 2005, N=2,624,959.
Principal Findings: Compared to 1995-1996, the relative difference in neonatal deaths in
Philadelphia increased between 1997-1999 compared to the six urban control areas (odds ratio
(OR) 1.39, 95% CI 1.12-1.74). The relative change in neonatal deaths in Philadelphia was not statistically different in 2000-2002 or 2003-2005.
The rate of increase in Cesarean section deliveries was smaller in Philadelphia compared to the control areas between 1997-1999 (OR 0.88, 95% CI 0.84-
0.92) and 2003-2005 (OR 0.95, 95% CI 0.91-0.99).
The difference in deliveries < 32 wks did not change significantly in Philadelphia over the 3 postclosure periods. However, the difference in the odds of a neonatal birth complication in
Philadelphia increased during all 3 post-closure epochs (1997-1999: OR 1.31, 95% CI 1.18-1.45;
2000-2002: OR 1.15, 95% CI 1.02-1.29; 2003-2005:
OR 1.34, 95% CI 1.19-1.51), and the odds of a maternal birth complication increased in the latter 2 post-closure epochs (2000-2002: OR 1.09, 95% CI
1.01-1.19; 2003-2005: OR 1.17, 95% CI 1.07-1.27).
Conclusion: Using a difference-in-difference model to control for secular trends in other cities without large-scale obstetric unit closures, these results suggest that a reduction in obstetric supply in
Philadelphia may have been associated in decreased Cesarean section rates but increased rates of neonatal deaths and birth complications.
Implications for Policy, Delivery or Practice:
Reductions in obstetric supply may result in adverse outcomes, especially for complications of pregnancy that may be reduced with careful monitoring of the woman. Public health agencies should monitor these outcomes when obstetric units close to ensure that the remaining supply of obstetric units can adequately manage the pregnant population in a given geographic region.
Funding Source(s): AHRQ, Maternal and Child
Health Bureau
Theme: Public Health
● Generating Patient Accounts of Quality using
Photography & Voice
Laura Lorenz, Ph.D., M.Ed.; Jon Chilingerian, Ph.D.
Presented by: Laura Lorenz, Ph.D., M.Ed.,
Program Manager Executive Education, The Heller
School for Social Policy & Management, Brandeis
University, 415 South Street Mail Stop 035,
Waltham, MA 02453, Phone: (781) 736-3847;
Email: llorenz@brandeis.edu
Research Objective: In health care, quality of care is critical to patient well-being and organizational accountability. This paper uses visual and narrative methods to elicit and understand quality of care from the patient’s perspective.
Study Design: Acquired brain injury (ABI) survivors took photographs of their lives and explored the meaning of their images in photo-elicitation interviews. Narrative analysis methods were used to analyze visuals and interview texts. A patient case study was selected to illustrate study methods and findings.
Population Studied: Adult acquired brain injury survivors accessing outpatient rehabilitation services or living in the community and members of a brain injury survivor support group.
Principal Findings: The patient’s visual illness account challenges policies and practices of health care providers, organizations, and systems. The account shows a brain injury survivor’s difficulty adhering to physician-recommended treatment, and provides a vivid depiction of her efforts to advocate for appropriate care. Her narrative resists current health care payer policy, which limits provision of brain injury rehabilitation services to six to 12 months post-injury.
Conclusion: Photo elicitation and narrative analysis methods could encourage clinicians and organizations to view illness and quality of care through patients’ eyes. Listening to brain injury survivor accounts could help to improve the quality of care for persons suffering with chronic conditions. The methods encourage provision of holistic, patient-centered care that addresses patients’ real lives.
Implications for Policy, Delivery or Practice:
This paper contributes to the accountability literature by translating patient accounts of their lived experience with illness and care in terms of quality of care. The findings demonstrate the potential for incorporating patient perspectives into health care policies and practices that support longterm healing for persons with chronic conditions,
address quality of care from patients’ perspectives, and promote innovation and learning.
Theme: Organizational Performance and
Management
● Geriatric Care Coordination: Comparing the
United States & the United Kingdom
Annesha Lovett, Pharm.D., M.S.; Pharm.D., M.S.
Presented by: Annesha Lovett, Pharm.D., M.S.,
Graduate Student, Health Services Research,
Management and Policy, University of Florida, P.O.
Box 100195, Gainesville, FL 32610-0195, Phone:
(404) 771-6546; Email: whiteann@ufl.edu
Research Objective: The purpose of this study is to review the literature published within the last decade related to the provision of geriatric care coordination in the United States and the United
Kingdom.
Study Design: An Internet search was conducted using the key words geriatric care coordination. The abstract and/or full text of each article was reviewed and articles were excluded if they did not relate directly to geriatric care coordination in the two countries. Information posted between January
1997 – November 2007 was reviewed. The following was noted from each article: study location, source, publication date, study design, population studied and summary of the article’s key points.
Principal Findings: The three main differences between the United States and the United Kingdom are (1) the provision of home care, (2) the delivery of coordinated services, and (3) the perceptions of the quality of geriatric care provided. The key driving forces for the differences are culture, financing and medical education.
Conclusion: Variations in the provision of geriatric care coordination exist between the United States and the United Kingdom. Both countries can benefit by exploring these variations. For example, exemplary efforts have been shown by the United
Kingdom in their development of cost saving initiatives to promote home care and by the United
States in their utilization of multidisciplinary teams.
As the number of elderly individuals in the population increases, both countries will face the challenge to provide geriatric care that is balanced in cost and quality.
Implications for Policy, Delivery or Practice:
Seniors in both countries can look forward to increased efforts in this area. Successful geriatric care coordination has been shown to decrease hospitalizations, decrease emergency room visits and increase quality of care. Given the expected growth of this population, efforts should be undertaken to enhance geriatric care coordination and to provide empirical evidence to allow for informed policy decisions.
● Prescription Drug Plan Enrollment & Cost-
Related Nonadherence in Medicare Part D
Beneficiaries with Diabetes
Kristin Lowe Geonnotti, B.S.; Mary Roth McClurg,
Pharm.D., M.H.S.; Tim Carey, M.D., M.P.H.;
George Pink, Ph.D.; Marisa Domino, Ph.D.; Morris
Weinberger, Ph.D.
Presented by: Kristin Lowe Geonnotti, B.S.,
Doctoral Candidate, Health Policy & Management;
Sheps Center for Health Services Research,
University of North Carolina at Chapel Hill, 725
Martin Luther King Jr. Boulevard, CB #7590,
Chapel Hill, NC 27599-7590, Phone: (919) 966-
6446; Email: klg@unc.edu
Research Objective: Medicare Part D aims to provide seniors with affordable prescription drug coverage. Annual out-of-pocket costs can vary widely between plans. This study will determine the proportion of individuals who are enrolled into lowest-cost prescription drug plans (PDP), and whether being in a lowest-cost PDP is associated with a lower likelihood of experiencing cost-related nonadherence.
Study Design: This observational, cross-sectional study combines data from participant surveys, medical records, and publicly-available CMS plan data. Study participants are surveyed about their
Medicare Part D PDP, cost-related nonadherence, and PDP enrollment decisions. PDP information is used to calculate out-of-pocket medication costs using the CMS Prescription Drug Plan Finder tool.
Lowest-cost PDP is defined as the plan in which a beneficiary spends the minimum out-of-pocket amount to obtain all Part D drugs annually.
Differential costs are the difference between annual out-of-pocket spending in a beneficiary’s current
PDP and the annual out-of-pocket spending that would be incurred in a lowest-cost PDP. Medical record data include patient characteristics, diagnoses, and prescribed medications. Descriptive and multivariate logistic regression analyses will be used to determine the proportion of participants in lowest-cost plans and the association between lowest-cost plan enrollment and cost-related nonadherence.
Population Studied: Study participants are drawn from diabetes registries at the UNC-Chapel Hill
Internal Medicine and Family Medicine practices.
Participants include Medicare Part D beneficiaries who: (1) have diabetes; (2) are age 65 and older;
(3) are community-dwelling; (4) have had at least 1 primary care visit in the past 12 months; and (5) are able to complete a telephone interview.
Principal Findings: To date, 76 participants have been enrolled and have completed telephone surveys. The majority of participants are female
(76.3%), and 27.6% are dually-eligible for Medicare and Medicaid. 32.9% of participants report experiencing cost-related nonadherence. Out-ofpocket costs and lowest-cost PDP enrollment have
been estimated for 18 participants. Preliminary results indicate that only 1 of 18 participants (5.6%) was enrolled into a lowest-cost PDP. Four more participants (22.2%) were enrolled into plans within
$100 of the lowest-cost PDP. Differential costs were substantial (mean ± SD = $557.41 ± $539.79; median = $454.00. Data collection and analyses will be completed by the annual research meeting.
Conclusion: Preliminary results suggest that the majority of beneficiaries are not enrolled in a lowest-cost PDP. Among those not in lowest-cost
PDPs, differential costs are sizable.
Implications for Policy, Delivery or Practice: If lowest-cost PDP enrollment is associated with a lower likelihood of cost-related nonadherence, policy interventions can target plan enrollment. For instance, the random auto-assignment of duallyeligible individuals could be replaced with beneficiary-centered assignment. This would consider medication coverage needs when assigning beneficiaries into plans. Additional strategies could be developed to simplify the complex enrollment decision-making process facing other beneficiaries and their caregivers.
Funding Source(s): AHRQ, NRSA Pre-doctoral
Fellowship
● Tough Time, Tough Choice – Health Benefit for Employees of County Governments
Ning Lu, Ph.D., M.P.H; Michael Samuels, Dr.P.H.;
Elmer Whitler, M.A., M.P.A.; Phillip Kletke, Ph.D.
Presented by: Ning Lu, Ph.D., M.P.H, Associate
Professor, Health Administration, Governors State
University, 1 University Parkway, University Park,
KY 60466, Phone: (708) 534-4416; Email: nlu@govst.edu
Research Objective: Many county governments find it increasingly difficult to provide health benefits to their employees because of rising premium costs, which often comprise a large portion of county budgets. Thus, many county governments must either find new cost containment strategies or reduce employees’ health benefits. This study provides a profile of the health benefits provided to county employees in the state of Kentucky. Our intention is to provide county governments with needed information on health insurance alternatives that will contain costs while maintaining employee health benefits.
Study Design: The data were collected through a mail survey of county treasurers in the 118
Kentucky counties that were headed by county judge executives. (Two large metropolitan counties with merged city-county governments were excluded from the study.) The survey was fielded between July 14 and October 23 of 2006. A second mailing was sent to those who had not responded after a month. There were 55 survey completions for a response rate of 47%.
Population Studied: Employees of county governments in the state of Kentucky.
Principal Findings: The average number of county employees was 162 (lowest 35, highest 1278). Over two-thirds (68%) of the county employees were regular fulltime (working 35 hours or more per week); 22% were regular part-time; and 10% were seasonal or temporary workers. Over half (54%) earned less than $25,000, among which 24% earned less than $10,000 per year. Almost all fulltime county employees (99%) were eligible for health insurance, but only 78% took up the option.
All the county governments offered employee health benefits. However, almost all (98%) reported a decrease in the generosity of health benefits over the past 3 years – increased copayments and deductibles, fewer benefits offered, fewer providers from which to choose, increased employees’ share of premiums, etc. Over two-thirds (69%) had changed insurance carriers over the previous five years, among which 71% had changed more than once. Almost three-fourths (72%) of the counties provided health insurance through an insurer, 11% by participating in the state government insurance program, and 13% by self-funding and administration, and 5% by shifting to HSAs.
Counties with self-funded plans were less likely to assume increases in the cost of health insurance.
The annual cost of employee health insurance increased by an average of 34% between 2003 and
2006. Over 8% of the 2006 fiscal-year budget went to employee health benefits, an increase from 6% in
2003.
Conclusion: About 22% of eligible employees did not take up health insurance, probably because coverage was available from other sources.
Offering the option of a cash payment in lieu of health insurance could be a cost-effective alternative for county governments. Such alternatives might ease pressures on the county government. County employees may prefer this option rather than reducing or foregoing employee benefits all together.
Implications for Policy, Delivery or Practice:
County governments are challenged by efforts to provide an increasingly expensive employee benefit, health insurance, which is not taken up by a significant percentage of county employees.
Alternatives, such as offering a cash payment option, may be more attractive to employees and less expensive for county governments.
Funding Source(s): HRSA
Theme: Coverage and Access
● Patient Perceptions about Electronic Health
Records (EHR)
Armine Lulejian, M.S., M.P.H., C.H.E.S; Farzad
Mostashari, M.D., M.S.; Taafoi Kamara, M.P.H.
Presented by: Armine Lulejian, M.S., M.P.H.,
C.H.E.S, Research Scientist, Primary Care
Information Project, New York City Department of
Health & Mental Hygiene, 161 William Street, New
York, NY 10038, Phone: (212) 788-5680; Email: alulejia@health.nyc.gov
Research Objective: Patients may be a significant driver of adoption of Electronic Health Records
(EHR) systems by their medical providers, but little is known regarding patient knowledge and attitudes regarding EHRs. This study examined pre-adoption patient attitudes as part of a large community EHR extension project.
Study Design: Paper-based surveys were distributed and collected following patient visits at ten randomly selected primary care clinics who had not yet adopted EHRs. Surveys were available in
English, Spanish and Chinese, and trained interviewers facilitated data collection for patients with low literacy levels.
Population Studied: Patients scheduled to see the primary care provider for a selected week prior to
EHR implementation were invited to participate in the study. Overall, 736 patients participated in the study with a response rate of 61.5%.
Principal Findings: Majority of the patients were highly satisfied with their existing care with a mean overall rating of 8.7 out of 10 (0 = worst doctor possible; 10 = best doctor possible). More than half of the patients (54.6%) reported no prior knowledge of EHRs. Following a brief factual description, the majority of the patients listed the following as likely benefits: providers having better access to their medical information (64.1%); patients having better access to their own information (37.1%), better care
(28.7%), and fewer errors (32.9%). Main listed concerns included less time with provider (16.3%) and medical records being accessed inappropriately (33.1%). Only one out of four respondents stated that when choosing a provider, the providers’ office having an EHR would positively impact their decision. In general, prior familiarity with EHRs was significantly correlated with noting more benefits, having fewer concerns, and being more likely to weigh that as a likely factor when choosing a provider. However, patients with prior knowledge of EHRs were more likely (40% vs.
28%) to be concerned about inappropriate access to their records.
Conclusion: Most patients were not aware of
EHRs, and are generally very satisfied with the care they are already receiving. In general, patients' reported likely benefits outweighed their concerns, although increased familiarity with EHRs raised, rather than decreased concerns regarding privacy and confidentiality of their records.
Implications for Policy, Delivery or Practice:
Although patients are very satisfied with their existing care, they may be recruited as advocates for EHR adoption by their providers. However, in order to do so, patient concerns regarding privacy and confidentiality of electronic medical records must be more fully addressed.
Funding Source(s): New York City Department of
Health and Mental Hygiene
Theme: Health Information Technology
● Assessing Barriers & Roles of Local Health
Departments Toward a Successful Obesity
Control in Local Communities — A Pilot Study
Huabin Luo, Ph.D.; Christopher Dyer, Ph.D.; James
Roosen, M.P.H.; Bankole Olatosi, Ph.D.
Presented by: Huabin Luo, Ph.D., Assistant
Professor, Mount Olive College, 634 Henderson
Street, Mount Olive, NC 28365, Phone: (919) 971-
8891; Email: hluo@moc.edu
Research Objective: To identify the barriers and roles expected of local health departments (LHDs) in implementing policy and environmental changes for obesity control in the communities.
Study Design: A survey questionnaire was designed, including 28 questions on potential barriers and 16 questions on roles expected of
LHDs in implementing policy and environmental changes for obesity control.
Population Studied: A pilot survey was conducted in three counties in North Carolina. Thirty-five staff members of LHDs were recruited to participate in the survey.
Principal Findings: Thirty-five survey questionnaires were collected. Reliability tests of the sets of questions on barriers and roles of LHDs
(Cronbach’s alpha coefficient of 0.93 and 0.95, respectively) suggest the survey questionnaire is a reliable instrument. Lack of funding, lack of time, competing work priorities, no immediate benefits, difficulty reaching diverse populations, and lack of advocate in the community were among the top barriers. Coordinating/facilitating, planning, being a partner, advocating, and monitoring/providing data were the important roles expected of LHD. The perceived importance of these barriers and roles varied as a result of jurisdiction characteristics.
Conclusion: Environmental and policy interventions have the advantages over individualbased interventions for obesity control as it can influence a larger population and create a supportive environment for developing healthy behaviors. However, considerable challenges are encountered by LHDs in implementing these strategies.
Implications for Policy, Delivery or Practice: It is essential to strengthen LHDs’ capacities if they are to be successful in conducting obesity prevention programs in their jurisdictions. Efforts to overcome the barriers should be made with consideration of local conditions.
Theme: Obesity Prevention and Treatment
● A Multivariate Model of Health Literacy
Nicole Lurie, M.D., M.S.P.H.; Laurie Martin, Ph.D.;
Teague Ruder, M.A.; Bonnie Ghosh Dastidar,
Ph.D.; Dan Sherman, Ph.D.; Jose Escarce, M.D.,
Ph.D.
Presented by: Nicole Lurie, M.D., M.S.P.H., Senior
Natural Scientist & Paul O'Neill Alcoa Professor of
Policy Analysis, Health, RAND Corporation, 1200
South Hayes Street, Arlington, VA 22209, Phone:
(703) 413-1100 x 5127; Email: lurie@rand.org
Research Objective: Low health literacy (HL) is a formidable barrier to reducing gaps in quality and improving outcomes. Health literacy screening of individuals is cumbersome, and population level estimates of health literacy that could inform priorities for community intervention are lacking.
We developed a predictive model of HL using variables available in the US Census.
Study Design: We used data from the 18,581 respondents to the 2003 National Assessment of
Adult Literacy (NAAL), a nationally representative survey that measured functional HL on a 0-500 point scale. The NAAL also categorized HL of individuals as below basic, basic, intermediate or proficient. We used linear and probit regression, respectively, to predict mean HL scores or the probability of having ‘above basic’(AB) HL.
Independent variables included: age, sex, race, education, poverty level, living in an MSA, language spoken, and years in US. Then, applying coefficients to census data, we estimated the mean
HL score and the % with AB HL for across the population for census tracts in Missouri (MO). We used these estimates to generate maps of population-level HL by tracts and larger areas built up from tracts.
Population Studied: US Census Data from the state of Missouri, and respondents to the 2003
National Assessment of Adult Literacy (NAAL) survey.
Principal Findings: The linear and probit models explain about 30 and 21% of the variance in HL respectively, compared to 15% and 9.5% explained by education alone, the single best predictor of health literacy in the NAAL. All variables contributed significantly to in the model at p<.05 or less (Table
1). Models did not differ significantly across regions of the country or age groups. The estimated % of population AB within census tract ranged from 30% to 87%; most tracts had 60-70% of the population at the AB level. The map shows the estimated mean
HL score for each census tract in MO. The maps also discriminate low HL areas from those with low income or education alone.
Conclusion: A multivariable predictive model explains twice as much of the variance in HL as education alone. Identifying the ‘hot spots’ of low
HL can help determine where community-level interventions to address HL might best be placed.
Implications for Policy, Delivery or Practice: The models provide a method by which researchers can approximate health literacy within their own data.
As a result, these predictive models will advance the study of health literacy, its contribution to health disparities and health outcomes, and the identification of pathways linking literacy skills and health.
Funding Source(s): Missouri Foundation for
Health
Theme: Disparities
● STEER: Encouraging Breast Cancer
Screening Behaviors among Rural Women
Using a Tailored Reminder System
Mia Liza Lustria, Ph.D.; Robert Glueckauf, Ph.D.;
Robert Hawkins, Ph.D.; Ebrahim Randeree, M.S.;
Michelle Kazmer, Ph.D.; Ivee Rosario, M.Ed.
Presented by: Mia Liza Lustria, Ph.D., Assistant
Professor, College of Information, Florida State
University, 270 Louis Shores Building, 142
Collegiate Loop, Tallahassee, FL 32306, Phone:
(850) 644-6237; Email: mlustria@ci.fsu.edu
Research Objective: Florida ranks third in the U.S. for total number of new cases and deaths from breast cancer. Although mammography screenings have helped stabilize breast cancer incidence and mortality nationwide, about one million women over
40 years old have never had a mammogram in
Florida. Rural women, especially those who are of lower socio-economic status, who are less educated and who are non-white, have far poorer access to cancer services and are less likely to have had a mammogram than urban women. Lack of physician recommendation and perceived barriers to accessing diagnostic and treatment services have been major factors that have hindered the uptake of regular mammography screenings among these women. While the factors contributing to disparities in breast cancer incidence and mortality are numerous and complex, experts have pointed to failures in the healthcare system to deliver adequate and timely cancer care as a critical issue. This new investigator research (NIR) aims to:
(1) use participatory approaches to design a clinicbased tool to help build rural health providers’ capacity to improve screening rates and access to breast cancer care resources within rural non-white and low-SES populations; and (2) evaluate the system’s usability, sustainability, and perceived benefits in a rural health clinic setting , as well as its effects on increasing physicians’ breast cancer screening referrals and rural patients’ adherence to breast cancer screening and treatment referrals.
The proposed system, STEER (System for
Tracking, Empowering, Equipping, and Reminding) will: send regular alerts to physicians to provide timely referrals in accordance with breast cancer screening guidelines; generate screening reminders that will include information and recommended actions tailored to patients’ specific needs or perceived barriers to cancer care; provide a tool to collect information from at risk patients and enable better tracking of patients throughout the cancer
care continuum; and provide a tool to monitor physician and patient adherence to breast cancer screening guidelines.
Study Design: The current paper reports on preliminary results of PHASE 3 of this three-year project. We are conducting a three-arm randomized controlled trial of STEER in two rural health clinics and will collect and analyze data regarding rural clinicians’ screening referral practices and patient adherence to breast cancer screening referrals 1,3, and 6 months after system launch in each site.
Women who consent to participate in the trial will be randomized into three conditions: a tailored condition (this group will receive tailored reminder letters); a generic condition (to receive generic reminder letters), and a waitlist group (which will in turn be randomized to the first two conditions in month 3).
Population Studied: The three-arm randomized controlled trial will be conducted in the Spring of
2009 in two rural communities: Gadsden and
Wakulla counties. The eligible patient population will consist of women 45 years old and above who have not had a mammogram in the past 2 years mainly identified through patient lists of the
Tallahassee Memorial Family Medicine Clinics. We are hoping to recruit at least 100 women (50 in each community) who we will be following up with over 12 months.
Principal Findings: This is a three-year threephase project and we are currently in our second year. Preliminary results of the randomized controlled trial will be reported in June.
Conclusion: Increasing early detection of breast cancer through regular mammography screening is a key public health goal and has been significantly linked to declines in breast cancer mortality over the years. Women living in less-educated, lower income rural communities, are far less likely to adhere to breast cancer screening recommendations and that lack of physician recommendation and perceived barriers to accessing screening services have been major factors that have hindered their uptake of regular mammography screenings. Computerbased tools (e.g., decision support systems and automated reminder systems) have shown good results in terms of increasing screening rates and patient adherence to adjuvant therapy appointments, but their sustainability especially within rural healthcare settings still needs investigation. STEER is an IT system that is not only be comprehensive and extensible but is also user-friendly and sustainable within rural healthcare providers’ daily operations.
Implications for Policy, Delivery or Practice:
STEER is an integrated system that can build rural providers’ capacity to address these issues in several ways. First, STEER is an automated reminder system that will provide regular prompts to providers and patients to adhere to national breast cancer screening guidelines. Second, STEER can help address perceived access barriers by generating tailored lists of ancillary breast cancer resources (e.g., information about transportation, childcare or financial support) and personalized cancer care recommendations that can help move patients faster through the cancer care continuum.
Third, STEER is a tracking system that will facilitate the monitoring of breast cancer screening uptake of at-risk patients at rural healthcare clinics and enable the identification of patients who may be at most need of follow-up and support. Fourth, the system provides an opportunity to collect data about screening trends in a largely underserved rural population, as well as data about rural providers’ and patients’ adherence to screening guidelines and perceived issues preventing adherence.
Funding Source(s): Florida Department of Health
Bankhead Coley Cancer Research Program
Theme: Health Information Technology
● Ready4Changes.com: A Computer-Tailored
Approach for Human Sexuality Education for
Middle Schoolers
Mia Liza Lustria, Ph.D.; Juliann Cortese, Ph.D.;
Ivee Rosario, M.Ed.; Casey McLaughlin, B.S.
Presented by: Mia Liza Lustria, Ph.D., Assistant
Professor, College of Information, Florida State
University, 270 Louis Shores Building, 142
Collegiate Loop, Tallahassee, FL 32306, Phone:
(850) 644-6237; Email: mlustria@ci.fsu.edu
Research Objective: Current pedagogical philosophy is increasingly supportive of the notion that tailoring health communications to the needs of each individual can enhance learning and improve comprehension of health content. Limited research in this area has generally indicated that the tailored approach can be successful at engaging individuals, promoting learning and fostering positive attitudes about behavior change. The
“effects” of tailoring health communications are based on the premise that individualizing the approach and the content to the individual can increase their motivation to attend to the messages, engage in the content and eventually engage in elaborative processing of the information presented.
The primary purpose of this study is to develop and evaluate a Tailored Web-Based Educational
System to help middle-schoolers learn about puberty and human sexuality. Specifically, the study seeks to explore the following research questions:
1. What is the relationship between computertailoring of online health content and elaboration and comprehension?; 2. What variables might moderate such an effect?
Study Design: A pretest-treatment-posttest experimental design will be used to test differences in the main dependent variable (comprehension) among middle schoolers randomly assigned to each experimental group or treatment. About 400 middle schoolers will be randomly assigned to
explore one of three conditions: a generic site, a tailored site and a “placebo” tailored site, in three
50-minute sessions. This design allows us to examine the critical components as they work together in a fully tailored website. At the end of the trial, subjects will be given a post-test to assess their elaboration of the content and knowledge gain and also asked to respond to some scenario-based and problem-solving questions to assess their comprehension of the topic area.
Population Studied: The trial is being conducted in
Leon County middle schools (in Tallahassee, FL) using a sample of 7th and 8th graders (13-15 year olds).
Principal Findings: This is a two-phase project: phase 1 focuses on the development of the tailoring system and educational website, while phase 2 involves the randomized controlled trial. We are currently pretesting the system and should have preliminary results of the trial by June.
Conclusion: Advances in artificial intelligence and computing technologies has made it entirely possible to provide personalized health education and individualized learning on a mass scale.
Interactive web technology not only provides an opportunity to present tailored health information in a number of formats, but it also provides the greatest ability to toggle between modalities, further enhancing learners’ experiences and understanding of the material. However, development and deployment of these types of approaches often requires a high capital overlay or initial investment.
This cost, however, is often justified based on wide assumptions about its effectiveness for achieving learning outcomes (e.g., mastery and deeper understanding of the course material). But, while there is growing evidence about the efficacy of computer-tailoring and adaptive learning environments on learning, we have yet to fully understand the mechanisms by which this affects how individuals process information.
Implications for Policy, Delivery or Practice:
Taking a closer look at how individual users access interactive information systems (e.g., their cognitive skills, ability to solve problems and form searches, etc.) will have a significant bearing on our ability to fully exploit this technology for health education purposes. Understanding differences in the way learners process similar content delivered using different levels of specificity and peripheral cue complexity (e.g., tailored vs. generic content) will also better inform us about the nature of tailored web-based health interventions and its critical components and how to design adaptive algorithms more effectively to tailor health content for specific learners. The project is one of the first attempts to systematically tease out the effective components of a tailored health education website and to examine its effects singly and in combination on the comprehension of a complex health domain.
Funding Source(s): Florida State University
Center for Research and Creativity
Theme: Health Information Technology
● A Meta-Analysis of Tailored Web-Based
Intervention Studies
Mia Liza Lustria, Ph.D.; Juliann Cortese, Ph.D.;
Seth Noar, Ph.D.; Robert Glueckauf, Ph.D.; Junga
Lee, M.S.; Stephanie Van Stee, M.S.
Presented by: Mia Liza Lustria, Ph.D., Assistant
Professor, College of Information, Florida State
University, 270 Louis Shores Building, 142
Collegiate Loop, Tallahassee, FL 32306, Phone:
(850) 644-6237; Email: mlustria@ci.fsu.edu
Research Objective: There is growing evidence for the use of tailoring in behavioral health interventions. First, tailored interventions evoke favorable perceptions from individuals. Tailoring allows for very personal and direct content presentation based on elements such as likes/dislikes, needs, and current health behaviors or behavioral intentions. Tailored interventions have led to positive outcomes across a variety of interventions, such as binge drinking, nutrition and diet, and smoking. Moreover, recent reviews and meta-analyses have reported incremental benefits for tailored versus non-tailored interventions in the literature as a whole. Despite these promising results, we still have limited knowledge about how tailoring works and what moderates the effects of tailoring on health outcomes. The paper reports on the results of a meta-analysis that seeks to examine whether tailored web-based behavioral interventions have effects on health behaviors or clinical outcomes as well as to examine several sets of moderators that may impact the effects of tailoring, including the following: 1. Participant features: Have outcomes of tailored web-based behavioral interventions varied with regard to demographic characteristics, such as age, gender, race, education level, and country of sample? 2.
Type of behavior: Have outcomes of tailored webbased behavioral interventions varied by health behavior and/or health behavior type (i.e., preventive vs. screening vs. vaccination)? 3.
Intervention and methodological features: Have outcomes of tailored web-based behavioral interventions varied when the comparison condition was a comparison (i.e., generic/targeted) message versus when it was a no-treatment control condition? What is the effect of tailoring (i.e., tailored message vs. comparison message)? Have outcomes varied based on the use of different modalities, number of intervention contacts with participants, type of recruitment strategy, length of follow-up, and publication year? 4. Theoretical concepts: Have outcomes of tailored web-based behavioral interventions varied depending on which theoretical concepts have been tailored upon? In addition, does tailoring on more theoretical concepts and other variables (e.g., demographics, behavior) result in better outcomes?
Study Design: We conducted an extensive search for web-based behavioral intervention studies published in English-language peer-reviewed journals from 1996 to December 2008. Two independent coders coded each study and calculated effect sizes on a number of variables of interest (see objectives above).
Population Studied: A total of 50 efficacy studies were included in the meta-analysis. We only included studies which were: published from 1996-
December 2008 in English-language, peerreviewed journals randomized controlled trials or quasi-experimental studies efficacy studies of online health interventions, which: had at least one web-based component (delivered over the Internet, e.g., using e-mail, website, etc.), used a computer algorithm for tailoring, had at least one health behavior or clinical outcome measure (NOT solely focused on knowledge or attitudinal measures), and focused on a patient population (NOT focused on caregivers or providers).
Principal Findings: Tailored, self-guided health interventions delivered via the Web to date have involved a great diversity of features and formats.
While some programs have been relatively brief and simple, others have involved complex, theorybased tailoring with iterative assessment, tools for development of self-regulatory skills, and various mechanisms for providing feedback. We are currently analyzing results of the actual metaanalysis and should be able to report on the results during the conference.
Conclusion: Our ability to fully optimize the use of computer-assisted tailoring will depend on the development of empirically-based guidelines for tailoring across populations, health foci, health behaviors and situations.
Implications for Policy, Delivery or Practice:
This study seeks to enhance our understanding of how and under what conditions computer-tailoring leads to positive health outcomes in online behavioral interventions.
Funding Source(s): Florida State University
Center for Research and Creativity
Theme: Health Information Technology
● Participatory Design of an Electronic
Reminder System for Breast Cancer Screening:
Physician Perspectives
Mia Liza Lustria, Ph.D.; Michelle Kazmer, Ph.D.;
Robert Glueckauf, Ph.D.; Robert Hawkins, Ph.D.;
Ebrahim Randeree, M.S.; Ivee Rosario, M.Ed.
Presented by: Mia Liza Lustria, Ph.D., Assistant
Professor, College of Information, Florida State
University, 270 Louis Shores Building, 142
Collegiate Loop, Tallahassee, FL 32306, Phone:
(850) 644-6237; Email: mlustria@ci.fsu.edu
Research Objective: Florida ranks third in the U.S. for total number of new cases and deaths from breast cancer. Although mammography screenings have helped stabilize breast cancer incidence and mortality nationwide, about one million women over
40 years old have never had a mammogram in
Florida. Rural women, especially those who are of lower socio-economic status, who are less educated and who are non-white, have far poorer access to cancer services and are less likely to have had a mammogram than urban women. Lack of physician recommendation and perceived barriers to accessing diagnostic and treatment services have been major factors that have hindered the uptake of regular mammography screenings among these women. While the factors contributing to disparities in breast cancer incidence and mortality are numerous and complex, experts have pointed to failures in the healthcare system to deliver adequate and timely cancer care as a critical issue. This new investigator research (NIR) aims to:
(1) use participatory approaches to design a clinicbased tool to help build rural health providers’ capacity to improve screening rates and access to breast cancer care resources within rural non-white and low-SES populations; and (2) evaluate the system’s usability, sustainability, and perceived benefits in a rural health clinic setting , as well as its effects on increasing physicians’ breast cancer screening referrals and rural patients’ adherence to breast cancer screening and treatment referrals.
The proposed system, STEER (System for
Tracking, Empowering, Equipping, and Reminding) will: send regular alerts to physicians to provide timely referrals in accordance with breast cancer screening guidelines; generate screening reminders that will include information and recommended actions tailored to patients’ specific needs or perceived barriers to cancer care; provide a tool to collect information from at risk patients and enable better tracking of patients throughout the cancer care continuum; and provide a tool to monitor physician and patient adherence to breast cancer screening guidelines. The proposed paper reports on the results of Phase 1 of this 3-year project ... results of indepth interviews with clinicians and staff of rural clinics to: (a) identify rural clinicians’ perceived needs and issues in terms of providing breast cancer care and timely screening referrals to their patients; (b) determine rural patients’ perceived breast cancer care needs and barriers to seeking adequate cancer care; (c) determine and collect ancillary breast cancer care resources that will be archived in the STEER system; and (d) conduct chart abstractions of recent patients at these clinics to help populate STEER. As such, we have completed a total of 6 focus groups with physicians, clinic staff, and breast cancer patients and now have a better understanding of various stakeholders’ knowledge of breast cancer, perceived barriers to breast cancer screenings, and their suggestions for how we might encourage more women to seek and adhere to screening recommendations in rural counties.
Study Design: We conducted focus group interviews with rural clinicians and staff of two rural clinics (a total of 28 individuals). The preliminary analysis of the focus groups was completed by two researchers and used the focus groups notes as the primary data source, with the recordings used as backup to clarify anything that was unclear in the notes.
Population Studied: Clinicians and staff of two rural clinics
Principal Findings: Our formative research thus far has provided us with interesting insights that should help us conceptualize an ideal system for each clinic. Interestingly, both communities we are working with are drastically different. Wakulla county is a largely white, more educated community albeit the TMFM clinic in this county seemed to have less resources than Gadsden. Meanwhile,
Gadsden county is largely African-American, with more low income, and low literate residents.
Ironically, the TMFM clinic in this community seemed to have more resources (i.e., larger staff; better connectivity; better support; etc.) than its counterpart in Wakulla. Both clinics have a variety of “systems” in place for reminding insured women of screenings needed. But for underinsured and uninsured women, there is no system in place for tracking and reminding them about the need for mammograms. Overall, there was no coordinated system for tracking adherence to screening reminders, which on the outset, typically happened only during routine clinic visits. As such, given competing demands and priorities, there were often very little opportunities to educate patients fully about the importance of screenings. Physicians and clinic staff appreciated the need for more education but also stressed that without adequate access to resources (e.g. transportation, financial assistance) it would be difficult to convince women in their communities to get mammograms more regularly.
They agreed that developing a system that would enable them to keep better track of their patients and serve as a ready database for cancer care resources would be very useful in helping them deliver more timely screening reminders. Our research also shows how we can link these rural clinics’ efforts with educational and outreach efforts of the Tallahassee Memorial Healthcare’s Breast
Cancer Patient Navigator (BCPN) Team. In essence, through STEER, the BCPN team can have ready access to lists of patients in this community who will need extra support in making screening decisions. At the same time, our research team has started conceptualizing how additional telephonic support and ties with local resources such as urgent care centers and the local DOH offices might help strengthen our efforts to reach more vulnerable populations in this community. We hope to solidify our action plans after we’ve completed all patient interviews and before we finalize the design of the prototype system.
Conclusion: Participatory design is an enhanced user-centered approach to designing information systems through the active involvement of users throughout the design process. A number of scholars in the areas of socio-technical design and human computer interaction have stressed the importance of user participation in system design.
Substantive user involvement in the design process has been found to have significant benefits in terms of: increasing user commitment to system adoption; increased ownership of the system; improved learning and understanding; improved system quality, and user satisfaction.
Implications for Policy, Delivery or Practice:
Computer-based tools (e.g., decision support systems and automated reminder systems) have shown good results in terms of increasing screening rates and patient adherence to adjuvant therapy appointments, but their sustainability especially within rural healthcare settings still needs investigation. The decision to use the participatory approach to develop an electronic reminder system was based on our intent to develop a usable system that would be appropriate for rural health clinics.
Understanding the context within which the system will be used as well as the main issues it will be designed to address (e.g., timely provision of screening recommendations, patient tracking, etc.) is an important step towards developing appropriate solutions.
Funding Source(s): Florida Department of Health
Bankhead Coley Cancer Research Program
Theme: Health Information Technology
● Health IT: A Study on Variables that Influence the Use of Technology Over Time
Mia Liza Lustria, Ph.D.; Carlos Torres, M.I.S.; Eric
Smith, B.S.; Leon Green, B.S.
Presented by: Mia Liza Lustria, Ph.D., Assistant
Professor, College of Information, Florida State
University, 270 Louis Shores Building, 142
Collegiate Loop, Tallahassee, FL 32306, Phone:
(850) 644-6237; Email: mlustria@ci.fsu.edu
Research Objective: Despite an increase in information technology adoption, there remains a paucity of research exploring physician satisfaction and IT use. The purpose of this study is to explore the relationship between the variables that influence the use of IT over time.
Study Design: This study utilized data collected through the Community Tracking Study (CTS) which was funded by the Robert Wood Johnson
Foundation. We utilized the data from 2000-2001
(N = 12,406) and 2004-2005 (N = 6,628) studies to explore patterns of technology use by physicians over time.
Population Studied: Physicians in the US.
Principal Findings: Limited number of variables influence the use and adoption of technology over time.
Conclusion: Has not been determined.
Implications for Policy, Delivery or Practice:
This study will provide an insight to the variables that influence the use of technology over time.
Different groups will be able to better understand what patterns influence the use and adoption of technology over time, thus having an impact on the technology that will be offered in the future and ways to market the product to reach a wider audience.
Theme: Health Information Technology
● So You Really Want to Improve Patient Care
Experience? The Role of Patient Engagement in
Improving Health Care Services
Karen Luxford, B.Sc., Ph.D.; Tom Delbanco, M.D.;
Dana Gelb Safran, Ph.D.
Presented by: Karen Luxford, B.Sc., Ph.D.,
Harkness Fellow in Health Policy & Practice,
Division of General Medicine, Beth Israel
Deaconess Medical Center, 1309 Beacon Street,
Brookline, MA 02446, Phone: (617) 977-5068;
Email: kluxford@bidmc.harvard.edu
Research Objective: To investigate the interplay of patient care experience, patient engagement and organization leadership in a range of exemplar health care institutions.
Study Design: Qualitative case studies of 8 health care organizations and interviews with 6 national experts in patient-centered care across the USA.
Institutions with demonstrated improvement in patient care experience ratings included in the study: 3 acute, inpatient hospitals; 3 medical groups/ambulatory care; and 2 group/staff model health management organizations. Studies focus on the role of leadership and patient engagement as key organizational factors. Thematic analysis of semi-structured in-depth interviews.
Population Studied: Senior staff of 8 organizations including the CEO, Senior VP Quality, Senior
Clinician Chief of Staff, an Administration Director and Patient Advisory Committee representative - total 40 interviews. Six experts with a national reputation in the field of patient-centered care.
Principal Findings: Findings presented will highlight critical success factors consistently reported by respondents as contributing to improved patient care experience. Organizational factors include board-level focus, attributes of senior leaders employed, prioritization of patient care experience as an over-arching service quality issue encompassing traditional clinical issues, degree of patient engagement eg involvement in service policy & design and new staff appointments, and responsiveness to patient feedback data.
Strategies for sustainability of improvements in patient care experience from exemplar sites will be discussed. The authors will discuss the blue sky visions by leaders in the field of patient-centered care.
Conclusion: The study findings will be the basis for recommendations designed to assist patients, health professionals, and those involved in health care governance to improve care experience.
Implications for Policy, Delivery or Practice: In
March 2008, the US Centers for Medicare and
Medicaid Services made hospital patient care experience ratings publically available via the website Hospital Compare. Similar state-based reporting systems are available for primary care physician practices. In a field previously dominated by clinical quality indicators, the introduction of such reports and associated financial incentives has accelerated provider and funder interest in patient experience performance data. Health care organizations are seeking new approaches to improving patient care experience.
Funding Source(s): CWF
Theme: Organizational Performance and
Management
● Reducing Disparities in Washington, D.C.:
Achieving Program Goals & Meeting Client
Needs as Perceived by Healthy Start
Community Health Workers
Meagan Lyon, M.P.H., C.H.E.S.; Anne, Rossier
Markus, J.D., Ph.D., M.H.S.; Ashley Foster, M.P.H.
Presented by: Meagan Lyon, M.P.H., C.H.E.S.,
Research Associate, Department of Health Policy,
The George Washington University, 2021 K Street,
Northwest Suite 800, Washington, DC 20006,
Phone: (202) 530-2316; Email: meagan.lyon@gwumc.edu
Research Objective: Since 1993, the federal
Healthy Start HS program has served to fund local programs in communities with disproportionate levels of undesirable birth outcomes, such as infant mortality and low birth weight, particularly among minorities. In DC, women in HS are insured either by the Medicaid or the Alliance programs and represent a small, but sentinel, percentage of all publicly-covered pregnant women, who are typically considered at higher risk of having undesirable birth outcomes. This study sought to understand the perceptions of HS community health workers about their ability to meet client needs and the barriers they often see their high-risk, publicly-insured clients face in accessing quality medical and social services in DC.
Study Design: Qualitative semi-structured interviews with CHWs employed by the two HS programs in DC.
Population Studied: DC HS employees, including nurses and CHWs n= 34.
Principal Findings: The greatest client needs/challenges ranged from social issues, such as inadequate housing and finances, and lack of employment opportunities, to immediate in-home needs, such as supplies for an infant, such as cribs, diapers, and food, health education, and emotional
support. CHWs felt most capable of meeting immediate in-home client needs, but limited in their abilities to meet several of the primary goals of
Healthy Start, including the reduction of barriers to accessing care and the provision of adequate pre- and postnatal care. Though CHWs were able to successfully schedule/refer clients to pre- and postnatal care, they were unable to ensure clients have sufficient transportation for their visits and receive appropriate language services once at the hospital or clinic. Additional barriers included racism, stigma as a Medicaid recipient and long wait-time in clinics. However, CHWs cited uncoordinated or lack of transportation and limited language support services as barriers, which can be eliminated.
Conclusion: CHWs identified numerous barriers to achieving program goals and meeting client needs.
They saw providing transportation services to and from pre- and postnatal care for one program and having access to improved language services for the other program as critical and necessary solutions to achieving program goals and meeting client needs.
Implications for Policy, Delivery or Practice:
Public insurance programs, such as Medicaid, have been credited for contributing to a reduction in health disparities. Improved coordination between
HS and Medicaid/Alliance, particularly between HS
CHWs and the managed care plans participating in
Medicaid/Alliance and their case managers, would increase pregnant and postpartum women’s ability to receive medically necessary prenatal and postnatal care, transportation and language services and contribute to the improvement of perinatal outcomes and the reduction of health disparities in DC.
Funding Source(s): District of Columbia
Department of Health Care Finance
● The Analysis of the Affects of Knowledge
Management to the Organizational
Chenchung Ma, Ph.D.; Chien-Chang, Yang, M.H.A.
Presented by: Chenchung Ma, Ph.D., Assistant
Professor, Healthcare Administration, I-Shou
University, 8, Yida Road, Yanchao Township,
Kaohsiung County, 824, TW, Phone: (07)3836187;
Email: up000238@isu.edu.tw
Research Objective: A medical institution is a health industry with capital intensity, human resources intensity and knowledge intensity, and the above resources were used to conduct the study on the connection between knowledge management and organizational innovation.
Study Design: This study used a cross-sectional design with data collected from 25 hospitals in
Taiwan
Population Studied: 500 questionnaires were distributed to medical staffs, nursing staffs, alliedhealth staffs, and administrative staffs, and 337 effective questionnaires were returned resulting in a response rate of 67.4%.
Principal Findings: From the study, we discovered that there are significant differences between the
Acquire Creation, Circulation Proliferation, Storage
Internalize, Application sharing, Management
Innovation, and Technological Innovation. We also learned that Academic Medical Centers are more superior than regional hospitals, and there are significant positive relations between every phase in knowledge management and organizational innovation. The "application sharing" from the knowledge management shows the biggest affect on the organizational innovation.
Conclusion: This study demonstrates that knowledge management can develop significant positive influences on organizational innovation.
Each hospital should combine knowledge management and organizational goals for developing a learning organization to reinforce the sharing and application of knowledge among the hospitals.
Implications for Policy, Delivery or Practice:
Through the positive support from the managers on the development of knowledge management, they will be able to bring all the workers to participate in knowledge management and bring out the benefits of the organizational innovation.
Theme: Health Care Workforce
● Factors Associated with Tocolytic
Hospitalizations in Taiwan: Evidence from a
Population-Based & Longitudinal Study from
1997 to 2004
Ke-Zong Ma, Ph.D.; Edward Norton, Ph.D.; Eing-
Mei Tsai, M.D., Ph.D.; Shoou-Yih Daniel Lee, Ph.D.
Presented by: Ke-Zong Ma, Ph.D., Assistant
Professor, Graduate Institute of Healthcare
Administration, Kaohsiung Medical University, 100
Shih-Chuan 1st Road, Kaohsiung, 80708, Taiwan,
Kaohsiung, 80708, TW, Phone: +886-7-3131101-
2781; Email: kezong@kmu.edu.tw
Research Objective: Previous studies have showed that tocolytic treatment accounts for onequarter to nearly one-half of all antenatal hospitalizations, and tocolytic hospitalizations also impose a significant health, economic, and psychosocial burden on pregnant women. This paper aims to investigate individual, institutional, and area factors that affect the use of tocolytic hospitalizations in Taiwan where fertility has rapidly declined in recent years. Our main hypothesis is that the change in market conditions may encourage obstetricians/gynecologists to admit more women for tocolytic treatments for reasons unrelated to clinical need.
Study Design: Longitudinal data from the 1996 to
2004 National Health Insurance Research
Database in Taiwan were used to identify tocolytic
hospitalizations. Two-part models were used to estimate factors associated with tocolytic hospitalizations. The regional-year general fertility rate was included as the main explanatory variable, and it is surely exogenous. Other explanatory variables fell into four categories: individual, physician, institutional, and area factors. A full set of regional and year dummies was also included to control for the regional fixed effects and time fixed effects.
Population Studied: The study population was defined as women who gave birth from January 1,
1997, through December 31, 2004. Hospitalizations for early pregnancy loss (ICD-9-CM 630-634 and
637-639) or elective termination (ICD-9-CM 635 and 636) were excluded from the study population.
To make the results from this study comparable to research from other countries, we further excluded the following observations: women above 50 and below 15 years of age, attending obstetrician/gynecologist’s age below 25 and above
75, and women whose deliveries involved more than one child. Pregnancies in women under 15 or over 50 years old are atypical and people rarely qualify as obstetricians/gynecologists before the age of twenty-five and after seventy-five. In addition, women with multiple pregnancies might have different obstetric considerations as to mode of delivery compared with women have singleton gestations. During the study period, there were a total of 1,979,311 singleton primiparous and multiparous women in Taiwan from 1997 to 2004, and the unit of observation in this study was a pregnancy.
Principal Findings: Results from two-part models indicated that decline in general fertility rate increased the probability of tocolytic hospitalizations. Declined fertility also increased the length of stay and inpatient expenditures on tocolytic hospitalizations. Under the fee-for-service payment scheme in Taiwan, the volume of services is heavily dependent on the behavior of each individual obstetrician/gynecologist, with the obstetrician/gynecologist serving as an agent in the physician-patient relationship.
Obstetricians/gynecologists may exploit this agency relationship by providing excessive care services in order to maintain, or increase, their level of income.
Conclusion: Tocolytic hospitalizations are closely related to the conditions that accompany the declined fertility observed in Taiwan, so the significant correlations between declining fertility and increasing use of tocolytic hospitalizations empirically make sense. On the other hand, as a result of continued decreasing fertility, women will have fewer babies and thus put a high expectation on their babies. Thus, the need to produce a
“perfect baby” is obvious. Woman with pregnancyassociated problems may tend to request more antenatal and tocolytic hospitalizations to guarantee a better birth outcome, and the relationship between low fertility and high use of tocolytic hospitalizations possibly reflects higher social value of newborns.
Implications for Policy, Delivery or Practice: The rapidly declining fertility rates have evoked throughout many developed countries, and our study offers a precautionary note to those countries where the privatization of health care and its financing is ushering in ingenious ways of cost containment. The analysis of tocolytic hospitalizations in Taiwan also hold major lessons for the many countries contemplating or having universal health insurance coverage with a similar mix of providers. Future research can expand to evaluate the relationship between tocolytic care with consequential maternal and neonatal health outcomes.
Funding Source(s): National Science Council in
Taiwan
Theme: Global Health
● Availability & Implementation of Formal
Women’s Mental Health Services in the
Veterans Health Administration (VA): Build It &
They Will Come?
Casey MacGregor, M.S.W.; Elizabeth Yano, Ph.D.,
M.S.P.H.; Ismelda Canelo, M.P.A.; Danielle Rose,
Ph.D.; Isomi Miake-Lye, B.A.
Presented by: Casey MacGregor, M.S.W., Health
Research Scientist, VA Greater Los Angeles Health
Services Research & Development Center of
Excellence, Greater Los Angeles VA Healthcare
System, 16111 Plummer Street (152), North Hills,
CA 91343, Phone: (818) 891-7711; Email: casey.macgregor@va.gov
Research Objective: Although women veterans have historically represented less than 5% of VA patients, policy changes have transformed women into one of the fastest-growing segments of new users. National VA organizational data suggest growth of separate and formal mental health (MH) services targeting women veterans, prompting exploration of their development and structure.
Study Design: We conducted semi-structured telephone interviews of a purposive geographicallystratified sample of VA MH clinicians and administrators from facilities reporting distinct women’s MH programs based on results from a national organizational survey. Transcribed interviews were analyzed and coded by a multidisciplinary team of researchers.
Population Studied: We interviewed 35 VA MH clinicians and administrators drawn from all VA hospitals (VAMCs) and large Community-Based
Outpatient Clinics (CBOCs) which indicated the presence of separate women's outpatient mental clinics (n=24, 12.3% of national sample) and designated women's mental health providers in general outpatient mental health clinics (n=67,
34.4% of national sample).
Principal Findings: VAMCs and CBOCs provide an array of routine outpatient mental health services. While some programs offer a formal women's mental health clinic or provider, informal arrangements are more typical. While fully one-third of VA facilities reported separate gender-specific mental health services, few formal programs existed. Many sites, however, described ongoing women's groups for Military Sexual Trauma (MST).
Respondents' attitudes regarding the necessity of separate women's mental health services varied across sites. Some MH clinicians and administrators expressed skepticism about the need for formalized gender-specific programs or providers, emphasizing the importance of treating
“all veterans” equally. Many described challenges in tailoring services to such a small population of their patients. ites with formal women veterans’ MH service programs or providers often depicted processes and a trajectory wherein service utilization increased after sites had developed gender-specific programs and begun to offer specialized women’s MH services.
Conclusion: Formal women's mental health programs and providers have not been implemented widely in VA settings, which may be due, in part, to local mental health leaders lack of concurrence about the need for or potential value of gender-specific mental health care. The logistical challenges in establishing such services for a minority population also serve as a barrier to further development. However, VA facilities that have established such programs have experienced increased utilization, suggesting that these services do in fact address unmet mental health needs among women veterans.
Implications for Policy, Delivery or Practice:
Women veterans are expected to comprise approximately 10% of the total user population of the VA by 2010, virtually doubling the numbers seen at individual VA facilities. Most growth is among returning veterans from Iraq and
Afghanistan, who have substantial mental health burdens. While national organizational data suggested increased development of separate clinics or designated providers to attend to women veterans’ mental health needs, closer study revealed that few sites offer such formal programs.
Implementation of the VA’s national Mental Health
Benefits package will require systematic efforts to increase local awareness of gender-specific needs, and strategic support for remedying service gaps and barriers to mental health care women veterans regardless of their minority status.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Survival, Weight Change & Medication
Discontinuation of Veterans Undergoing
Bariatric Surgery in VA
Matthew Maciejewski, Ph.D.; Christopher Bryson,
M.D., M.S.; Edward Livingston, M.D.; Maren Olsen,
Ph.D.; Leila Kahwati, M.D., M.P.H.; David
Arterburn, M.D., M.P.H.
Presented by: Matthew Maciejewski, Ph.D.,
Investigator, Center for Health Services Research in Primary Care, VA Health Services Research &
Development Service, 508 Fulton Street, Mail Stop
152, Durham, NC 27705, Phone: (919) 286-0411 ext. 5198; Email: matthew.maciejewski@va.gov
Research Objective: Bariatric surgery results in substantial weight loss and improved health, but less is known about its impact on survival, weight change and medication discontinuation for diverse populations, including men and older adults. The purpose of this analysis was to examine patient factors associated with these outcomes for obese veterans undergoing bariatric surgery in 2000-2006.
Study Design: Retrospective cohort study.
Population Studied: Using data from the VA
National Surgical Quality Improvement Program
(NSQIP), we identified 856 veterans (73% male) who had bariatric surgery in one of twelve approved
VA bariatric centers in 2000-2006. For each surgical case, we obtained NSQIP variables for patient age, gender, race, body mass index (BMI), smoking status, and physical status as measured by the American Society of Anesthesiologists (ASA) class. We adjusted for comorbidities via the
Diagnostic Cost Group (DxCG) risk adjuster.
Survival was estimated via Cox regression. Weight change at six and twelve months was determined from Corporate Data Warehouse (CDW) data, and discontinuation of diabetes medication and lipidlowering medications at one year was determined from Pharmacy Benefits Management (PBM) data.
Weight change and medication discontinuation outcomes were estimated via logistic regression models.
Principal Findings: Fifty-five veterans (6.4%) died by the end of 2006. The mortality rate varied across years, but generally declined over time. BMI>50
(p=0.04), an ASA class of four or five (p=0.03), and a DxCG score ? 2 (p<0.001) were associated with an increased risk of death. Fifty-five percent of veterans taking oral hypoglycemic agents (OHAs) discontinued one year after surgery and 44% discontinued lipid-lowering medications. Veterans taking OHAs or taking fibrates were more likely to discontinue within a year after surgery than veterans taking insulin or statins. Veterans lost an average of 35 pounds 12 months after surgery.
Super obese veterans (BMI>50) were more likely to lose significant amounts than veterans with lower
BMI levels.
Conclusion: Bariatric surgery provides significant health improvements within one year and enables veterans with less severe chronic conditions to discontinue medication. The mortality risk is generally low, except for higher risk veterans who are super-obese or have multiple comorbidities.
Implications for Policy, Delivery or Practice:
Bariatric surgery is beneficial for lower risk morbidly obese veterans. Data systems are needed to monitor the longer-term health outcomes of patients who undergo bariatric surgery.
Funding Source(s): VA
Theme: Obesity Prevention and Treatment
● Update on Cost-Related Nonadherence Two
Years After Implementation of Medicare Part D
Jeanne Madden, Ph.D.; Amy Johnson Graves,
M.P.H.; Fang Zhang, Ph.D.; Becky Briesacher,
Ph.D.; Dennis Ross-Degnan, Sc.D.; Stephen
Soumerai, Sc.D.
Presented by: Jeanne Madden, Ph.D., Instructor,
Department of Ambulatory Care & Prevention,
Harvard Medical School, 133 Brookline Avenue, 6th
Floor, Boston, MA 02215, Phone: (617) 509-9953;
Email: jeanne_madden@hphc.org
Research Objective: Cost-related medication nonadherence (CRN) has been a persistent problem for elderly and disabled Americans.
Previously, we found modest overall declines in
CRN and in forgoing basic needs to afford medicines after implementation of the Medicare prescription drug benefit (Part D) in 2006, but no declines in CRN among those in poor health. We recently examined survey data from 2007, the second year of the Part D program, to assess whether the earlier findings were stable over time.
Study Design: Full-population analysis of four years of survey data, with multivariate logistic regression, comparing rates of cost-related nonadherence (skipping or reducing doses, not obtaining prescriptions) and spending less on basic needs (such as food, heat) in order to afford medicines during two years before and after implementation of Part D.
Population Studied: Community-dwelling samples,
2004 to 2007, of the nationally representative
Medicare Current Beneficiary Survey (MCBS; unweighted unique n=29,023; total person-year observations 58,737).
Principal Findings: The unadjusted, weighted prevalences of cost-related nonadherence in 2004,
2005, 2006, and 2007 were 15.2%, 14.1%, 11.5%, and 10.7%, with confidence intervals around these rates of 13.8-16.6%, 13.1-15.2%, 10.9-12.2%, and
9.9-11.6%, respectively. CRN prevalence did not change significantly between 2004 and 2005, but the significant decline observed between 2005 and
2006 when Medicare Part D was implemented remained stable from 2006 to 2007. Findings were similar for spending less on basic needs, for which the prevalences over the 4 years were 10.6%,
11.1%, 7.6%, and 7.8%, respectively. Logistic regression analyses adjusting for individual demographic and health characteristics confirmed the stability of effects for both CRN and spending less on basic needs in the second year of the program.
Conclusion: The initial declines in CRN and forgoing basic needs in the overall population of
Medicare beneficiaries that we previously found to be associated with the implementation of Part D in
2006 were sustained between 2006 and 2007.
Implications for Policy, Delivery or Practice:
These results validate our earlier finding of a significant improvement in overall access to medicines among Medicare beneficiaries following
Part D program initiation. Whether these stable 2year effects varied by health and income status is a question of major policy importance which we are currently investigating. Our results highlight the persistence of the medication cost burden on patients. Future policy adjustments should ensure that needed therapies medications are affordable for all Medicare beneficiaries.
Funding Source(s): NIA
Theme: Medicare
● Trends & Predictors for Asthma Treatment &
Control: 2002 to 2007
Jeanne Madden, Ph.D.; Yury Vilk, Ph.D.; Allyson
Abrams, M.S.; Michael Schatz, M.D.; Ken
Kleinman, Sc.D.; Greg Sawicki, M.D./M.P.H.
Presented by: Jeanne Madden, Ph.D., Instructor,
Department of Ambulatory Care & Prevention,
Harvard Medical School, 133 Brookline Avenue, 6th
Floor, Boston, MA 02215, Phone: (617) 509-9953;
Email: jeanne_madden@hphc.org
Research Objective: National asthma treatment guidelines recommend regular use of controller medications (including inhaled corticosteroids, longacting beta-agonists, and leukotriene modifiers).
We investigated rates, trends, and risk factors for controller use, uncontrolled asthma events, and controller underuse prior to events.
Study Design: Retrospective, longitudinal study using GLM models and 7 years of claims data linked to US Census and physician registry information.
Population Studied: Asthma patients were
Harvard Pilgrim Health Care members 0–56 years with 2+ outpatient asthma diagnoses or 2+ asthma medication fills per base year. Asthma control status and controller medication use was measured during the 2 years subsequent to each base year
(individual patient may contribute multiple 3-year series). Uncontrolled asthma events were an asthma-related emergency room [ER] visit or hospitalization in a study year (i.e., 3rd year), or high use of rescue therapies (2+ oral steroid courses or 5+ short-acting beta agonist [SABA] courses) within a 12-month period ending in a study year. A total of 28,595 unique patients contributed
54,653 patient study years.
Principal Findings: Acute uncontrolled asthma events were rare and stable over 5 study years
(July 2002 to June 2007; hospitalizations 0.7% of patients on average; ER visits 3.4%). Acute events were associated with young childhood, female gender, and neighborhoods with fewer whites and lower educational attainment. Only ER visits were also associated with neighborhood poverty. By contrast, uncontrolled asthma defined by high use of SABA rescue therapy declined from 32% in
2002-03 to 19% in 2006-07. High SABA use was associated with older age, male gender, and low education, but not with poverty or race. High use of oral steroid rescue therapy was more common among young children and in low education neighborhoods, and also declined over time (16% to 12%). Use of any controller medication per year was higher among patients with uncontrolled asthma (78%) than those with controlled status
(42%), likely reflecting unmeasured differences in underlying illness severity. Nevertheless, underuse of controllers was high prior to uncontrolled asthma events, across age groups. In the 3 months preceding hospitalizations and ER visits, 43% and
53% of patients, respectively, had zero controller medication fills. During 12-month periods with high use of SABAs or oral steroids, 46% and 46% of patients had fewer than 4 controller fills. Use of any controller increased slightly over 5 study years
(52% to 54%), and shifts in controller regimens were noted, including increased use of leukotriene modifiers among children and long-acting beta agonists among adults. Analyses were clustered by patient and primary care provider, with strong clustering effects.
Conclusion: From July 2002 to June 2007, uncontrolled asthma defined by use of rescue therapies declined in this population.
Improvements in preventive treatment may help explain these declines, but underuse of controllers was widespread and persistent.
Implications for Policy, Delivery or Practice:
Secular changes and demographic disparities in uncontrolled asthma status, underuse of preventive treatments, and differences by provider all suggest opportunities for better control of asthma in the future. Further investigation of the demographic and clinical risk factors predicting both underuse and poor control is warranted.
Funding Source(s): AstraZeneca LP
Theme: Prevention and Treatment of Chronic
Illness
● Rehospitalization among Home Health Care
Patients with Heart Failure
Elizabeth Madigan, Ph.D., R.N., F.A.A.N.; Tanya
Scharpf, Ph.D.; Jennifer Riggs, M.S.N, R.N.
Presented by: Elizabeth Madigan, Ph.D., R.N.,
F.A.A.N., Associate Professor, Bolton School of
Nursing, Case Western Reserve University, 10900
Euclid Avenue, Cleveland, OH 44106-4904,
Phone: (216) 368-8532; Email: elizabeth.madigan@case.edu
Research Objective: Older patients with heart failure (HF) have the highest rates of hospital use among all older patients yet the reasons for the high use are not well understood. In addition, patients with HF represent a substantial proportion of patients receiving home health care and also experience high rates of rehospitalization. The purpose of the present study is to describe rehospitalization among patients with HF receiving home health care and determine the explanatory factors associated with rehospitalization among these patients.
Study Design: Using the population of patients with HF receiving home health care in 2005, the following data sets were merged: Outcome and
Assessment Information Set (OASIS) providing demographic and clinical data, MedPAR for hospitalization data, home health standard analytic file for home health care claims information,
Provider of Service file to describe home health care agency characteristics, and Area Resource
File to describe the geographic area factors.
Population Studied: The 2005 national population of Medicare-certified home health agency patients with a primary diagnosis of heart failure.
Principal Findings: There were 108,579 patients in the analytic data set with mean age of 81, 62% were female, and 83% were white, 13% black, 2%
Hispanic and 2% other racial and ethnic minorities or unknown. A substantial proportion (91%; 98,238) had at least one hospital stay in the six months prior to beginning home health care and, of those, 70%
(68,277) had a hospital stay within 14 days of starting home health care. Rehospitalization (having an index stay within 14 days of starting home health care and a subsequent hospital stay within 6 months) occurred among 58% (40,154) but only a small number happened while the patient was receiving home health care (5%; 3,134). Using logistic regression to predict rehospitalization, there were a number of factors significantly associated with more probability of rehospitalization: increasing diagnosis severity (OR = 1.03), increasing dyspnea
(OR = 1.09), increasing anxiety (OR = 1.02), increased levels of ADL (OR = 1.06) and IADL dependency at baseline (OR = 1.06), increasing number of comorbid conditions (OR = 1.12), and an increasing number of depressive symptoms (OR =
1.05). Lower risk of rehospitalization was associated with female gender (OR = .96), older age (OR = .98), poorer rehabilitation prognosis (OR
= .84), and poorer cognitive functioning (OR = .95).
Conclusion: Consistent with other research, this population has very high levels of hospital use with almost all (91%) having one hospital stay prior to starting home health care and many occurring within two weeks of starting home health care
(70%). Rehospitalization is relatively frequent (58%) but most often occurs after home health care has concluded. The factors associated with probability of rehospitalization are consistent with the clinical
presentation of a chronic, progressive disease although some factors associated with less rehospitalization are consistent with increasing frailty or decline (i.e. poorer rehabilitation prognosis or poorer cognitive functioning).
Implications for Policy, Delivery or Practice: The policy implications include a better understanding of those patients at most risk for rehospitalization.
Home health care agencies may want to target services (telephonic follow up or telehealth) to these patients to potentially reduce rehospitalization even after home health care concludes.
Funding Source(s): NIH NHLBI
Theme: Prevention and Treatment of Chronic
Illness
● An Evaluation of a Patient Reported
Outcomes HIV Quality of Life Assessment
Jared Lane Maeda, M.P.H.; DerShung Yang, Ph.D.;
Richard Novak, M.D.
Presented by: Jared Lane Maeda, M.P.H., Ph.D.
Candidate, Division of Health Policy &
Administration, University of Illinois at Chicago
School of Public Health, 1036 North Dearborn
Street, #809, Chicago, IL 60610-7855, Phone:
(617) 877-4561; Email: jared.maeda@gmail.com
Research Objective: The number of individuals infected with HIV/AIDS continues to rise with an estimated 56,000 new cases reported in 2006.
Although significant advancements have been made to treat and manage patients with HIV/AIDS, the health-related quality of life of these individuals remains a concern as they continue to live longer.
As part of the National Institutes of Health’s
PROMIS (Patient-Reported Outcomes
Measurement Information System) initiative to better measure patient-reported symptoms associated with chronic diseases, the Patient
Reported Outcomes IT (PROsIT) Project, developed by Bright Outcome, sought to promote the implementation of patient-reported outcomes for
HIV-infected patients in clinical settings through the use of a comprehensive, practical, and integrated system. The objective of this study was to assess the usefulness of the PROsIT HIV quality of life assessment by comparing the instrument using
Classical Test Theory (CTT) and Rasch analysis methods.
Study Design: A convenience sample was used to recruit and enroll study participants. All participants completed PROsIT, a computerized146-item survey questionnaire on HIV-related quality of life. Data analysis using Classical Test Theory (CTT) methods of reliability and factor analysis were performed using SPSS v. 16.0. In addition, a
Rasch analysis using WINSTEPS was performed to further assess the PROsIT instrument and compare the resultant findings.
Population Studied: A total of 400 adult participants with HIV/AIDS were recruited from outpatient infectious disease clinics at two major academic medical centers in Chicago, Illinois from
2006 thru 2007.
Principal Findings: Evaluating the survey instrument under traditional Classical Test Theory
(CTT) methods yielded a Cronbach’s alpha reliability coefficient of 0.979 indicating high internal consistency. Additionally, factor analysis with varimax rotation revealed one large factor that explained most of the variance in the data (22%) and potentially one other smaller factor. A Rasch analysis of the data revealed a person reliability of
0.95 and a person separation of 4.59, also indicating high reliability of the instrument. In addition, one large factor (27.8%) was found confirming the unidimensionality of the measure.
Further inspection of the data under the Rasch analysis however, revealed several misfitting response categories and items that performed poorly. These items included concealing one’s HIV status and being fearful of disclosing one’s HIV status to others.
Conclusion: Using CTT methods, PROsIT was found to have high reliability and unidimensionality.
A subsequent analysis using Rasch methods confirmed the findings obtained under CTT, however the Rasch analysis also exposed a few response categories and items that did not function optimally.
Implications for Policy, Delivery or Practice:
PROsIT can be a useful and convenient tool for healthcare providers to implement in the routine clinical setting to assess HIV-infected patients.
Additionally, PROsIT can help to further the understanding of HIV health-related quality of life and improve patient care.
Funding Source(s): NIMH
● Peer Mentoring in In-Patient Rehabilitation:
Multiple Stakeholders' Perspectives on Benefits
& Barriers
Susan Magasi, Ph.D.; Christina Papadimitriou,
Ph.D.; Lisa Rosen, M.S.; Kristine Cichowski, M.S.,
T.R.S.
Presented by: Susan Magasi, Ph.D., Research
Scientist & Research Assistant Professor, Center for Outcomes, Research & Education & Department of Physical Medcine & Rehabilitation, NorthShore
University HealthSystems & Northwestern
University, 1001 University Place, Suite 100,
Evanston, IL 60201, Phone: (224) 364-7437;
Email: smagasi@northshore.org
Research Objective: To examine the peer mentoring process in in-patient rehabilitation from the perspectives of multiple rehabilitation stakeholders, including: rehabilitation in-patients, peer mentors with disabilities, and rehabilitation providers. Perceived benefits, extent of use and barriers to using peer mentors as an adjunct to clinical practice were evaluated.
Study Design: This study used a three phase mixed methods design including in-patient satisfaction surveys, peer mentor focus groups, and an online provider survey.
Population Studied: Rehabilitation stakeholders from a free standing rehabilitation hospital in a large
Midwestern American city, including: 1) in-patients with spinal cord injuries (n=50), 2) trained peer mentors with spinal cord injuries, strokes, amputations, and traumatic brain injuries (n=19), and 3) rehabilitation professionals (n=251).
Principal Findings: All stakeholder groups viewed peer mentoring as an important adjunct to in-patient rehabilitation offering people with disabilities first person insights on what it is like to live with a disability. Peer mentors provide informational, emotional and appraisal support that can help ease disability transitions, including return to community living. Rehabilitation professionals identified a lack of information about peer mentoring processes, the lack of communication between peers and clinical staff, and concerns over the adequacy of mentor training as primary barriers to use of peer mentors in in-patient rehabilitation.
Conclusion: Peer mentoring can be an important adjunct to in-patient rehabilitation for people with acquired disabilities. There is a need for greater communication with and education of rehabilitation professionals on how to effectively use peer mentors in their practice.
Implications for Policy, Delivery or Practice:
Accrediting agencies, such as the Commission on
Accreditation of Rehabilitation Facilities (CARF), mandate that peer mentoring be offered part of a comprehension rehabilitation services. For such programs to be fully integrated into rehabilitation practice there needs to be greater communication and education of rehabilitation providers about the benefits, training and processes of peer mentoring.
Funding Source(s): National Institute on Disability and Rehabilitation Research
Theme: Prevention and Treatment of Chronic
Illness
● Multitrait-Multimethod Analysis of Three
Generic Preference-Based Health-Related
Quality of Life Measures in the National Health
Measurement Study
Gregory Maglinte, B.S., M.P.H., Ph.D.; Robert
Kaplan, Ph.D.; Ron Hays, Ph.D.
Presented by: Gregory Maglinte, B.S., M.P.H.,
Ph.D., Student, Health Services, University of
California, Los Angeles School of Public Health,
5430 Royer Avenue, Woodland Hills, CA 91367,
Phone: (626) 264-2409; Email: greg.maglinte@ucla.edu
Research Objective: Generic preference-based health-related quality of life (HRQOL) measures are required to calculate quality-adjusted life expectancy for cost-utility analysis. However, different measures may not yield the same results, challenging the meaning of comparisons of incremental cost-effectiveness ratios (ICERs) in league tables. This study compares corresponding domain scores from different HRQOL measures to evaluate the extent to which they tap into the same or unique constructs.
Study Design: The National Health Measurement
Study (NHMS) is a random-digit dial telephone survey of adults in the United States. The crosssectional survey was conducted by the University of
Wisconsin Survey Center from June 2005 to August
2006. The Medical Outcomes Study Short Form-36 version 2 (SF-36), EuroQol (EQ-5D), and Health
Utilities Index Mark 3 (HUI-3) were administered to all study participants. Mean domain and preferencebased scores were calculated. Repeated measures one-way analysis of variance (ANOVA) followed by
Bonferroni-corrected paired t-tests were conducted to analyze within-group differences between preference-based scores of the SF-36, HUI-3, and
EQ-5D. Convergent and discriminant validity were evaluated by a multi-trait multi-method (MTMM) analysis of the three HRQOL instruments across the attributes of physical functioning (SF-36 physical functioning, HUI-3 ambulation, and EQ-5D mobility), mental health (SF-36 mental health, HUI-
3 emotion, and EQ-5D anxiety/depression), and pain (SF-36 bodily pain, HUI-3 pain, and EQ-5D pain/discomfort).
Population Studied: A total of 3,844 adults from residential households in the continental United
States participated in the NHMS. Participants were
35-89 years of age and represented various health states, chronic conditions, and personal characteristics.
Principal Findings: Mean scale scores of the SF-
36, HUI-3, and EQ-5D, respectively, ranged from
47.2 (+/-10.7) to 53.4 (+/-10.2) [46.9 (+/-11.0) and
53.0 (+/-10.1) for PCS and MCS], 0.928 (+/-0.121) to 0.996 (+/-0.028), and 1.0171 (+/-0.269) to 1.596
(+/-0.587). Mean preference-based scores obtained from the SF-36 (calculated from the SF-6D), HUI-3, and EQ-5D were 0.765 (+/-0.144), 0.766 (+/-0.275), and 0.838 (+/-0.173), respectively (p < 0.001 for repeated measures one-way ANOVA; p < 0.001 for
Bonferroni-corrected paired t-tests comparing EQ-
5D to both SF-36 and HUI-3). The MTMM matrix had average validity and off-diagonal correlations of
0.622 and 0.404, indicating high convergent validity.
MTMM analysis also showed 91% of convergent validity correlations to be significantly larger (p <
0.05) than relevant other correlations in the MTMM matrix, providing substantial support for discriminant validity.
Conclusion: Although the mean preference-based score obtained from the EQ-5D was different from the SF-36 and HUI-3, MTMM analysis demonstrated good support for construct validity among analogous domains of these measures.
Further analysis is required to better understand
how similar elements of different generic HRQOL instruments impact preference-based scores.
Implications for Policy, Delivery or Practice:
When comparing treatment choices based on incremental cost-effectiveness ratios, careful consideration should be given to the preferencebased measure used in quality-adjusting life expectancy.
Theme: Consumer Choices in Health Care
● Perceived Neighborhood Quality, Individual
SES, Civic Participation & Health, Mental Health
& Drug Abuse/Alcohol Services: An Analysis of
Adults in the City of Los Angeles, CA
Jennifer Magnabosco, Ph.D.; Jennifer,
Magnabosco, Ph.D.; Brianne Barclay, M.A.;
Stephen Nuno, Ph.D.
Presented by: Jennifer Magnabosco, Ph.D.,
Associate Director & Senior Research Associate,
Leavey Center for the Study of Los Angeles, Loyola
Marymount University, 516 San Vicente Boulevard,
#304, Santa Monica, CA 90402, Phone: (310) 948-
1758; Email: jlmagnabosco@yahoo.com
Research Objective: While it is well established that neighborhoods matter for health, limited attention has been paid to understanding how collective properties of neighborhood environments explain such quality of life outcomes. Nonetheless, research shows that perceptions of neighborhood quality impact on well-being and that community processes like collective efficacy (e.g., civic participation) may also be linked to health outcomes. Using the 2007 Los Angeles Riots
Telephone Survey, relationships between neighborhood quality, SES and civic participation on adult ratings, A-F, of health, mental health and drug abuse/alcohol services in the City of Los
Angeles were investigated.
Study Design: Cross-sectional stratified random telephone survey of 1,651 Los Angeles City residents in March/April 2007 in English, Spanish and Korean, with a margin of error +/- 3 percent.
Two overall models were tested using probit regression controlling for perception of overall direction of Los Angeles, and direction of neighborhood.
Population Studied: Residents in the City of Los
Angeles.
Principal Findings: The model controlling for perception of direction of neighborhood was significant. A positive perception of neighborhood direction was associated with higher grades of health, 0.114, p<0.05, and drug/alcohol, 0.135, p<0.05, services. U.S. born were likely to grade health, 0.045, p<0.01, mental health, 0.267, p<0.05, and drug/alcohol, 0.316, p<0.05, services higher, whereas non-citizens were likely to grade only mental health services higher, 0.296, p<0.05. Older respondents were likely to grade drug/alcohol services higher 0.084, p <0.01. Respondents who perceived that neighborhood councils-NCsencouraged new people to be involved in neighborhood issues, or improved quality of life, were likely to grade mental health, 0.243, p<0.05 and health services higher 0.270, p<0.05, respectively. Participation in NCs was asociated with a lower health services grade -0.261, p<0.05.
Conclusion: Neighborhood and collective efficacy factors-such as perception that NCs improve quality of life, participation in NCs, perception of neighborhood direction and perception that NCs encourage new participants--may matter for health, drug/alcohol and mental health services in the city of Los Angeles. Individual factors that continue to matter for services include citizenship status and age.
Implications for Policy, Delivery or Practice:
Since quality of neighborhood life, SES and civic participation were associated with a range of grades for health, mental health and drug-alcohol servces, meeting quality of life needs in Los
Angeles/similar urban settings, suggests a comprehensive community-based approach. NCs may be a collective mechanism to help improve perceptions of human services and their access, and therefore increase utilization.
Funding Source(s): Research Center Annual
Contributions
Theme: Behavioral Health
● Consequences of Inefficient Medical Practice:
Uninsurance & Practice Variations
Bisundev Mahato, A.B.; Sherry Glied, M.A., Ph.D.;
Bisundev Mahato, A.B.
Presented by: Bisundev Mahato, A.B., Senior
Programmer, Deaprtment of Health Policy &
Management, Columbia University, 600 W 168th
Street, 6th Floor, New York, NY 10032, Phone:
(212) 305-4031; Email: bm2140@columbia.edu
Research Objective: There has been much research done documenting significant variation in health care practice patterns across the United
States. Previous research has also found that communities with higher health care costs have higher rates of uninsurance. This study examines the relationship between practice patterns and health insurance coverage.
Study Design: Data Sources, Study Setting, and
Method: Our analysis is based on 1996-2005 data from the Dartmouth Atlas Project (DAP) and 1996-
2007 data from the Current Population Survey
(CPS). Three- year moving average county level data from the CPS are matched to each year of
Health Referral Region (HRR) level data from DAP.
For example, we match average data from 2003-
2005 CPS years to the data from the 2003 DAP.
DAP makes available the zip-code to HRR match.
We merge zip-code-to-county data (from 1990
Census) with the county level CPS data, and then we merge CPS data with the HRR data using zip-
code as the link variable. Out of the total 306 HRRs, the number of HRRs that had matched data from
CPS ranged from 171 in 2002 to 218 in 2003-2005 years.
Study Design: Ordinary Linear Regression (OLS) and instrument variable techniques are used to study the effects of variation in health care costs on health coverage pattern as well as the effect of coverage patterns on health care costs. In analyses of the effects of health care costs on insurance coverage, we use days of ICU care in the last six months of life among Medicare patients as an instrument for overall health care costs. In analyses of the effects of coverage on costs, we use small firm and self-employment rates at the county level as an instrument for the uninsurance rate in analyses of the effect of insurance coverage on Medicare expenditures. We treated the data as repeated cross-sections, clustering on HRR.
Population Studied: Variation in insurance coverage patterns was measured using data on the
22-64 year old population. Variation in health care costs and practices was measured using data on elderly Medicare beneficiaries (65+). We controlled for demographic characteristics of the 22 to 64 years old population in the analyses.
Principal Findings: Using firm size and selfemployment as instrumental variables for uninsurance, we find that variation in the rate of uninsurance among 22-64 year olds has no effect on Medicare costs and practice patterns among the elderly population. We find that an increase of one day in ICU days during the last six months of life is associated with an increase of 0.56 percentage points in the uninsurance rate for an average HRR,
CI =(0.36, 0.76), p<0.01. Using ICU days as an instrument variable for total Medicare expenditure, we find that a $1,000 increase in total Medicare expenditure leads to a 1.2 percentage point increase in the uninsurance rate for an average
HRR, CI=(0.8, 1.6), p<0.01.
Conclusion: Health insurance coverage is significantly affected by health care practice patterns.
Implications for Policy, Delivery or Practice:
Improving the efficiency of health care delivery is not only likely to reduce health care cost, but it is also likely to reduce the rate of uninsurance.
Funding Source(s): CWF
Theme: Coverage and Access
● Impact of Regulation on Inpatient
Rehabilitation Facilities: 2002-2007
Trudy Mallinson, Ph.D.; Trudy, Mallinson, Ph.D.;
Larry Manheim, Ph.D.; Orit Almagor, M.A.; Kyle
Seanor, B.A.; Holly DeMark, B.A.; Allen
Heinemann, Ph.D.
Presented by: Trudy Mallinson, Ph.D., Associate
Director, Center for Rehabilitation Outcomes
Research, Rehabilitation Institute of Chicago, 345
East Superior Street, Onterie 960, Chicago, IL
60611-2654, Phone: (312) 238-1623; Email: trudy@northwestern.edu
Research Objective: Inpatient rehabilitation facilities (IRFs) provide comprehensive, intensive rehabilitation services to patients with physical disabilities. To be Medicare-certified, IRFs must provide 24-hour rehabilitation nursing and MD services, and 3 hours of rehabilitation therapy per day to each patient. Medicare reimburses IRFs using aprospective payment system (PPS) and applies a “75% rule" to define an IRF. The latter, phased in since 2005, requires that 75% of patients admitted be in one of 13 diagnostic groups although compliance levels are currently set at 60%. This rule has made it increasingly difficult for IRFs to admit sufficient patients to maintain historic occupancy levels.
Study Design: This retrospective study identified closings and change in bed-size among IRFs 2002 through 2007. Data were extracted from the
Hospital Cost Report Information System (HCRIS) and supplemented with data from MedPAR, IRF-
PAI, and area resource file data sets. For IRFs existing in 2002, logistic regression methods predict odds of closing, as a function of IRF status, percent compliance, occupancy rate, for-profit status, bed size, urban/rural location, proximity to SNF, and geographic region.
Population Studied: Inpatient rehabilitation facilities certified by Medicare between 2002 through 2007.
Principal Findings: There has been a significant drop in inpatient bed days since 2005 and also in occupancy rates. IRFs least likely to close had a higher percentage of compliant cases, were more likely to have higher occupancy rates, and more likely to have smaller change in inpatient days.
Conclusion: The progressive implementation of the 75% rule had a negative impact on the supply of
IRF services. By mandating admission of patients in generally more severe diagnostic categories, this rule creates greater competition among providers for a limited number of patients resulting in fewer bed days and a higher probability of closures.
Implications for Policy, Delivery or Practice: The direct regulation of which patients could go to IRFs had much larger effects on the availability of IRF services than what we previously found for the introduction of prospective payment. The subsequent negotiations over these regulations will likely be the major determinant the availability of inpatient rehabilitation facility services in the future.
Funding Source(s): National Institute for Disability and Rehabilitation Research
● Developing Qualitative Methods to Study
Nurse/Physician Communication in the
Intensive Care Unit
Milisa Manojlovich, Ph.D., R.N., C.C.R.N.; Jane
Forman, Sc.D., M.H.S.; Carol Fletcher, Ph.D., R.N.;
Rosalind Keith, M.H.S.A.; Sanjay Saint, M.D.,
M.P.H.; Sarah Krein, Ph.D., R.N.
Presented by: Milisa Manojlovich, Ph.D., R.N.,
C.C.R.N., Assistant Professor, School of Nursing,
University of Michigan, 400 South Ingalls, Ann
Arbor, MI 48109, Phone: (734) 936-3055; Email: mmanojlo@umich.edu
Research Objective: Ineffective communication between nurses and physicians can adversely affect patient care, especially in a complex environment such as an intensive care unit (ICU).
Unfortunately, we know little about communication between nurses and physicians in the ICU setting, how work culture characteristics might affect communication, or how to study these important but not easily measured issues. We therefore sought to first determine feasibility, then develop qualitative procedures, and conduct preliminary data analysis for understanding how and what information is communicated between nurses and physicians in the ICU.
Study Design: We conducted 4 observations and semi-structured interviews with 4 nurses and 4 physicians in 3 ICUs in the same academicallyaffiliated facility: a medical ICU, surgical ICU, and a thoracic ICU. As the study progressed we developed, tested, and refined a data collection protocol and tools, and preliminary data analysis techniques based on experience in the field and team discussion. A pair of observers: one nurse, who focused on communication content, and one non-nurse, who focused on communication context conducted observations. Observed nurses and at least one of the physicians with whom they communicated participated in retrospective, semistructured interviews. Interviews focused on general communication issues and on observed communication, and were audio-recorded and transcribed. Data analysis was iterative, and included codebook development, case-based analysis, and identification of themes for subsequent exploration.
Population Studied: The study population was registered nurses and all levels of physicians (first year residents to attending physicians).
Principal Findings: The protocol we developed includes procedures specific to the ICU environment for entry into the field, obtaining informed consent before observation periods, and conducting observation of team interactions. Our observation tool records observer description and interpretation of interactions around each patient, and semi-structured interview guides are tailored to type of interviewee (i.e., nurse or physician). Our analytic approach includes collation of multiple data sources, and several units of analysis (patient, unit, and clinical issue) to identify and better understand communication events in the ICU, and the interpretation of those events by the nurse and physician involved. The observation process suggested the following: 1) communication patterns among the 3 units were markedly different; 2) opportunities for communication differed in each unit; 3) certain issues (e.g., nutrition status, pain and/or sedation management, patient “road trips”) may contribute to communication problems; and 4) nurses did not feel they could speak freely in some situations and therefore withheld important clinical information.
Conclusion: Although communication is a fundamental part of the care process it is not easily studied using many of the research methods commonly used in health services research.
Innovative strategies that employ qualitative methods, including observations of nurses and physicians followed by interviews, can help characterize and understand this vital process.
Implications for Policy, Delivery or Practice:
Better understanding the dynamics of communication and the context in which it occurs, using rigorous qualitative methods, provides an opportunity for developing strategies to promote effective communication between nurses and physicians. Our findings suggest that opportunities for communication are being missed and patient care may thereby suffer.
Funding Source(s): VA
Theme: Quality and Efficiency: Measurement
● Understanding Pathways to Disparities in
Hypertension Control: Race & Treatment
Intensification
Meredith Manze, M.P.H.
Presented by: Meredith Manze, M.P.H., Research
Coordinator, Health Policy & Management, Boston
University School of Public Health, 801
Massachusetts Avenue, CT 2098, Boston, MA
02118, Phone: (617) 414-8441; Email: meredith.manze@bmc.org
Research Objective: Our objectives were to explore the extent of racial disparities in treatment intensification (TI) in hypertension care and elucidate how patient characteristics, beliefs and behaviors, provider characteristics, and patientclinician interactions may contribute to racial disparities in TI.
Study Design: Racial disparities in hypertension control and hypertension-related outcomes persist despite efforts to improve control and reduce disparities. TI, the increase in medication for patients with elevated blood pressure (BP), has been linked to improved BP control. Disparities in
TI may contribute to racial disparities in BP control, but this issue has not yet been examined. In addition, TI is best viewed as something that patients and providers accomplish in collaboration.
A more complete understanding of relational determinants of TI might help in designing interventions to increase TI and thus improve BP control. TI was our main dependent variable,
calculated using the formula: (visits with medication changes – visits with elevated BP) / number of clinic visits. Independent variables were collected via patient and provider questionnaires and medical record, and included patient and provider sociodemographics, patient clinical characteristics, health beliefs, perceptions of provider, dietary habits, medication adherence, and provider counseling. We performed a series of linear regressions to investigate the effect of race upon TI, first alone, and then after the addition of patientlevel covariates.
Population Studied: We collected data from 819
Black (58%) and White (42%) patients with hypertension, managed in primary care at an urban safety-net hospital.
Principal Findings: Black patients had higher rates of uncontrolled BP (p<.01). In unadjusted analyses,
Black patients had significantly less TI compared to
Whites, equivalent to approximately one fewer therapy increase per 14 clinic visits (-0.31 vs. -0.24, p < 0.001). After including patient sociodemographic variables, clinical characteristics, and provider age/gender into our regression model, race remained significantly related to TI (p=0.04).
In the final model, after including patient beliefs, perceptions of provider, dietary habits, provider counseling and medication adherence, race was no longer significant (p=0.46). This final model revealed several determinants of TI. Increased patient concern about BP medications was related to lower TI (p=0.05), as was more provider counseling (p=0.04). Patients viewing their hypertension less seriously received more TI
(p=0.01). Older patients or those with private insurance had less TI, while patients with hyperlipdemia or public insurance had increased TI.
Conclusion: Black patients have significantly lower rates of TI compared to Whites, even after accounting for differences in their clinical and sociodemographic status. These differences may contribute to racial disparities in BP control and hypertension-related cardiovascular outcomes.
However, racial disparities in TI were explained by patient characteristics, health beliefs, and provider counseling. Patient concerns about BP medications, and more provider counseling, seem to inhibit rather than promote TI, indicating provider reluctance to intensify treatment when patients express concerns, and suggesting that providers may substitute counseling for TI. The racial differences in BP control in this sample suggest that this is not an effective strategy.
Implications for Policy, Delivery or Practice:
Improved patient-provider communication, targeted towards addressing patient concerns about medications, may have the potential to reduce racial disparities in TI and ultimately, BP control.
Funding Source(s): National Heart, Lung & Blood
Institute
● Mammography Result Notification Letters:
Can Most Women Understand Them?
Erin Marcus, M.D., M.P.H.; Yanisa Del Toro, M.D.;
Margaret Pereyra, Dr.P.H.; Ada Pat Romilly, M.D.;
Maria Victoria Velasquez, M.D.; Lee Sanders, M.D.,
M.P.H.
Presented by: Erin Marcus, M.D., M.P.H.,
Associate Professor of Clinical Medicine, Institute for Women's Health, University of Miami Miller
School of Medicine, 1120 Northwest 14th Street,
NW, Suite 1140 (M-716), Miami, FL 33136, Phone:
(305) 243-3576; Email: emarcus@med.miami.edu
Research Objective: This study's purpose was to assess the readability of the sample notification letters that many mammography centers use as a template for the letters they send patients about their results. The 1998 Mammography Quality
Standards Reauthorization Act requires that all licensed mammography facilities mail each patient such letters in an effort to prevent patients from being lost to follow-up. The law specifies that the letters should be written "in terms easily understandable by a lay person."
Study Design: To assess text readability, we used the Flesch Kincaid reading ease and grade level scales, which incorporate total words per sentence and total syllables per word. A Flesch reading ease score of 60 or above and grade level score of 4th to
6th grade are usually considered acceptable for health materials. To assess document suitability, we used the Suitability Assessment of Materials
(SAM), which measures content clarity, layout, graphics and typography. A score of >40 % is considered adequate. Two independent observers completed the SAM on each document, and mean scores were used for analysis. One-way analysis of variance with Bonferroni multiple comparison tests was used to assess differences in the primary outcomes by diagnostic category (as indicated by
BI-RADS level) or by the source of the letter (ACR,
MRS or Penrad).
Population Studied: We analyzed 55 sample mammography result notification letters made available to imaging centers by the American
College of Radiology (ACR) as well as by the two most widely used mammography transcription services, the MRS and Penrad systems.
Principal Findings: 55 letters were analyzed: 8 from the ACR, 10 from the Penrad system, and 37 from the MRS system. The Flesch reading ease ranged from 25 to 61, with a mean of 42 (S.D., 7.9).
The Flesch Kincaid grade level score ranged from
6.7 to 13.5, with a mean of 10 (S.D., 1.2). The mean SAM score was 29%, with a range of 24% to
38% (S.D., 2 %). Reading ease, grade level, and
SAM score did not vary significantly by diagnostic category. Mean reading ease, grade level, and
SAM scores did differ by source of letter (all p<0.05), but none were in acceptable readability range.
Conclusion: The letters evaluated in this analysis were written at readability and suitability levels that are too difficult for many patients to understand.
These findings suggest there is a need to develop alternate ways of communicating mammography results in a way that accounts for low health literacy.
Implications for Policy, Delivery or Practice:
Lack of timely mammography follow-up remains problematic, especially in safety-net systems serving low-income women. Other research suggests that communication failures contribute to lack of follow-up. Our findings suggest mammography result notification letters are not written at a level that most patients can understand.
More research is needed to assess patient understanding of mammography results and the role of patient understanding in appropriate followup, and to develop alternative means of notifying women about their results.
Theme: Gender Health
● A Multidisciplinary Women's Clinic to Improve
Patient Satisfaction in an Academic Safety-Net
Health System
Erin Marcus, M.D., M.P.H.; Yvonne Diaz, M.D.;
Jorge Garcia, M.D.; Meghan Mayhew, M.P.H.;
Margaret Pereyra, Dr.P.H.; Thomas Hooton, M.D.
Presented by: Erin Marcus, M.D., M.P.H.,
Associate Professor of Clinical Medicine, Institute for Women's Health, University of Miami Miller
School of Medicine, 1120 Northwest 14th Street,
Suite 1140 (M-716), Miami, FL 33136, Phone:
(305) 243-2576; Email: emarcus@med.miami.edu
Research Objective: The purpose of this study is to assess the effect of a multidisciplinary women's health clinic on female patients' satisfaction with their care in a large academic safety-net health system in Miami.
Study Design: The Women´s Comprehensive
Clinic (WCC) was developed to address the diverse health needs of women in Miami's large safety-net system, Jackson Memorial. The clinic occurs once weekly and is staffed by first year internal medicine residents on their ambulatory rotation. A gynecologist and two women´s health-focused general internists, together with rotating general internal medicine (GIM) faculty, supervise. The clinic is open to women 18 and older who are new to the Jackson system and have not had recent gynecologic screening. Patients in the WCC and
GIM clinics were verbally administered a survey that included items on demographics, perception of education received during the visit, as well as the
Primary Care Satisfaction Survey for Women
(PCSSW), a validated survey tool. Two-sample t testing was performed to compare responses.
Population Studied: 223 female patients completed the survey (104 WCC, 119 GIM). 70 % of surveyed WCC patients were uninsured and 6 % received Medicaid. 84 % were Hispanic and 14 % were black. WCC patients´ average age was 46 years. Patients in the two clinics did not differ significantly in age nor in self-reported general health status.
Principal Findings: WCC patients were more likely to report high satisfaction with their health professional´s ability to answer questions sensitively (69% vs. 53%, p<0.05) and clearly (68% vs. 57%, p<0.05), to take their concerns seriously
(70% vs. 57%, p<0.05), to help them feel comfortable talking about their concerns (69 % vs.
58%, p<0.05), and to explain different care options
(68% vs. 56%, p<0.05). They were also more likely to report higher satisfaction with the chance to get both general and gynecological health care at the clinic (p<0.05) and the health professionals´ knowledge of women´s health issues (p<0.05).They were more likely to report having discussed sexually transmitted infections, calcium intake, personal safety and menopausal management with their physicians (all p<0.05).
Conclusion: A multidisciplinary approach, in which gynecologists and internists work together in one setting, can significantly improve the satisfaction and education of female patients in an ethnically diverse, safety-net health center.
Implications for Policy, Delivery or Practice:
Women´s health care in large safety-net settings is often fragmented, requiring patients to attend multiple appointments at different clinics for their care. Cross-sectional surveys in other settings have demonstrated higher satisfaction levels among female patients attending comprehensive women´s clinics versus those attending traditional primary care clinics. This study demonstrates that such an approach also can significantly improve patient satisfaction in an ethnically and linguistically diverse group of patients in a safety-net setting. Future studies will explore the clinic´s impact on receipt of recommended screening and prevention measures as well as on physician attitudes and skills in women´s health.
● Physician Satisfaction with Chronic Illness
Care: A Randomized Trial of Guided Care
Jill Marsteller, Ph.D., M.P.P.; Yea-Jen Hsu, M.H.A.;
Lisa Reider, M.H.S.; Lya Karm, M.D.; Daniel
Scharfstein, Sc.D.; Chad Boult, M.D., M.P.H.,
M.B.A.
Presented by: Jill Marsteller, Ph.D., M.P.P.,
Assistant Professor, Health Policy & Management,
Johns Hopkins Bloomberg School of Public Health,
624 South Broadway, Room 433, Baltimore, MD
21205, Phone: (410) 614-2602; Email: jmarstel@jhsph.edu
Research Objective: Primary care physicians’ satisfaction with their care of chronically ill older patients is poorly understood, but may be suboptimal because meeting the complex needs of
such patients is often difficult. New models may be needed to care for this population. Our objective is to evaluate the effect of Guided Care on primary care physicians’ satisfaction with their care of older patients with chronic conditions, as well its effects on physicians’ knowledge about such patients, time spent on their care, and coordination of their care.
Study Design: We conducted a cluster-randomized controlled trial of Guided Care. All eligible primary care physicians in 14 “pods” (teams of 2-5 physicians and their multi-morbid older patients) agreed to participate (n=49). Pods were randomly assigned to provide either Guided Care or a usual care. All physicians were surveyed at baseline and a year later. Analysis included a comparison of baseline characteristics in the intervention and control groups. Several multivariate linear regression models evaluated the relationship between intervention or control status and one-year satisfaction, time spent on chronic care tasks, knowledge of the patient and coordination outcomes, adjusted for baseline values of the outcomes and whether the practice was HMOowned or not. We calculated effect sizes using
Hedges’ d (to account for the small sample size and the multivariate context).
Population Studied: Eighty-four percent of physicians responded at both baseline and one year. Data from the 38 physicians who completed both surveys were analyzed to assess the effects of
Guided Care.
Principal Findings: Compared to the physicians in the control group, the physicians in the Guided
Care group rated their satisfaction with patient/family communication and their knowledge of their chronically ill patients’ clinical conditions significantly higher (p<0.05). Each of these showed medium-to-large effect sizes (d > 0.70). Guided
Care physicians’ ratings on all other scales and items also exceeded those of the control physicians, but these differences were not statistically significant. Contrary to hypotheses, there were no significant effects of the intervention on satisfaction with management of chronic care, physicians’ time spent on chronic care tasks, knowledge of the patient’s personal characteristics, or care coordination activities.
Conclusion: Guided Care provides significant benefits to physicians and physician organizations in the areas of communication with chronically ill patients and their families and physicians’ knowledge of their chronically ill patients’ clinical conditions. We may not have been able to detect weaker effects of Guided Care due to the small sample of physicians and the short (one year) timeframe of this particular comparison.
Implications for Policy, Delivery or Practice:
Physicians who participate in demonstrations of the
“medical home” may wish to consider implementing
Guided Care in their practices to improve their care of older patients with chronic illness and their satisfaction with the care they are able to provide.
Funding Source(s): John A. Hartford Foundation
Theme: Prevention and Treatment of Chronic
Illness
● Association Between Surgeon Volume &
Reoperation Rates Following Lumbar
Decompression Operation for Herniated Disc
Brook Martin, M.P.H.; Richard Deyo, M.D., M.P.H.;
Patrick Heagerty, Ph.D.; David Flum, M.D., M.P.H.;
Thomas Wickizer, Ph.D.; Sohail Mirza, M.D.,
M.P.H.
Presented by: Brook Martin, M.P.H., Health
Services Researcher; Ph.D. Student, Orthopaedics
& Sports Medicine; Health Services, University of
Washington, 325 Ninth Avenue, Box 359736,
Seattle, WA 98104, Phone: (206) 744-1803; Email: bim@u.washington.edu
Research Objective: For many common surgical procedures, high-volume surgeons have lower rates of reoperation; however, this relationship has not been examined in lumbar spine decompression surgery for herniated discs. Repeat operations generally imply persistent symptoms, progression of disease, or treatment complications.
Understanding any association between surgical volume and reoperations may help to improve surgical safety, inform patients about the risks of surgery, and be used with pay-for-performance, regionalization and quality improvement programs.
We examined whether surgeon procedure volume was associated with 4-year, 1-year and 30-day reoperation rates following lumbar decompression surgery for a herniated disc.
Study Design: We calculated the cumulative incidence of second lumbar spine operations in a retrospective cohort from an inpatient discharge registry that included all non-federal acute-care hospitals in Washington State. Lumbar spine surgeries were identified using a validated algorithm based on the International Classification of Disease,
Version 9, Clinical Modification (ICD-9-CM). We categorized surgeons into three equal numbered groups, based on the volume of lumbar surgery they performed from 2000-2002. We then examined the association between surgeon volume and reoperation rate using a multilevel mixedeffects logistic regression model, clustering for patients within surgeons, and adjusting for age, sex, co-morbidity and insurance. To provide meaningful provider level information, we excluded surgeon who performed fewer than 5 lumbar spine procedures per year.
Population Studied: We included adults who underwent an initial inpatient lumbar decompression operation for a herniated disc from
2000 through 2002. We excluded those with previous spine surgery, cancer, spinal deformity, arthritis or pregnancy. From 2000-2002, a total of
8,090 patients received lumbar decompression for herniated discs; operations were performed by 153
surgeons. The mean number of patients per surgeon in the high-volume group was 303, compared to 111 and 35 in the middle and lowvolume groups, respectively.
Principal Findings: The high-volume surgeon group had a lower proportion of patients on workers’ compensation and public insurance compared to other groups, but were similar with regard to age, sex, and comorbidity. The adjusted multilevel mixed-effects logistic regression found that relative to low-volume surgeons, high-volume surgeons had a significantly higher odds of reoperation at 4-years (OR 1.42 95% CI 1.08 –
1.86), and non-significantly higher odds at 1-year
(OR 1.19 95% CI 0.84 – 1.70) and 30-days (OR
2.82 95% CI 0.68 – 11.83).
Conclusion: The likelihood of reoperation following a lumbar decompression for herniated disc is substantial, and high-volume surgeons had the highest rate of reoperation, even after adjusting for patient characteristics. We cannot exclude the possibility that differences in unmeasured characteristics account for some of the association we observed. However, these counterintuitive results are contrary to other studies of surgical volume-outcome relationships. They may indicate that high-volume surgeons have a lower threshold for performing both initial and subsequent operations, or have greater use of techniques associated with higher complication rates.
Implications for Policy, Delivery or Practice:
These findings suggest a need to improve indications for initial and repeat disc surgery; to involve patients in informed decision-making; and to determine the safest and most effective surgical techniques.
Funding Source(s): Institutional endowment
Theme: Quality and Efficiency: Policies and
Incentives
● Variation in Reoperation Rates Following
Decompression Surgery for Lumbar Herniated
Disc
Brook Martin, M.P.H.; Richard Deyo, M.D., M.P.H.;
Patrick Heagerty, Ph.D.; David Flum, M.D., M.P.H.;
Thomas Wickizer, Ph.D.; Sohail Mirza, M.D.,
M.P.H.
Presented by: Brook Martin, M.P.H., Health
Services Researcher, Orthopaedics & Sports
Medicine; Health Services, University of
Washington, 325 Ninth Avenue, Box 359736,
Seattle, WA 98104, Phone: (206) 744-1803; Email: bim@u.washington.edu
Research Objective: Laminectomy and diskectomy are spinal decompression procedures performed to correct herniated discs or to alleviate back pain or sciatica caused by neural impingement. Repeat operations following lumbar decompression surgery may sometimes indicate poor quality of care or missed opportunities to better coordinate care. Variation in surgeon reoperation rates following lumbar decompression for herniated discs has not previously been reported, but may be useful to inform policies aimed at improving surgical safety and provide a benchmark for examining outcomes. We sought to determine the rate of reoperation following decompression surgery for herniated discs in
Washington State, and report the variation in these rates among individual surgeons using several common measures of variability.
Study Design: We calculated the 4-year, 1-year, and 30-day cumulative incidence of subsequent lumbar spine surgery using a retrospective cohort from an inpatient discharge registry that included all non-federal acute-care hospitals performing lumbar decompression surgery for herniated discs from
2000-2002. To account for clustering of patients
“within” surgeons, we used a multilevel mixedeffects logistic regression, adjusting for age, sex, co-morbidity and insurance type, to estimate the statewide and individual surgeon reoperation rates.
To provide meaningful analysis for provider profiling, we excluded cases performed by surgeons who performed fewer than 5 lumbar surgeries per year. Several common measures of variability for the reoperation rates were reported.
Population Studied: We included adults who underwent an initial inpatient lumbar decompression operation for a herniated disc. We excluded those with previous spine surgery, cancer, arthritis or pregnancy. A total of 8,090 patients received inpatient lumbar decompression for herniated discs; operations were performed by 153 surgeons. The mean age of the cohort was 46 years (SD 13.2), with 39% female; 21% on received workers’ compensation insurance, and 14% reported any comorbidity. The average number of decompressions for herniated discs performed by surgeons over the three years was 231, ranging from 16 to 573.
Principal Findings: After adjusting for patient differences and clustering, we found a greater than four-fold difference in the variation among surgeon reoperation rates at 4-years (ranging from 9.5% to
43.6%); a three-fold variation at 1-year (ranging from 4.2% to 13.4%); and a 2.2-fold variation at 30days (ranging from 0.5% to 1.1%). The estimated
4-year, 1-year, and 30-day cumulative incidences of reoperation were 20.5% (SD 0.05), 7.6% (SD 0.18),
0.7% (SD 0.09), respectively.
Conclusion: Our study provides a benchmark for monitoring surgical safety following decompression surgery for herniated disc. The high variation in reoperation rates that we observed with long term follow-up may reflect uncertainty regarding patient selection, varying complication rates, progression of disease, or treatment failures.
Implications for Policy, Delivery or Practice:
Comparing reoperation following decompression surgery may allow comparisons of safety outcomes across providers. This is an attractive target for
reporting programs because variations may reflect overuse of services, variation in the amount and quality of care, and variations in outcomes. Further research to identify factors that account for the variation in surgeon reoperation rates may provide an opportunity to improve clinical care and reduce unnecessary surgery.
Funding Source(s): UW Endowment
Theme: Quality and Efficiency: Policies and
Incentives
● Trends in Health Care Expenditures,
Utilization & Health Status among U.S. Adults with Spine Problems: 1997-2006
Brook Martin, M.P.H.; Sohail Mirza, M.D., M.P.H.;
Judith Turner, Ph.D.; Michael Lee, M.D.; Bryan
Comstock, M.S.; Richard Deyo, M.D., M.P.H.
Presented by: Brook Martin, M.P.H., Health
Service Researcher; Ph.D. Student, Orthopaedics &
Sports Medicine; Health Services, University of
Washington, 325 Ninth Avenue, Box 359736,
Seattle, WA 98104, Phone: (206) 744-1803; Email: bim@u.washington.edu
Research Objective: Medical care expenditures for spine-related conditions amounted to $86 billion in
2005, increasing 65% since 1997. It is important to understand changes in the average individual burden of services relative to changes in the health status of individuals with spine problems. We sought to study trends from 1997 to 2006 in peruser expenditures for spine-related inpatient, outpatient, pharmacy, and emergency services; and to compare these trends to changes in health status.
Study Design: We analyzed data from the Medical
Expenditure Panel Survey, a multistage sampled survey designed to provide unbiased national estimates of health care utilization and expenditure.
Spine-related hospitalizations, outpatient visits, prescription medications and emergency department visits were identified using International
Classification of Disease, Version 9, Clinical
Modification (ICD-9-CM) diagnosis codes.
Regression analysis controlling for age-, sex-, comorbidity-, and time (years) were used to estimate trends in inflation-adjusted expenditures, utilization, and self-reported health status.
Population Studied: An average of 1,774 respondents with spine problems were surveyed per year, reflecting an increase in the number of people with spine problems in the U.S. from 14.8 million in 1997 to 21.9 million in 2006. The average age of people who reported spine problems increased slightly from 1997 to 2006 (47.7 to 50.4 years), as did the proportion who relied only on public health insurance (13.3% to 15.6%).
Principal Findings: From 1997 to 2006, national expenditures for spine problems increased an average of 7.0% per year, while population measures of mental health, social and physical functioning worsened. The mean adjusted per-user expenditure were the largest component of increasing total costs for inpatient hospitalizations, prescription medications and emergency department visits, increasing 37% (from $13,040 in
1997 to $17,909 in 2006), 139% (from $166 to
$397), and 84% (from $81 to $149) respectively. A
49% increase in the number of patients seeking spine-related care (from 12.2 million in 1997 to 18.2 million in 2006) was the largest contributing factor to increased outpatient expenditures.
Conclusion: For spine-related inpatient, pharmacy, and emergency care from 1997-2006, the rise in per-user expenditure outpaced the rise in the number of people seeking these services. In contrast, increases in spine-related outpatient expenditures were primarily the result of an increase in the number of people seeking care.
Implications for Policy, Delivery or Practice: The discordance in population-level changes in health status versus spine expenditures raises concerns about value for therapies for spine problems.
Increasing per-user costs may indicate greater use of expensive technology and treatments. Within inpatient, pharmacy and emergency visit categories, the rise in per-user costs suggest that a greater level of scrutiny may be needed in evaluating treatments and introducing new spinal technologies. Broader implementation of chronic care models and technology assessments may be an important strategy for slowing expenditure growth in back pain. Rising per-user costs may also affect accessibility of spine care and lead to a widening of health disparities across socioeconomic levels independent of whether population measures of health improve. Efforts to control costs must also assure patient safety and effectiveness.
Theme: Public Health
● Health Insurance Coverage among United
States Children, with Emphasis on Hispanic
Sub-Groups: Findings from the National Health
Interview Survey, 2003-2007
Michael Martinez, M.P.H., M.H.S.A.; Debra
Blackwell, Ph.D.; Heather Free, M.P.H.
Presented by: Michael Martinez, M.P.H., M.H.S.A.,
Epidemiologist, Division of Health Interview
Statistics, National Center for Health Statistics,
3311 Toledo Road, Room 2332, Hyattsville, MD
20782, Phone: (301) 458-4758; Email: bmd7@cdc.gov
Research Objective: There is much concern in the
United States about the health insurance coverage status of children under 18 years of age. Previous research has shown that Hispanic children were less likely to have private health insurance coverage and more likely to be uninsured than either non-Hispanic white or non-Hispanic black children. Prior research on Hispanic adults shows variation in health insurance coverage among sub-
groups. However, little research has been conducted on the health insurance coverage status of Hispanic children by sub-groups. Accordingly, the objective of this analysis is to compare health insurance coverage status for children of different races and Hispanic sub-groups.
Study Design: This study used data from the 2003-
2007 National Health Interview Survey (NHIS), a nationally representative sample of the United
States civilian non-institutionalized population, conducted in face-to-face household interviews.
Population Studied: Children under 18 years of age with known health insurance coverage status comprised the study population.
Principal Findings: Hispanic children were less likely to have private health insurance coverage than non-Hispanic white and non-Hispanic black children. Hispanic and non-Hispanic black children were more likely to have public coverage than non-
Hispanic white children. In addition, Hispanic children were more likely to be uninsured than non-
Hispanic white and non-Hispanic black children.
Within the Hispanic sub-groups, children of Cuban ancestry were more likely to have private health insurance coverage than other Hispanic subgroups. Children of Dominican ancestry were more likely to have public coverage than other Hispanic sub-groups. Finally, children of Mexican and
Central or South American ancestry were more likely to be uninsured than other Hispanic subgroups.
Conclusion: Differences exist in health insurance coverage status between Hispanic and non-
Hispanic children. Furthermore, there are differences in health insurance coverage status among children of different Hispanic sub-groups.
Implications for Policy, Delivery or Practice:
These results indicate that disaggregating estimates of health insurance status by type of coverage and by Hispanic sub-groups provides useful information for assessing disparities in health care access and utilization.
Funding Source(s): CDC
● An Evaluation of the Impact of Enhanced
Income & Citizenship Verification on Public
Health Insurance Enrollment
James Marton, Ph.D.; Angela Snyder, Ph.D.; Mei
Zhou, M.S., M.A.
Presented by: James Marton, Ph.D., Associate
Professor, Economics, Georgia State University,
P.O. Box 3992, Atlanta, GA 30033, Phone: (404)
413-0256; Email: marton@gsu.edu
Research Objective: The Deficit Reduction Act of
2005 (DRA) was signed by President Bush in
February 2006 and is projected to generate $39 billion in federal entitlement expenditure reductions between 2006 and 2010. One aspect of the DRA that has garnered recent attention is the increase in documentation requirements for Medicaid enrollment and recertification. Beginning in July
2006, the DRA would require most new Medicaid applicants, as well as most current beneficiaries at re-determinations, to document their citizenship.
Acceptable documentation includes a U.S. passport, birth certificate or driver’s license from a state that verifies social security numbers. Some states, including Georgia, also made changes to their SCHIP documentation requirements in order to bring the SCHIP requirements more in line with the new Medicaid requirements. The purpose of this paper is to examine the impact of these changes, as well as increases in required income verification documentation, on the public health insurance coverage of children using state administrative
Medicaid and SCHIP data from Georgia.
Study Design: In response to the DRA, Georgia started in January 2006 requiring original documentary evidence of a child’s citizenship when applying for Medicaid, or, if the child was already enrolled, when recertifying coverage. In order to bring PeachCare requirements more in line with
Medicaid, Georgia began in July 2007 to require families to provide documentary evidence of citizenship and income, though, unlike Medicaid, faxed copies of documents are considered acceptable. In this paper, we use Medicaid and
PeachCare eligibility and claims data between 2004 and 2007 to evaluate the impact of these changes in citizenship and income verification on enrollment in the public health insurance programs in Georgia.
For ease of comparison, we focus on enrollees under the age of 19 within the Medicaid program.
Because of the difference in the timing of the policy changes, enrollees in one program can serve as a quasi-control group for children in the other. Using simple counts of monthly enrollment and new entry into the programs, we estimate time series models to analyze the magnitude of these policy changes on enrollment in both Medicaid and PeachCare.
While simple enrollment counts can shed some light on the impact of these policy changes, the aggregate nature of the data does not allow for an analysis of differential responses by children of different ages and races, in different parts of the state, and with different levels of health. Using state administrative data on individual enrollees, we further analyze spells of enrollment in both
Medicaid and PeachCare using hazard models to address the differential impacts of these policy changes.
Population Studied: Children enrolled in Georgia’s
Medicaid and PeachCare program between
January 2004 and December 2007.
Principal Findings: Our preliminary time series results suggest that the implementation of increased income and citizenship requirements in
Georgia Medicaid did not have a statistically significant impact on the number of monthly children enrolled, but did result in a statistically significant reduction in the number of new child enrollees per month. There appears to be some
preliminary evidence of a reduction in both the total number of enrollees and the number of new enrollees in Georgia’s PeachCare program but it is difficult to disentangle the impact of the new income and citizenship documentation requirements from a state implemented freeze on PeachCare enrollment that occurred at about the same time.
Conclusion: This preliminary evidence suggests that there are potential enrollment impacts associated with changes in income verification and citizenship documentation requirements. Our hazard analysis will allow us to examine this in more detail and look for sub-group effects as well as changes in the duration of dis-enrollment from both programs.
Implications for Policy, Delivery or Practice: If a programmatic priority of SCHIP reauthorization is to increase the coverage of children that are income eligible but not covered, then an important policy question that needs to be addressed is how much of an effect new income and citizenship requirements have on program participation. Our analysis will answer this question by examining whether or not enrolled children are more likely to exit coverage after these verification changes and whether or not children that exit coverage are less likely to re-enroll. If the analysis suggests that these new verification procedures increase the number of income eligible but uncovered children, then these results may highlight the need for more research on how best to implement stricter verification of income and citizenship while at the same time minimizing the impact on program take up.
Funding Source(s): Georgia Department of
Community Health
Theme: Medicaid, SCHIP and State Health Reform
● Which Response Rates to Report in Health
Surveys? Implications of Different Response
Rate Calculations & Reporting in Health
Services Research
Grant Martsolf, M.P.H., R.N., Ph.D.; David Johnson,
Ph.D.; Kurt Johnson, Ph.D.; Dennis Scanlon, Ph.D.
Presented by: Grant Martsolf, M.P.H., R.N., Ph.D.,
Graduate Assistant, Health Policy & Adminstration,
Penn State University, 504 Ford Building, University
Park, PA 16802-6500, Phone: (814) 863-0875;
Email: grm153@psu.edu
Research Objective: We examine the patterns of awareness and use of public provider quality reports in a population of people with chronic illnesses by testing the following hypotheses: 1) individuals with chronic illnesses are more likely to be aware of and use physician reports than hospital reports; 2) those who are more “activated” are more likely to report greater awareness and use of both physician and hospital reports; 3) individuals with chronic illnesses for which there is more quality information in reports specific to their illnesses are more likely to use physician quality reports; and 4) the more reports that are available and the longer the reports have been available, the more likely are people to be aware of and use them.
Study Design: This paper presents a general overview of various response rate calculation techniques with particular emphasis on the most widely used techniques in health services research:
AAPOR and CASRO. AAPOR is a more conservative calculation technique used primarily by public policy researchers. CASRO is used to calculate response rates for the Behavioral Risk
Factor Surveillance Survey (BRFSS) which is a commonly cited and analyzed survey. To illustrate the importance of the choice between these two methods, we calculate both the AAPOR response rates as well as CASRO response rates for a large random digit dial telephone survey of individuals with chronic disease.
Population Studied: Adults residing in 14 Aligning
Forces for Quality (AF4Q) communities, who are 18 years or older and have at least one of the five following five chronic conditions: diabetes, hypertension, heart disease, asthma, and depression (N=7,337). The random-digit dial (RDD) telephone survey was conducted from June 2007 through August 2008 to collect data from the target population.
Principal Findings: We find that there are large differences in response rates depending on the method employed. For the entire survey, response rates using AAPOR were equal to 27.1% while response rate using the CASRO methodology were
48.4%. We also find that there is wide variation in the difference between CASRO and AAPOR by sampling region. For example, in Cincinnati the
AAPOR response rate was equal to 55.7% while the CASRO response rate was equal to 25.8%
(54% lower) whereas the CASRO and AAPOR response rates in Minneapolis/St.Paul were 47.1 and 34.2 respectively (27% lower). This shows that the differential between the rates is not consistent suggesting that CASRO rates cannot be easily transformed into AAPOR rates and vice versa without the original response data. This makes the comparisons of response rates by readers and editors even more complicated.
Conclusion: Because the willingness of journals to publish and the ability of readers to interpret findings are at least partially conditional on response rates, this study suggest that the calculation methodology matters. Some implications of this study include: 1) Health services researchers using survey data must better understand the differences between the response rate calculation techniques and the implication this has for response rate reporting; 2) These differences must also be communicated to other academic and lay consumers of survey results for better interpretation of findings; 3) Because journal editors do consider response rates in publication decisions, better standardization and
communication of response rate calculations and expectations is important.
Funding Source(s): RWJF
● Employee Trust as a Proxi Measure for
Workforce QI Initiatives
William Mase, Dr.P.H., M.P.H., M.A.; James
Holsinger, Jr., M.D., Ph.D; Richard Ingram, M.Ed.
Presented by: William Mase, Dr.P.H., M.P.H.,
M.A., Research Assistant Professor, Public Health
Sciences, University of Cincinnati, P.O. Box
670840, Cincinnati, OH 45267-0840, Phone: (513)
558-2710; Email: william.mase@uc.edu
Research Objective: To demonstrate if differences in employee self reported measures of supervisor trust are statistically significant when controling for gender and/or race.
Study Design: cross-sectional
Population Studied: Employees of a large urban health department
Principal Findings: 1) Gender is statistically significant for the overall trust measure with a point estimate of 0.54 and a confidence interval of [0.08,
1.00] 2) Race is statistically significant for the overall trust measure with a point estimate of 0.58 and a confidence interval of [0.20, 0.97]
Conclusion: 1) Men employed in the LHD reported statistically significantly lower overall supervisor trust as compared to women. 2) Blacks employed in the LHD reported statistically significantly lower overall supervisor trust as compared to whites.
Implications for Policy, Delivery or Practice:
Quality Improvement interventions need to be developed and implemented to increased employee
-- manager trust. Measureable trust disparities exist by gender and race. Worksite specific initiatives need to be developed to maximize workforce effectiveness.
Funding Source(s): RWJF
Theme: Public Health
● Physician Mobility & Work Location Choice:
Exploring Generational Differences
Maria Mathews, Ph.D.; Maureen Seguin, M.A.;
Robert Card, M.D.
Presented by: Maria Mathews, Ph.D., Associate
Professor, Community Health & Humanties,
Memorial University, 2387, Health Science Centre,
300 Prince Philip Drive, Saint John's, NL, A1B 3V6,
CA, Phone: (709) 777-7845; Email: mmathews@mun.ca
Research Objective: How often do physicians change work locations? What influences physicians’ choice of work location and do these reasons change over a physician’s career? Using qualitative interviews, we examined physician mobility, highlighting differences in the factors affecting work location choice in different generations of physicians.
Study Design: We interviewed 48 randomly selected physicians who had graduated from either the University of Saskatchewan or Memorial
University of Newfoundland in the 1960s, 70s, 80s, and 90s. We asked participants about the number of locations where they worked over their career and their reasons for choosing each location.
Interviews were recorded, transcribed and coded thematically. We used Barer et al.’s framework to code six groups of factors related to practice location decisions: 1) personal background factors,
2) professional education factors, 3) professional practice factors, 4) personal/family factors, 5) community factors, and 6) economic factors.
Population Studied: Physicians who had graduated from either the University of
Saskatchewan or Memorial University of
Newfoundland
Principal Findings: Most physicians had worked in two or fewer locations over the course of their career, regardless of when they graduated.
Personal/family factors (i.e. being close to family, lifestyle) and professional education factors (i.e. completed residency there) were the most commonly cited reasons for selecting a specific location, regardless of year of graduation. For older generation physicians, wanting to work abroad or political crises were important in their choice or work location whereas younger generation physicians cited lifestyle as an important consideration. With the exception of return for service agreements and some specialists, economic factors were not generally cited as primary reasons for choosing a work location.
Conclusion: Factors influencing physicians choice work locations are largely stable over different physician generations among graduates from two
Canadian medical schools.
Implications for Policy, Delivery or Practice:
While economic factors have largely been the focus of recruitment and retention initiatives, our study highlights the importance of personal background and professional education factors in provincial physician resource strategies.
Funding Source(s): CIHR, IRIF, NLCAHR, SK
Learning
Theme: Health Care Workforce
● Why Physicians Leave: Implications for
Physician Retention Policies
Maria Mathews, Ph.D.; Maureen Seguin, M.A.;
Robert Card, M.D.
Presented by: Maria Mathews, Ph.D., Associate
Professor, Community Health & Humanties,
Memorial University, 2387, Health Science Centre,
300 Prince Philip Drive, Saint John's, NL, A1B 3V6,
CA, Phone: (709) 777-7845; Email: mmathews@mun.ca
Research Objective: What influences a physician to leave a community? Using qualitative interviews, we examined the factors associated with a physician leave work highlighting differences between generations of physicians.
Study Design: We interviewed 48 randomly selected physicians who had graduated from either the University of Saskatchewan or Memorial
University of Newfoundland in the 1960s, 70s, 80,s and 90s. We asked participants about the number of locations where they worked over their career and their reasons for choosing each location.
Interviews were recorded, transcribed and coded thematically. We used Barer et al.’s framework to code six groups of factors related to practice location decisions: 1) personal background factors,
2) professional education factors, 3) professional practice factors, 4) personal/family factors, 5) community factors, and 6) economic factors.
Population Studied: Physicians who had graduated from either the University of
Saskatchewan or Memorial University of
Newfoundland.
Principal Findings: Poor personal satisfaction and professional practice factors, specifically workplace polices, models of care, and/or poor relationships with partners or administration were commonly reported by physicians of all generations for why physicians left a practice or would consider leaving their current practice. Some older generation physicians cited political policies or crises for leaving a location. Few physicians said they left a position solely because of economic factors. For physicians in Saskatchewan and Newfoundland and Labrador, many noted that lower remuneration was offset by lower cost of living and better lifestyle.
Economic factors however were often cited as the
“last straw” when physicians experienced poor professional practice factors.
Conclusion: Professional practice factors and personal satisfaction are more influential than remuneration in physicians’ decisions to remain in a practice location.
Implications for Policy, Delivery or Practice:
Policies aimed at improving work environments may lead to improved long-term retention than initiatives that focus on economic factors alone.
Funding Source(s): CIHR, IRIF, NLCAHR, SK
Learning
Theme: Health Care Workforce
● Using Health Plan Data to Support & Inform
Comparative Effectiveness
Richard Mathis, Ph.D.; John Barnes, M.B.A., M.A.
Presented by: Richard Mathis, Ph.D., Manager,
Medical Policy R&D, Medical Policy R&D,
BlueCross BlueShield of Tennessee, 801 Pine
Street, 2 East, Chattanooga, TN 37402, Phone:
(423) 535-5801; Email: richard_mathis@bcbst.com
Research Objective: To determine the extent to which Health Plan data can be used to support and target a national comparative effectiveness effort.
Study Design: Quasi experimental. The poster draws upon a BCBST diabetes study that identifies gaps in diabetes care as well as HEDIS and MEDai data related to other chronic conditions.
Population Studied: Commercially insured health plan members identified with chronic conditions in a
PPO benefit design network.
Principal Findings: Findings confirm that gaps exist in diabetes care and with respect to the care of patients with other chronic conditions.
Conclusion: This presentation demonstrates how health plan data may be used to support the creation of a comparative effectiveness effort. It also shows how plan data my be used to identify conditions showing the greatest variation in care and thus most likely to benefit from a comparative effectiveness program promoting increased standardization.
Implications for Policy, Delivery or Practice: A national comparative effectiveness program is supported by health plan data showing variation in practice against evidence based guidelines. Such a program could benefit from plan data suggesting conditions where variation is the greatest.
● Disparities in Health & Access to Services by
Ethnicity
Jaya Mathur, B.A.; James Maxwell, Ph.D.; Tom
Mangione, Ph.D.; Jack Boesen, Esq.; Christine
Bishop, Ph.D.
Presented by: Jaya Mathur, B.A., Research
Associate, Health Care Policy & Management
Research, JSI Research & Training Institute, Inc.,
44 Farnsworth Street, Boston, MA 02210, Phone:
(617) 482-9485; Email: jmathur@jsi.com
Research Objective: The Massachusetts Personal
Care Attendant (PCA) program is funded by the state Medicaid program and helps elderly and disabled consumers live independently in their homes. Hispanics comprise a significant and growing share of the PCA consumer population.
The purpose of this study is to evaluate the extent to which differences exist in health status, satisfaction with PCA hours, and use of additional formal support services by ethnicity.
Study Design: Questions were taken from evaluations of other state PCA programs and prior research on home-based care. A random sample of consumers was drawn from the PCA program consumer population. For consumers less than 18 years of age or those incapable of completing the survey, a proxy completed the interview. Survey data were analyzed in SAS. Chi-square tests for categorical variables and t-tests for continuous variables were used to test for significant differences, p less than 0.05.
Population Studied: Data for 111 Hispanic consumers and 391 non-Hispanic consumers of the program.
Principal Findings: Hispanic respondents were more likely to be older, female, and have less than a high school education than non-Hispanic consumers. Eighty-three percent of Hispanic consumers reported fair/poor health compared to
55 percent of non-Hispanic consumers. Further, 62 percent of Hispanics reported that their health status as somewhat or much worse than a year ago, compared to 37 percent of non-Hispanics. A larger percentage of Hispanics responded that their disability status had progressed over time, 84 percent vs. 70 percent. However, their reported
ADLs and IADLs were similar to non-Hispanics.
Hispanics were more likely to rely on a single PCA than non-Hispanics, 74 percent vs. 49 percent, and the main PCA was usually a family member,
Hispanics, 41 percent vs. non-Hispanics, 26 percent. Thus, Hispanic consumers also had lower turnover rates than non-Hispanics, 11 percent vs.
17 percent. Even though Hispanics reported worse health and made requests for additional hours at rates similar to non-Hispanics, a greater percent of
Hispanics were allocated LTE 20 hours of PCA time than non-Hispanics, 45 percent vs. 29 percent.
Additionally, Hispanics were more likely than non-
Hispanics to report that the allocated time was never enough to meet their needs, 16 percent vs. 6 percent. Finally, 8 percent of Hispanics received services from other support agencies compared to
32 percent of non-Hispanics.
Conclusion: Hispanic respondents were older, had less formal education, and reported much worse health than non-Hispanic consumers. These factors place Hispanics at risk for potentially worse outcomes, while the hours allocated for PCA time does not meet their needs.
Implications for Policy, Delivery or Practice:
This initial analysis of Hispanic PCA consumers suggests the need for increased monitoring and reevaluation of their health status and limitations by the PCA program over time, especially given the progressive nature of their disabilities and worsening of health status. Efforts should be made to connect Hispanic consumers with other formal services that could supplement PCA hours and provide care without additional cost to the PCA program. This study raises the question as to whether health disparities should be examined in other segments of the community-based long-term care system.
Funding Source(s): PCA Quality Home Care
Workforce Council
Theme: Disparities
● Development of a Tool for Measuring
Elements Critical to Organization
Transformation
Alexis Maule, M.P.H.; Carol VanDeusen Lukas,
Ed.D.; Barbara Lerner, M.S., C.G.C.; Michael
Shwartz, Ph.D.; Irene Cramer, Ph.D., M.S.S.A.;
Martin Charns, D.B.A.
Presented by: Alexis Maule, M.P.H., Research
Analyst, Center for Organization, Leadership &
Management Research, VA Boston Healthcare
System, 150 South Huntington Avenue (152M),
Jamaica Plain, MA 02130, Phone: (857) 364-5098;
Email: alexis.maule@us.army.mil
Research Objective: Our objective was to create a tool to measure the presence of -- or fidelity with -- the elements of the Organization Transformation
Model (OTM) in ten US healthcare systems. OTM was developed from the evaluation of RWJF’s
Pursuing Perfection initiative, which aimed to substantially transform seven healthcare systems to provide perfect patient care. OTM identifies five elements critical to organization-wide patient-care quality improvement: (1) Impetus to transform; (2)
Leadership commitment to quality; (3) Improvement initiatives that actively engage staff; (4) Alignment of organization goals; and (5) Integration across boundaries. A key challenge in measurement is determining whether the model is present in each system. It is difficult because improving care across an organization calls for complex changes.
To measure fidelity with OTM, we needed a tool that captured its core elements while recognizing that they would not be operationalized in exactly the same way across organizations.
Study Design: We developed a fidelity rating tool to measure the five critical elements. Two to five questions were developed to measure operational features of each element. Following all interviews in a hospital, each question was rated by interviewers on a 4-point scale, defined at the top by full presence of the question’s operational features throughout the organization and at the bottom by no evidence of the question’s features.
Ratings were initially done independently by two interviewers; each numeric rating was accompanied by qualitative extracts to support the chosen score.
Afterwards, raters discussed and reached a consensus on final ratings. To create final fidelity ratings, raters averaged questions to get element scores and added element scores to get an overall fidelity score.
Population Studied: We conducted telephone interviews at ten US hospitals selected on two criteria: (1) bed size (>399 versus 100-399); and
(2) performance on CMS measures (top versus third quintile). At each hospital, 10-15 employees, ranging from top leadership to frontline providers, were interviewed for a total of 118 interviews.
Principal Findings: The fidelity rating tool provided a reliable basis for comparing the degree to which each model element was present in each hospital.
Qualitative evidence provided for each question was consistent between raters, and quantitative ratings had a high degree of rater agreement, within one point in all but a few cases. The overall fidelity
scores discriminated between hospitals in the two performance groups.
Conclusion: The fidelity tool is a viable instrument for quantifying the presence of a complex organizational model. Anchoring the ratings in qualitative evidence gathered from each hospital also allowed us to consider how each hospital uniquely operationalized the model. Using the tool, we were able to compare and distinguish among hospitals based on the presence of the five critical elements.
Implications for Policy, Delivery or Practice: The
OTM fidelity tool can be used by hospital leadership and researchers as a diagnostic tool or an evaluative tool to determine the degree to which the
OTM elements exist in the organization. It can also be used as a general approach to measure the presence of complex organizational models.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Measurement
● Do Rural Reproductive-Aged Women Receive
Less Preventive Healthcare Consulting than
Urban Women?
Jennifer McCall-Hosenfeld, M.D., M.Sc.; Carol
Weisman, Ph.D.
Presented by: Jennifer McCall-Hosenfeld, M.D.,
M.Sc., Assistant Professor of Medicine & Public
Health Sciences, Medicine & Public Health
Sciences, Penn State Milton S. Hershey Medical
Center, 500 University Drive, Mailcode: HU-15,
Hershey, PA 17033, Phone: (717) 531-8161;
Email: jsm31@psu.edu
Research Objective: Preventive health screening is less frequent among rural women compared to urban women. Preventive counseling is equally important for high quality primary healthcare delivery, but little prior research has explored whether rurality impacts receipt of preventive counseling. We investigate the impact of rurality on preventive counseling among reproductive-aged women.
Study Design: Guideline-concordant preventive healthcare counseling was based on United States
Preventive Services Task Force recommendations before 2004: screen and counsel all adults for tobacco use and alcohol misuse, provide routine counseling about effective contraception, and provide intensive counseling to promote weight loss for obese patients. Subjects self-identified demographics, rurality was classified by Rural-
Urban Commuting Area codes, and obesity was calculated from self-reported height and weight.
Subjects identified whether they had discussed smoking, alcohol/drug use, birth control, nutrition, weight management, or physical activity with a health professional within the past year. We assessed bivariate relationships between receipt of counseling and rurality, then developed multivariate models to investigate the independent contribution of rurality to each of the following: counseling for 1) smoking, 2) alcohol/drug use, 3) contraception, and, among obese subjects only, counseling for 4) diet,
5) weight management, and 6) exercise.
Multivariate analyses employed logistic regression, controlling for age, race, education, and regular healthcare provider.
Population Studied: The Central Pennsylvania
Women´s Health Study, a cohort of 2002 women aged 18-45. Data were collected during 2004-
2005.
Principal Findings: 768 (39%) of the women were rural. Compared to urban women, rural women were more likely to be of the oldest age group (35-
45 years) (54% versus 49%, p<.05), white (97% versus 85%, p<.001), have a regular healthcare provider (91% versus 88%, p=.05), and have less than or equal to a high school education (45% versus 37%, p<.001). There were no significant differences between urban and rural women in household income or gaps in health insurance. In unadjusted models, rural women were less likely to receive counseling for alcohol/drugs (10% versus
13%, p=.03) and birth control (30% versus 36%, p=.01), but equally likely to receive tobacco counseling. Among 444 obese women, rural women were less likely to receive exercise counseling (53% versus 66%, p<.01), but equally likely to receive diet and weight management counseling. In six multivariate models, rurality was independently associated with lack of preventive counseling in only one (exercise counseling among obese women.) Adjusting for demographics fully attenuated the effect of rurality in each of the five remaining models. Older age was independently associated with nonreceipt of preventive counseling in three models (tobacco use, alcohol/drug use, birth control) a finding that was preserved in analyses controlling for smoking and binge drinking.
The interaction of poverty and rurality is explored.
Conclusion: In multivariate analysis, obese rural women were less likely to receive exercise counseling than obese urban women. Rurality did not affect counseling for tobacco, alcohol and birth control. Older age may be associated with decreased preventive counseling, regardless of locality.
Implications for Policy, Delivery or Practice: All healthcare providers should ensure counseling for alcohol, tobacco use and birth control to older reproductive-aged women. Rural providers should ensure comprehensive weight management counseling for obese patients.
Funding Source(s): Pennsylvania Department of
Health
Theme: Gender and Health
● Quality of Care in Hospitalizations for Sickle
Cell Disease in the U.S.: Do Low Patient
Volumes Mean Higher Mortality?
Timothy McCavit, M.D.; Song Zhang, Ph.D.; Chul
Ahn, Ph.D.; Glenn Flores, M.D.; Charles Quinn,
M.D.
Presented by: Timothy McCavit, M.D., Fellow in
Pediatric Hematology-Oncology, Pediatrics,
University of Texas Southwestern Medical Center,
5323 Harry Hines Boulevard, Dallas, TX 75390-
9063, Phone: (214) 456-6914; Email: tim.mccavit@childrens.com
Research Objective: Sickle cell disease (SCD) affects between 70,000-100,000 Americans and results in approximately 100,000 hospitalizations annually at a cost of $488 million. Little is known about the quality of care for patients hospitalized with SCD. The study aim, therefore, was to examine whether four potential determinants of quality care, 1) hospital volume, 2) hospital teaching status 3) patient socioeconomic status (SES), and
4) patient insurance coverage are associated with three quality of care (QOC) indicators for SCD patients: 1) mortality, 2) length of stay (LOS), and 3) cost of hospitalization.
Study Design: We conducted a cross-sectional analysis of the 2003-2005 National Inpatient
Sample (NIS) datasets. In the NIS, hospital volumes are categorized as small, medium, or large depending on location and teaching status. We also calculated SCD-specific hospital volume quartiles by ascertaining the total number of SCD discharges for each hospital. SES was determined using median household income for the patient’s ZIP code. Initial analyses included descriptive statistics and bivariate analyses of associations between all dependent and independent variables. Multivariable analyses controlling for 8 covariates, including 4 measures of disease severity, were performed to examine the adjusted associations among the 4 potential QOC determinants and mortality, LOS, and cost.
Population Studied: The NIS includes data on nearly 8 million hospitalizations annually from 1000 hospitals in 37 states. It is an all-payer data set comprising a 20% stratified sample of US community hospitals. We identified cases using
ICD-9CM codes for any form of SCD as either the primary or a secondary diagnosis in patients of all ages.
Principal Findings: There were 70,748 SCD discharges and 425 deaths yielding a mortality rate of 0.6%. The median LOS was 4 days (5th-95th percentile range: 1-15 days), and the median hospitalization cost was $4,100 (5th-95th percentile range: $1,200-20,500). Bivariate analyses revealed an increased odds of death in hospitals with lower
SCD volume (Quartile 1 vs 4: odds ratio [OR] 3.1,
95%CI: 2.4-4; Quartile 2 vs 4: OR 2.1, 95%CI: 1.8-
2.5; Quartile 3 vs 4: OR 1.8, 95%CI: 1.6-2.0).
Multivariable analyses revealed increased adjusted odds of death in hospitals with lower SCD volume
(Quartile 1 vs 4: OR 1.5, 95%CI: 1.1-2; Quartile 2 vs 4: OR 1.2, 95%CI: 0.98-1.5; Quartile 3 vs 4: OR
1.3, 95%CI: 1.2-1.5) and in SCD patients with selfpay as primary payer (Self-pay vs Private/HMO OR
1.8, 95%CI: 1.4-2.3). Non-teaching hospitals had lower adjusted costs of care than teaching hospitals(p=.003.) SES was not associated with the
QOC indicators.
Conclusion: Lower SCD-specific hospital volumes and lack of health insurance coverage are associated with significantly increased mortality in
SCD patients.
Implications for Policy, Delivery or Practice:
Additional research is needed to determine structures and processes in high volume hospitals that are responsible for the volume-outcome association in SCD. Recognition of centers with the best outcomes using a Centers of Excellence model and a report-card approach might improve SCD outcomes. Because lack of health insurance is associated with greater mortality, more attention should be paid to providing improved care and health insurance for all uninsured SCD patients.
Funding Source(s): NHLBI
Theme: Prevention and Treatment of Chronic
Illness
● Marking Time in the Emergency Department
Mark McClelland, M.N., R.N.; Vickie Sears, M.S.,
R.N.; Bonnie Norton, B.A.; Bruce Siegel, M.D.,
M.P.H.
Presented by: Mark McClelland, M.N., R.N., Senior
Research Scientist, Center for Health Care Quality,
The George Washington University School of Public
Health & Health Services, 2021 K Street, NW, Suite
800, Washington, DC 20006, Phone: (202) 994-
4262; Email: mark.mcclelland@gwumc.edu
Research Objective: Emergency Department crowding has emerged as one of the key problems facing America’s health systems. Decreasing capacity, increasing demand and more medically complex patients have created a perfect storm scenario resulting in ambulance diversion, patient boarding and suboptimal care. With 114 million patients visiting the ED annually, millions of
Americans may not be receiving safe and timely treatment. Despite these problems there are no widely used measurements of ED care that can inform patients, providers and policy makers about the effectiveness of ED organizational performance.
The literature is bereft of ED length of stay information with the associated measures of variability. This project is field testing a set of metrics that could form the basis of a national quality improvement initiative that will have implications for future ED based public reporting and pay-for-performance. The principal objectives of this study are to provide the Center for Medicare and Medicaid Services with estimates of variability in ED LOS and patient boarding times and information essential to define sample size
requirements for national implementation of standardized ED performance measures.
Study Design: Standard ED performance measures have been recently developed by a CMS
Technical Expert Panel and endorsed by the
National Quality Forum. Three of these standards are used in this project. Six study sites were obtained through a national competitive process utilizing the AHRQ ACTION network. Hospitals are reporting monthly LOS data for admitted and discharged ED patients. They also report the length of time from the decision to admit to the patient leaving the ED. In addition to the throughput measures, metrics related to timeliness of pain management and ancillary testing are also reported. Hospitals report challenges and time requirements related to collecting performance measure data. Structured interviews with the project directors and data abstractionists will be performed after the hospitals have had 8 months experience with the performance measures.
Population Studied: Of the six ACTION network hospitals included in the study four are private, notfor-profit, one is public and one is investor owned.
Bed size ranges from 230 to 765 beds and ED volume ranges from 2900 to 7900 visits per month.
Three hospitals are academic teaching hospitals, and three are community hospitals. Hospitals were evenly divided between urban and suburban settings.
Principal Findings: The collection and reporting of the CMS generated, NQF endorsed standard ED performance measures does not present an onerous burden to hospitals. The study hospitals reported previous experience collecting all or subsets of the required measures. The final analysis of this study data will provide LOS information and estimates of variability from six very diverse EDs.
Conclusion: Hospitals can adapt their in-house performance measurement to facilitate national reporting of ED performance measures.
Implications for Policy, Delivery or Practice:
Widespread use of standardized performance measures will increase transparency of ED operations for national benchmarking, public reporting, pay for performance and regional emergency response planning. As reimbursement for health services increasingly incorporates performance and quality domains the knowledge gained from this research contributes information essential for the establishment of valid and reliable standards.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Measurement
● Patient, Context & Symptom Characteristics
Associated with Presentation more than 12 hours after Symptom Onset among Veterans
Kelly McDermott, M.A., Ph.C.; Charles Maynard,
Ph.D.; Elliott Lowy, Ph.D.; Stephan Fihn, M.D.,
M.P.H.
Presented by: Kelly McDermott, M.A., Ph.C.,
Student, Health Services, University of Washington,
337 17th Avenue East, Seattle, WA 98112, Phone:
(206) 366-5271; Email: kmcdermo@u.washington.edu
Research Objective: Increased time from symptom onset of acute myocardial infarction (AMI) to hospital presentation is strongly associated with increased morbidity and mortality. This is particularly concerning to the Department of
Veteran Affairs (VA) where the majority of AMI patients present =12 hours after symptom onset
(hereinafter =12 hours will be “late presentation” and <12 hours will be “early presentation”). We sought to identify characteristics associated with late presentation among VA patients presenting to
VA facilities with AMI.
Study Design: This is a retrospective study of veterans with AMI (with or without ST-elevation) admitting to selected VA facilities between FY05-
FY06. Based on the literature, we developed an exhaustive conceptual model detailing characteristics and behaviors related to late presentation. We compared these characteristics for early and late presenters using abstracted medical records and patient survey data from the
Acute Coronary Syndrome (ACS) Study linked with data regarding symptoms within the 24 hours prior to presentation from the Cardiovascular Care
Follow Up Study database. We used standard descriptive statistics adjusted for multiple comparisons and multivariate logistic regression to identify characteristics associated with late presentation. In addition, we performed a sensitivity analysis using multiple imputation methods to evaluate potential selection bias from missing data.
Population Studied: Of the 505 patients with AMI enrolled in the ACS Study, we were able to obtain symptom data for 259 (51%). Patients with complete data were all male, 81% white, and averaged 65±11 years old. Thirty eight percent of these patients were obese, 32% were current smokers and 66% were late presenters. Thirty two percent of patients had had a prior AMI, 41% had diabetes and the mean Charlson comorbidity index score was 1.65±1.1
Principal Findings: All patient characteristics including demographic, medical history, AMI knowledge, prior health care experience, and self concept were similar between the two groups. All contextual characteristics including living alone, distance traveled, admitting facility and nighttime presentation were also similar. There were higher proportions of late presenters with =3 angina episodes in the prior 24 hours (9.4% vs. 38% p<0.001) and symptoms of heart failure (2.3% vs.
16.9%, p=0.001). Multivariate analysis on patients with complete data revealed that late presenters had a lower odds of having 1 or 2 angina episodes
in the prior 24 hours relative to none (OR=0.18
95%CI[0.04-0.97]) and a higher odds of having at least one atypical symptom (OR=2.45, 95%CI[1.05-
5.75]). Sensitivity analysis revealed late presenters had increased odds of older age, and increased angina within the prior month and prior 24 hours.
Conclusion: While patient and contextual characteristics were not associated with late presentation in our sample, late presenters may have different symptom experiences.
Unfortunately, our sample may be subject to selection bias due to missing data.
Implications for Policy, Delivery or Practice:
Based on this analysis, strategies to reduce late presentation in the VA might be more effective if focused on education regarding symptoms rather than targeting particular population subgroups. In addition, the large proportion of missing data that we encountered should be investigated further.
Funding Source(s): AHRQ
● Depression among Uninsured, Insured &
Medicaid-insured United States Adults Aged 18 to 64 years
Margaret McDonald, Ph.D.; Robin Hertz, Ph.D.;
Michael Lustik, M.S.; Alan Unger, Ph.D.
Presented by: Margaret McDonald, Ph.D., Pfizer
Inc, 235 East 42nd Street Mail Stop 235/7/12, New
York, NY 10017, Phone: (212) 733-3093; Email: margaret.m.mcdonald@pfizer.com
Research Objective: To assess the current prevalence of depression among uninsured, insured, and Medicaid-insured United States adults aged 18 to 64 years.
Study Design: Analysis of nationally representative data collected from adults aged 18 to 64 years (N=
4,209) participating in the National Health and
Nutrition Examination Survey (NHANES) 2005-
2006.
Population Studied: Uninsured, insured (without
Medicaid), and Medicaid-insured United States adults aged 18 to 64 years.
Principal Findings: Depression is significantly more prevalent in uninsured adults (11.1% age and gender adjusted) than in insured adults (4.6%). At
21.3%, Medicaid- insured adults also are significantly more likely than the insured to screen positive for depression. Uninsured adults aged 40 to 64 years are 3 times more likely to screen positive for depression than their same-age insured counterparts (16.3% vs 5.1%, P<0.001). Among men, depression rates are significantly higher among the uninsured (10%) and Medicaid-insured
(17.9%) compared with the insured (3.5%).
Uninsured and Medicaid-insured women are also significantly more likely to screen positive for depression (12.2% and 24.6%) compared with insured women (5.6%). Depression rates among uninsured non-Hispanic whites and Hispanics
(11.4% and 10.9%) are significantly higher than their insured counterparts (3.8% and 3.4%).
Conclusion: Depression is highly prevalent among the uninsured, particularly among the middle-age group; and prevalent among women insured through Medicaid.
Implications for Policy, Delivery or Practice:
These findings call for increased recognition among providers and targeted interventions to address the burden of depression in the US in general and particularly among the uninsured and Medicaid populations.
Theme: Coverage and Access
● Medicare’s Payment Policy for Hospital-
Acquired Conditions: What is the Impact on
Safety Net Hospitals?
Megan McHugh, Ph.D.; Kevin Van Dyke, M.P.P.;
Timothy Martin, Ph.D.; John Orwat, Ph.D.
Presented by: Megan McHugh, Ph.D., Director,
Research, Health Research & Educational Trust,
One North Franklin, Chicago, IL 60606, Phone:
(312) 422-2634; Email: mmchugh@aha.org
Research Objective: As of October 1, 2008, CMS will no longer pay hospitals reimbursed under the inpatient prospective payment system (IPPS) for twelve hospital acquired conditions (HACs) (e.g., blood incompatibility, object left in surgery, air embolism). With the exception of demonstration projects, this represents Medicare’s first valuebased purchasing (VBP) policy that ties reimbursement to hospital performance. Safety net hospitals provide a key role in the provision of care to vulnerable populations, often with few resources.
Understanding the potential impact of this Medicare payment policy is important because safety net hospitals may be differentially affected, leading to more limited access for vulnerable populations.
Obtaining early information is particularly important as these initiatives continue to expand in both the public and private sectors. The purpose of this study is to compare the incidence of cases that meet Medicare’s HAC criteria at safety net and nonsafety net hospitals to explore whether safety net hospitals are more likely to be impacted by the policy than non-safety net hospitals.
Study Design: Following the approach used by
Zuckerman et. al. (2001) and Bazzoli et. al. (2005), we categorized hospitals as core, voluntary, or non safety net providers based on two criteria: (1) hospitals’ volume of uncompensated care, and (2) hospitals’ contribution to the total volume of uncompensated care provided in the community.
Data on uncompensated care were obtained from the 2006 American Hospital Association Annual
Survey. We identified discharges meeting the criteria for a HAC from 2006 Medicare claims data following the approach used by CMS in the FY
2008 Inpatient Prospective Payment System Final
Rule. T-tests were used to examine differences in
rates of HACs between core, voluntary and non safety net hospitals.
Population Studied: Our analysis includes data on
Medicare discharges in 2006 from 3,734 hospitals subject to Medicare’s payment policy for hospitalacquired conditions (i.e., hospitals reimbursed under the Inpatient Prospective Payment System).
Principal Findings: Our findings show statistically significant differences in the incidence of hospitalacquired conditions between core, voluntary, and non safety net hospitals; however differences were not sizable. Core safety net hospitals had an average of 41.4 HACs per 1,000 Medicare discharges compared to 35.8 at voluntary safety net hospitals and 41.7 at non safety net hospitals.
Smaller hospitals had lower rates of HACs than larger hospitals and non-teaching hospitals had lower rates of HACs than teaching hospitals.
Conclusion: Voluntary safety net hospitals, which tend to be smaller, non teaching facilities, may be less likely to be impacted by Medicare’s new payment policy than core safety net hospitals and non safety net hospitals.
Implications for Policy, Delivery or Practice:
Medicare’s new payment policy for hospitalacquired conditions is likely to impact some hospitals more than others. CMS should monitor the impact of the policy, particularly at core safety net hospitals, which tend to be financially vulnerable.
Funding Source(s): CWF
Theme: Quality and Efficiency: Policies and
Incentives
● Quality Improvement Initiatives & HIV and
Hepatitis C (HCV) Testing in VA Substance Use
Disorder Programs
D. Keith McInnes, M.S., Sc.D.; Jenny Hyun, Ph.D.;
Jodie Trafton, Ph.D.; Steven Asch, M.D., M.P.H.;
Allen Gifford, M.D.
Presented by: D. Keith McInnes, M.S., Sc.D.,
Health Scientist, Center for Health Quality
Outcomes & Economic Research, Veterans Affairs,
200 Springs Road (152), Bedford, MA 01730,
Phone: (781) 687-3507; Email: keith.mcinnes@va.gov
Research Objective: Many veterans treated in substance use disorder (SUD) programs have never been tested for HIV or HCV, despite being at high risk. We surveyed all Veterans Affairs (VA)
SUD programs to assess HIV and HCV testing rates and examine associations between quality improvement initiatives (e.g. computerized reminders for testing, provider feedback) and testing rates.
Study Design: Cross-sectional survey of all VA
SUD programs in the U.S., between October 1,
2005 and January 31, 2006. We estimated associations between quality improvement initiatives and testing rates using multivariate regression.
Population Studied: Surveys were sent to program directors of all 232 SUD programs
(inpatient, residential, intensive outpatient, and outpatient) in the VA. They reported on the treatments and services provided to the patients in their programs.
Principal Findings: Survey response rate was
100% (N=232 programs). Reported testing rates were lower for HIV (35%) than for HCV (57%) (P
=.01). Residential programs had the highest testing rates (51% of patients tested for HIV and 73% tested for HCV), while outpatient programs had the lowest (23% tested for HIV; 45% tested for HCV).
SUD programs placed a lower priority on HIV testing than HCV testing (mean score 1.7 vs. 1.8,
P<.01). Of the six quality improvement systems assessed, each was used less for HIV than for
HCV. SUD programs used computerized reminders most, on average by 41% of programs for HIV testing and 74% for HCV testing (P < .001). Next most frequently used was provider education (55% of programs for HIV testing; 59% of programs for
HCV testing, P=.57), followed by computerized templates (41% for HIV testing; 48% for HCV testing, P=.32), designated clinician (35% for HIV testing; 38% for HCV testing, P=.66), performance profiling (14% for HIV testing; 18% for HCV testing,
P=.45), and clinical champion (9% HIV; 15% HCV,
P=.19). In multivariate models three of the six quality improvement systems were positively associated with testing rates. Presence of a designated clinician was positively associated with both HIV (beta = 10.66, P = .057) and HCV (beta =
16.14, P = .01) testing. Use of computerized templates was positively associated with HIV testing only (beta = 10.33, P = .06), while use of computerized reminders was positively associated with HCV testing only (beta = 15.43, P = .023).
Conclusion: Quality improvement initiatives to increase HIV and HCV testing rates– especially computerized reminders, computerized templates, and designated clinician – were associated with more testing. The finding that testing rates for HIV were 22% lower than for HCV, may be due to provider perceptions that HIV is a less important condition, and to VA policies that make HIV testing more difficult than HCV testing, including requiring written informed patient consent, and pre- and posttest counseling. Focused VA initiatives to increase
HCV screening may have contributed to higher
HCV testing rates.
Implications for Policy, Delivery or Practice: To increase testing for infectious diseases, SUD programs should consider system changes, i.e. implementing computerized reminders, computerized templates, and use of a designated clinician. The VA should also continue to explore ways to make HIV testing easier to perform and more routine.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Patient & Provider Perceptions of Email &
Secure Messaging to Encourage HIV Testing
D. Keith McInnes, M.S., Sc.D.; Jeffrey Solomon,
Ph.D.; Barbara Bokhour, Ph.D.; Jane Burgess,
A.C.R.N., M.S.; Kim Nazi, F.A.C.H.E.; Allen Gifford,
M.D.
Presented by: D. Keith McInnes, M.S., Sc.D.,
Health Scientist, Center for Health Quality
Outcomes & Economic Research, Veterans Affairs,
200 Springs Road (152), Bedford, MA 01730,
Phone: (781) 687-3507; Email: keith.mcinnes@va.gov
Research Objective: Despite the importance of
HIV testing in Veterans Affairs (VA) Medical
Centers, less than 50% of high risk veterans have been HIV-tested. The VA is examining the feasibility of electronic communication with veterans to increase HIV testing rates. We explored patient and provider perceptions of electronic communication methods designed to encourage veterans to talk with their providers about HIV testing.
Study Design: We conducted focus groups with
VA patients and providers and used grounded thematic analysis. We initially focused on perceptions of diagnostic testing (including HIV testing), direct-to-patient emails, and the VA electronic personal health record (PHR) system, called “MyHealtheVet”. In a second round of focus groups we shared draft messages with participants for their evaluation.
Population Studied: Patients at the Bedford,
Massachusetts VA medical center (2 groups), and primary care providers at the Manchester, New
Hampshire VA medical center (2 groups).
Principal Findings: Patient themes identified were confidentiality, comprehension of the PHR, and HIV message content. Provider themes were confidentiality, message content, and provider workflow. Patients were uncomfortable with direct emails: “I've given away computers … [The new owner] can find stuff that you left in there. I'm not taking that chance... I'm very careful about what goes in my computer”. But they seemed willing to receive messages via the secure VA electronic
PHR system: “Just tell me to go to My HealtheVet website, log in and I get messages there. I'd rather see a message there than coming into [my personal email]”. Some were unsure how the electronic PHR worked: “Are you telling me…if I sign up for this password…I can get my medical records via the
Internet?” Patients wanted messages to have factual information about HIV, cite authoritative sources, and have a positive tone, e.g. indicating that treatments enable infected individuals to lead healthy lives.Providers agreed that secure message systems were better than unsecure emails, and that message content should be informative and positive. They agreed that HIV testing is important, however, they had concerns that electronic messages broadcast to veterans could create unmanageable workflow, including increased phone inquiries and testing requests. They suggested that additional designated staff would be needed to respond to patient questions, offer education, and provide pre- and post-test HIV counseling: “The only way I can see this working would be if there were a stand-alone clinic or person that [patients] could go to with questions…a person that they could call. That person could provide the education and pre- and post-test counseling.”
Conclusion: Patients and providers appear uncomfortable with unsecure email messages to veterans, but accepting of secure messages (e.g. through the PHR) that would provide veterans with health-related information, including about HIV testing. Patients were unclear about available functions of the PHR. Providers were concerned about substantial increases in workflow due to broadcast electronic messages that encourage health screening.
Implications for Policy, Delivery or Practice:
Electronic communication may be a powerful means of providing targeted health messages to veterans, even on potentially sensitive topics such as HIV. Health systems that implement such communication systems should be prepared to address provider concerns about increased workflow.
Funding Source(s): VA
Theme: Military and Veterans Heath Care
● Back Pain in Canadian Family Practice:
Characteristics & Determinants of Episodes of
Care Using Electronic Medical Record Data
Megan McKeown, M.S.; Moira Stewart, Ph.D.;
Amardeep Thind, M.D., Ph.D.
Presented by: Megan McKeown, M.S.,
Student/Research Assistant, Deliver Primary Health
Care Information Project, Centre for Studies in
Family Medicine, University of Western Ontario,
245-100 Collip Circle, UWO Research Park,
London, ON N6G 4X8, CA, Phone: (519) 878-
6220; Email: mmckeown@uwo.ca
Research Objective: In Canada and internationally, back pain is one of the most common reasons patients visit their family physician. The episode of care is an epidemiologic concept of health care consumption and physician management of a health condition, and one that is not well researched in back pain. An episode of care is a series of one or more physician visits for a specific health condition. The objectives of this research are two-fold: 1) to characterize episodes of care for back pain in a Canadian context, including length of episode and management of back pain care; 2) to identify differences in episode of care characteristics based on patient sex, age,
comorbidities, history of physician visits for back pain, and back pain severity.
Study Design: A longitudinal, observational study using quantitative data from an EMR database.
Population Studied: The Deliver Primary Health
Care Information (DELPHI) Project data was used for this research. The DELPHI database consists of
36 months of de-identified EMR data from 10 family practices across south-western Ontario. A random subset of 10% of these patients has been coded using the International Classification of Primary
Care (ICPC), a classification system that describes important features of a primary care physician visit, including diagnosis. ICPC diagnostic codes were utilized to identify back pain cases. A population of
464 back pain patients was identified.
Principal Findings: From the population of back pain patients, 456 episodes of care for back pain were identified and analyzed. The mean episode of care length in days was 15.03; 20.80 for males and
10.10 for females (p=0.002); 19.54 for patients with prior back pain visits and 11.21 for patients without prior back pain visits (p =0.014); 18.94 for patients with severe back pain and 9.65 for patients with moderate back pain (p =0.007). Age and chronic, musculoskeletal, and psychosocial conditions were not significantly associated with episode of care length. On average, there were 1.53 physician visits for back pain per episode of care; 1.79 for males and 1.30 for females (p =0.001); 1.67 for patients with severe back pain and 1.32 for patients with moderate back pain (p =0.022); 1.40 for patients with one or more chronic condition(s) and 1.70 for patients with no chronic conditions (p =0.047). Age, history of back pain visits, and musculoskeletal and psychosocial conditions were not significantly associated with the number of physician visits for back pain within an episode of care. Further analysis will investigate referrals, investigations, and prescriptions within the back pain episode of care.
Conclusion: The length of episodes of care for back pain, in days and number of back pain physician visits, is associated with patient characteristics. Further analysis will determine if these patient characteristics are associated with the physician management of back pain within episodes of care.
Implications for Policy, Delivery or Practice:
Knowledge of back pain episodes of care can assist policy makers and practitioners in planning the use of scarce resources for such a prominent condition in primary care.
Funding Source(s): Deliver Primary Health Care
Information (DELPHI) Project, Centre for Studies in
Family Medicine, University of Western Ontario
● Obesity & Health Services Utilisation
Meghan McMahon, M.Sc.; Audrey Laporte, Ph.D.;
Paul Grootendorst, Ph.D.; Peter Coyte, Ph.D.
Presented by: Meghan McMahon, M.Sc., Senior
Projects Officer, Faculty of Law, University of
Toronto, Institute of Health Services & Policy
Research, 39 Queen's Park Crescent East,
Toronto, M5R2X3, CA, Phone: (416) 978-5172;
Email: meghan.mcmahon@utoronto.ca
Research Objective: The rise of obesity as a global health concern reflects, in part, the substantial economic burden it imposes on health sectors. Few studies have investigated the complex relationships between obesity and health service utilisation using multivariate regression techniques with controls for competing health risk behaviours on nationally representative individuallevel surveys. This study examines relationships between obesity (measured using the body mass index) and health service utilisation in adults.
Study Design: Cross-sectional data from the 2005
Canadian Community Health Survey were used to estimate a two-part model for general practitioner and specialist visits. Models control for competing health risk behaviours and socioeconomic status and a secondary analysis also controls for the number of chronic conditions suffered by the individual.
Population Studied: The study population was comprised of adults that were non-pregnant, aged
18-75, and resided in Canada’s 10 provinces
(n=88,020 individuals weighted to represent
18,726,091 Canadians 18-75 years of age).
Principal Findings: Results indicate that the propensity to use physician services significantly increases with body mass index (BMI), independent of competing health risk behaviours and socioeconomic status. The relationship between obesity and health service utilisation is, however, attenuated by the number of chronic conditions suffered by the individual, particularly among older individuals. The part-two (intensity model) results reveal that the intensity of physician service utilisation (both with GPs and specialists) signficantly increases with BMI.
Conclusion: The results suggest that there is a signficant relationship between obesity and the use of health care and this is most apparent with GP services.
Implications for Policy, Delivery or Practice:
Expanding our understanding of the patterns and intensity of health services utilisation stemming from varying degrees of obesity will contribute to developing effective models of care provision and informing resource allocation decisions involving an increasingly obese population.
Funding Source(s): CIHR and OGS
Theme: Obesity Prevention and Treatment
● Bridging the Gap Between Research & Policy:
Partnerships for Health System Improvement
Meghan McMahon, M.Sc.; Colleen Flood, L.L.M.,
S.J.D.; Ellen Melis, M.S.; Krissy Davidge, M.S.;
Chris McCutcheon, M.A.
Presented by: Meghan McMahon, M.Sc., Senior
Project Officer, Faculty of Law, University of
Toronto, Institute of Health Services & Policy
Research, 39 Queen's Park Crescent East,
Toronto, M5R 2X3, CA, Phone: (416) 978-5172;
Email: meghan.mcmahon@utoronto.ca
Research Objective: While decision-making in health care organizations is a complex process, research evidence has an important role to play. It is becoming increasingly evident that the translation and uptake of research into policy, management practice and program planning requires continuous long-term interaction between researchers and decision makers. Partnerships for Health System
Improvement (PHSI) is an integrated knowledge translation (KT) program that supports partnerships between researchers and decision makers who are committed to working together on applied and policy-relevant health policy research that is deemed high priority by decision makers. The program aims to increase the likelihood that the research it supports is used to improve health policy, health care management and program planning. The purpose of the study is to describe the PHSI program, its intended outcomes, and the initial findings regarding the program’s effects on evidence-informed decision making.
Study Design: Since 2004, the PHSI program has supported 68 teams conducting research in priority areas as deemed relevant by health care decision makers in Canada. This study took the form of a preliminary program evaluation to assess 1) the generation of new knowledge; 2) the strength of the relationships between researchers and decision makers; and 3) the impact of the research on policy, practice, or program planning. A document review of mid-term and final grant reports was conducted to determine progress according to these three dimensions. In order to gain in-depth understanding of the PHSI programs’ intended effects, semi-structured telephone interviews were conducted with the principal investigators and principal decision makers of ten research teams, which were identified through purposive sampling.
Case studies were used to highlight particularly innovative success stories.
Population Studied: The PHSI program supports researchers and decision makers to conduct applied health systems and services research.
Decision makers included individuals at the provincial and regional levels of government, as well as directors and managers of hospitals, health care organisations and voluntary health organisations.
Principal Findings: Preliminary findings are encouraging. It appears that the PHSI program has been successful in supporting the generation of new knowledge and the establishment of productive and meaningful partnerships between health services researchers and decision makers. These partnerships have fostered research evidence that is considered by participating decision makers to be applicable and relevant to the decisions they face.
Progress along the three evaluation dimensions appears to depend partly on factors that lie outside of the program’s influence, such as the organizational context within which the decision makers operate.
Conclusion: The findings of this preliminary evaluation are encouraging, for they suggest that the PHSI program is achieving its intended outcomes of supporting the generation of new and policy-relevant research evidence that is considered useful by decision makers.
Implications for Policy, Delivery or Practice:
This study may be of interest to health research funding bodies or other organisations interested in developing programs that contribute to evidenceinformed health care decision-making.
Funding Source(s): Canadian Institutes of Health
Research
Theme: Organizational Performance and
Management
● Human & Organizational Characteristics of
High & Low Performing Nursing Homes: A
Comparative Study Using Mixed Methods
Darcy McMaughan Moudouni, M.S.; Bita Kash,
M.B.A., Ph.D.
Presented by: Darcy McMaughan Moudouni, M.S.,
Graduate Assistant Researcher, Health Policy &
Management, Texas A&M Health Science Center
School of Rural Public Health, Adriance Way,
College Station, TX 77840, Phone: (979) 294-3983;
Email: dkmoudouni@srph.tamhsc.edu
Research Objective: The purpose of this exploratory study was to identify differences in human and organizational characteristics in high and low performing nursing homes in an effort to 1) understand what constitutes adaptive and maladaptive nursing home care practices from the perspective of long-term care consumers and 2) validate a system for nursing home performance evaluation
Study Design: Quantitative data of relevant performance measures were retrieved from two secondary data sources (Medicaid cost reports and
MDS based quality data) and used as selection criteria to identify low and high performing nursing homes. Facility designation was based on cutoff scores (marking top and bottom twenty performers in the region) generated from performance measures related to people, processes, quality, service, and finance. After identification of high and low performing facilities focus groups with residents and family were administered and qualitative data was analyzed based on an inductive methodology.
Six nursing home facilities (three “high performing” and three “low performing”) from the two
comparison groups participated in the qualitative study.
Population Studied: The study population included a total of 45 long term care consumers (family members and cognitively-intact residents) of six nursing homes in the Southwestern United States.
Principal Findings: Low and high performing facilities selected via the quantitative analysis of quality measures also differed in residents’ and families’ perspectives of human and organizational factors. Prominent themes related to staffing, management practices, and physical environment emerged from the comparative focus group analysis. Themes differed in tone along the lines demarking low and high performing facilities.
Participants from low performing homes expressed fear and concerns for safety, lack of attention by staff to basic needs, chronic low staffing levels, and limited familial involvement. Participants from high performing homes expressed comfort with their surroundings, fulfillment of basic and higher needs by staff, and intense familial involvement.
Conclusion: Long-term care consumers can provide valuable insight into work and organizational practices that raise (or in some instances lower) perceived quality of nursing home care. These insights point to differences between low and high performing facilities that are amendable to change, rather than traditional classifications of homes not easily changed (such as profit status or geographic location).
Recommendations include efforts to improve communication between staff and consumers, attention to food preferences, attention to interior facility aesthetics as experienced by residents and development of a sense of ownership in the facility.
Implications for Policy, Delivery or Practice:
Identification of practices in low and high performing nursing homes as perceived by the nursing home care consumer can be utilized for making recommendations to nursing home administrators on improving quality of care, and to validate current nursing home evaluation practices using a consumer point of view.
Funding Source(s): Texas A&M HSC Research
Development Grant
Theme: Long Term Care
● The Impact of Formal Care Authorization on
Informal Care Utilization: Can One Replace the
Other?
Darcy McMaughan Moudouni, M.S., E.D.; Robert
Ohsfeldt, Ph.D.; Charles Phillips, Ph.D., M.P.H.
Presented by: Darcy McMaughan Moudouni, M.S.,
E.D., Graduate Assistant, Research, Health Policy
& Management, Texas A&M Health Science Center
School of Rural Public Health, 1266 TAMU, College
Station, TX 77843, Phone: (979) 204-3983; Email: dkmoudouni@srph.tamhsc.edu
Research Objective: As formal health care costs climb, informal care (unpaid care) is explored as a means for reducing formal care consumption and expenditures. Using informal care to reduce formal care utilization assumes the two forms of care function as substitutes, meaning provision of informal care reduces formal care use. Although previous research provides some support for a substitutive relationship, other findings suggest informal and formal care differ in terms of availability and utility to individuals in need and are therefore complementary. This study aims to ascertain the impact of level of formal care authorized on informal care utilized.
Study Design: Data for the cross-sectional analysis were collected using a variant of the interRAI Community Health Assessment Tool and analyzed using OLS and instrumental variable analysis. Instrumental variable analysis was utilized to control for potential bias due to endogeneity. This study differs from other studies of informal/formal care by conceptualizing informal care as the dependent variable, including informal care from all informal caregivers (not just children or spouses), and using formal care hours allocated (as opposed to utilized) as the main treatment variable.
Population Studied: This research explores the relationship between formal and informal care among 472 adult Medicaid home health care recipients residing in four areas in Texas.
Principal Findings: A positive and significant relationship was obtained between formal care and informal care, suggesting that informal and formal care are complements. If informal care and formal care are substitutes, the estimate of ß1 should be negative. Although the impact of formal care eventually becomes statistically insignificant after controlling for access to informal care, need for care, and client demographics the relationship maintains its direction. There remains a positive relationship between formal and informal care indicating the two forms of care are complements.
This complementary relationship is maintained throughout different model iterations, from OLS to
IV.
Conclusion: Results indicate that, after controlling for a variety of care-recipient characteristics formal and informal care are not substitutes for one another, but rather complements. As the hours of formal care authorized increases we would expect the hours of informal care utilized to likewise increase (all other influences held constant). In this sample, the provision of formal care through the
Medicaid program did not reduce the level of informal care provided by friends and family.
Implications for Policy, Delivery or Practice:
Public policy geared towards reducing formal care expenditures by inhibiting formal care utilization, with the rationale that informal care can replace formal care may result in unmet health care needs.
Formularies for allocation of formal care sometimes include amount of informal care utilized to ‘adjust’
the amount of formal care. The amount of informal care a client receives may be used reduce the amount of formal care allocated based on the supposition that formal and informal care are substitutes. Such care allocation policies including informal care received as an adjustment for formal care bear revisiting, as informal care does not appear to fulfill the same needs as formal care.
Theme: Long Term Care
● Spillover Effects of the Uninsured
Stacey McMorrow
Presented by: Stacey McMorrow, University of
Pennsylvania, 3641 Locust Walk, Philadelphia, PA
19104, Phone: (215) 840-2767; Email: staceymc@wharton.upenn.edu
Research Objective: This paper explores the relationship between local uninsurance rates and the quality of inpatient care provided to Medicare beneficiaries. Changing uninsurance rates in a market can alter the incentives for providers to invest in quality and can thus impact patients with all types of insurance. An inability to spread some fixed cost of quality or an adjustment of practice patterns to reflect the high cost of uncompensated care may result in lower quality for Medicare beneficiaries in markets with high uninsurance rates. The potential for such spillover effects at the market-level will depend on the ability of providers to segment the market based on insurance coverage or specialize in a particular type of patient. Smaller markets, where this segmentation is apt to be incomplete, will be more likely to exhibit spillover effects.
Study Design: Changes in MSA-level uninsurance rates over time (1996-97 to 2002-03) are used to identify the effect of the insurance distribution on
Medicare quality. Data from the Current Population
Survey is used to identify uninsurance rates for the
100 largest MSAs in the US. Quality is measured by applying the AHRQ quality indicators to
Medicare discharge data. Discharge-level logistic regressions are then estimated on binary indicators of mortality from several conditions and procedures.
The models include individual-, hospital- and MSAlevel covariates as well as MSA fixed-effects to control for unobservable market characteristics that are constant over time. Separate models are estimated for markets of various sizes.
Population Studied: The study population consists of Medicare beneficiaries who live in and receive inpatient care from a hospital in one of the 100 largest MSAs. Stronger results would be expected in smaller MSAs.
Principal Findings: In smaller markets, evidence of a negative spillover effect from the uninsured to
Medicare beneficiaries is strong and consistent. All six models show a sign on the uninsurance rate which indicates that a larger growth in uninsurance results in a smaller improvement in quality over time. Four of the six are significant at the 5 percent level and the odds ratios indicate meaningful effects with a range from 3.8 for stroke to 68 for hip fracture. In larger markets, the results are somewhat weaker and less consistent. Five of the six models show a sign that indicates a positive spillover effect from the uninsured to Medicare beneficiaries, but only two are significant at the 5 percent level (hip fracture and pneumonia).
Conclusion: The findings for small markets are generally as expected. As uninsurance rates rise, provider incentives to invest in quality appear to be altered in such a way as to create negative spillover effects to Medicare beneficiaries. Larger markets exhibit more surprising findings which will require further investigation.
Implications for Policy, Delivery or Practice: The policy implications of this work are among its most significant contributions. The evidence that the uninsured exert a negative externality on Medicare beneficiaries in their local healthcare market provides a far stronger justification for an intervention to cover the uninsured than a typical moral or ethical argument. Further evidence on the spillover effects of the uninsured will continue to strengthen this argument.
Funding Source(s): AHRQ
Theme: Health Care Markets and Competition
● Costs of Admissions for Treating
Complications of Care
Peter McNair, B.N., M.P.H., M.H.S.; Associate
Professor Terri Jackson, Ph.D.; Danial Borovnicar,
Ph.D.
Presented by: Peter McNair, B.N., M.P.H., M.H.S.,
Visiting Fellow, Phillip R. Lee Institute for Health
Policy Studies, University of California, San
Francisco, 3333 California Street, Suite 265, San
Francisco, CA 94114, Phone: (415) 630-3295;
Email: peter.mcnair@dhs.vic.gov.au
Research Objective: To quantify the frequency of, and the costs and payments associated with, admissions for treatment of injuries and illnesses that are consequences of care.
Study Design: Secondary data analysis of admissions where a principal diagnosis specifies a complication of care.
Population Studied: Routinely coded 2005-06 public hospital inpatient data from Victoria, Australia
(1.25mil admissions) and corresponding patientlevel cost data (1.04mil admissions). Payments reflected DRG-based prospective rates.
Principal Findings: 1.5% (15,336) of the cost data admissions specifically treat an injury or illness arising from medical or surgical care. This consumed 2.74% of hospital prospective payments and represented $AUD 89.3 mil (2.84%) in total reported costs. 1.4% (17,429) of all public hospital admissions and 2.82% of hospital prospective payments (estimated cost - $AUD101.5mil/year)
were used to treat complications of care. Private residence was the source of 84.9% (14,804) of admissions to treat complications of care, which may have been sustained in private or public, inpatient or outpatient settings.
Conclusion: Admissions for treating complications of care represent a small, relatively expensive, proportion of hospital admissions and account for disproportionate levels of hospital costs and funding. Policy options providing incentives to manage complications and costs more effectively are discussed.
Implications for Policy, Delivery or Practice:
This study demonstrates considerable costs to health care systems arising from hospital-acquired injury and illness. While the incidence and cost of hospital acquired injury has been estimated across jurisdictions, the focus of inpatient payment modification policies has been on events that occur within a single discharge. Payers of hospital care have rarely focused on funding policies that address the economic burden of readmissions for those injuries and illnesses that are a direct consequence of medical and surgical care. For example, the Centers for Medicare and Medicaid services policy involving non-payment for Hospital
Acquired Conditions (HAC) attempts to modify IPPS payment without taking into account sequelae nor readmissions associated with the HAC to date, although the impact of HAC associated readmissions has been flagged as a priority for inpatient payment modification by the US Medicare
Payment Advisory Committee (MedPAC). How the payments should be modified is problematic. IPPS type payments for acute care episodes based on
DRG-specific cost weights provide a theoretical incentive to transfer cases with hospital-acquired complications as these cases are more costly than cases with no hospital-acquired complication
(results not shown). Moreover, more than one hospital could potentially claim reimbursement for treatment of the same adverse event when the patient is transferred. Monopsony funders like
Medicare could look to trace patients across episodes to identify all costs incurred by receiving hospitals and allocate the costs back to the original episode for estimation of DRG cost weights. This would result in funding the originating hospital based on the average cost of the entire complicated episode (including readmissions), with the receiving hospital’s costs to be fully recovered from the originating hospital. This would create a strong incentive for hospitals to minimize complications of care, although it may also drive efforts to avoid complex or high-risk patients more generally.
Funding Source(s): CWF
Theme: Quality and Efficiency: Policies and
Incentives
● Gender Differences in Psychosocial Issues
Related to Coronary Heart Disease
Holly Mead, Ph.D,; Christal Ramos, M.P.H.
Presented by: Holly Mead, Ph.D., Assistant
Research Professor, Department of Health Policy,
The George Washington University School of Public
Health & Health Services, 2021 K. Street, NW,
Suite 800, Washington, DC 20008, Phone: (202)
994-8615; Email: khmead@gwu.edu
Research Objective: Many patients experiencing chronic illness also struggle with psychosocial issues that impede their ability to care for their illness. Psychosocial barriers associated with CHD, however, may be different for men and women.
This paper examines the psychosocial obstacles that patients with CHD experience and seeks to better understand gender differences in an effort to develop and customize programs that support selfmanagement equally as well for men and women with chronic illnesses.
Study Design: We convened 33 focus groups in 10 cities across the country with a total of 387 participants to explore the barriers that patients’ experience as they care for their chronic heart conditions. Participants in the groups suffered from a heart condition or related chronic illness. We conducted content analysis to identify key psychosocial issues that illustrate patients’ experiences with their illnesses and to analyze differences by gender.
Population Studied: Participants in the focus groups were predominantly minority patients with low socio-economic status who suffered from a chronic heart condition.
Principal Findings: We identified 5 major categories that reflect the most common psychosocial issues discussed by patients’ as impediments to caring for their illness: 1) depression associated with a long-term illness; 2) fear of death and other serious health consequences resulting from CHD; 3) anxiety/stress related to job loss and self-sufficiency and 4) anger associated with feelings of frustration and lack of control and 5) isolation and loneliness.
Depression was the most common articulation of a psychological consequence of CHD. Participants of both genders were equally as likely to talk about depression related to their heart condition.
Differences, however, emerged when examining how the depression impacted their lives. Women were more likely to say it interfered with their ability to care for their disease, while men were more likely to talk about their depression as a physical aspect of their disease and therefore out of their control.
Men were also much more likely to talk about psychosocial problems in terms of anger/frustration and anxiety, while women experienced psychosocial issues in terms of depression, fear and isolation.
Conclusion: Analysis of the focus group interviews found that both men and women faced psychosocial issues that impeded their ability to manage the care of their CHD. Gender differences
did emerge in terms of how men and women identify and articulate psychosocial aspects of their illnesses and how these issues translate into significant barriers to self-management of their illness.
Implications for Policy, Delivery or Practice: The chronic care model provides capacity for the management of patients with comorbidities, such as coronary heart disease and depression. However, too often the psychological illness goes undiagnosed because there are no consistent mechanism in place to identify, monitor and manage these issues among patients with chronic illness. By identifying and revealing the gender differences of patients’ personal experiences dealing with the psychological impacts of CHD, this paper can help develop programs to identify the spectrum of psychosocial issues experienced by both men and women and support care for these problems.
Funding Source(s): RWJF
Theme: Gender and Health
● Medicare’s Home Medical Equipment Benefit
Before Competitive Bidding: How Satisfied Are
Consumers & Their Agents with the Services &
Products Furnished?
Ann Meadow, Sc.D.; Andrea Hassol, M.S.; Sarah
Shoemaker, Ph.D., Pharm.D.; John Christianson,
Ph.D.
Presented by: Ann Meadow, Sc.D., Social Science
Research Analyst, Research & Evaluation Group,
Centers for Medicare & Medicaid Services, 7500
Security Boulevard, Baltimore, MD 21244-1850,
Phone: (410) 786-6602; Email: ann.meadow@cms.hhs.gov
Research Objective: In a departure from traditional government-administered fee schedules, the
Medicare Modernization Act of 2003 mandated that
Medicare establish competitive bidding to set prices and to contract with suppliers furnishing selected items of durable medical equipment. Before the initial bidding round, we collected data to understand the processes and outcomes of DME provision under Medicare, and to measure quality and beneficiary satisfaction.
Study Design: We conducted focus groups of referral agents and mail surveys with phone followup of fee-for-service beneficiaries in Fall
2007. Focus group participants came from organizations employing staff who arrange for DME on behalf of beneficiaries (e.g., discharge planners, visiting nurses). From the National Claims History, we identified survey samples of beneficiaries who used Medicare’s DME benefit. The overall response rate was 57%.
Population Studied: We surveyed beneficiaries in five of the 80 largest MSAs, in which Medicare intends to conduct bidding initially. Samples are representative of fee-for-service, then-current users of equipment in five categories (oxygen, hospital beds, walkers, continuous positive airway pressure machines, and power wheelchairs) within each
MSA.
Principal Findings: Referral agents indicated that large numbers of suppliers compete aggressively for referrals, based primarily on fast and timely order fulfillment, proper fitting, training the patient and family in use/maintenance, and responding quickly to problems. In most organizations, referral agents distribute referrals among suppliers they trust. Under administered pricing, fees are irrelevant to making supplier distinctions, although some firms differentiate themselves by adding free accessories.
Beneficiary satisfaction varied by product category.
Asked whether the respondent would recommend the supplier to a friend, hospital bed, wheelchair, and CPAP users (85%-88%) were less likely to recommend their supplier than oxygen and walker users (95%-96%) (N=6,650 weighted for sampling probability and nonresponse, p < .0001, after adjusting for self-reported health status, spousal presence, education, income, racial/minority status, proxy respondent, and MSA). Similar differences existed when respondents were asked for global ratings (on a five-point scale) of the supplier.
Service quality indicators, including whether the respondent believes “you have the [equipment] that is right for you,” and the amount of "time and energy" it takes to arrange for equipment, also varied by product category. For example, hospital bed, wheelchair, and CPAP users (87%-92%) were less likely to believe they had the right equipment than oxygen and walker users (97%) (p < .0001).
The consumer satisfaction measures
(recommending supplier and global rating) were strongly related to several quality indicators, particularly the two indicators addressing service
(which raised the odds of recommending the supplier 12.6 and 6.6 times, respectively), as opposed to two indicators of equipment reliability
(OR=5.3 and 1.6, respectively). After controlling for the two service quality measures, only hospital bed users were statistically less satisfied than users of the other equipment categories.
Conclusion: Consumer satisfaction reflected generally high levels, consistent with previous studies. Consumers apparently distinguish service from equipment quality when evaluating their suppliers. Survey-reported instances of poor firm performance suggest that, even amid a typically generous supply of DME sources, hospital-based referral agents may have limited awareness of quality problems, since many do not follow the patient after return to home.
Implications for Policy, Delivery or Practice:
Aggressive competition among a multitude of suppliers to gain the favor of referral agents may be an indication that prices are not economically optimal prior to competitive bidding. Variation in survey measures of quality and consumer satisfaction, in the context of the impending national
bidding program, indicates there may be opportunities to strengthen the benefits derived from Medicare-sponsored assistive technology, particularly when this involves the types associated with lower service ratings. Whether it is feasible to increase accountability of suppliers for equipment reliability deserves study.
Funding Source(s): CMS
Theme: Medicare
● Team San Diego: Translating Chronic Care
Management Research into Practice
Mark Meiners, PhD., M.A., B.A.; Evalyn Greb,
M.S.W.; Sidney Johnson, M.S.
Presented by: Mark Meiners, PhD., M.A., B.A.,
Professor, Administration & Policy, George Mason
University, 4400 University Drive, MS 1j3, Fairfax,
VA 22030, Phone: (703) 930-1909; Email: mmeiners@gmu.edu
Research Objective: Team San Diego (TSD) is a community health education initiative designed to improve community care coordination for older, disables, and chronic care patients who need extensive medical and social services.
Study Design: This research project developed a concise online and in class curriculum to engage physicians, their office staff, and community-based health and social service providers in a training program to increase awareness of the target population's chronic care needs and available resources to serve those needs. It gives special emphasis to issues of consumer empowerment techniques and tools and the need for coordinating care across settings and providers with a virtual team approach. Those who take the training are surveyed to determine outcomes related to community chronic care improvement.
Population Studied: San Diego County based medical and social service providers who serve chronically ill, aged, and persons with disabilities.
Principal Findings: The translational research project resulted in a new knowledge base and strategy leading to broader community understanding of chronic care management improvements. The local aging services agency improved its ability to use the Network of Care web tool for referral resources, patient education, personal health record-keeping.
Conclusion: Collaborative team relationships between physician offices, ancillary providers and local neighborhood social service providers can be improved through a targeted training and outreach efforts.
Implications for Policy, Delivery or Practice:
Social marketing efforts to encourage improved chronic care management are needed in communities around the County. A basic community health education curriculum for medical and social services providers that emphasize the topics covered in TEAM SAN DIEGO gives such efforts a useful framework to work toward chronic care management improvements.
Funding Source(s): California Endowment and
Alliance Health Care Foundation
● Self Report Versus Pharmacy Data Adherence
Measures
Stephanie Melnyk, Pharm D.; Amy Jeffreys, M.
Stat.; Cynthia Coffman, Ph.D.; Carolyn Thorpe,
Ph.D.; Christopher Bryson, M.D.; David Edelman,
M.D.
Presented by: Stephanie Melnyk, Pharm D.,
Research Associate, VA's Health Services
Research & Development Service (152), Durham
VA Medical Center, 508 Fulton Street, Durham, NC
27705, Phone: (919) 286-6936; Email: stephanie.melnyk@va.gov
Research Objective: Numerous measures of patient medication adherence exist, but their relative efficacy is unknown. The objective of this study was to compare ReComp, a measure of adherence based on pharmacy data, to the Morisky
(MSR), a self reported adherence scale, and to compare differences in Hemoglobin A1c (A1c), Low
Density Lipoprotein (LDL), and systolic blood pressure (SBP) among all patients.
Study Design: We performed cross-sectional and longitudinal analyses on data from patients enrolled in a randomized, controlled trial of diabetes group medical clinics (GMC). ReComp was calculated from prescription data for oral hypoglycemics
(OHA), antihypertensives, and HMG-COA reductase inhibitors (statins) acquired during the year prior to enrollment. 212 patients completed a
Morisky survey and had enough data to calculate
ReComp. The ReComp score was truncated at 1.0 and then the average was taken at the patient level for each medication type and dichotomized into <
0.80 (not adherent) and >=0.80 (adherent). The
MSR was dichotomized into perfect (score=4) and imperfect (score < 4) adherence. McNemar’s Test was used to examine differences in the measurement of medication adherence between both measures and an unweighted kappa statistic was calculated to assess agreement. The first modeling step fit separate linear regression models for each outcome (A1c, LDL, SBP) to assess the effect of adherence using either the Recomp or
MSR measure on baseline outcomes adjusting for site, age, race, years since DM diagnosis, BMI and if on insulin or not (A1c outcome only). For the longitudinal analyses, we fit linear mixed-effects models (LMM) to account for the correlation in measures over time and to adjust for clustering within each GMC. We examined whether the outcomes changed differently over time between adherent and non-adherent patients for both adherence measures. We then fit LMMs to these outcomes to assess whether they changed
differently over time in adherent and non-adherent patients between the intervention and control arms.
Population Studied: 239 patients enrolled in primary care at the Durham or Richmond VA
Medical Centers and enrolled in a randomized, controlled trial of diabetes group medical clinics. All patients had poorly controlled diabetes (DM)
(A1c>= 7.5%) and hypertension (SBP) >=140 or diastolic BP >=90).
Principal Findings: Baseline mean LDL was 98 mg/dl, SBP was 153 mmHg, and A1c was 9.2%.
Adherence measured by ReComp was greater than
MSR (44% vs 35%, p=0.03); Kappa was 0.12 (95%
CI -0.01, 0.24). For ReComp and MSR measures respectively, adherent patients had lower baseline
LDL (-13 mg/dl, p<0.03 and -12 mg/dl, p<0.04) but no difference in A1c (-0.3 %, p=0.07 and 0.1 %, p=0.7) or SBP (0.1 mmHg, p=0.9 and -0.2 mmHg, p=0.9) compared to non-adherent. Over 12 months, patients who were adherent by ReComp improved
LDL by 3 mg/dl, whereas non-adherent patients improved 16 mg/dl (p=0.01). There was no similar difference in either A1c or SBP. At 12 months, patients adherent by MSR improved A1c by 1.1%, whereas non-adherent patients improved 0.5%
(p=0.01). There was no difference in either LDL or
SBP. Analyses incorporating treatment arm suggest that the differences were largely in the intervention arm but findings were not statistically significant.
Conclusion: There is poor agreement between
MSR and ReComp. Patients who are not filling their prescriptions may have more lipid control benefit from a GMC, presumably via improved adherence. In contrast, patients reporting perfect medication adherence may have more glycemic control benefit from a GMC, presumably by improving their medication regimens.
Implications for Policy, Delivery or Practice:
Researchers studying medication adherence may wish to choose measures carefully, as self-report and medication records may be measuring different domains of adherence.
Funding Source(s): VA
● The Effects of Introducing only PPO plans as a new Employer-Sponsored Insurance Option on Consumer Health Care Utilization & Out-of-
Pocket Spending: A Natural Experiment (2005-
2007)
Doug Melton, M.P.H., Ph.D.; Sandra Greene,
Dr.P.H.
Presented by: Doug Melton, M.P.H., Ph.D., Predoctoral Research Fellow, Health Policy &
Management, University of North Carolina at
Chapel Hill, P.O. Box 1574, Pittsboro, NC 27312,
Phone: (919) 260-1568; Email: lmelton@email.unc.edu
Research Objective: We examined whether there were observable differences in health care utilization and out-of-pocket (OOP) spending for employees when given the opportunity to switch to one of three PPO plans or an indemnity plan.
Employer acceptance of PPOs in the group insurance market has grown recently because
PPOs are seen as cost-containment strategies that cover the same set of services at a lower-cost than indemnities, while also offering consumers less utilization management. We will produce results revealing whether PPOs were effective in reducing employee health services utilization and OOP spending for those members who switched to a
PPO.
Study Design: We used a quasi-experimental design that combines NC State Employees Health
Plan claims data from October 1, 2005 to
September 30, 2006, when only an indemnity plan was offered, and claims data from October 1, 2006 to September 30, 2007, when three new PPOs were offered in addition to the indemnity plan. Two-
Stage Least Squares was used to estimate parameters in the two separate models: one model where utilization is the dependent variable and another model where OOP spending is the dependent variable. Utilization in this study is the number of services received, defined as the total number of Current Procedural Terminology Codes reported in the inpatient, outpatient, and professional claims. OOP spending is defined as the employee contribution (i.e., premium, deductible, copayment).The categorical independent variable of interest in both models was health plan type which consisted of the four available health plans (i.e., 3 PPOs and 1
Indemnity). We controlled for age, gender, socioeconomic status, and comorbidities.
Population Studied: The analysis was conducted on 328,172 commercially insured NC state employees and their family members between the ages of 1 and 64, who were continuously enrolled in the North Carolina SEHP from October 1, 2005 to
September 30, 2007. Following the introduction of three new PPO plans for coverage year October
2006 to September 2007, 31% of members remained in the indemnity plan, 4% switched to a
PPO Basic plan, 52% switched to a PPO Standard plan, and 13% switched to a PPO Plus plan. In the study sample, 63% of members were females. In terms of family additional family coverage, 20% of members were children-dependents and 5% of members were spousal-dependents covered under the health plan. 63% of members analyzed were females.
Principal Findings: Data analysis and results will be complete by annual research meeting.
Conclusion: Data analysis and results will be complete by annual research meeting.
Implications for Policy, Delivery or Practice:
Employers may be able to reduce employee OOP spending by providing multiple types of plans to selection from. Specifically, including plans with varying OOP requirements gives employees the
choice to pay more in out-of-pocket premiums to obtain the level of health services they desire.
Funding Source(s): AHRQ, NRSA Sheps Center for Research
● Predictors of Health Plan Selection among
Employees in the North Carolina State
Employees Health Plan
Doug Melton, M.P.H., Ph.D.; Sandra Greene,
Dr.P.H.
Presented by: Doug Melton, M.P.H., Ph.D.,
Predoctoral Research Fellow, Health Policy &
Management, University of North Carolina at
Chapel Hill, P.O. Box 1574, Pittsboro, NC 27312,
Phone: (919) 260-1568; Email: lmelton@email.unc.edu
Research Objective: Extensive literature exists on employee health plan decision-making when given multiple new health plan designs (e.g. Point-of-
Service Plan, PPO, CDHP, Indemnity) to choose from. Less is known about employee decisionmaking when the only new health plan design being offered is a PPO. We examined whether employees health services utilization during the 2005-2006 coverage period was associated with selecting a new Preferred Provider Organizations (PPO) plan or deciding to remain in the pre-existing indemnity plan, for the 2006-2007 coverage period.
Study Design: We used a longitudinal study design that combines NC State Employees Health Plan
(SEHP) claims data from October 1, 2005 to
September 30, 2006, when only an indemnity-plan was offered, and claims data from October 1, 2006 to September 30, 2007, when three new PPOs were offered in addition to the indemnity plan. A multinomial logit model was used to estimate parameters and the dependent variable was the choice of one of four available health plans (i.e. 3
PPOs and 1 Indemnity). All four plans cover at least
97% of providers in the state. The plans differ by level of out-of-pocket spending; such as the no premium and high-deductible plans (i.e. PPO Basic and Indemnity), the no premium and moderate deductible plan (i.e. PPO Standard), and the monthly premium and low-deductible plan (i.e. PPO
Plus). The primary independent variable was a Pre-
PPO utilization variable based on previous year (i.e.
Oct. 2005-Sept. 2006) utilization. Utilization in this study is defined as the total number of Current
Procedural Terminology Codes reported in the inpatient, outpatient, and professional claims. We controlled for age, gender, socioeconomic status, and comorbidities.
Population Studied: 328,172 commercially insured
NC state employees and their family members between the ages of 1 and 64, who were continuously enrolled in the North Carolina SEHP from October 1, 2005 to September 30, 2007 were analyzed. Descriptive statistics reveal that following the introduction of three new PPO plans, 31% of members remained in the indemnity plan, 4% switched to a PPO Basic plan, 52% switched to a
PPO Standard plan, and 13% switched to a PPO
Plus plan. In terms of family additional family coverage, 20% of members were childrendependents and 5% of members were spousaldependents covered under the health plan. 63% of members analyzed were females.
Principal Findings: Data analyses will be complete by annual research meeting. However, preliminary multivariate analysis reveal that Pre-PPO highutilizers of care residing in urban areas of the state were more likely to switch to a PPO plan (ß=.07, p<.001) then were Pre-PPO high-utilizers of care residing in rural areas of the state.
Conclusion: Results and conclusions will be complete by annual research meeting.
Implications for Policy, Delivery or Practice:
Differences in employee plan selection by residential location suggests to policymakers that information dissemination efforts to encourage employees to switch to a PPO may not have penetrated well to those employed in rural areas of the state.
Funding Source(s): AHRQ, NRSA Sheps Center for Research
● The Demand for Agency- & Consumer-
Directed Personal Assistance Services:
Evidence from a Randomized Experiment
Hongdao Meng, Ph.D.; Brenda Wamsley, Ph.D.
Presented by: Hongdao Meng, Ph.D., Assistant
Professor, Preventive Medicine, Stony Brook
University Medical Center, HSC, Level 3, Room
071, Stony Brook, NY 11794-8338, Phone: (631)
444-7281; Email: hongdao.meng@stonybrook.edu
Research Objective: To estimate the effect of an experimental voucher benefit on the demand for
Personal Assistance Services (PAS) under agencydirected and consumer-directed care models among Medicare beneficiaries with functional limitations.
Study Design: This study is a randomized controlled trial involving 1441 older Medicare beneficiaries from the Medicare Primary and
Consumer-Directed Care Demonstration
(MPCDCD). Among these participants, 754 were randomly assigned to receive a voucher benefit (Up to $250 per month for two years) that could be used toward either personal assistance services provided by a home care agency (agency-directed model) or by an independent worker (consumer-directed model). We estimated two-part models to identify the likelihood of PAS use (Part I, logit models) and the number of PAS hours used conditional on any use (Part II, ordinary least squares models). We controlled for individual baseline characteristics
(socio-demographic, health and functional status, and prior use of health services) in these models.
Population Studied: Older Medicare beneficiaries with functional limitations from 19 counties in
Western New York, West Virginal, and Ohio.
Principal Findings: The mean age of the sample was 80 years and more than two-thirds were female. Eight percent of the participants were enrolled in Medicaid and 60% reported annual household income of $20,000 or less. The average participant had a mean ADL and IADL dependency score of 2.4 and 3.6, respectively. During the twoyear follow-up, 41% of participants used agencydirected workers and 45% reported using consumer-directed workers. Voucher led to a 1.2 hour-per-week increase in agency-directed PAS and a 0.8 hour-per-week increase in consumerdirected PAS (p<0.05 in both cases). The effects of voucher were very different in urban and rural areas. Participants in urban areas were more likely to use agency-directed care while participants in rural areas were more likely to use consumerdirected care.
Conclusion: Voucher benefit lowered the effective price of PAS, thus led to increases in the hours of services received by Medicare beneficiaries.
However, urban participants seemed to prefer agency-directed PAS providers while rural participants seemed to prefer consumer-directed
PAS providers.
Implications for Policy, Delivery or Practice:
Evidence from the Cash and Counseling
Demonstration found that a voucher-like program reduced unmet needs and improved patient satisfaction and overall well-being. This study provided evidence that urban and rural Medicare beneficiaries with disabilities may have very different responses to changes in price for personal assistance. Benefit design of such programs to improve access to PAS need to take into account the availability of either type of PAS providers to achieve the intended policy effect.
Funding Source(s): CMS
Theme: Long Term Care
● What is the Impact of a New Law Requiring
California HMOs & Health Insurers to Cover HIV
Testing Regardless of a Primary Diagnosis?
Ying-Ying Meng, Dr.P.H.; Janet Coffman, Ph.D.;
Jay Ripps, F.S.A., M.A.A.A.; Chankyu Lee, M.B.A.;
Gerald Kominski, Ph.D.
Presented by: Ying-Ying Meng, Dr.P.H., Denior
Research Scientist, Center for Health Policy
Research, University of California, Los Angeles,
10960 Wilshire Boulevard, Suite 1550, Los
Angeles, CA 90024, Phone: (310) 794-2931;
Email: yymeng@ucla.edu
Research Objective: In October 2008, California
Governor Arnold Schwarzenegger signed into law a first-in-the-nation piece of legislation, which requires private health insurers to cover the cost of HIV testing regardless of whether the testing is related to a primary diagnosis for the encounter. This study is designed to estimate the impacts of the bill on coverage, utilization, and cost for Californians with commercial health insurance.
Study Design: We surveyed the seven largest health plans and insurers in California regarding their coverage and benefit levels for HIV testing.
We estimated the minimum and maximum effects of the expansion of coverage on costs and access one year after the mandate. The minimum effect assumed only expansion of coverage to include testing for asymptomatic patients with unknown risk of infection and who receive testing from an innetwork emergency or urgent care service provider, even if the testing is unrelated to the emergency or urgent care episode; while the maximum effect assumed expansion of coverage to meet the
Centers for Disease Control and Prevention (CDC) testing guidelines that recommend routine screening for all adults and adolescents regardless of perceived risk.
Population Studied: There are 22,190,000 individuals in California under age 65 with coverage in group and individual health insurance plans
(HMOs) or health insurance policies who would be affected by the mandate.
Principal Findings: All affected individuals had coverage for HIV testing before the mandate. Most affected individuals (65%) had coverage for HIV testing if medically necessary; an additional number of individuals (30%) had coverage for testing according to the United States Preventive Services
Task Force (USPSTF) guidelines (only testing those at high risk of infection, plus all pregnant women); and a small number of individuals (5%) had coverage of testing in accordance with the
CDC testing guidelines. If minimum expansion of coverage occurs (mainly to cover testing done by an in-network emergency or urgent care service providers), overall healthcare expenditures are estimated to increase by only $554,000 due to the new law. If testing broadens to comply with CDC testing guidelines, expenditures are estimated to increase by about $10,151,000 in the year following the effective date of the law.
Conclusion: The cost of expansion in coverage is small, which is about 0.01% of the total expenditures, even if the CDC guidelines are followed. Further studies are needed to assess the actual impact of this legislation on HIV testing after the law is enacted in 2009.
Implications for Policy, Delivery or Practice:
This policy change could serve as an important step towards making HIV testing a routine screening.
However, the impact of such a coverage mandate on access to HIV testing could be minimal if no efforts were made to increase providers and enrollees’ awareness and willingness to adopt the
CDC guidelines.
Funding Source(s): The California Health Benefits
Review Program (CHBRP)
Theme: Coverage and Access
● Organizational Culture & Employee & Patient
Satisfaction: A Case of Mediation
Mark Meterko, Ph.D.; Kelly Stolzmann, M.S.; Martin
Charns, D.B.A.
Presented by: Mark Meterko, Ph.D., Investigator,
Health Services Research & Development Center for Organization, Leadership & Management
Research, Veterans Health Administration, VA
Medical Center (152M), 150 South Huntington
Avenue, Boston, MA 02130, Phone: (857) 364-
4608; Email: mark.meterko@va.gov
Research Objective: Organizational culture has been described as a critical determinant of performance. However, efforts to empirically demonstrate the impact of culture on performance have yielded mixed results. One possible explanation is that the effect of culture may be indirect. To investigate this possibility, we examined whether employee job satisfaction was a mediator of the relationship between organizational culture and patient satisfaction.
Study Design: The present study was conducted in the Veterans Health Administration (VA). From the
FY07 All Employee Survey (AES) we obtained a single-item measure of overall job satisfaction and multi-item scale scores for four dimensions of organizational culture: group, entrepreneurial, hierarchical and rational. From the inpatient and outpatient versions of the FY07 Survey of
Healthcare Experiences of Patients (SHEP) we obtained single-item measures of overall satisfaction with care. We hypothesized that culture would influence staff job satisfaction, which would in turn affect patient satisfaction. Following the analytic model described by Barron and Kenny, multiple regression was used to test the hypothesized links by estimating three models: culture predicting patient satisfaction, culture predicting employee job satisfaction, and both culture and employee job satisfaction as predictors of patient satisfaction. The latter serves the dual purpose of estimating both the strength of (a) the relationship between the mediator and the outcome while controlling for the possibility that that relationship may be an artifact of their joint relationship to the primary predictor (culture); and
(b) the relationship between the primary predictor
(culture) and the outcome (patient satisfaction) after controlling for the influence of the mediator (job satisfaction).
Population Studied: Two populations were included in the present study: VA employees and patients. The employee sample consisted of those who responded to the 2007 AES, an annual national voluntary and anonymous census survey conducted to obtain staff perceptions of a variety of aspects of work life. The patient sample consisted of those veterans who were either discharged from an inpatient stay or had an ambulatory care visit and were randomly selected to receive a SHEP mail survey.
Principal Findings: The 2007 AES response rate was 76% and the SHEP response rates were 41% and 55% for the inpatient and outpatient surveys, respectively. All data were aggregated to the medical center level (n=129). Results consistent with a mediated relationship to performance were observed for two of four culture dimensions – hierarchical and group. For example, hierarchical culture was a significant predictor of both outpatient satisfaction and staff job satisfaction, but the relationship to outpatient satisfaction was substantially diminished when job satisfaction was included in the same model.
Conclusion: The impact of group and hierarchical culture on the outcome of outpatient satisfaction was mediated through employee satisfaction.
Implications for Policy, Delivery or Practice:
Examining the relationship between organizational culture and outcomes using a mediation model suggests that culture may be more important than implied by the results from models of the simple direct effects of culture. Although culture may be relatively resistant to change, knowledge of the indirect impact of culture provides managers at all levels with potential leverage points for influencing outcomes.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Team Work, Case Difficulty & Non-routine
Events in Operating Rooms
Ann Minnick, Ph.D., R.N., F.A.A.N.; Beth
Donaghey, M.A.; Jason Slagle, Ph.D.; Matt
Weinger, M.D.
Presented by: Ann Minnick, Ph.D., R.N., F.A.A.N.,
Senior Associate Dean for Research, Julia Eleanor
Chenault Professor of Nursing, School of Nursing,
Vanderbilt University, 461 21st Avenue South,
Nashville, TN 37240, Phone: (615) 343-2998;
Email: ann.minnick@vanderbilt.edu
Research Objective: To describe the differences and similarities in operating room (OR) nurses’, anesthesia providers’ and surgeons’ beliefs about surgical team function, case difficulty and nonroutine event (NRE) causation. A NRE is a deviation from optimal care in a specific care context. Understanding the meaning providers attach to these terms will inform their use in quality and workload assessment tools and in the design of peri-operative safety interventions. Although not every NRE involves harm, decreasing NREs is one step in preventing harm. Each specialty provider in a team can inform how team interaction during the task influences NREs.
Study Design: An interviewer, using a modified nominal group technique, conducted one-hour focus groups and interviews organized by specialty.
The pre-tested four question sessions were audio-
recorded and conducted using the Kreuger approach. Rating cards and audio transcripts were examined for themes and the identification of key events and sequences.
Population Studied: Surgeons, nurses and anesthesia providers (anesthesiologists and
CRNAs) in three Veterans Administration hospitals in two geographic areas of the US. The volunteer sample consisted of 34 nurses, 14 anesthesia providers and 10 surgeons.
Principal Findings: Similarities among provider types included (1) an understanding of case difficulty that was not heavily reliant on patient characteristics or procedural complexity (2) the belief that NREs have a cascade of causation that begins well before the day of surgery and (3) that
NRE prevention is partially outside of the roles of the provider types present during an operation.
Differences included: (1) level of awareness of other professions’ case difficulty (2) degree of emphasis on particular NRE causations and on effective NRE interventions/prevention, and (3) attributions/recognition of the role of personal performance in NRE occurrence. The data suggest a “perfect storm” of factors that increase the likelihood of an NRE. The data also demonstrate provider differences in team definition and function that may influence NRE occurrence. Although team consistency (regular members versus interchangeable professionals) was identified as an important potential factor in NRE occurrence, the majority of underlying causes of NRE were outside the traditionally defined surgical team in surgical scheduling and supply/equipment support.
Conclusion: There tends to be more similarities than differences by provider type in case difficulty definition and beliefs about NRE occurrence.
Opportunities for NRE prevention exist in and outside of the OR team triumvirate.
Implications for Policy, Delivery or Practice:
Opportunities to prevent NREs in the OR may exist in pre-operative scheduling and equipment/supply preparation, areas outside of typical OR team responsibilities that influence event sequencing and team function inside the OR. If these pathways are confirmed, safety improvement efforts will need to expand beyond current OR error prevention approaches. Other implications include: (1) providers’ ratings of case difficulty as a risk adjustment variable cannot be construed to represent technical complexity or the state of the patient (2) the influence of the fixed / interchangeable nature of surgical teams should be explored in seeking to determine how teamwork influences NREs and (3) differences in team definition and ratings of team performance by provider type should be explored to determine if resolution of these difference changes team performance.
Funding Source(s): VA
Theme: Health Care Workforce
● Strengthening Public Health Infrastructure:
The Role & Capacity of Public Health Institutes
Karen Minyard, Ph.D.; Mary Ann Phillips, M.P.H.;
Naima Wong, Ph.D., M.P.H.; Jane Branscomb, B.E.
Presented by: Karen Minyard, Ph.D., Executive
Director, Georgia Health Policy Center, Georgia
State University, 14 Marietta Street, Suite 221,
Atlanta, GA 30303, Phone: (404) 413-0301; Email: kminyard@gsu.edu
Research Objective: Over the past decade, the number of Public Health Institutes (PHIs) in the U.S. has expanded three-fold. With this growth, federal public health agencies and philanthropic organizations have become increasingly interested in understanding how PHIs can help strengthen the national public health infrastructure. This research aimed to 1) understand the overall capacity of existing PHIs, 2) assess the potential for mature
PHIs to provide technical assistance to emerging institutes, and 3) explore the contextual conditions in states without statewide PHIs.
Study Design: The study used mixed qualitative and quantitative methods. Qualitative methods included in-depth interviews with a sample of existing PHIs. Inductive and deductive methods were used to analyze thematic patterns among several factors including organizational development, technical assistance capacity, and content area focus. To examine the conditions in states without a PHI, contextual data for each state were collected. Variables for this contextual analysis included Commonwealth and United
Health Foundation rankings, proportion of uninsured population, and number of academic public health institutions. Logistic regression was used to estimate the likelihood that any of the select factors could predict the existence of a PHI.
Population Studied: The primary unit of analysis was PHIs (N=32) that were either members or affiliates of the National Network of Public Health
Institutes (NNPHI).
Principal Findings: Thematic analysis revealed five core elements that are essential for PHI success: 1) vision, 2) entrepreneurial leadership, 3) partnerships, 4) resources, and 5) organizational and programmatic capacity. When PHIs were able to combine these elements, they could respond nimbly to state-level public health needs and served as neutral conveners of diverse stakeholders. A major challenge for most PHIs was their limited access to core and unrestricted funding. PHIs' roles and impact within their respective states, along with their technical assistance capacity, varied according to maturity level, access to core funding, and state contextual factors. Regression analysis suggested no predictive relationship between the selected contextual factors and the existence of a PHI in a given state.
Conclusion: Findings from the assessment informed the development of recommendations for
strengthening PHIs. First, any such efforts should view these institutions as integral components of an enabling system. Second, they should address the core elements of vision, leadership, relationships and funding, in addition to organizational and programmatic capacity. Third, performance analysis of PHIs is needed to distinguish, quantify and convey their contributions to the public’s health.
Implications for Policy, Delivery or Practice:
PHIs are emerging as a critical facet of the nation’s public health infrastructure by serving as a bridge between private and governmental public health stakeholders. These institutes operate under conditions of high innovation, emerging development, and uncertainty. The current vulnerability of state budgets makes this an opportune time for national stakeholders to invest in enhancing the capacity of PHIs to leverage state funds to greatest effect. Future efforts to understand the contributions of PHIs and enhance their development will strengthen our ability to collectively address the complex health issues facing the nation.
Funding Source(s): NNPHI
Theme: Public Health
● Hospital Performance Assessment with
Existing Administrative Database: Enhancement of Case-Mix In-Hospital Mortality Risk-
Adjustment for Benchmarking
Hiroaki Miyata, Ph.D.; Hideki Hashimoto, M.D.,
D.P.H.; Hiromasa Horiguchi, Ph.D.; Shinya
Matsuda, M.D., Ph.D.
Presented by: Hiroaki Miyata, Ph.D., Assistant
Professor, Healthcare Quality Assessment, The
University of Tokyo, The University of Tokyo
Hospital, 7-3-1, Hongo, Bunkyo-ku, Tokyo, 113-
8655, JP, Phone: +81-3-5800-9121; Email: hm@umin.net
Research Objective: In-hospital mortality is an important performance measure for quality improvement, though it requires proper risk adjustment. Though disease-specific risk adjustment models have been well developed in several specialties, case-mix model for more generic use of outcome evaluation is rather underdeveloped. Few studies examine whether case-mix risk adjustment is evenly applicable to hospitals with various characteristics and case mix.
In this study, we developed a case-mix-based inhospital mortality prediction model, and evaluated its performance for benchmarking with nation-wide records of discharge cases from various types of hospitals.
Study Design: We generated and validated inhospital mortality prediction model using 50/50 splitsample validation. Based on value of each hospital’s c-index, we classified them into two group
(hospitals c-index->0.8; hospitals c-index<0.8) and examine differences of characteristics.
Population Studied: We used electronic standardized data of 1,878,767 discharged patients, provided by 469 hospitals during the period July 1,
2006 and October 31.
Principal Findings: Older age, male gender, use of ambulance at admission, and emergency admission status showed a significantly higher odds ratio. Hospitalization for examination and Planned short-term admission showed a significantly lower odds ratio. Charlson Index scores, Performance status, Fletcher Hugh-Jones Classification exhibited a linear increasing trend in odds ratio as the score went higher. The prediction model showed good discrimination with a c-index 0.882 in both test dataset and development dataset. In this study 446 hospitals (95%) had c-index ->0.8 and were classified into higher c-index group. Significantly higher rate of Lower c-index group were “Special hospitals” and “hospital with convalescent wards”.
Conclusion: The case mix model developed in this study exhibited a good degree of prediction accuracy and had adequate performance for benchmarking. As the case-mix model might not well applicable to particular kind of hospitals such as specialized hospitals or health facilities for recuperation, general hospitals would be appropriate for evaluation.
Implications for Policy, Delivery or Practice:
Though the all variables included in this study were easily accessible in administrative data and met international standards, model performance was still excellent. The framework of our case-mix model might be applicable and useful in other countries.
● The Impact of Asthma Action Plans on
Healthcare Utilization & Healthcare Costs
Praveen Mogill, B.S.; Briana Jegier, Ph.D., M.S.;
Kimberly Skarupski, Ph.D., M.P.H.; Theresa
Burkhart, M.S., M.H.A., F.A.C.H.E.; Brian Smith,
M.H.A.; Mark Yoder, M.D.
Presented by: Praveen Mogill, B.S., Student,
Health Systems Management, Rush University
Medical Center, Chicago, IL 60612, Phone: (847)
722-3008; Email: Praveen_Mogill@Rush.edu
Research Objective: Asthma is a prevalent chronic disease in the United States; treating asthma is costly for both the individual and healthcare providers. It is estimated that $16Billion dollars are spent annually treating asthma and asthma related conditions. Proper disease management can improve patient outcomes and lower healthcare costs. One physician-patient asthma management tool, the Asthma Action Plan (AAP), has been shown to improve patient outcomes as part of a larger asthma education program. However, its effectiveness alone in reducing healthcare utilization and its cost-effectiveness has not been studied. The purpose of this study is to examine the association between AAP completion and healthcare utilization, including Emergency
Department (ED) visits, hospitalizations, and cost, at a large urban academic medical center in
Chicago, IL.
Study Design: This study is a secondary analysis of 210 asthma patients who had asthma-related physician visits that occurred from July 1st, 2007 to
June 31st, 2008. Data for this study will come from electronic claims data and paper chart reviews from four faculty physician practices associated with the medical center. The variables included in these analyses include completion of the asthma action plan, ED visits, hospitalizations, costs, and patient characteristics including gender, age, race, asthmaseverity, and co-morbidities.
Population Studied: The inclusion criterion for these data included patient visits to any of the four faculty practices with a primary diagnosis of asthma. The four faculty practices were selected from the employed faculty physician practices of the academic medical center and represent both primary care practices and specialty care practices.
Analyses of these data are underway and will be completed in March, 2009.
Implications for Policy, Delivery or Practice:
This study will be the first to examine if the AAP reduces healthcare utilization as well as its costeffectiveness. The AAP is currently a component in
Medicare’s voluntary Physician Quality Reporting
Initiative and pay-for-reporting program and this study will provide evidence to support the inclusion of the AAP in broader pay-for-performance programs. This study will also provide actuarial evidence for reimbursement for physicians and healthcare institutions from other third-party payers outside of Medicare. Finally, this study may provide evidence to improve physician compensation strategies in academic medical center owned faculty practices.
Theme: Prevention and Treatment of Chronic
Illness
● The Influence of Service Lines on Physician’s
Professional Interest Identification
David Mohr, Ph.D.; Martin Charns, D.B.A.; Mark
Meterko, Ph.D.; Gary Young, J.D., Ph.D.
Presented by: David Mohr, Ph.D., Health Scientist,
VA Center for Organization, Leadership &
Management Research, Department of Veterans
Affairs, 150 South Huntington Avenue (152M),
Boston, MA 02130, Phone: (857) 364-5679; Email: david.mohr2@va.gov
Research Objective: The goal of this study was to examine the extent that working in a clinical service line has on physician’s professional interest activities. It has been speculated that service lines may decrease professional identification, but this has not been tested to date. Many hospitals have embarked on significant re-organization efforts to create service line structures as a way to provide care to a specific patient population (e.g. children) or disease (i.e. cardiology). As compared to a service focus (e.g. nursing), in a service line structure, heath care professionals need to coordinate activities with a greater number of other professionals with different backgrounds and across a larger number of departments to ensure a full array of services are provided for patients. A possible result of this broad and multi-disciplinary focus in service lines may be a reduction in professional identification with one’s own occupation.
Study Design: We obtained data from a survey conducted in 2000 at Veterans Health
Administration. The survey randomly sampled approximately150 respondents at 141 medical centers across the United States. A total of 7,058 completed responses were received and 417 respondents (7%) identified themselves as physicians. One section of the survey asked four questions on professional interest identification (i.e. reading journals, attending conferences, professional idealism) using a five point-scale ranging from “Strongly Disagree” to “Strongly
Agree”. We found a one factor solution among those items, a Cronbach’s alpha of .76 and a normal response distribution. Respondents to the survey also indicated if they worked in a service line, clinical setting, authority level, tenure, and type of facility (i.e. medical center, community clinic).
We also linked a variable indicating hospital teaching affiliation to the individual-level responses.
Population Studied: Physicians sampled during a
2000 national survey administration within the
Veterans Health Administration.
Principal Findings: Physicians reported higher scores on the professional interest identification scale (M = 4.11, SD=.67) compared to all other respondents (e.g. RNs, psychologists, non-clinical)
(M = 3.41, SD=.84). This finding was expected due to the emphasis placed on continuing medical education credits for physicians. In the regression model, we found a negative association between working in a service line and professional interest identification (b = -.25, p < .01). We also found a positive association between the variable of
“section chief” and professional interest identification (b = .18, p = .05). No other variables were significant in the model; r-squared =.07,
F(10,398)=2.71, p<.01.
Conclusion: The results of this study indicated that compared to other hospital staff, physicians expressed higher scores on professional interest identification. Working in a service line appeared to have a negative effect on professional interest identification for physicians.
Implications for Policy, Delivery or Practice:
Although service line structures may provide a competitive marketplace advantage, one consequence may be a decreased sense of professional interest identification for physicians.
This may present a challenge when recruiting new physicians to join the health care organization.
Funding Source(s): VA
Theme: Health Care Workforce
● Health Information Exchange Economic
Sustainability
Adil Moiduddin, M.P.A.; Jessica Krondstadt, M.P.P.;
Dan Gaylin, M.P.A.; Michael O'Grady, Ph.D.
Presented by: Adil Moiduddin, M.P.A., Senior
Research Scientist, Associate Director Health Care
Research, Health Care Research, National Opinion
Research Center at University of Chicago, 4350
East-West Highway, Bethesda, MD 20814, Phone:
(301) 634-9419; Email: moidudden-adil@norc.org
Research Objective: Health information Exchange
(HIE) holds the potential to increase the efficiency of our health care system, improve the quality of care, facilitate disease surveillance, and support research in clinical effectiveness and the evaluation of health care services. Despite these potential benefits, there are relatively few examples of sustainable exchanges of health information.
Grounded in an assessment of the current challenges to HIE, this project sought to develop and analyze a set of policy approaches for promoting more widespread engagement in the exchange of health information.
Principal Findings: Informed by a literature review and deliberations of a panel of experts, this study assessed four policy approaches: 1) Federal government focus, which includes HIE incentives and requirements for all federally underwritten health care programs; 2) State government focus, which provides states resources to build the HIE infrastructure and otherwise support exchange; 3)
Subsidizing the market, which offers tax incentives, loans, and grants to providers, payers, and insurers; and 4) Supporting the market, which extends current federal efforts to convene stakeholders and advance privacy and interoperability. The analysis placed a particular focus on one of the most critical challenges for
HIE—a lack of financial incentives or a misalignment between which entities stand to benefit from HIE and which ones bear the costs. To address this challenge, several of the approaches envisioned changes in reimbursement policy. This could involve direct incentives—bonus payments for engaging in HIE—or indirect ones—rewards for better health outcomes or demonstration of care that is coordinated among a patients’ different providers. One of the conclusions from analyzing the approaches is that when promoting HIE, generating demand for HIE from providers and patients must be complemented by actions to ensure that the infrastructure is in place to allow for the exchange of health information. Also, incentives must be structured such that they do not promote
HIE for HIE’s sake, but incentive HIE that delivers value.
Implications for Policy, Delivery or Practice: As health care consumes a growing share of the country’s GDP and medical errors continue to occur with alarming frequency, HIE holds the potential to help reign in spending and improve the safety and quality of care. As the market for HIE has been slow to evolve on its own, it is important to explore and analyze potential policy levers for widening its adoption. HIE also may play an important role in potential changes in the way health care is financed in this country. Not only may a more outcomesfocused payment structure provide incentives for
HIE, but the type of data that can be generated through exchange may be essential to implement new payment systems. For example, easy access to up-to-date data is a key ingredient in pay-forperformance systems. As policy makers contemplate methods for increasing the number of
HIE participants, consistent with the goals of the new administration, it is important to evaluate the tradeoffs that exist among potential policy approaches.
Funding Source(s): Office of the National
Coordinator for Health IT
Theme: Health Information Technology
● PHR Use in Medicare Health Plan Members:
Evaluating a Pilot Program
Adil Moiduddin, M.P.P.; Dan Gaylin, M.P.A.; Alison
Muckle, B.A.; Shamis Mohamoud, M.A.
Presented by: Adil Moiduddin, M.P.P., Senior
Research Scientist, Associate Director Health Care
Research, Health Care Research, National Opinion
Research Center at the University of Chicago, 4350
East-West Highway, Bethesda, MD 20814, Phone:
(301) 643-9419; Email: moiduddin-adil@norc.org
Research Objective: We set out to describe and assess the value of personal health records (PHR) for Medicare MA/PDP (managed care plan) beneficiaries and identify key barriers and enablers of PHR adoption and use based on the pilot experience.
Study Design: The evaluation includes a review of the functionality available through each of the PHRs employed as part of the pilot, analysis of utilization data provided by each of the health plans on a monthly basis, focus groups with beneficiaries representing users and non-users of PHRs and a survey of over 2,000 beneficiaries that have accessed their PHR.
Population Studied: Medicare beneficiaries enrolled in health plans under Part C or Part D.
Principal Findings: Currently, we have findings available from all phases except the survey.
Findings from discussions with beneficiaries reveal huge variatio in attitudes and attitudes towards
PHRs as well as PHR use. Higher rates of use and satisfaction tend to relate to higher levels of functionality especially having clinical lab values avialable through the PHR, being able to access
high quality health education materials, teh abiltiy to email or electronically message physicians and the ability to graph clinical measures.
Funding Source(s): AHRQ
Theme: Health Information Technology
● Using Safety Net Networks to Support HIT /
HIE: Findings from Six Site Visits
Adil Moiduddin, M.P.P.
Presented by: Adil Moiduddin, M.P.P., Senior
Research Scientist, Health Care Research, National
Opinion Research Center, 4350 East West
Highway, Bethesda, MD 20814, Phone: (301) 634-
9419; Email: moiduddin-adil@norc.org
Research Objective: Given the current emphasis on electronic health record (EHR) promotion and health center expansion, there is increasing interest in the use of EHRs by these providers to improve quality and efficiency of care delivered to the underserved. One approach to EHR adoption among health centers has been collaboration across health centers to form health center controlled networks (HCCNs). A series of site visits conducted under contract with the U.S. DHHS, have examined barriers and enablers of health IT adoption among these networks including implementation costs and challenges, role of health IT in quality improvement (QI), impact of health IT on efficiency and financial outcomes, network sustainability and the role of the safety net in health information exchange (HIE).
Study Design: Each site visit includes a series of in person and telephone meetings with leadership from HCCNs, health center medical directors, IT leadership and executive management. To date we have met with nearly100 stakeholders and conducted site visits in Chicago, IL; San Diego, CA; southeastern North Carolina, northern Utah and southern New Mexico with visits to Boston, MA and
San Francisco.
Principal Findings: To the extent that health centers have been able to pursue and achieve QI following EHR adoption, they have applied significant effort from clinical staff to customize EHR applications to offer specific decision support and reporting functionality not available through basic products. Many HCCNs have developed specialized data warehouses to replicate registry reports in an EHR framework or to allow for benchmarking and tracking of key metrics over time. Findings also show limited participation in regional HIE initiatives even among the most advanced health centers. Further analyses will elaborate the relationship between network characteristics, level of staffing and expertise, sources of funding, governance and organization and geography and outcomes of interest such as implementation success, evidence of impact and participation in HIE. Finally, we will present conclusions to assist policy makers and program officials in directing investments moving forward, help the informatics community understand the unique challenges facing effective adoption and use of health IT in the safety net and leave safety net stakeholders with important “take aways” on the effectiveness of different strategies represented by each site visit.
Theme: Health Information Technology
● Improving Direct Care Compensation:
Medicaid Wage Pass Through Adoption, 1999-
2004
Vincent Mor, Ph.D., M.P.A.; Lili Wang, Ph.D.,
M.P.H.; Zhanlian Feng, Ph.D.; Edward Miller
Presented by: Vincent Mor, Ph.D., M.P.A.,
Taubman Center for Public Policy, Brown
University, 67 George Street, Box 1977,
Providence, RI 02912-1977, Phone: (401) 863-
9311;
Research Objective: Although a well-trained, stable workforce is a prerequisite for quality longterm care, providers have had an increasingly difficult time recruiting and retaining direct care staff. State governments have sought to improve recruitment and retention through wage passthrough policies in which increases in Medicaid reimbursement are directed toward wages and benefits for nurse aides and other workers. The objective of this study is to identify factors that promote or impede the adoption of wage-pass through programs for nursing homes.
Study Design: This study uses a pooled crosssectional time series design to identify the determinants of wage-pass through adoption. The analysis extends from 1999 to 2004 and includes
282 observations across 47 states. Data derive from a variety of sources, including periodic surveys of state Medicaid officials. Robust standard errors are used to account for the clustering of observations across states over time. Yearly fixed effects are included as well.
Population Studied: The unit of analysis is the state-year.
Principal Findings: Six states had a wage passthrough program in 1999, 12 in 2000, 11 in 2001,
13 in 2002, 11 in 2003, and 10 in 2004. Findings indicate that states with more professional legislatures (b=.291, p<.05) and capable public welfare bureaucracies (b=1.191, p<.05) were more likely to adopt a wage-pass through program.
States were also more likely to adopt if they had higher Medicaid nursing home reimbursement rates
(b=.021, p<.10) and greater numbers of home health agencies per 100,000 population (b=.192, p<.10). By contrast, adoption was less likely to occur in states with higher unemployment rates (b=-
.849, p<.05) and more powerful nursing home lobbies (b=-.034, p<.05).
Conclusion: This study suggests that states with greater governing capacity were more likely to
adopt wage-pass through programs. It also suggests resistance to wage pass-through programs on the part of the nursing home industry, possibly because facilities prefer to retain the freedom to allocate revenue as they see fit and because of the additional burdens that often accompany implementation. That states with higher reimbursement were more likely to adopt may reflect greater concern among better paying states with how that money is spent. That states with higher unemployment rates were less likely to adopt may reflect the incentive to refrain from increasing reimbursement under adverse fiscal circumstances, not matter whether nursing homes are told how to spend that money or not. Nursing homes in states with greater home care penetration face greater competition for direct care staff. This may increase the likelihood of adoption as states seek to aid nursing homes’ efforts to recruit and retain sufficient numbers of workers.
Implications for Policy, Delivery or Practice:
Especially important for successful wage-pass through adoption is gaining the buy-in of the nursing home industry, perhaps by better balancing the reimbursement increases obtained with the additional administrative and auditing burdens imposed. State officials should recognize that expanding home- and community-based options may have the unintended effect of spurring spending increases via wage-pass through adoptions in an effort to help nursing homes compete for the limited pool of labor available.
Funding Source(s): NIA
Theme: Long Term Care
● Long-Term Employment Effects of Surviving
Cancer
John Moran, Ph.D.; Pamela Farley Short, Ph.D.;
Christopher Hollenbeak, Ph.D.
Presented by: John Moran, Ph.D., Assistant
Professor, Health Policy & Administration, Penn
State University, 504 Ford Building, University Park,
PA 16802, Phone: (814) 865-8893; Email: jrm12@psu.edu
Research Objective: As survival rates have dramatically improved, cancer is being transformed from a fatal disease into a chronic illness. As with other chronic illnesses, successful treatment of cancer must be measured in terms of the long-term quality of life of patients. Our research objective was to examine one important aspect of quality of life, namely, the long-term effects of cancer on employment. Because earlier research on this question has focused more on workers diagnosed at older ages, we were particularly interested in employment effects for younger workers and in comparing effects for younger and older workers.
Study Design: Four telephone interviews were conducted annually with 1173 adult cancer survivors, who were 25 to 62 years of age and working when diagnosed with cancer in 1997-1999.
A non-cancer comparison group of 4723 similarly aged adults working in June 1998 was identified from a national longitudinal survey, the Panel Study of Income Dynamics (PSID). Probit/Tobit regression and propensity score matching were used to compare employment between the two groups in 2002, 3-5 years after diagnosis for the cancer survivors. Three measures of employment were considered: the percent working, the percent working full-time (35+ hours per week), and average hours per week (including non-workers).
The regression and matching estimators controlled for other factors likely to affect employment, including socio-demographic characteristics, baseline job characteristics, and the presence of other health conditions.
Population Studied: The cancer survivors were identified from the cancer registries of four medical centers in Pennsylvania and Maryland. Included were survivors of all cancer types, except for common skin cancers and Stage 4 cancers where extended survival was unlikely. The PSID comparison group was nationally representative of adults in the U.S. population, excluding cancer survivors. Separate comparisons were made for four groups: older males (55-65 years of age in
2002), younger males (27-54 years of age in 2002), older females, and younger females. Additionally, employment effects were estimated separately for survivors who remained cancer-free and those with recurrences or second cancers.
Principal Findings: All measures of employment were significantly lower for cancer survivors in each demographic group, except for the employment rate among older women where the difference was smallest and less precisely measured. There were no significant differences in employment effects across demographic groups. Average employment effects ranged from 4.4 to 9.9 percentage points for any employment, 9.8 to 17.3 percentage points for full-time employment, and 4.6 to 6.2 hours per week. There was little effect on the probability of working for survivors who were cancer-free, but significant reductions in full-time employment and hours were observed for all groups except older women.
Conclusion: Even over the long-term, cancer survivors work less than other adults. Effects on the employment of workers who are nearing retirement are not very different from effects on younger workers. Effects on cancer-free survivors are largely confined to hours worked, not the likelihood of working at all.
Implications for Policy, Delivery or Practice: Like other chronic illnesses, cancer has long-lasting effects on patients’ quality of life that should be monitored and addressed as part of their ongoing care.
Funding Source(s): American Cancer Society
Theme: Prevention and Treatment of Chronic
Illness
● A Conceptual Framework for Targeting
Information to Medicare Beneficiaries: It’s Not
Black or White
Robert Morgan, Ph.D.; Cayla, Teal, Ph.D.; Debora
Paterniti, Ph.D.; Christi Murphy, B.F.A.; Robert
Morgan, Ph.D.
Presented by: Robert Morgan, Ph.D., Professor,
Management, Policy & Community Health,
University of Texas School of Public Health, 1200
Herman Pressler, Room E-343, Houston, TX
77030, Phone: (713) 500-9187; Email:
Robert.O.Morgan@uth.tmc.edu
Research Objective: The increased complexity of the Medicare system has resulted in significant research on how Medicare beneficiaries acquire and use knowledge about their health care system.
Medicare beneficiaries’ knowledge about their health benefits has been studied extensively with regard to health plan choice, and has been linked to their ability to access care, their perceptions about their access to care, and their overall health status.
Prior research provides insight into how educational efforts regarding Medicare benefits can help increase beneficiary knowledge and ensure beneficiary understanding. Studies of intervention effectiveness demonstrate positive though modest impacts on beneficiary knowledge and understanding. Researchers have suggested that educational materials could be more effective if targeted to specific groups. Identifying beneficiary factors other than socio-demographic factors along which individuals could be profiled should enhance the likelihood of successfully matching information to individuals as they are making choices regarding
Medicare.
Study Design: We conducted in-depth semistructured interviews to identify patterns in beneficiary cognition and behavior that were linked to observed differences in knowledge and understanding. An iterative group review process was utilized to develop thematic codes from the interview text. Team members agreed on thematic categories, positive and negative cases and example text segments for each code, resolving disagreements through extensive discussion, further exemplification and consensus.
Population Studied: Study participants were 21
Medicare beneficiaries from a large, southern urban area. The sample was purposively selected to be balanced by sex, race - white, African American, and Hispanic - and by Medicare managed care enrollment status.
Principal Findings: Six beneficiary-level factors emerged from our interview data: motivation, selfefficacy, experience, trust, information source, and information-seeking style. These factors were highly interrelated, and were found among both
FFS and MCA Medicare benefit program beneficiaries. The six factors did not appear to vary by race-ethnicity or gender.
Conclusion: These factors are consistent with previous research. However, our findings regarding the interrelationships between them illustrate the complexity of Medicare beneficiary knowledge acquisition, understanding, and utilization.
Implications for Policy, Delivery or Practice: We share the belief that targeted interventions to improve patient knowledge are needed. However, we posit that interventions based on broad demographic factors, such as race or age, might be minimally effective, primarily because beneficiaries in those demographic groups are very heterogeneous with respect to other factors, such as those in our study, that can influence beneficiary knowledge or understanding. We suggest that the six factors observed in our study could be explored as avenues for differentially targeting beneficiaries for tailored information.
Funding Source(s): NIA
Theme: Consumer Choices in Health Care
● Mobility of North Carolina Physician
Assistants & Nurse Practitioners Between
Primary Care & Non-primary Care Specialties: A
Longitudinal Study
Perri Morgan, Ph.D.; Erin Fraher, M.P.P.
Presented by: Perri Morgan, Ph.D., Director of
Physician Assistant Research, Community & Family
Medicine, Duke University Medical Center, DUMC
3848, Durham, NC 27710, Phone: (919) 681-6791;
Email: perri.morgan@duke.edu
Research Objective: Whether the U.S. faces a physician shortage is actively debated, but there is general consensus that the nation faces an inadequate supply of primary care physicians.
Primary care workforce shortages could be met by the deployment of providers who can change specialty to fill emerging workforce needs. In the
U.S., physicians are generally bound to one specialty for the course of their careers. Nurse practitioners (NPs) are educated and certified within broad categories of practice, such as pediatrics, geriatrics, internal medicine, and family medicine and physician assistants (PAs) obtain generalist training and certification. Although PAs and NPs are able to change practice specialties, little is known about how often they actually make these shifts.
This study examines trends in specialty distribution and transitions between primary care and nonprimary care specialties among PAs and NPs in
North Carolina.
Study Design: Data were collected by the North
Carolina Board of Medical Examiners as part of the annual license renewal process. We coded practice specialty as primary care or not primary care and analyzed whether individuals changed between these two categories, as well as the direction of these specialty changes. Additional
analyses examined the association of specialty change with gender, age, years since completion of basic professional education, and location in rural versus urban counties.
Population Studied: PAs and NPs licensed in
North between 1996 to 2007.
Principal Findings: In 2007, there were 3,054 PAs and 2,933 NPs in active practice in North Carolina, compared to 18,913 physicians. In 1996, 48% of
PAs and 50% of NPs reported a primary care specialty. By 2007, the proportion of PAs in primary care had decreased to 41% while the proportion of
NPs in primary care had increased to 59%. Moving between primary care and non-primary care specialties was more common for PAs (24%) than
NPs (13%). Among PAs and NPs who made these transitions, PAs were more likely to change from primary care to specialty practice (58% of PAs who changed) while NPs were more likely to change from specialty to primary care (53% of NPs who changed).
Conclusion: This study shows that NPs are more likely than PAs to be found in primary care specialties and that they are more likely to move into primary care. PAs move into and out of primary care specialties more often than NPs, but they are increasingly specializing.
Implications for Policy, Delivery or Practice: If primary care physician shortages materialize, PAs and NPs may help fill this gap. While physicians require extensive training for each specialty, PAs and NPs, who are able to move among specialties with informal or on-the-job training, may serve as a flexible workforce able to meet emerging needs.
More research is needed to elucidate the factors that influence specialty changes and to test policy incentives that could be designed to increase the movement of PAs and NPs into primary care.
Funding Source(s): Duke University Physician
Assistant Division, North Carolina Area Health
Education Centers (NC AHEC) Program
Theme: Health Care Workforce
● How are PAs & NPs Included in State
Workforce Studies?
Perri Morgan, Ph.D.; Justine Strand, Dr.P.H.
Presented by: Perri Morgan, Ph.D., Director of
Physician Assistant Research, Community & Family
Medicine, Duke University Medical Center, DUMC
3848, Durham, NC 27710, Phone: (919) 681-6791;
Email: perri.morgan@duke.edu
Research Objective: Over the past decade, United
States workforce analysts have suggested that physician workforce planning should account for the impact of physician assistants (PAs) and nurse practitioners (NPs). Among recent national physician workforce studies, the majority have addressed the effect that the growing role of PAs and NPs may have on future physician requirements. Other national studies have not included PAs and NPs, arguing that their impact will be negligible, while still others assert that flaws in data and methodology prohibit their inclusion. Due to predicted physician shortages, many states have recently conducted state-level physician workforce assessments designed to promote or inform state programs to ensure an adequate medical workforce. This study examines the range of recent treatment of PAs and NPs in state workforce analyses, with the purpose of elucidating best practices.
Study Design: Twenty-seven state level workforce studies were identified from a listing compiled by the Association of American Medical Colleges, a
Medline search, and an internet search. We identified the specific purpose of each study, the entity that conducted the study, and whether PAs and NPs served on the study panel. Studies were also reviewed to determine whether the impact of
PAs/NPs was discussed, whether PAs/NPs were enumerated, what data sources were used, how
PAs/NPs were figured into provider supply estimates, and whether recommendations included
PAs and NPs.
Population Studied: State medical workforce studies in the U.S.
Principal Findings: Most state studies defined their purpose as assessing the adequacy of the physician workforce. The majority of study advisory panels did not include PA or NP representatives, and half did not mention PAs or NPs. One third included PAs and NPs in provider supply estimates, and all of these used state licensing data. These states found physician:PA/NP ratios in the range of
3:1. Only a few state reports included recommendations pertaining to PAs and NPs, and these tended to focus on the need for further research. One state (North Carolina) differed significantly from the others by consistent inclusion of PAs and NPs in its defined purpose, on its advisory panel, in provider supply projections, and in recommendations. Since there is not consensus on how to convert PA and NP labor into physicianequivalents, this state used both a high and a low estimate for the physician substitution ratio.
Conclusion: Despite wide agreement that physician shortages are pending and clear evidence that the roles of NPs and PAs are expanding, most state medical workforce studies did not assess the impact of PAs and NPs.
Although some states cited data and methodology concerns as their reason for not including PAs and
NPs, other states were able to successfully use state licensing data and others plan to collect more detailed data in the future through the state licensing process.
Implications for Policy, Delivery or Practice:
Findings from this study may be useful to states as they continue to develop methodology for assessing state medical workforce adequacy
Funding Source(s): Duke University Physician
Assistant Division
Theme: Health Care Workforce
● Variation in Prescription Drug Coverage, Use
& Costs Across Canada: Evidence from the
Canadian Rx Atlas
Steve Morgan, Ph.D.; Colette Raymond, Pharm.D.;
Dawn Mooney, B.A.; Daniel Martin
Presented by: Steve Morgan, Ph.D., Associate
Professor, Centre for Health Services & Policy
Research, University of British Columbia, #201 -
2206 East Mall (LPC), Vancouver, BC, V6T 1Z3,
CA, Phone: (604) 822-7012; Email: morgan@chspr.ubc.ca
Research Objective: Medicines are one of the most important components of modern health care, yet Canadian drug pharmaceutical policy varies considerably in terms of which populations and products are covered under public drug benefit program. As a consequence, residents of some
Canadian provinces spend 50% more on prescription drugs than residents of other provinces.
We used unique market research datasets to estimate the magnitude and causes of such variation in spending on this important component of health care.
Study Design: We analyzed variations and trends in spending on prescription drug as well as differences in public drug coverage and formulary listings. We combined year-2007 data from two datasets of IMS Health, Canada Inc to produce province-specific estimates of the number, size, and cost of prescriptions filled by persons in four age groups (0-19, 20-44, 45-64, and 65+ years of age).
We analyzed the IMS data with economic measurement indexes that separate the effect of population age on per capita spending from the impact of six age-adjusted factors that fall into three broad categories of pharmaceutical cost-drivers:
Volume Effects, Therapeutic Choices, and Price
Effects. We analyzed public coverage (formulary listings) and the impacts of population age and ageadjusted cost-drivers on spending for 20 therapeutic categories and 10 Canadian provinces.
Population Studied: Our study is an analysis of population-level data pertaining to the health system and pharmaceutical utilization patters of persons residing within each of Canada’s 10 provinces.
Principal Findings: In 2007, Canadians spent
$578 per capita on retail purchases of prescription drugs, approximately $19 billion in total. On average, spending on prescriptions for Canadians age 65 and older was more than twice that of
Canadians aged 45–64 and over six times that of
Canadians aged 20–44.
After figures were adjusted for differences in population age, spending per capita varied by over
55% across provinces, from $418 in British
Columbia to $655 in Quebec.
Differences in the extent of public drug coverage or number of drugs covered by provincial drug plans do not appear to explain interprovincial variations in prescription drug spending. Furthermore, population characteristics such as socioeconomics, health status, and health system structure do not point to clear explanations of interprovincial variations in age-standardized spending.
Conclusion: Even after adjusting for population age, the use and cost of medicines varies considerably across Canadian provinces. These findings underscore the importance of efforts to build better data systems for monitoring pharmaceutical utilization, expenditures, and health outcomes.
Implications for Policy, Delivery or Practice:
Analyses into the clinical and economic consequences of major variations in pharmaceutical practice patterns across Canada are needed. In
July 2008, Health Canada announced plans to establish a Drug Safety and Effectiveness Network that would link centres of excellence in the evaluation of pharmaceutical use and health outcomes. The goal of this network is to increase knowledge about the safety and effectiveness of drugs based on their use in the "real world."
Funding Source(s): Canadian Institutes of Health
Research - CIHR
Theme: Quality and Efficiency: Policies and
Incentives
● Registration for Public Drug Benefits Across
Areas of Differing Ethnic Composition in British
Columbia, Canada
Steve Morgan, Ph.D.; Vivian, Leong, M.S.; Sabrina
Wong, Ph.D.; Gillian Hanley, M.A.
Presented by: Steve Morgan, Ph.D., Associate
Professor, Centre for Health Services & Policy
Research, University of British Columbia, #201 -
2206 East Mall (LPC), Vancouver, BC, V6T 1Z3,
CA, Phone: (604) 822-7012; Email: morgan@chspr.ubc.ca
Research Objective: In 2003, the government of
British Columbia introduced a universal drug benefit plan to cover drug costs that are high relative to household income. Residents were required to register in order to be eligible for the income-based benefits. Recognizing that the registration process may pose a barrier for certain populations, the government made concerted efforts to promote the program and facilitate registration among major ethnic groups. We aimed to determine whether registration rates varied across small geographic areas of differing ethnic composition and whether government communications encouraged registration among targeted populations.
Study Design: Using population-based administrative datasets, we conducted multivariate logistic regression analyses to determine whether the probability of registration for the public drug plan
varied across areas of differing ethnic composition.
The dependent variable was whether or not any member of a household had registered for the new drug plan at any time during 2003. Independent variables were selected based on a modified version of the Andersen-Newman framework that includes “contextual” variables that describe the ethnic concentration of areas in which members of our study population live.
Population Studied: Our study cohort included all
British Columbia households who were eligible for coverage under the new public drug benefit program in 2003. The unit of analysis is the household because the program (and registration for it) operates at the household level. We excluded households with persons aged 65 years and older because these households received public drug benefits as an age-entitlement prior to 2003 and were individually mailed registration packages for the new drug plan.
Principal Findings: The adjusted odds of registration did not differ across regions characterized by high concentrations (greater than
30%) of residents identifying as North American,
British, French or other European. Households located in areas with concentrations of ethnicities targeted in government communications had the highest adjusted odds of program registration:
Chinese (OR=1.21) and South Asian (OR=1.19).
Despite this positive finding, households residing in areas with relatively high concentrations of recent immigrants had lower adjusted odds of registering for the program (OR=0.97).
Conclusion: Our study identified ethnic variation in registration for a new public drug benefit program in
British Columbia. We found the ethnic populations that government had targeted communications toward were among those with the highest likelihood of program registration.
Implications for Policy, Delivery or Practice: Our findings suggest that government interventions can encourage program participation among ethnic minority groups. More needs to be done, however, to mitigate access barriers among relatively recent immigrants.
Funding Source(s): Canadian Institutes of Health
Research - CIHR
Theme: Disparities
● Regional Variation in Prescription Drug Use &
Cost in British Columbia, Canada
Steve Morgan, Ph.D.; Natasha de Sousa, M.A.;
Dawn Mooney, B.A.; Colette Raymond, Pharm.D.;
Daniel Martin
Presented by: Steve Morgan, Ph.D., Associate
Professor, Centre for Health Services & Policy
Research, University of British Columbia, #201 -
2206 East Mall (LPC), Vancouver, BC, V6T 1Z3,
CA, Phone: (604) 822-7012; Email: morgan@chspr.ubc.ca
Research Objective: Medicines are one of the most important components of modern health care; yet, relatively little is known about small-area variations in their use and cost. We used population-based linked administrative datasets to describe the age-, sex- and needs-standardized variations in prescription drug use and cost among small areas of British Columbia, Canada.
Study Design: Statistical models were constructed to estimate age-, sex- and needs-standardized average (private and public) spending per capita on prescription; age-, sex- and needs-standardized likelihood of use; and the age-, sex- and needsstandardized number of days of therapy received and cost per day of therapy received conditional on use. Fixed-effects models were run to test the significance of differences between predicted and observed values for each of 84 Local Health Areas of the province. Variation in the use and cost of medicines were computed for 20 therapeutic categories of prescription drug.
Population Studied: Year-2006 administrative data on prescription drug use, demographics, and health status were obtained for virtually every resident of BC. To ensure complete data capture, we excluded residents who lived outside of the province for more than 90 days (e.g., students who study outside of province or retirees who spend winter in warmer climates). This left us with a sample of 4.1 million individuals (approximately
96% of the provincial population).
Principal Findings: We found significant age-, sex- and needs-standardized variations in prescription drug use and cost among Local Health Areas of
BC. As might have been expected, variations in overall spending were smaller than variations by therapeutic category. In 10 of the categories studied, observed spending in at least 20% of the regions was at least 30% off of what would be expected given population age, sex and health status. Most of the variation in spending stemmed from variations in the likelihood of medicine use.
We found relatively little regional variation in the number of days of therapy prescribed to medicine users or the average cost per day of prescribed therapy.
Conclusion: Variation of prescription drug use and costs across geographic areas of BC are significant even after controlling for population demographics and health status. The local health system determinants and effects of this variation should be further investigated.
Implications for Policy, Delivery or Practice: We are working directly with policy makers to assist with the planning of initiatives that might mitigate the most concerning variations in categories: such as opioid analgesics, psycho-stimulants, antidepressants, and bisphosphonates.
Funding Source(s): Canadian Institutes of Health
Research - CIHR
Theme: Disparities
● International Perspectives on Current Issues in Pharmaceutical Licensing, Assessment,
Pricing & Reimbursement
Steve Morgan, Ph.D.; Colleen Cunningham, M.A.;
Gillian Hanley, M.A.
Presented by: Steve Morgan, Ph.D., Associate
Professor, Centre for Health Services & Policy
Research, University of British Columbia, #201 -
2206 East Mall (LPC), Vancouver, BC, V6T 1Z3,
CA, Phone: (604) 822-7012; Email: morgan@chspr.ubc.ca
Research Objective: Brand-name drugs with current sales of over USD$120-billion will face generic competition in the next 5 years due to expiring patents. Manufacturers are retooling corporate structures, R&D activities and sales forces to adjust. Governments must also be prepared for the opportunities and challenges of this changing pharmaceutical marketplace. The purpose of this study was to understand the challenges and priorities of pharmaceutical policy decision makers from a selection of OECD countries.
Study Design: This study employed participant observation and qualitative analysis of telephone interviews and focus group discussions to identify and classify policy challenges and priorities that were common and important across countries. Over a two year period, the principal investigator attended and participated in three invitational meetings for competent authorities in pharmaceutical licensing, assessment, pricing and reimbursement. In conjunction with the third meeting, focus group discussion sessions on specific pharmaceutical policy-related topics were arranged and digitally (audio) recorded. Over the same period, telephone interviews were also conducted with selected decision makers from five countries: the US, the UK, New Zealand, Canada and Australia. Focus group sessions and interviews were semi-structured around discussing major policy challenges and factors enabling successful assessment, pricing and reimbursement. All recordings and transcripts were obtained under informed consent and de-identified to protect the identity of speakers. Key themes were identified from transcripts using software-facilitated qualitative data analysis (NVivo version 8). We used a thematic analysis akin to simple grounded theory: coding for the themes and issues was generated through participation in the networking meetings and upon review of the transcripts; data were grouped and categorized by researchers; and key themes were extracted for public reporting.
Population Studied: Decision makers from a variety of countries participated in this study:
Australia, Austria, Canada, England, Hungary, Italy,
Mexico, the Netherlands, New Zealand, Norway,
Scotland, and the USA.
Principal Findings: A dominant challenge and priority reported by multiple pharmaceutical policy decisions makers was the need to identify tools and mechanisms to contain rising pharmaceutical costs and expenditures while ensuring equitable access to needed treatment. Decision makers also reported difficulty identifying the value for a drug and therefore an appropriate price, and the lack of an acceptable definition for “innovation” and “value” in the pharmaceutical context. Secrecy - - of scientific and pricing information - - was further identified as an increasing challenge to licensing, assessment, pricing and reimbursement policy. Finally, policy makers expressed concerns about the use of trial designs and outcome measures that generate data of limited or ambiguous value.
Conclusion: Pharmaceutical policy makers around the world face many common policy issues and challenges. Principal among these are the challenges of collecting adequate scientific information to inform decisions, and developing acceptable means of defining fair and reasonable prices for pharmaceutical products.
Implications for Policy, Delivery or Practice: The common policy challenges associated with defining and maximizing the value obtained from pharmaceuticals present many areas for potential international collaboration among researchers and policy makers.
Theme: Global Health
● Medicare Beneficiaries with Multiple
Sclerosis: Utilization of Rehabilitation Services
& Functional Impairment Levels
Melissa Morley, Ph.D.; Barbar, Gage, Ph.D.;
Melissa Morley, Ph.D.; Laura Coots, M.S.
Presented by: Melissa Morley, Ph.D., RTI
International, 1440 Main Street, Suite 310,
Waltham, MA 02451, Phone: (781) 434-1773;
Email: mmorley@rti.org
Research Objective: This study examines the utilization and costs of rehabilitation services for chronically ill Medicare beneficiaries with multiple sclerosis. Using Medicare administrative data, we profile the annual cost and utilization patterns of
Medicare beneficiaries with multiple sclerosis, including their use of rehabilitation services and other types of services during 2007. Outcome measures, such as changes in function are also examined for the subpopulation of beneficiaries with multiple sclerosis who are admitted to an inpatient rehabilitation facility during the year.
Study Design: This study uses national claims data for Medicare beneficiaries with at least one multiple sclerosis diagnosis. Beneficiaries are identified as a multiple sclerosis patient by examining inpatient and ambulatory Medicare claims data. Beneficiaries with multiple sclerosis are compared to a sample of beneficiaries without multiple sclerosis. Key measures include Medicare
payments, out-of-pocket payments, lengths of stay, and functional impairment.
Population Studied: Medicare beneficiaries with at least one multiple sclerosis diagnosis in 2007.
Principal Findings: Mean days of services and payments are presented for beneficiaries with a diagnosis for multiple sclerosis. Payment measures include Medicare payments and out-of-pocket costs for these chronically ill populations. Differences in annual utilization and spending are stratified by individual case mix measures including age, comorbidities, and urban versus rural differences.
In-depth analysis of beneficiaries admitted to
Inpatient Rehabilitation Facilities compares those with Multiple Sclerosis to other admissions with neuromuscular conditions. The findings demonstrate the increased utilization of services associated with multiple sclerosis. In 2004, nearly
4,000 inpatient rehabilitation admissions had at least one multiple sclerosis diagnosis and one percent of these admissions were for stays incurring outlier payments. The inpatient rehabilitation facility patient assessment instrument data is also used to estimate differences in functional levels, medical complexity, and other factors affecting outcomes between those with and without multiple sclerosis who are receiving inpatient rehabilitation services.
Conclusion: Annual costs for this chronicaly ill population are substantial. Average use varies by the number and type of comorbidities. This analysis documents the types of services most commonly used by beneficiaries with this chronic condition, and demonstrate the types of physical and cognitive impairments complicating their rehabilitation. Use of rehabilitation services and other medical care is related to progression of functional and cognitive impairments.
Implications for Policy, Delivery or Practice:
This work will have important implications for the field of rehabilitation medicine in illustrating the role of rehabilitation services for this chronically ill population. It is important to critically assess the value of these services and provide accurate information on their costs and benefits to people with multiple sclerosis, clinicians, and policy makers.
Funding Source(s): National Multiple Sclerosis
Society
● Race & Patient Perspectives on
Chemotherapy for Colorectal Cancer
Arden Morris, M.D., M.P.H.; Gwen Alexander,
Ph.D.; Margaret Ann Murphy, R.N.; Paula
Thompson, B.S.; Maria Silveira, M.D., M.S.E.; John
Birkmeyer, M.D.
Presented by: Arden Morris, M.D., M.P.H.,
Assistant Professor of Surgery, Surgery, University of Michigan, 211 N 4th Avenue, Suite 2A, Ann
Arbor, MI 48104, Phone: (734) 647-9710; Email: ammsurg@umich.edu
Research Objective: Previous data reveals that black Americans are 20% less likely to survive colorectal cancer after surgery than whites. From the healthcare side, this is most influenced by decreased use of chemotherapy among black patients. However, black and white Medicare patients are equally likely to consult an oncologist.
This study aimed to explain the racial discrepancy in use of chemotherapy after an oncology visit, by eliciting patient perceptions of colorectal cancer care and chemotherapy.
Study Design: Race- and gender-concordant focus groups were conducted in the Detroit area. Data were analyzed using grounded theory techniques in an iterative fashion. Codes were developed and revised to provide a coding manual. Themes were identified and refined until saturation was achieved.
Focus group themes were validated in two ways: conduct of individual in-depth interviews and computer analyses. Individual in-depth interviews were coded in a manner identical to those of the focus groups, using the same data dictionary, with revision and recoding until final consensus was reached. Computer analyses were complementary, including content analyses in ethnography, domain analysis, and perceptual mapping.
Population Studied: Study participants were recruited based on diagnosis of stage II or III colorectal cancer, previous surgical resection for cure, English language fluency, ability to hear, and previous conversation about chemotherapy with a health care provider. Subjects who declined were offered an individual in-depth interview for the same remuneration.
Principal Findings: Five focus groups (n=24) and
25 individual interviews were conducted. Resultant themes centered around three main topics:
(1)Preconceptions and emotionality about colorectal cancer, (2) Patient decision making, (3)and the physician:patient relationship. Patients who had a strong family history of colorectal cancer did not necessarily have accurate preconceptions but were less fearful and were more likely to use chemotherapy. Patients with limited or no previous cancer experience were less so. Patients with high emotionality (grief and anger) about their cancer diagnosis reported less trust in their oncologist and were also more likely to decline chemotherapy.
Positive and negative emotional factors (religiosity, apprehension) were a major influence on decision making, in turn a major influence on chemotherapy use. Instrumental factors (rent, time, transportation), were cited as barriers but never as sole reasons for non-use.
The physician:patient relationship was most important for white patients. In particular, white women were more likely to yield control and decision making to their physicians. Black women were more likely to cite trust in spirituality and prayer. Black men cited greatest trust in the sensation of physical health. For example, several
men stated that they felt well post-operatively and did not want to become ill again by using chemotherapy. Black and white men were most likely than women to rationalize their choices, sometimes based on misconceptions about risk.
Conclusion: Race and gender weigh heavily on how and whether patients make decisions about chemotherapy. White female patients were more likely to trust their physicians' recommendations without question, while black female patients reported relying on spirituality and personal relationships for their decisions. Both white and black men reported rationally assessing risks based on medical information, but the informational basis reported was not always correct.
Implications for Policy, Delivery or Practice:
Physicians may be more effective in communicating risks and benefits of chemotherapy to patients of any race or gender by (1) providing support opportunities for highly emotional or apprehensive patients, (2) exploring how or what patients trust and incorporating into the conversation, (3) supplying written information to serve as a memory aid regarding risks and benefits. Based on substantial differences in trust in health care, decision aids may be more useful for black patients than they are for whites.
Funding Source(s): Jeffrey A. Colby Foundation grant
Theme: Disparities
● Evidence of Improvement of the Nursing
Shortage in a Longitudinal Study of North
Dakota
Patricia Moulton, Ph.D.; Terri Lang, B.A.
Presented by: Patricia Moulton, Ph.D., Assistant
Professor, Center for Rural Health, University of
North Dakota, 1201 11th Avenue SW, Minot, ND
58701, Phone: (701) 858-6770; Email: pmoulton@medicine.nodak.edu
Research Objective: The ND Nursing Needs
Study, a longitudinal supply and demand study, is currently in its seventh year of data collection.
Research objective of this study include: 1) To determine whether ND is projected to have a shortage by 2020 of RNs. 2) Determine RN nursing supply trends. 3) Determine RN nursing demand trends.
Study Design: Supply and demand projections
(2007-2020) were determined using the Nursing
Supply Model and Nursing Demand Model both of which were developed for HRSA. The HRSA projections were modified using North Dakota specific information for the state nursing population, education upgrade and graduation totals. Primary data collection included five years (2002-2007) of
RN and health care facility surveys (hospitals, clinics, home health, long-term care and public health) The ND Board of Nursing licensure database was also used to determine county level nurse ratios.
Population Studied: The study examined several aspects of supply and demand for ND RNs.
Principal Findings: Projections indicate a small statewide surplus of RNs by 2020 (317 FTEs). The
HRSA model likely underestimates demand in states with large rural populations and with large elderly populations such as North Dakota. To reflect this concern, a 10% low to high demand range resulted in a surplus of 1,058 FTEs to a shortage of
424 FTEs (Skillman, Andrill, Hart, 2007). The number of licensed RNs per population increased from 12.90 RNs to 15.90 RNs per 1,000 people is greater than the national average of 8.48 per 1,000.
Maldistribution of RN supply remains a concern with
17 of 53 counties having less than the national average of RNs per 1,000 people. Enrollment in all
RN nursing education programs has resulted in an increase from 337 in 2000 to 491 annual graduates in 2007. The average age of RNs has decreased to
44 years; below the national average of 47 years.
After a two year climb, the vacancy rates have decreased over the last three years to 3% in 2007.
The percentage of RNs working full-time (36-40 hours/week) has increased from 11% in 2004 to
43% in 2007. The percentage of RNs with a second job has also declined from 33% in 2003 to 16% in
2007. The statewide turnover rate for RNs has increased to 19 percent.
Conclusion: There is evidence of a lessening of the RN shortage in North Dakota in the future according to supply and demand projections. This finding is supported by increases in enrollment and graduation of new nurses, decreasing vacancy rates, declines in multiple job RNs paired with increased full-time RNs.
Implications for Policy, Delivery or Practice:
Although findings provide an optimistic picture of
North Dakota RN supply and demand, the distribution of an inadequate number of nurses across rural areas of the state remains a concern.
Continued support of North Dakota’s nursing education programs in particularly support for the recent expansion of class size will play an important role in ensuring an adequate supply of nurses in the future. Increased turnover rates of nurses in health care facilities indicate a need to improve the work environment and maximize retention of nurses.
Funding Source(s): North Dakota Board of
Nursing
Theme: Health Care Workforce
● Disparities in Preventive Health Care:
Variations Across States
Ernest Moy, M.D., M.P.H.; Karen Ho, M.H.S.
Presented by: Ernest Moy, M.D., M.P.H., Medical
Officer, Center for Quality Improvement & Patient
Safety, Agency for Healthcare Research & Quality,
540 Gaither Road, Rockville, MD 20814, Phone:
(301) 427-1329; Email: ernest.moy@ahrq.hhs.gov
Research Objective: To assess 1) the feasibility of examining disparities in adult preventive health care across States and 2) variations in the magnitude of disparities related to education, race, and ethnicity across States.
Study Design: Data come from the 2006
Behavioral Risk Factor Surveillance System. Five adult preventive care services were examined: 1)
Mammogram among women age 40+; 2) Pap tests among women age 18-65; 3) Colorectal cancer screening among adults age 50+; 4) Flu shot among adults age 65+; and 5) Pneumococcal shot among adults age 65+. For each measure, differences in service receipt were examined between 1) Individuals with less than a high school education and individuals who graduated from college or technical school; 2) Non-Hispanic blacks and non-Hispanic whites; and 3) Hispanics and non-Hispanic whites. Differences were examined for all States in aggregate and for each individual
State (50 States, DC, Puerto Rico, and Virgin
Islands). Best (top 10%) and worst (bottom 10%) performing States were identified for each disparity.
Population Studied: Noninstitutionalized adults age 18 and over in the U.S.
Principal Findings: Educational disparity is widespread, ranging from a gap of 8.1% for pneumococcal shots to 16.4% for Pap tests. Racial and ethnic disparities are small for Pap tests and mammograms, but significant for colorectal cancer screening and immunizations. It is feasible to examine disparities in adult preventive health care across States. Sample sizes are sufficient to examine educational disparities across all five measures in all 53 States. Black-white and
Hispanic-white differences in the cancer screening services could be assessed for ~40 States. Blackwhite differences in immunizations could be assessed for ~30 States while Hispanic-white differences could be assessed for ~20. Wide variations in disparities across States are observed.
For example, educational disparity gaps in colorectal cancer screening ranged from under
9.3% in the Best States to over 20.8% in the Worst
States. Racial disparity in colorectal cancer screening ranged from under -0.1% (favoring blacks) in the Best States to over 10.9% in the
Worst States while ethnic disparity ranged from under 7.2% in the Best States to over 22.1% in the
Worst States.
Conclusion: It is possible to examine disparities in adult preventive health care at the State level. This may be important because wide variation in the magnitude of disparities is observed across States.
Implications for Policy, Delivery or Practice:
Individual States differ in the specific preventive services and populations most affected by disparities. Examining variations in disparities across States may permit better targeting of local disparities resources and establish practical benchmarks.
● How Interpreters Facilitate Communication
Between Limited
Mary-Rose Mueller, Ph.D., R.N.; Nadia Salas,
M.P.H.; Veronica VillReal, B.S.; Stergios Roussos,
Ph.D.; Linda Hill, M.D.,M.P.H.; Melbourne Hovell,
Ph.D.
Presented by: Mary-Rose Mueller, Ph.D., R.N.,
Associate Professor, Hahn School of Nursing &
Health Science, University of San Diego, 5998
Alcala Park, San Diego, CA 92110, Phone: (619)
260-4562; Email: mmueller@sandiego.edu
Research Objective: Shared decision making is characterized by the active engagement of both health care providers and patients in the exchange of information, consideration of available options, and expression of preference for and commitment to a clinical course of action. Although scholars acknowledge that other parties may be integral to shared clinical decision making, to date there has been little research on the role and or involvement of third parties in the process. Here we present findings from the first phase of a study which seeks to understand and possibly improve shared decision making on prostate cancer screening when third party interpreters are needed to facilitate communication between providers and patients from different language and cultural traditions. An objective of this phase of the research was to identify and describe, from the perspective of monolingual health care providers, limited English speaking Latino men, and language interpreters, how communication unfolds and decisions are made during ‘interpreted’ clinical encounters.
Study Design: An exploratory, descriptive, qualitative design was utilized. Data were collected from a large urban health center in a Southwestern region of the country. Data sources were focus group interviews and demographic survey responses. Accepted methods of qualitative data analysis, including coding, memo-writing, and interrate agreement, were employed.
Population Studied: The four groups of participants were: monolingual health care providers, consisting of physicians and nurse practitioners; medical assistants who were employed as Spanish language interpreters and clinic staff workers; and English and Spanish speaking Latino men ages 50-70 who had been screened for prostate cancer with a prostate specific antigen (PSA) test.
Principal Findings: Interpreters draw upon three approaches to facilitate communication between health care providers and patients from different language traditions: 1) direct translation, in which an interactional event is transmitted in a word for word, verbatim fashion; 2) approximate translation, in which the exact language used in an interactional
event is altered prior to transmission; and 3) adaptive translation, in which the both the direct and approximate approaches are used in transmitting an interactional event. The conditions under which interpreters deploy these approaches varies, and are contingent on four factors: 1) institutional policies, in that some settings favor and or encourage one approach over the others; 2) provider preferences, in that health care providers communicate to interpreters the approach they favor; 3) interpreter preferences, in that interpreters favor one approach over the others; and 4) interpreter discretion, in that interpreters select an approach they view is warranted by the interactional situation.
Conclusion: Interpreted encounters are complex interactional events. Interpreters play a critical role in facilitating communication and decision-making between providers and patients from different language traditions.
Implications for Policy, Delivery or Practice:
Knowledge of the approaches that facilitate communication during interpreted encounters may be useful to the design of shared decision-making interventions for providers and patients from different language traditions. Health service researchers may wish to involve interpreters in the development, planning and implementation of such programs.
Funding Source(s): RWJF
Theme: Health Care Workforce
● Nurses, Congressional Hearings & Workforce
Issues: 1995-2007
Ulrike Muench, R.N., M.S.N.; Moira O'Neill, M.S.N,
M.P.H.; Sally Cohen, Ph.D., M.S.N.; Emily Sheren,
M.S.N.
Presented by: Ulrike Muench, R.N., M.S.N., Ph.D.
Student, Health Policy & Health Care Systems,
Yale University, 100 Church Street South, New
Haven, CT 06511, Email: ulrike.muench@yale.edu
Research Objective: The purpose of this study was to analyze trends in nurses testifying on workforce issues before the U.S. Congress, 1995-
2007. It was the first study to document the political participation of nurses in congressional hearings.
Our research questions were: 1) On what issues did nurses testify before Congress between 1995 and 2007? 2) What trends are discernible over time in nursing’s political involvement at congressional hearings in terms of topics they addressed, committees, and chambers that sponsored the hearings, and characteristics of the nurses, themselves? 3) What are the implications of these findings for nursing’s ability to contribute to health care reform deliberations, including workforce issues?
Study Design: This descriptive study used the
Lexis-Nexis Congressional online data-base to identify nurses who testified before Congress between January 1995 and 2007. We collected data on testimony topics, hearing dates, committees, and characteristics of the nurse witnesses, such as position and education. Data were coded and entered on an Excel Spreadsheet.
Descriptive statistics were used to analyze nurses’ roles as witnesses on workforce issues before congressional committees.
Population Studied: Nurses (N=219) who testified before congressional committees, 1995-2007.
Principal Findings: Workforce issues figured prominently (58%) as a topic that nurses testified on before Congress. Specifically, 26% of nurses testifying focused on military workforce issues, including recruitment strategies and funding for military nursing services. Another 25% of nurses testifying addressed workforce solutions (nonmilitary), including funding for nursing research and education. House and Senate Appropriations
Committees were the panels before which most of the nurses testified (46%), with an additional 18% appearing before congressional committees authorizing major public health programs, and 9% testifying before committees on Veterans Affairs.
The largest proportion (45%) of nurses represented national nursing organizations, the majority of which were nursing specialty associations, such as those representing nurse anesthetists. Other nurses represented a range of roles, organizations, and settings.
Conclusion: Considering the prominence of the nursing shortage on congressional and public agendas, the participation of nurses as witnesses on workforce issues at congressional hearings was lower than what one might have expected.
Testimony on military workforce issues was surprisingly frequent. Funding for nursing education and research remains a high priority for the profession, often at the expense of other compelling workforce issues.
Implications for Policy, Delivery or Practice:
Nurses have great potential to influence health care reform, including workforce issues, through congressional hearings. Doing so could enhance congressional support of innovations in health care delivery and develop novel ways to recruit and retain a well-qualified nursing workforce. By analyzing testimony data, policymakers, health services researchers, and practitioners can gain insights regarding the relationship between nursing workforce issues and policy development.
Theme: Health Care Workforce
● Development & Use of an Electronic Chart
Abstraction Tool: Strengths, Weaknesses, &
Lessons Learned
Hillary Mull, M.P.P.
Presented by: Hillary Mull, M.P.P., Health Policy &
Management, Boston University School of Public
Health, Boston, MA , Email: hjmull@bu.edu
Research Objective: Chart review is considered the “gold standard” for obtaining patients’ clinical data. However, the process poses numerous concerns with respect to efficiency, patient privacy, and expense. Adoption of electronic health records
(EHRs) has improved chart abstraction by enabling faster and more focused review of patients’ clinical information. Combined with EHRs, electronic chart abstraction tools (ECATs) may help to reduce the problems associated with manual chart review.
Study Design: ECATs were created for a VA
HSR&D-funded study validating 15 of the Agency for Healthcare Research and Quality (AHRQ)
Patient Safety Indicators (PSIs). Paper abstraction tools were designed by AHRQ and modified by team clinicians for use with VISTA, the VA’s EHR.
Abstraction tools were complex, with approximately
50 questions per tool, including fields for each question’s data source. Four research nurses performed medical record review. Although the research team considered using the AHRQabstraction tools and transcribing results into an
Access database, the team instead elected to build
ECATs for each PSI using Microsoft InfoPath.
ECATs were developed in order to protect patient data, facilitate more reliable abstraction, and provide accessibility for data analysis. Based on a team member’s experience with InfoPath, the team estimated development of an ECAT would not take longer than entering paper-based results into a customized Access database.
Principal Findings: The ECAT conferred several benefits over paper-based abstraction methods.
Nurse abstractors could simultaneously view the
VISTA screen and the InfoPath form, resulting in greater efficiency and accuracy. Conditional formatting hid all unnecessary fields, simplifying abstraction. Missing data were coded as ‘absent from record’ or ‘skipped field’ in the ECAT, reducing the burden of cleaning the data and improving the quality of the results. ECATs were accessible online such that nurses could work offsite; physicians were able to review problem records quickly and efficiently. Errors associated with dual data entry were eliminated. Despite these advantages, the development of the first ECAT was far more time-consuming than anticipated. Delays occurred for several reasons: the InfoPath software was difficult to learn; redundant programming and abstraction resulted from programmer errors with field types or conditional formats, or because of changes to the clinical content of the tool; and the programming required for ECAT data extraction was significantly more difficult than expected. To resolve these problems, the team developed a process of ECAT review. After development of three ECATS, process efficiencies were observed.
Conclusion: ECATs improved the accuracy and efficiency of data abstraction and reduced data security risks. The ECAT design reduced missing data issues, saving time for analysts. While the abstraction of patient records was faster, the process of building abstraction tools was more time consuming than anticipated. Nonetheless, nurse abstractors and data analysts expressed preference for the ECAT over the use of paper-based tools.
Implications for Policy, Delivery or Practice:
ECATs are suitable replacements for paper chart abstraction tools, and, coupled with an EHR, can enhance the quality of the data and the timeliness of results. The effort to develop and use these tools can be considerable and demands the full commitment of the research team to ensure a successful product.
Funding Source(s): VA
● Assessing the Massachusetts Health Reform for Racial & Ethnic Minorities in Mental Health
Care
Norah Mulvaney-Day, Ph.D.; Margarita Alegria,
Ph.D.; Anna Nillni, B.A.; Sabrina Gonzalez, B.A.
Presented by: Norah Mulvaney-Day, Ph.D.,
Research Associate, Center for Multicultural Mental
Health Research, Cambridge Health Alliance, 120
Beacon Street, Somerville, MA 02143, Phone:
(617) 503-8448; Email: nmulvaneyday@charesearch.org
Research Objective: Overall, health reform in
Massachusetts (MA) has led to drops in rates of uninsurance. However, little information is available on how these reforms are working for individuals with mental health disorders. The purpose of this study was to assess the experience of health insurance reform for individuals from racial and ethnic minorities who had received mental health services through the Free Care pool (FC) prior to the heath care reform. We were particularly interested in assessing the experience of individuals switching into Commonwealth Care
(CC), the new subsidized health insurance plan in
MA.
Study Design: This paper reports on the first two waves of data collection. A purposive sample of 66 patients previously receiving mental health care via the FC pool was interviewed by telephone between
August and December of 2007. The semistructured interview included open and close-ended questions in the following domains: health insurance coverage; knowledge about health reform; process of switching; attitudes towards health insurance; access and delivery of health and mental health care; and demographics. A follow-up interview (n=63) was conducted 6-months later.
Respondents who had switched out of FC were selected, and descriptive analyses were conducted using a case study approach whereby each case was reviewed, basic demographics and insurance pattern summarized, and open-ended data from five key domains was analyzed: support with insurance shifts, education about reform, administrative clarity in the process, medical and mental health needs, and attitudes/values regarding insurance.
Population Studied: The majority of the respondents were female (77.2%), immigrant
(84.7%), Latino (74.6%) and mostly Spanishspeaking (57.6%). The sample was low income with 78.9% making less than $10,000 in the previous 6 months. About half the sample had received a high school education (51.5%). The respondents in the sample had unstable work situations, with almost half (45.5%) experiencing a change in their employment status between baseline and follow-up.
Principal Findings: Half (50.7%) had switched out of FC and into another payment mechanism: CC
(n=16), private insurance (n=7), public insurance
(n=6) or self-pay (n=3). Out of this group, about one third reported both difficulties with the insurance transition and having to cut back on mental health care without wanting to do so.
Complications for individuals with mental health disorder included difficulty paying copayments for medications, difficulty affording mental health care for those with co-occurring physical health disorders, and gaps in coverage due to poor understanding of redetermination requirements.
Conclusion: Individuals with co-occurring physical and mental health disorders, those who speak
Spanish only, and those who experience significant life events such as job loss may be vulnerable to disruption in needed mental health care during the process of health reform. It will be important to assess whether these disruptions resolve themselves over time as the health reform continues to roll out.
Implications for Policy, Delivery or Practice:
Ensuring that the shift to universal coverage does not exacerbate health concerns for those with mental health disorder may require targeted policy interventions from health and mental health care professionals.
Funding Source(s): NIMH
Theme: Behavioral Health
● Moral Exemplars, Outpatient Medical Clinic
Climate, Temporal Affect & Patient Care Errors
Deborah Mullen, M.S., Ph.D.
Presented by: Deborah Mullen, M.S., Ph.D.,
Consultant, Health Policy & Management & Health
Informatics, University of Minnesota &
HealthPartners, 8170 33rd Avenue South, Mailstop:
21108Q, Bloomington, MN 55123, Phone: (952)
883-5753, Email: mull0284@umn.edu
Research Objective : Specific Aims 1. Create a safety climate instrument for use in the outpatient medical clinic setting. 2. Determine if there is any correlation between the safety climates characteristics of outpatient medical clinics, moral exemplarism traits of practitioners and their interaction (practitioner temporal affect) on patient care error rates.
Study Design : The study includes cognitive interviews for survey instrument development, pilot testing of the instrument and a random survey of nursing professionals in outpatient medical clinics.
Population Studied : Nursing professionals (NP,
RN, CNM, LVN, LPN) working within outpatient medical clinics in Minnesota
Principal Findings : The instrument is currently being fielded and results will be completed by the time of the conference. The hypotheses to be tested are:1. Clinics with higher levels of safety climates will have fewer patient care errors. 2.
Clinics with one or more moral exemplars will have fewer patient care errors. 3. The lower the safety climate level in a clinic, the larger the effect of the presence of a moral exemplar.
Conclusions : The cognitive interviews and pilot testing have been completed. The instrument is currently being fielded and results will be completed by the time of the conference.
Implications for Policy, Delivery or Practice :
Patient safety is a national priority and while much has been done to survey hospital staff about safety cultures, little is know about the outpatient clinic setting. This instrument blends not only an assessment of the error reporting systems and outcomes, but additionally psycho-social factors.
Using instruments such as this in clinics allows for assessment of strengths and weaknesses which can be utilized in creation of thoughtful programs to address issues.
● Calculating Cost Savings for Care
Management Programs
Shannon Murphy, M.A.; Martha Sylvia, M.S.N,
M.B.A.; Linda Dunbar, Ph.D.; Michael Griswold,
Ph.D.
Presented by: Shannon Murphy, M.A.,
Biostatistician, Care Management, Johns Hopkins
HealthCare LLC, 6704 Curtis Court, Glen Burnie,
MD 21060, Phone: (410) 762-5239; Email: smurphy@jhhc.com
Research Objective: There is little evidence to suggest that Care Management (CM) programs significantly reduce health care costs. Results from prior studies are questionable due to the use of prepost designs, no comparison groups, small sample sizes, inadequate observation lengths, and oversimplified statistical methods. Studies lacking in methodological rigor tend to report higher ROIs than studies with more rigorous designs. Our goal is to develop a consistent framework for producing sound CM costs savings estimates.
Study Design: We conducted a quasi-experimental longitudinal data analysis from monthly observations on over 20,000 study participants enrolled in a commercial health plan from 2002 through 2007. Our CM program began in mid-2005 and consisted of high-intensity (nurse case manager) and low-intensity (education only)
interventions. A pseudo-control comparison group was defined as health plan members eligible for CM who never enrolled. We compared cost trajectories before and after CM implementation for both the intervention and pseudo-control groups to evaluate
CM cost savings.
Population Studied: Our study population included adult commercial health plan members eligible for
CM programs due to their health condition(s). A total of 624 study members were in the highintensity CM program, 1,218 were in low-intensity, and 18,880 members who never enrolled in CM served as the comparison group.
Principal Findings: Our framework generates estimates that can account for temporal change, selection bias, and subject-specific regression to the mean effects and can additionally be updated regularly for evaluating the CM program.
Conclusion: When calculating CM program cost savings, we recommend creating a longitudinal data set with frequent (e.g., monthly) repeated observations over multiple years on a large number of participants. We also suggest using a randomized design or identifying an appropriate comparison group for comparing cost trajectories with the costs for CM participants. Additionally, savings estimates should be generated regularly
(e.g., quarterly) to monitor changes in the savings rate over longer implementation periods.
Implications for Policy, Delivery or Practice:
National, State, and Private payers who hold the risk for health care expenditures are hopeful that
CM programs will help to reduce unmanageable rising health care costs. It is imperative that we use a sound framework for evaluation of CM cost savings to provide a confident basis for investment in these programs.
Funding Source(s): Internal (Johns Hopkins
HealthCare)
Theme: Health Care Markets and Competition
● Demographic and Health Predictors of
Obesity among older American Indians and
Alaska Natives
Kyle Muus, Ph.D.; Leander McDonald, Ph.D.
Presented by: Kyle Muus, Ph.D., Research
Associate, Center for Rural Health, School of
Medicine & Health Sciences, 501 North Columbia
Road, Grand Forks, ND 58203, Phone: (701) 777-
4048; Email: klmuus@medicine.nodak.edu
Research Objective: Delineate demographic and health predictors of obesity among older American
Indians/Alaska Natives (AI/ANs).
Study Design: This study used data from a nationwide cross-sectional survey conducted in
2004-2005 involving 7,656 AI/ANs aged 55+. The survey was conducted by the UND National
Resource Center on Native American Aging and funded by the Administration on Aging.
Population Studied: 7,656 AI/ANs aged 55+ from communities throughout the U.S.
Principal Findings: AI/AN Elders with obesity tended to: be female; were younger; were married; possessed higher education; were employed; and possessed higher incomes. Regarding health factors, elder with obesity tended to: not smoke cigarettes; have difficulty walking; avoid recent drinking; avoid binge drinking; have moderate/high nutritional risk; have a functional limitation; have fair/poor health; and engaged in <2 vigorous exercises per month. Obesity increased the risk of having asthma, hypertension, diabetes, depression, congestive heart failure and arthritis.
Conclusion: Obesity is linked with poorer health among older AI/ANs. Having multiple comorbidities and certain conditions, namely hypertension, diabetes and arthritis, are problematic for obese
AI/AN elders.
Asthma, diabetes and CHF, although lower in prevalence, were strongly related to obesity especially in the presence of other chronic conditions. Lifestyle factors were closely associated with obesity among younger elders, whereas lifestyle and sociodemographics influenced obesity among the older cohort.
Implications for Policy, Delivery or Practice:
Future studies on obesity among AI/ANs are needed in the following areas: prevalence, longitudinal trends; U.S. regional patterns; barriers to physical activity and proper nutrition; psychosocial aspects; and social/cultural links.
● Impact of Transfer on Outcomes for Rural
Hospital Patients with Acute Myocardial
Infarction
Kyle Muus, Ph.D.; Alana Knudson, Ph.D.; Marilyn
Klug, Ph.D.; Joshua Wynne, M.D., M.B.A., M.P.H.
Presented by: Kyle Muus, Ph.D., Research
Associate, Center for Rural Health, University of
North Dakota School of Medicine & Health
Sciences, 501 North Columbia Road, Grand Forks,
ND 58203, Phone: (701) 777-4048; Email: klmuus@medicine.nodak.edu
Research Objective: Compare predictors of inhospital acute myocardial infarction (AMI) mortality between rural hospital transferred and nontransferred patients.
Study Design: Cross-sectional retrospective of
Medicare hospital inpatient data for years 2003-
2005. Patients with a principal diagnosis of AMI and admitted to a rural hospital were identified and assigned to wither of these cohorts: those who were transferred to a larger hospital; and those who were not transferred to a larger hospital.The prevalence of all variables for each cohort was described and compared using Chi-Square and independent t-test statistics. Relative risk for mortality was estimated for the comorbidities. All variables were entered into a logistic regression. Due to frequent interaction of
the related variables and transfer status, separate logistic regressions were conducted for each cohort and results were compared.
Population Studied: Medicare patients admitted to a rural hospital with a principal diagnosis of AMI for years 2003-2005 in OR, UT, ND, IA or ME.
Principal Findings: Transfer patients were younger, more likely male, had fewer comorbidities and were less likely to expire (6.6 vs. 18.3%). This increase in mortality of non-transfer patients was strongly associated with older age; after adjusting for age, the excess mortality rate was reduced to
33%. We used logistic regression to compare demographic, hospital and comorbidity predictors of
AMI mortality for the transfer and no-transfer cohorts. Shock (Odds Ratio [OR]=7.65), acute renal failure (OR=3.55) and length of stay <5 days
(OR=3.51) were the strongest mortality predictors for both cohorts.
Conclusion: Transfer was associated with a greater likelihood of in-hospital AMI survival, largely but not fully explained by transfer patients being younger with fewer comorbid conditions.
Implications for Policy, Delivery or Practice:
This study demonstrates the importance of transfer and indicates key potentially mutable factors that are important for improving outcomes of rural AMI patients. For rural health care providers, this study emphasizes the importance of considering patient demographics in relation to the nature, severity and mix of comorbidities when deciding upon the appropriate course of AMI treatment, including transfer. For urban health care providers, these findings stress the importance of inquiring about and determining various aspects and circumstances of impending AMI patient transfers such as hospital origin of transfer, transferring patients’ demographics and health status including presence and severity of potentially serious comorbid conditions like shock, acute renal failure and CHF.
For rural residents, it is important to be acutely aware of local hospitals’ cardiac care capabilities, contemplate how and where they wish to be provided care in the event of an adverse cardiac event and share these intentions with their local health care providers and family members.
Funding Source(s): AHRQ
● Potentially Preventable Readmissions for
Veterans with Congestive Heart Failure: A
Sociodemographic Profile
Kyle Muus, Ph.D.; Alana Knudson, Ph.D.; Marilyn
Klug, Ph.D.; Mary Sarrazin, Ph.D.
Presented by: Kyle Muus, Ph.D., Research
Associate, Center for Rural Health, University of
North Dakota School of Medicine & Health
Sciences, 501 North Columbia Road, Grand Forks,
ND 58203, Phone: (701) 777-4048; Email: klmuus@medicine.nodak.edu
Research Objective: To determine and compare sociodemographic profiles of veterans with CHF who did and did not have a hospital readmission within 15 days of initial admission.
Study Design: Retrospective, cross-sectional analyses of Veterans Affairs inpatient hospital patient data for 2006-2007. Veterans with CHF who were hospitalized within the VA system of care were identified and assigned to either of two cohorts: those who had a hospital readmission with
15 days; and those who did not have a hospital readmission within 15 days. Hospital readmissions were identified using 3M Potential Preventable
Readmissions Grouping Software. The prevalence of all variables for each cohort was described and compared using Chi-Square and independent t-test statistics.
Population Studied: 36,888 patients of Veteran
Affairs hospitals with congestive heart failure in
2006 and 2007.
Principal Findings: Potentially preventable hospital readmissions among veterans with CHF yielded differences between rural and urban residence. Readmissions among urban veterans were associated with being male, older age and being married.
Conclusion: Study findings stress the importance for veterans with CHF to actively engage in ongoing self-care and have regular visits to their primary care provider to avoid serious complications and comorbidities.
Implications for Policy, Delivery or Practice: The
Department of Veterans Affairs should consider using 15-day hospital readmission for measuring, monitoring and improving of health care quality for patients with CHF. Additional efforts are needed for patient education on the importance of appropriate and ongoing self-care of CHF among veterans and to increase access to and utilization of primary health care for veterans.
Funding Source(s): VA
● Gender Differences in Health Insurance
Coverage: The Case for Women with Disabilities
Ari Mwachofi, B.Ed., M.A., M.A., Ph.D.; Allen
Heinemann, Ph.D., A.B.P.P. (R.P.), F.A.C.R.M.;
A.F. Al-Assaf, M.S., M.P.H.
Presented by: Ari Mwachofi, B.Ed., M.A., M.A.,
Ph.D., Assistant Professor, Health Administration &
Policy, University of Oklahoma Health Sciences
Center, 801 Northeast 13th Street, CHB Room 351,
Oklahoma City, OK 73104, Phone: (405) 271-2114;
Email: amwachof@ouhsc.edu
Research Objective: What are the gender differences in health insurance coverage for people with disabilities? What factors influence health insurance coverage for people with disabilities?
Study Design: This study applies two complementary frameworks, the social model of disability and the capabilities model, as the analytic
frame of reference. Within this framework, the study uses case management data from the
Rehabilitation Services Administration (RSA) to compare and contrast health insurance coverage for men and women with disabilities. The study analyzes data from 617,149 cases closed by VR in
2006 in all states. It examines differences in insurance coverage before and VR interventions and assesses factors that influence health insurance coverage.
Population Studied: The focus population is people with disabilities who accessed or tried to access vocational rehabilitation services.
Specifically the study focuses on people with disabilities whose cases were closed by state vocational rehabilitation services in 2006.
Principal Findings: The study finds significant differences in insurance coverage between men and women with women being more dependent on coverage from public sources. The study also finds a positive and significant association between VR intervention duration, VR per capita expenditures for services and health insurance coverage. After
VR interventions, men have a higher likelihood of acquiring their own-employment based health insurance.
Conclusion: VR is more effective in improving insurance coverage for men that for women. This is probably due to the fact that VR is more effective in improving employment and earnings for men that for women.
Implications for Policy, Delivery or Practice:
There is a need to implement policies or practices that ensure equity in access to services which might translate into more gender-equitable employment and earnings outcomes and insurance coverage.
Funding Source(s): N/A
Theme: Gender and Health
● The Tennessee Health Care Safety Net
Carole Myers, Ph.D., M.S.N., B.S.N.; Julia Phillippi,
M.S.N.
Presented by: Carole Myers, Ph.D., M.S.N.,
B.S.N., Assistant Professor, College of Nursing,
University of Tennessee-Knoxville, 1200 Volunteer
Boulevard, Knoxville, TN 37996, Phone: (865) 414-
7218; Email: cmyers9@utk.edu
Research Objective: Evaluate the readiness of the
Tennessee health care safety net at the time of the
TennCare disenrollments. Examine the contextual factors that influenced the status of the Tennessee health care safety net beginning with the inception of TennCare in 1994 to the present. Examine planned, unplanned and unanticipated safety net outcomes using a multi-perspective view.
Study Design: A single case study design was used to study the Tennessee health care safety net, relying on multiple data sources and the examination of diverse perspectives. Interviews were conducted with twenty-one diverse public figures and a variety of documents were reviewed.
These data streams were used to triangulate converging lines of evidence.
Population Studied: A case study is an intensive and in-depth examination of a single entity or phenomenon. Rather than a population, a case is studied. For this study, the case was the
Tennessee health care safety net.
Principal Findings: Three conceptual themes were identified from the data. These are: 1) sustainability, the ability or willingness to support and maintain health care coverage and services or funding over time without compromising other key priorities of future funding; 2) scalability, the ability to increase health care service capacity readily and effectively which implies a adequate supply of resources is, or could be, effectively deployed; and
3) sufficiency, the availability of ample resources to support access to an adequate amount of needed services to meet objectives.
Conclusion: Safety net services are inadequate, underfunded, and patchy in the state of Tennessee.
While intended to be healthcare of last resort, the safety net is being used to support an increasing patient burden without adequate infrastructure support. Affordability is a major concern. The safety net is not sustainable, scalable or sufficient.
There are shortages of primary care providers willing to care for uninsured individuals, individuals who often have extensive social needs. The safety net program is under-funded and most of the funds designated for the program were non-recurring.
Beneficial models of care for these individuals are often not as effective on a large scale. There are also major obstacles to securing specialty care. The problems with the safety net are multi-factorial touching policy, individual determinants of health and the health care delivery system.
Implications for Policy, Delivery or Practice: The problems with the safety net need to be addressed with broad-based policy changes. Strategies are needed to recruit and retain advanced practice nurse and physician providers to provide continuous and stable care to underserved populations without insurance. An adequate infrastructure and funding is needed to provide efficient and effective care that meets health needs in the least intensive setting possible. The unique needs of underserved individuals and the special toll providing care to them entails must be considered in developing strategies. Currently effective models of care are localized and specific to niche patient populations. While much can be learned from these models, it is impractical to assume that effective clinics can be easily expanded or copied.
Funding Source(s): University of Tennessee
Theme: Coverage and Access
● Pseudo or Quasi Convergence? U.S. & Dutch
Health Care since the 2006 Health Insurance Act
Pooya Naderi, B.A., M.A.; Brian Meier, B.A., M.A.
Presented by: Pooya Naderi, B.A., M.A., Graduate
Teaching Assistant, Sociology, University of
Kansas, 716 Fraser Hall, 1415 Jayhawk Boulevard,
Lawrence, KS 66045, Phone: (785) 917-1755;
Email: psdn@ku.edu
Research Objective: In 2006, the Netherlands passed the Health Insurance Act requiring all legal residents to obtain health insurance from private insurance companies. The reform created 1) a national health insurance system guaranteed to all citizens regardless of income or labor force status, and 2) a market orientation that makes private insurance companies the sole providers of health insurance. How does the new policy compare to the
U.S. model of private health insurance provision? Is this reform evidence of a shift toward the American model of private health insurance?
Study Design: We use a comparative case study method to compare the new Dutch system to the private insurance system in the United States.
Principal Findings: We find that although the
Dutch system includes market solutions similar to the U.S. model, it still provides a universal guarantee of coverage to all of its citizens and should be viewed as privatization within the Dutch context rather than a cooptation of American health policy.
Conclusion: By placing greater control of health care organization and provision in the hands of private insurers, and by emphasizing managed competition, it has been suggested that the Dutch are moving closer in logic and practice to the U.S. model of health insurance provision. Our analysis suggests that despite these changes, the Dutch system continues to stand apart from the U.S. in significant ways. First, the Netherlands offers statemandated, compulsory health insurance that encompasses the entire population (98.5%), regardless of employment or occupational status. In contrast, the U.S. houses the highest percentage and number of uninsured—now approaching 50 million—among OECD countries, and the system of voluntary coverage continues to tie economics and employment in male-dominated labor markets to health insurance coverage.
Implications for Policy, Delivery or Practice:
Researchers claim that the Dutch reforms provide a unique and interesting opportunity for comparative policy analysis and policy learning, noting that
Germany and Switzerland have already taken an active interest in implementing features of the Dutch design (van de Ven et al. 2008; Helderman, Schut, van der Grinten, and van de Ven 2005). Might the
U.S. be next? Expansion of private insurance coverage within a market framework of privatization and competition appears to complement the
“exceptional” nature of U.S. social provision.
● Evaluation of Health Plans’ HEDIS Measures by Ethnicity, Language & Geographic Region for California’s SCHIP (Healthy Families
Program)
Muhammad Nawaz, Ph.D.; Mary Watanabe, B.A.;
Shelly Rouillard, B.A.
Presented by: Muhammad Nawaz, Ph.D.,
Research Manager II, California Managed Risk
Medical Insurance Board, California Managed Risk
Medical Insurance Board, 1000 G Street,
Sacramento, CA 95814, Phone: (916) 324-7444;
Email: mnawaz@mrmib.ca.gov
Research Objective: Comparative evaluation of the quality of care provided to Healthy Families
Program (HFP) children by 24 managed health care plans and to examine the differences between ethnicity, language and geographic regions for 12
Healthcare Effectiveness Data and Information Set
(HEDIS) measures.
Study Design: The California Managed Risk
Medical Insurance Board (MRMIB) analyzed data for 12 HEDIS measures using Proc Logistic, Proc
Multtest, CA test, Bonferroni’s Method and Contrast
Statement tests in SAS.
Population Studied: Data from 24 health plans for
12 HEDIS measures collected by administrative and hybrid methods following the National
Committee for Quality Assurance (NCQA) guidelines were used. Nearly 700,000 HFP children under 19 years of age were included in this study. HFP is California’s State Children’s Health
Insurance Program (SCHIP) and is the largest
SCHIP in the country.
Principal Findings: The percentage rates for
Childhood Immunization Combination 3, Well-Child
Visits in the First 15 Months of Life and, Adolescent
Well Care Visits improved by at least 3% from
2006. However, average percentages for 4 measures remained the same or showed slight improvements. The HFP averages for Childhood
Immunization, Well-Child Visits and Access to
Primary Care Practitioners exceeded the national
Medicaid average, whereas the averages for Well-
Child Visits exceeded the national Commercial average. The Adolescent Well-Care Visits averages increased by 4% and 7% from 2006 and
2005, respectively. Significant* improvements of 6% from 2006 and 23% from 2005 were observed in the number of children that received all of the recommended Combination 3 vaccinations.
Demographic analysis revealed that Chinese speaking children received child immunization and
Well Child Visits at the highest rates (89% and
67%, respectively), while English speaking children had the lowest rates of 78% and 51% in the same measures. In ethnic comparisons, African
American, Asian/Pacific Islander and Hispanic children received the highest rates child immunizations (85%, 82% and 82%, respectively) compared to White children who received the lowest (71%). Similarly, Asian/Pacific Islander children had the highest rates of Well Child Visits at
58% compared to 49% for Whites. Comparison between regions showed that the Bay Area region had 85% child immunization rates compared to
69% for Northern and 78% for Los Angeles regions.
Similarly, the Bay Area had significantly* higher
(66%) rates of Well Child Visits compared to 54% for the Los Angeles region. Among gender, males received all recommended visits and Adolescent
Well-Care Visits at a higher rate than females.
However, males over age 16 received a well-care visit at a significantly* lower rate than females.
MRMIB also studied three new measures introduced in 2007: Appropriate Treatment for
Children with Upper Respiratory Infections,
Appropriate Testing for Pharyngitis, and Chlamydia
Screening. Significant* differences were observed for all three new measures across ethnicity, language and geographic regions. Chlamydia
Screening was highest for African American,
English speakers and Bay Area children (58%,
41%, and 52%, respectively) compared to 36%
Whites, 29% Chinese speakers, and 29% for
Northern region. For ranking, health plans were evaluated relative to the HFP average by assessing
12 individual HEDIS measures and by combining these measures into to a single cumulative HEDIS measure using cluster analysis. Based on this analysis, Alameda Alliance for Health, Kaiser
Permanente North and Central Coast Alliance for
Health were up to 20% superior to other health plans for the cumulative HEDIS performance measure. *(p<.01)
Conclusion: HFP children received the recommended services at a rate that is higher than the national Medicaid average and close to national
Commercial averages. The percentage of cumulative measure varied from 56% for poor cluster to 78% for the superior cluster.
Implications for Policy, Delivery or Practice:
Significant differences observed across ethnicity, language, and geographic regions suggest more research is needed to establish whether true disparities in the quality of care exist or whether the differences are due to cultural factors, access, or other issues. The 20% difference between the superior and poor clusters suggests that there is plenty of room for improvements for health plans in average, below average, and poor clusters.
Theme: Child Health
● Preventable Hospitalizations: Does Rurality or
Non Physician Clinician Supply Matter?
Preethy Nayar, M.D., Ph.D.; Anh Nguyen, M.P.H.;
Wanqing Zhang, M.D., M.S.; Kelly Shaw-
Sutherland, M.P.A.; Dan Hummel, B.A.; Keith J
Mueller, Ph.D.
Presented by: Preethy Nayar, M.D., Ph.D.,
Assistant Professor, Health Services Research &
Administration, University of Nebraska Medical
Center, 984350 Nebraska Medical Center, Omaha,
NE 68198-4350, Phone: (402) 559-1981; Email: pnayar@unmc.edu
Research Objective: The purpose of this study is to examine the effects of rurality and the supply of
Non physician clinicians (Physician assistants and
Nurse practitioners practicing in primary care) on the rates of hospitalization for Ambulatory Care
Sensitive Conditions (ACSCs) in Nebraska counties. Prior studies have found that impoverished rural areas and sprawling suburban areas have similar rates of hospitalization for
ACSCs. Greater prevalence of Non-Physician clinicians and International medical graduates were also found to be associated with lower rates of
ACSCs especially in poor rural areas (Mobley et al.
2006). However, others have found that physician supply is not associated with hospitalizations for
ACSCs in rural areas (Laditka et al, 2005). Isolated rural or frontier communities may be at a higher risk of preventable hospitalizations due to worse access to primary care. In frontier areas that have difficulty in recruiting and retaining primary care providers, physician extenders or Non physician clinicians
(NPCs)could be addressing a vital need in the provision of primary care. This study examines the relationship between rurality (non-metro and isolated rural or frontier designation) as well as the proportion of Non physician clinicians (NPCs) to
MDs and DOs and county hospitalization rates for
ACSCs.
Study Design: The study design is a Crosssectional observational study of county level factors that affect the county level rates of hospitalization for ACSCs.
Population Studied: The study population includes all Nebraska counties. The key independent variables are rurality (metro/non-metro and frontier designation) and prevalence of NPCs as a proportion of total primary care provider supply.
Rural counties are defined as non-metro counties using the Office of Management and Budget 2003 definition of metropolitan and non-metropolitan counties. Only nine of Nebraska’s 93 counties are designated as metropolitan. Of the 84 non-metro counties 38 counties are frontier or isolated rural counties, using the conventional definition of frontier as counties having less than 7 persons per square mile. The data are analyzed using multivariate regression models. Control variables included in the regression models are county poverty rate, diversity
(% non-white population), unemployment rate and elderly population rate. Data on county hospitalization rates for ACSCs are obtained from the 2007 Nebraska Hospital Discharge data set.
County level control variables are obtained from the
US Census Bureau 2007 population estimates.
Data on Primary care provider supply are obtained from the University of Nebraska Medical Center
Health Professions Tracking System (HPTS) 2007 data.
Principal Findings: The rate of preventable hospitalizations was positively related to the percent of population over 65 years of age and the proportion of non physician clinicians to total primary care providers. The supply of primary care providers was not significantly related to the county rates of preventable hospitalizations. Rurality (Nonmetro and isolated rural or frontier county designation) was also not significantly associated with the rate of preventable hospitalizations.
Conclusion: Counties with a higher percentage of elderly population in Nebraska have higher rates of preventable hospitalizations. Primary care provider supply or isolated rural (frontier) status were not found to be associated with the rate of preventable hospitalizations. Contrary to the findings of earlier studies, counties with higher proportion of non physician clinician had higher rates of preventable hospitalizations.
Implications for Policy, Delivery or Practice: The findings suggest that the elderly population face problems with access to primary care that are not solely explained by the supply or availability of primary care providers. The accessibility of providers, including travel impedance (distance and time to travel as well as availability of transportation) may be an issue affecting the access to primary care of elderly populations. The study also calls in to question whether a greater reliance on non clinician providers is a solution to the access problems that rural counties face considering that counties with a higher proportion of non physician clinicians have higher rates of preventable hospitalizations.
Funding Source(s): University of Nebraska
Foundation
Theme: Health Care Workforce
● An Evaluation of Forecasting Techniques for
Physician Workforce Supply Projections in
Nebraska
Preethy Nayar, M.D., Ph.D.; Anh Nguyen, M.P.H.;
Kelly Shaw Sutherland, M.P.A.; Dan Hummel, B.A.;
Keith Mueller, Ph.D.
Presented by: Preethy Nayar, M.D., Ph.D.,
Assistant Professor, Health Services Research &
Administration, University of Nebraska Medical
Center, 984350 Nebraska Medical Center, Omaha,
NE 68198-4350, Phone: (402) 559-1981; Email: pnayar@unmc.edu
Research Objective: The objectives of this study are: 1)To use econometric forecasting models for time series data to forecast the supply of physicians in Nebraska 2010 to 2020 2)To evaluate and compare the different forecasting models to identify the best physician workforce forecasting model.
Study Design: The study design is a time series design of physician supply from 1980 to 2005.
Population Studied: The data on physician supply
(total active non-federal physicians) for years 1980 to 2005 are obtained from the Area Resource File.
Data on Nebraska’s population 1980 to 2005 and population projections 2010 to 2020 are obtained from the US Census Bureau population data. The data are analyzed using Stata 10 software. The time series data are analyzed by different econometric forecasting techniques including
Exponential smoothing, double exponential smoothing, Holt Winters and Moving Averages models. The different models are compared to identify the model with the best forecasting accuracy.
Principal Findings: Forthcoming: Expected
Completion date of the project March 1
Conclusion: Forthcoming
Implications for Policy, Delivery or Practice: The
Bureau of Health Professions National Center for
Health Workforce Analysis Physician Supply and
Requirements models forecast that the United
States will have a projected shortage of physicians in 2020. However, the models do not make statespecific forecasts of physician supply and demand.
This study applies econometric smoothing and forecasting techniques to provide state specific forecasts of physician workforce supply in Nebraska that will better inform state policy makers in their physician workforce planning efforts.
Funding Source(s): University of Nebraska
Foundation
Theme: Health Care Workforce
● Did Medicare's Prospective Payment System make Long Term Care Hospitals more Efficient?
Preethy Nayar, M.D., Ph.D.; Fang Yu, Ph.D.
Presented by: Preethy Nayar, M.D., Ph.D.,
Assistant Professor, Health Services Reseach &
Administration, University of Nebraska Medical
Center, 984350 Nebraska Medical Center, Omaha,
NE 68198-4350, Phone: (402) 559-1981; Email: pnayar@unmc.edu
Research Objective: In October 2002, Medicare changed its mode of payment to long term acute care hospitals (LTCHs) from a cost based to a prospective payment system (PPS). The objective of this study is to analyze the effect of Medicare’s
Prospective Payment System (PPS) on the operating efficiency of LTCHs.
Study Design: Research Design: The research design is a pre-post test design of within hospital change in operating efficiency before and after the implementation of PPS.
Population Studied: The study population includes all LTCHs that were operational during the four – year period from 2001 to 2004. Measures: The impact of the payment mechanism (PPS), market and organizational variables on the operating efficiency (operating costs per inpatient day) of
LTCHs is evaluated using Growth curve analysis.
Principal Findings: Results: The study found no significant association of PPS and the operating
efficiency of LTCHs. A secular trend in operating efficiency was noted with higher operating costs per inpatient day in years 2002, 2003 and 2004 as compared to the base year 2001. County per capita income and population estimate were found to be significantly positively associated with the operating costs per inpatient day of the LTCHs. Larger
LTCHs had lower operating efficiency (higher operating costs per inpatient day).
Conclusion: Medicare's PPS has cost containment and efficiency enhancement incentives.
However,with the shift in reimbursement from cost based system to PPS, LTCHs paid under PPS did not have higher operating efficiency as would be expected. LTCHs in markets with greater resources as reflected by county level per capita income and population estimate had lower operating efficiency.
Implications for Policy, Delivery or Practice:
This study does not find support for Medicare’s shift to a prospective payment system for LTCHs having achieved its intended cost containment objective, in the first two years following the implementation of
PPS. The results are significant in an era of spiraling health care costs for post acute care and an ongoing search for better cost containment tools.
Theme: Medicare
● Trigger Systems for Iatrogenic Adverse
Events: A Typology with Implications for Design
& Evaluation
Jonathan Nebeker, M.S., M.D.; Stephanie Shimada,
Ph.D.; Michael Harrison, Ph.D.; Lucy Savtiz, Ph.D.;
Haytham Kaafarani, M.D.; Amy Rosen, Ph.D.
Presented by: Jonathan Nebeker, M.S., M.D.,
Associate Professor of Medicine, University of Utah
School of Medicine, VA Salt Lake City Geriatrics
Research, Education, & Clinical Center, 500 Foothill
Drive, Salt Lake City, UT 84148, Phone: (801) 584-
2522; Email: Jonathan.Nebeker@hsc.utah.edu
Research Objective: Over the last 20 years, enormous progress has been made in the development of trigger systems. Trigger systems use clinical algorithms or “triggers” to flag patients who may have experienced iatrogenic adverse events. They may also provide methods for investigating, confirming, and responding to such events. Initially developed for mitigating adverse drug events (ADEs), trigger systems have been adapted to address other types of events and purposes. However, there is considerable variation in how goals, design features, and evaluation metrics of these systems match their implementations. It is very difficult for researchers, practitioners, and policymakers to agree upon—or even to discuss—future work on trigger systems without a common understanding of these domains.
Study Design: In response to these concerns, the
Agency for Healthcare Research and Quality
(AHRQ) recently sponsored a meeting on trigger systems with experts in this area. An outcome of the meeting was development of a typology for trigger systems. The resulting typology, presented in this study, refers primarily to systems for the detection of adverse drug events (ADEs) and surgical complications, but it is applicable to all trigger systems.
Principal Findings: This typology defines and distinguishes triggers, trigger systems, trigger tools, and Targeted Injury Detection Systems (TIDS). It also describes key design features of trigger systems, including the clinical specificity of triggers and the timing of alert investigation. The objectives of each individual trigger system, such as rate estimation, case finding, and clinical interventions, determine which design features it should include.
We also explore implications of this typology for evaluating the performance of trigger systems.
These implications include understanding triggersystem performance measures, estimating performance in the context of implementation, selecting the evaluation cohort, determining a true event, and determining the appropriate unit of analysis.
Conclusion: This typology explicates different types of trigger systems, key features of their design, and principles of their evaluation.
Implications for Policy, Delivery or Practice: The presented typology provides an essential framework with which to advance the evidence base for trigger systems. Secondarily, the typology provides a common language to support dialogue among policy makers, practitioners/users, and researchers. These contributions are essential to moving the field to more systematic, productive, and effective efforts that enhance patient safety in practice.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Measurement
● Hospital Religious Affiliation, Service
Provision & Outcomes for Low Birthweight
Infants
Melissa Nelson, Ph.D.; Kirk Allison, M.S., Ph.D.;
Bryan Dowd, Ph.D.
Presented by: Melissa Nelson, Ph.D., Postdoctoral
Fellow, Institute for Health, Health Care Policy &
Aging Research, Rutgers University, 30 College
Avenue, New Brunswick, NJ 08805, Phone: (732)
932-1309; Email: mnelson@ifh.rutgers.edu
Research Objective: Despite the prevalence of religiously affiliated hospitals in the United States, little is known about the extent to which hospital religious affiliation (HRA) impacts service provision or patient outcomes. Differences between religiously affiliated (RA) and non-RA hospitals should be more likely where affiliated religions have clear positions regarding appropriate care, such as at the beginning of life. We examined relationships among HRA, ventilation, nutrition delivery, and discharge status for low birthweight (LBW) infants.
Study Design: We analyzed 2004 data from the
American Hospital Association Annual Survey and the Healthcare Cost and Utilization Project (HCUP)
State Inpatient Databases for five states (Arizona,
New Jersey, North Carolina, Washington, and
Wisconsin); we included records indicating newborn admission; age = 28 days; and LBW, short gestation, or fetal growth retardation (Clinical
Classification Software [CCS] code = 219).
Bivariate logistic regressions probed relationships among respiratory intubation/ventilation or enteral/parenteral nutrition (CCS procedures 216,
223), ownership (government, church not-for-profit
[NFP], other NFP, private for-profit), Catholic affiliation, congenital anomalies or birth-related maternal complications, and sociodemographics.
Relationships among discharge status (routine, to another inpatient facility or against medical advice, died), ventilation, nutrition delivery, HRA, and patient characteristics were examined using multinomial logit regression.
Population Studied: AZ, NC, NJ, WA, and WI
LBW infants born in 2004 (N = 57,345).
Principal Findings: Ventilation and nutrition delivery support were associated with HRA.
Ventilation was more likely in Catholic than non-
Catholic hospitals (OR = 1.23, 95% CI =[1.11,
1.37]). Both ventilation and especially nutrition delivery were less likely in for-profit than in churchowned NFP hospitals (ventilation: OR = 0.74, 95%
CI =[0.61, 0.90]; nutrition: OR = 0.23, 95% CI
=[0.17, 0.31]). Nutrition delivery was also less likely in government-owned and other NFP hospitals than in church-owned NFP hospitals (government: OR =
0.43, 95% CI =[0.37, 0.50]; other NFP: OR = 0.63,
95% CI =[0.55, 0.71]). Discharge status varied slightly with HRA. Church-owned NFP hospitals, government and other NFP hospitals had similar probabilities for discharge type, but routine discharges were more likely than deaths in for-profit hospitals than church-owned NFP hospitals (OR =
1.77, 95% CI =[1.09, 2.89]). Routine discharges were less likely than transfers to other inpatient facilities for Catholic hospitals versus non-Catholic hospitals (OR = 0.88, 95% CI =[0.79, 0.99]).
Conclusion: Interventions for LBW infants are used more often in church-owned hospitals, but
HRA other than Catholic does not appear to significantly affect patients’ discharge statuses.
Catholic hospitals appear more likely to transfer patients to other inpatient facilities than to discharge patients routinely, even after controlling for service differentials, but post-hospital discharge outcomes are unknown. Further studies should include postdischarge measures of infant growth and health over time to see whether HRA-related practice differences have differential health effects for LBW infants, and they should control for patient (parent) self-selection into hospitals by RA.
Implications for Policy, Delivery or Practice:
Service provision for LBW infants differs by hospital ownership structure, but it is yet unclear whether
HRA differences have lasting effects on health. A clearer understanding of relationships among services, outcomes, and HRA could facilitate informed consumer choice of health care providers, networks, and systems.
Funding Source(s): MacLaurin Institute
Theme: Organizational Performance and
Management
● Health Insurance Coverage among Diverse
Family Types
Justine Nelson, M.A.; Jean Marie Abraham, Ph.D.
Presented by: Justine Nelson, M.A., Doctoral
Student, Family Social Science, University of
Minnesota, 1463 Blair Avenue, Saint Paul, MN
55104, Phone: (651) 917-0690; Email: nels2177@umn.edu
Research Objective: To investigate the relationship between demographic characteristics, economic factors, and family structure on the probability that a family has health insurance for all family members.
Study Design: The family was the unit of analysis in this study. Families were defined as individuals living with others they are related to by blood, marriage, or in a marriage-like relationship.
Individuals not in such a relationship were counted as one-person families. Families were divided by structure into three categories. The first group of families had relationships that usually make them eligible for dependent coverage on employersponsored insurance (ESI). These eligible relationships included only spouses, and parents and minor children (or full-time college students).
The second group was families with at least one relationship that is generally not eligible for dependent status on an ESI policy (eg kinship families. The third group was one-person families.
Multivariate logistic regression was used to determine whether family structure was associated with families having health insurance for all members. The following societal variables were controlled for: income ceiling on public health insurance eligibility in the resident’s state, MSA, and region. Employer variables included employer size and hours worked. Family variables included education, comfort speaking English, whether anyone has a serious medical condition, family income, presence of a child in the family, and the employee (or reference person’s) age and race.
Population Studied: The 2005 Medical
Expenditure Panel Survey provided the data for this nationally representative sample of U.S. families.
Principal Findings: In December 2005, 69% of all
U.S. families had health insurance coverage for all members. This was true for 78% of families with relationships that generally make them eligible for dependent coverage, but only 45% of families with at least one relationship that is generally not eligible. Seventy-three percent of one-person
families had coverage for themselves. After controlling for regional, statewide, employer and other family variables, families with generally eligible relationships were 4.5 times more likely to have health insurance coverage for everyone in the family than were families with at least one generally ineligible relationship.
Conclusion: Family relationships are a significant predictor of health insurance status for families.
Implications for Policy, Delivery or Practice: The dominant role played by employer-sponsored health insurance (ESI) in the US puts people living in diverse families at risk for not having access to health insurance. Policymakers should keep in mind that expansions in ESI will not benefit these families unless families are defined as more than just spouses, and parents and minor children.
Theme: Coverage and Access
● National & Regional Growth in the Availability of High Cost Medical Technology at U.S.
Hospitals, 2001-2007
Jason Nelson, M.P.H.; Thomas Concannon, Ph.D.;
Jason Nelson, M.P.H.; John Griffith, Ph.D.
Presented by: Jason Nelson, M.P.H., Statistical
Associate, Institute for Clinical Research & Health
Policy Studies, Tufts Medical Center, 800
Washington Street, #63, Boston, MA 02111,
Phone: (617) 636-0546; Email: jnelson2@tuftsmedicalcenter.org
Research Objective: The appropriateness of highcost technology diffusion throughout U.S. hospitals has been the subject of wide ranging debate, and discussion of technology change in the treatment of heart disease has been especially intensive. While it is clear that medical technology has added significantly to the health of the nation overall, it is widely observed that incentives to adopt a technology are not always aligned well with improvement in patient outcomes. The purpose of this project was to identify national rates, regional variation and local factors associated with growth in technology at U.S. hospitals over a five year period
(2002-2006). We studied growth in the rates of two technologies: primary percutaneous coronary intervention (PCI) labs and Positron Emission
Tomography (PET) scanners.
Study Design: American Hospital Association
Annual Survey data were used to identify rates of growth in the percent of US hospitals with PCI labs and PET scanners. The AHA datasets include hospitals' longitude and latitude coordinates, allowing for integration with a geographic information system (GIS). For every US-county
(N=3,141) we estimated population-adjusted 5-year growth rates in hospitals with PCI labs and PET scanners (per 100,000 people). To account for overdispersion due to zero growth in most US
Counties we ran negative binomial regressions separately for growth in PCI labs and growth in PET scanners. We adjusted estimates of growth for each of four regions (North, Midwest, South and West) with county-level estimates of the number of hospitals without PCI labs and PET scanners (the
“risk” for growth), with an estimate of the % urban land cover in a county (“urbanicity”), and with county-level estimates of socioeconomic status
(SES) drawn from the US Census (% foreign born,
% above 200% of the poverty level, % completed college, and % white).
Population Studied: U.S. residents.
Principal Findings: In 2006, 13% of all hospitals with a PCI lab (N=1,932) acquired the lab after
2001. In 2006, 56% of all hospitals with a PET scanner (N=1,114) acquired the scanner after 2001.
We found significant pair-wise differences in mean growth rates between the Northeast and all other regions, but none elsewhere (p<0.01, adjusted for multiple comparisons). After adjusting for county level SES, and urbanicity, regional effects all but disappeared (faster growth in hospitals with PET scanners remained marginal in the North - West comparison). After adjustment, urbanicity was a significant predictor of growth in PCI labs (p<0.001) but not for PET scanners (p=0.33). The risk for growth was significant predictor of growth for both
PCI labs (p<0.001) and PET scanners (p<0.001).
Conclusion: In summary, we found differences in the rates of growth from region to region but no clear regional effects after adjusting for county level characteristics such as SES, urbanicity and risk for growth. We found effects at the county level that are suggestive of small area variation.
Implications for Policy, Delivery or Practice: To understand this small area variation, more research is needed in the immediate environment surrounding US hospitals. Specifically, we suggest that further investigation is needed to understand the roles of the local regulatory and competitive environments hospitals face in their decisions to acquire high cost medical technology.
Theme: Health Care Markets and Competition
● Utilization of Mental Health Care by
Community-Dwelling Older Adults
Melissa Nelson, Ph.D.; Robert Kane, M.D.; Merrie
Kaas, D.N.Sc., R.N., C.S.; Rosalie Kane, Ph.D.
Presented by: Melissa Nelson, Ph.D., Postdoctoral
Fellow, Institute for Health, Health Care Policy &
Aging Research, Rutgers University, 30 College
Avenue, New Brunswick, NJ 08901, Phone: (732)
932-1309; Email: mnelson@ifh.rutgers.edu
Research Objective: Older adults use less mental health care (MHC) than younger adults, and when they do seek care, it is often from a primary care provider (PCP). Older adults also report more somatic and fewer psychological symptoms than younger adults. Despite this, most MHC research focuses on younger adults. In older adults, correlates of utilization beyond current symptoms
and system characteristics are often ignored. We examined relationships among probability of MHC and perceived need for care, history of physical and mental illness, social support, and current mental illness among older adults.
Study Design: Data were from the Collaborative
Psychiatric Epidemiology Surveys (CPES), nationally representative, cross-sectional surveys about mental illnesses and treatment. Logistic regression with SUDAAN examined the probability of any MHC (from PCPs and/or specialists) in the past year. Models included chronic physical condition and alcohol abuse history, current and past depression and/or anxiety, social support, cognition, mobility, self-care abilities, and sociodemographics. The sample was analyzed as a whole and stratified by history of depression/anxiety. Another logistic regression examined respondents reporting perceived need for care in the past year.
Population Studied: Community-dwelling adults
<= 65 years old with and without mental illnesses
(N=1,874).
Principal Findings: Of the entire sample, 6% received MHC. Eighty-three percent of those who perceived a need for care and 25% of those with current depression and/or anxiety received care. In the complete sample, odds of utilization increased with history of chronic conditions or alcohol abuse, private insurance, and current major depressive disorder (MDD), generalized anxiety disorder
(GAD), or other anxiety. Relationships with utilization changed in the stratified model. Among those who never had depression/anxiety, utilization odds increased with symptoms reported during worst lifetime depression episode (OR =
1.36/symptom, 95% CI=[1.15, 1.61]) and private insurance (OR = 4.05, CI=[1.07, 15.25]). Among those with past but not current depression/anxiety, utilization odds increased with symptoms reported during worst lifetime anxiety episode (OR =
1.25/symptom, CI=[1.06, 1.48]). By contrast, for respondents with current depression/anxiety, symptoms were not significant, but utilization increased with physical conditions (OR =
1.89/condition, CI =[1.30, 2.73]), age (OR =
1.07/year, CI = [1.01, 1.12]), and private insurance
(OR = 3.60, CI = [1.10, 11.77]), and decreased with poorer self-care abilities. Among respondents with perceived need for care, neither MDD nor anxiety was related to utilization. Instead, utilization odds increased with age, household size, and education, and decreased with poorer self-care abilities and subthreshold GAD.
Conclusion: Many older adults who met criteria for depression/anxiety did not use MHC, and past depression/anxiety symptoms were related to current utilization for respondents without either condition, suggesting that formal diagnostic criteria do not reflect individual experiences. Physical and mental health were not related to utilization among respondents with perceived need for care, suggesting that health status may influence perceived need, but other factors influence utilization among those with perceived need.
Implications for Policy, Delivery or Practice:
Providers should be sensitive to depression/anxiety symptoms presented by older patients, even if they are below diagnostic thresholds. Future studies should examine how psychosocial factors influence older adults with perceived need for MHC to use or avoid care.
Funding Source(s): NIH Predoctoral and NIMH
Postdoctoral Training Grants
Theme: Behavioral Health
● Age & Race-Related Differences in Perceived
Purpose for Antidepressant Use
Melissa Nelson, Ph.D.
Presented by: Melissa Nelson, Ph.D., Postdoctoral
Fellow, Institute for Health, Health Care Policy &
Aging Research, Rutgers University, 30 College
Avenue, New Brunswick, NJ 08901, Phone: (732)
932-1309; Email: mnelson@ifh.rutgers.edu
Research Objective: Older adults and members of racial/ethnic minority groups report more somatic than emotional symptoms of depression, affecting likelihood and type of mental health care use. It is unclear if differences in somatic/emotional attributions also occur in explanations for antidepressant use. Furthermore, while evidence shows that antidepressant adherence relates to beliefs about necessity of antidepressants and fear of side effects, it is unclear how the perceived purpose of antidepressants might affect adherence.
People taking antidepressants for localized physical symptoms may become frustrated when symptoms are not mitigated quickly, affecting adherence. This study examined relationships among age, race, perceived purpose for antidepressants, and adherence among community-dwelling adults taking antidepressants.
Study Design: Data were from the Collaborative
Psychiatric Epidemiology Surveys, nationally representative surveys on mental illness and service use, which oversampled racial/ethnic minority populations. Multinomial logit regression in
SUDAAN examined probability of reasons for taking antidepressants (physical but not mood [P], mood but not physical [M], physical and mood [MP], and neither). Physical reasons were problems with energy, appetite, sleep and/or pain. Mood-related reasons were sadness, anxiety, panic, mania, anger, and/or suicidal thoughts. Analyses controlled for probable diagnoses of depression and anxiety (via the WHO Composite International
Diagnostic Interview), chronic physical condition history, social support, mental health services use, and sociodemographics. Bivariate analyses between perceived purpose for antidepressants and non-adherence (days medication not taken as
instructed or non-sanctioned cessation) were also performed.
Population Studied: Community-dwelling adults who reported taking one to three antidepressants and no other psychotropic medications in the past year (N = 678, ages 18-84).
Principal Findings: Older people were more likely to endorse neither M nor P than the MP group (OR
= 1.05/year, 95% CI= [1.01-1.09]). African-
American and Afro-Caribbean respondents were more likely than Whites to be in the P group than the M or MP groups (OR = 2.42, CI= [1.11, 5.27];
OR = 4.32, CI= [1.51, 12.36], respectively), and
Latinos were more likely than Whites to endorse neither category than either the M or MP groups
(OR = 5.26, CI= [1.46-18.88]; OR= 10.04, CI=
[2.55-39.50], respectively). There were no significant differences in attributions between Asian and White respondents. Respondents reporting mental health care use in the past year were more likely to endorse the M and MP groups than the P group (OR= 7.85, CI= [3.51, 17.53]; OR= 6.03, CI
=[2.42, 15.03], respectively). Bivariate relationships between perceived purpose and non-adherence were non-significant.
Conclusion: Members of racial/ethnic minority groups were more likely than Whites to cite physical symptoms and less likely than Whites to cite moodrelated symptoms as reasons for taking antidepressants. Among respondents receiving mental health services, emotional attributions were more likely than physical attributions, but this was not related to adherence. Future studies need to examine attitudes toward antidepressants among people who did not fill their prescriptions to understand if somatic/emotional attributions play a role in decisions to fill prescriptions.
Implications for Policy, Delivery or Practice:
Efforts to improve adherence among those who have filled prescriptions for antidepressants should focus more on beliefs about their necessity and fear of side effects than on emotional/somatic characteristics of depression.
Funding Source(s): NIMH
Theme: Disparities
● Performance Improvement in Health Care
Organizations: The Role of Interorganizational
Learning in Quality Improvement Collaboratives
Ingrid Nembhard, S.M., Ph.D.
Presented by: Ingrid Nembhard, S.M., Ph.D.,
Assistant Professor, Yale University, 60 College
Street, P.O. Box 208034, New Haven, CT 06520,
Email: ingrid.nembhard@yale.edu
Research Objective: Quality improvement collaboratives - organized programs in which teams from multiple organizations work together to improve practices around a specific topic (e.g. decreasing infection rates) – have been called
“arguably the health care delivery industry’s most important response to quality and safety gaps”
(Mittman 2004). In part, their presumed effectiveness stems from their provision of an infrastructure for interorganizational learning about best practice implementation. However, empirical research has not examined the extent to which interorganizational learning within collaboratives occurs and is associated with performance improvement. The present research investigates the use and performance consequences of interorganizational learning activities in collaboratives.
Study Design: This study involved cross-sectional examination of collaborative participants’ interorganizational learning activity. Data on organizations’ participation in six collaborativesponsored interorganizational learning activities
(e.g. monthly collaborative conference calls) was obtained via self-administered surveys of their improvement teams. Performance improvement data was obtained from the Institute for Healthcare
Improvement (IHI), while demographic data was obtained from the American Hospital Association
(AHA) Directory, Profiles of U.S. Hospitals, and organizations’ websites. We used multiple regression analyses to statistically analyze relationships.
Population Studied: Our sample consisted of 31 health care organizations (hospitals and clinics) in the U.S. and Canada that had participated in one of two 2004-2005 IHI Breakthrough Series collaboratives: Improving Access and Efficiency in
Primary Care and Reducing Complications from
Ventilators and Central Lines in the ICU.
Principal Findings: Organizations varied in their use of collaborative-sponsored interorganizational learning activities. Moreover, their use of these activities was positively associated with performance improvement in two respects. The more an organization used these activities, the more quality-improving practices it successfully implemented and the more likely it experienced significant improvement on outcomes measures in the collaborative topic area.
Conclusion: The interorganizational learning opportunities that collaboratives provide are beneficial, but only to those organizations that take advantage of them. Simply being a member of a collaborative does not ensure significant performance improvement. Rather, active participation in the collaborative’s activities is required to reap the full benefits of collaborative membership.
Implications for Policy, Delivery or Practice:
This research affirms the importance of interorganizational learning for quality improvement in health care. By extension, it suggests the importance of implementing policies that facilitate and encourage interorganizational learning among healthcare organizations. Additionally, from a practice perspective, it suggests that organizations aiming to improve their performance quality should
seek and utilize interorganizational learning opportunities, in addition to traditional, intraorganizational learning opportunities.
Funding Source(s): Harvard Business School
Division of Research
Theme: Quality and Efficiency: Organized
Processes
● Obesity, Disease Burden & Health Disparities in American Indian/Alaskan Native & Native
Hawaiian/Pacific Islander Medicare Populations
Judy Ng, Ph.D.; Sarah Hudson Scholle, Dr.P.H.;
Rachel Collins, M.P.H.
Presented by: Judy Ng, Ph.D., Research Scientist,
Research, National Committee for Quality
Assurance, 1100 13th Street, NW, Suite 1000,
Washington, DC 20005, Phone: (202) 657-6567;
Email: ng@ncqa.org
Research Objective: Despite well-documented disparities in health status among non-white racial and ethnic minorities, much less is known about
Medicare beneficiaries of American Indian/Alaskan
Native and Native Hawaiian/Pacific Islander race and ethnicity. These beneficiaries face risk of adverse health outcomes associated with their race and ethnicity, such as higher rates of diabetes and alcoholism, as well as with age or disability. This study describes obesity, disease burden and health disparities in American Indian/Alaskan Native
(AI/AN) Medicare beneficiaries and in Native
Hawaiian/Pacific Islander (NH/PI) Medicare beneficiaries.
Study Design: A nationally-representative sample of AI/AN and NH/PI Medicare beneficiaries was analyzed using data on individuals’ body mass index, chronic conditions, functioning and health status from the 2007 Medicare Health Outcomes
Survey, an annual health survey of beneficiaries in all active Medicare managed care health plans
(response rate = 66%).
Population Studied: In 2007, the Health Outcomes
Survey included a random sample of 182,255
Medicare beneficiaries with information on race and ethnicity, covering almost 200 plans from all 50 states. Of these, a total of 1,158 respondents selfidentified as AI/AN (n=716) or as NH/PI (n=442), and 131,669 self-identified as white.
Principal Findings: Approximately 32% of AI/AN and 31% of NH/PI Medicare beneficiaries reported a BMI in the obese range, compared to 26% of whites (p<0.0001). NH/PI and white beneficiaries reported a similar number of mean chronic conditions (3 conditions) and Activities of Daily
Living impairments (1 impairment), but AI/AN beneficiaries reported higher means for both (4 conditions, 2 impairments). In addition, AI/AN and
NH/PI beneficiaries were more likely to have only fair or poor perceived health (40% AI/AN and 34%
NH/PI versus 29% in whites, p<0.0001), and more likely to face difficulties with their reading and hearing ability. Differences in symptoms were also prominent. Similar rates of NH/PI and white beneficiaries reported severe arthritis pain
(approximately 9%) or problems from lower back pain (approximately 17%); however, AI/AN beneficiaries reported much higher rates of severe arthritis pain (17%) or problems from lower back pain (24%, both p<0.0001). Moreover, 27% of
AI/AN and 19% of NH/PI beneficiaries reported presence of depressed mood, compared to only
14% of white beneficiaries (p<0.0001). Multivariate analyses will adjust for differences according to beneficiaries’ age, disability, region, and socioeconomic status.
Conclusion: Although NH/PI and white Medicare beneficiaries report similar rates of functioning and some aspects of health, both AI/AN and NH/PI beneficiaries continue to have greater obesity rates, higher disease burden, and worse health status in general than white beneficiaries in the Medicare population.
Implications for Policy, Delivery or Practice:
These findings indicate that while some differences in health status are narrower, AI/AN and NH/PI beneficiaries in Medicare face troubling lower health status compared to whites. Efforts to encourage
Medicare managed care plans to track disparities in health and health care and to ensure that AI/AN and NH/PI populations receive culturally appropriate care are needed. These concerns are especially important given the unique historical context and culture of AI/AN and NH/PI populations, the expected increase in numbers of older AI/AN and NH/PI, and the expected increase of all nonwhite minorities in the Medicare population.
Funding Source(s): CMS
Theme: Medicare
● Quality of Care for Aging "Baby Boomers" &
Implications for Medicare
Judy Ng, Ph.D.; Sarah Hudson Scholle, Dr.P.H.;
Shakeh Kaftarian, Ph.D.
Presented by: Judy Ng, Ph.D., Research Scientist,
Research, National Committee for Quality
Assurance, 1100 13th Street, NW, Suite 1000,
Washington, DC 20005, Phone: (202) 657-6567;
Email: ng@ncqa.org
Research Objective: In 2011, the first of the post-
World War II “baby boom” generation will reach age
65 and become eligible for Medicare. The implications of the baby-boomers’ entry into
Medicare for costs have been widely discussed, but less attention has been paid to the implications for health care quality. This analysis of nationallyrepresentative data explores the extent of health care quality provided to “baby boomers”, between ages 45 to 64, and older adults ages 65 or over, using established quality of care measures for two of the most common, costly conditions among older adults: cardiovascular disease and diabetes. We
also pay special attention to potential racial and gender disparities in quality of care.
Study Design: Using quality of care measures from the National Medical Expenditures Panel Survey
(MEPS), this study aggregated multiple years of data (2004-2006) to examine care for the 45-64 age group, and for the 65 and over population. Reports of quality across racial/ethnic (white, black, and
Hispanic) and gender subgroups are also examined. The available measures focused on access and effectiveness of care for cardiovascular disease and diabetes.
Population Studied: Older adults ages 45 to 64, and ages 65 and over, with self-reported cardiovascular disease (n=23,235) and diabetes
(n=5,077).
Principal Findings: There were strong, significant differences in access to care based on age and gender. Individuals age 45-64 more often reported delays in care, and women more often than men met problems with one of four access issues: getting medical care as soon as one wanted, delays in critical medical care, delays in dental care, and delays in necessary prescriptive medications. This was consistent across all racial/ethnic groups. For example, among cardiovascular disease patients, the relative rates for delays in critical medical care among respondents under 65, compared to those
65 or over, were 1.90 for white men and 2.59 for white women. Relative rates for a composite of all four access problems ranged from 1.22 for black men to 1.91 for Hispanic men. Whites generally had an advantage over blacks or Hispanics, particularly for receiving diabetic care (such as eye examinations), flu shots, and dental care.
Multivariate analyses will adjust for differences according to private insurance coverage, health and socioeconomic status.
Conclusion: “Baby boomers” ages 45 to 64 are at special risk for delays in care. Black and Hispanic older adults with cardiovascular disease and diabetes also face disparities in quality of care.
Implications for Policy, Delivery or Practice:
Older adults who face delays in care are at risk of adverse health resulting from their access problems. Furthermore, older adults ages 45 to 64 who face delayed care for their chronic health conditions, such as cardiovascular disease or diabetes, may enter Medicare with worsened health and ability to manage their care, and risk needing costlier care. This could have expensive, harmful implications for the Medicare program. In addition, the number of racial/ethnic minorities is growing, while their quality of care remains behind that of whites. This could also be unhealthy not just for these older adults, but have expensive implications for the Medicare program as more racial/ethnic minorities enter Medicare.
Funding Source(s): AHRQ
Theme: Disparities
● The Effects of General Practice Size on
Quality of Care
Charis Ng, B.A.; Kok Ping, Ng, B.S.
Presented by: Charis Ng, B.A., Research Officer,
Health Services & Outcomes Research, National
Healthcare Group, 6 Commonwealth Lane, #04-
01/02 GMTI Building, Singapore, 149547, SG,
Phone: (656) 496 6933; Email: charis_wl_ng@nhg.com.sg
Research Objective: Primary health care involves the provision of primary medical treatment, preventive healthcare and health education. 80% of primary health care in Singapore is provided through some 2,000 private medical clinics, and the remaining 20% through 18 public polyclinics. In the
Primary Care Survey 2005 commissioned by the
Ministry of Health, Singapore, 55% of the 490 participating private clinics were solo practices, and
45% were group practices, having 2 or more doctors in their clinics. The relationship between practice size and quality of care has been of great interest to healthcare providers. In the United
Kingdom, there is a trend towards general practice consolidation into large multi-partner practices. This review aim to determine if there is an association between general practice size and quality of care.
Study Design: A systematic review was done. Five databases (CINAHL, Cochrane, CRD, Embase,
PubMed) were searched to identify primary studies that examined the relationship between practice size and quality of care, using the search terms
(with some variations across databases): 1) MeSH terms: family practice, practitioner, primary health care, quality of healthcare, outcome and process indicators, 2) Keywords: solo, single handed, practice size, list size, caseload and patient volume.
Articles were independently reviewed, and eventually assessed vis-à-vis by two investigators.
Population Studied: The initial search resulted in
240 articles. Following selection criteria and assessment for methodological quality, only eight studies were included in this review.
Principal Findings: In seven studies looking at clinical effectiveness, larger practices showed either similar or better performance compared to smaller practices, though the differences were usually small. Larger practices were similar to or marginally better than smaller practices for quality indicators for diabetes, coronary heart disease (CHD), CHD with left ventricular dysfunction (LVD), hypertension and stroke. Practices with three or more physicians delivered higher quality of clinical care. On the other hand, the only study on patient satisfaction reported that larger practices were associated with poorer patient satisfaction.
Conclusion: There is some evidence to support the hypothesis that larger general practices perform better in terms of clinical processes and outcomes.
However, patient satisfaction may be compromised.
More research is needed to identify the differences
between large and small practices that make a difference in their quality of care.
Implications for Policy, Delivery or Practice: The majority of the private practices in Singapore are small. If there is a policy shift towards consolidating small practices into larger practices, patient satisfaction may be compromised. The cost and effort for restructuring may also outweigh the savings from cost-sharing, and modest improvement in clinical processes and outcomes.
Unlike in the UK where practices are rewarded for their clinical performance, patients are a unique source of income for private practices in Singapore.
Patient satisfaction may play a bigger role than achievement of clinical indicators in the decision to form partnerships.
● Does Pay-for-Performance Improve Inpatient
Quality of Care?
Lauren Nicholas, Ph.D., M.P.P.
Presented by: Lauren Nicholas, Ph.D., M.P.P.,
Research Fellow, Institute for Social Research,
University of Michigan, 426 Thompson Street, Ann
Arbor, MI 48104, Phone: (734) 764-2562; Email: lnichola@isr.umich.edu
Research Objective: Pay-for-Performance (P4P) mechanisms have become increasingly common in health care despite limited empirical evidence of their effectiveness. Theoretical literature highlights the challenges in designing incentive-based compensation for multidimensional tasks such as health care delivery. Performance pay can improve overall quality of hospital care if incentives to achieve performance thresholds in measured quality domains spillover to treatment of patients with other types of hospitalizations. Alternatively, performance pay may adversely affect quality of care by causing hospitals to shift resources away from areas of care that do not contribute to performance measurement. This paper tests whether the Centers for Medicare and Medicaid
Services Premier P4P demonstration program caused changes in quality of care for surgical conditions which are not subject to incentive pay.
Study Design: I use a difference-in-difference estimator to compare patient outcomes at hospitals receiving performance pay to non-recipients before and after the experiment began. Use of a control group allows identification of the effect of performance pay separately from the impact of public reporting on quality of hospital care since all hospitals begin reporting quality measures during the study period. Quality is measured using riskadjusted mortality rates for several medical and surgical conditions validated by the Agency for
Healthcare Research and Quality.
Population Studied: All patients hospitalized for heart attacks, heart failure, pneumonia, abdominal aortic aneurysm repair, coronary artery bypass graft, craniotomy, esophageal resection, hip/knee replacement, and pancreatic resection in 43 P4P hospitals and 415 non-P4P hospitals in Florida and
New York between 2000-2006.
Principal Findings: Although P4P hospitals show larger gains in incentivized process of care measures for medical conditions than nonparticipant hospitals, these gains do not translate into improvements in risk-adjusted mortality rates. I find no evidence that incentivizing performance on a small subset of quality tasks improved quality on unincentivized aspects of surgical care.
Conclusion: The CMS P4P demonstration did not cause improvements in risk-adjusted inpatient mortality rates for either incentivized or unincentivized admissions.
Implications for Policy, Delivery or Practice:
Targeting incentives to a narrow area of performance does not appear to be an effective way to improve hospital quality broadly. Future demonstration projects should test different incentives to design a Medicare Pay-for-
Performance program that improves inpatient outcomes.
Funding Source(s): NIA
Theme: Quality and Efficiency: Policies and
Incentives
● The Effect of Food Stamp Receipt on Blood
Sugar Control for Elderly Diabetics
Lauren Nicholas, Ph.D., M.P.P.
Presented by: Lauren Nicholas, Ph.D., M.P.P.,
Research Fellow, Institute for Social Research,
University of Michigan, 426 Thompson Street, Ann
Arbor, MI 48104, Phone: (734) 764-2562; Email: lnichola@isr.umich.edu
Research Objective: Diabetes prevalence is rapidly increasing amongst older adults, raising concerns about the medical and financial burden of elderly diabetics, particularly to the publicly funded
Medicare and Medicaid systems. As food consumption plays a critical role in diabetes management, this study examines the effect of participation in United States Department of
Agriculture’s Food Stamp Program (FSP) on blood sugar and cholesterol management for elderly diabetics.
Study Design: This study use including newly collected physiological indicators of diabetes control and survey data from the Health and Retirement
Study (HRS) to assess the role of the FSP in meeting American Diabetes Association treatment goals for blood sugar (Hemoglobin A1c level) and
HDL cholesterol levels of elderly diabetics.
Propensity score matching and instrumental variables techniques are used to overcome nonrandom participation in the FSP.
Population Studied: 1,311 elderly diabetics participating in the 2006 wave of the Health and
Retirement Study for whom valid bloodspot data were available. Compared with non-recipient
diabetics, Food Stamp recipients are younger (66 vs. 69), more likely to be Black or Hispanic (56% vs.
29%), less educated (56% vs. 26% have less than a high school education) and more likely to report fair or poor health (74% vs. 41%).
Principal Findings: Relative to other diabetics in the HRS sample, food stamp recipients are less likely to meet treatment guidelines for blood sugar
(64% vs. 72%). There is no significant difference in cholesterol control (87% vs. 89%). Observed differences in blood sugar are largely explained by sociodemographic characteristics. In multivariate analysis, Food Stamp receipt improves glycemic control in the three month period prior to bloodspot collection, reducing measured Hemoglobin A1C levels by 0.62 (p = 0.05).
Conclusion: Food stamp receipt can contribute to improved glycemic control in elderly diabetics, potentially by improving dietary quality.
Implications for Policy, Delivery or Practice: The
USDA should take steps to increase FSP participation by the elderly and to improve participants’ nutritional choices as a way of combating the diabetes epidemic. Medicare and
Medicaid should consider innovative strategies to ensure access to high quality diets for elderly diabetics as a cost-effective way of reducing treatment costs for health consequences of poor glycemic control.
Funding Source(s): USDA
Theme: Prevention and Treatment of Chronic
Illness
● Recommended Cancer Screening Rates among California HMO & Non-HMO Enrollees
Gina Nicholson, M.P.H.
Presented by: Gina Nicholson, M.P.H., Research
Associate, University of California, Los Angeles
Center for Health Policy Research, 10960 Wilshire
Boulevard, Suite 1550, Los Angeles, CA 90024,
Email: g.nicholson@ucla.edu
Research Objective: Studies have previously suggested that members in health maintenance organizations (HMO) have more access and greater utilization of screening tests than their fee-forservice counterparts. This study aims to examine differences in recommended cancer screening rates among HMO and non-HMO insured individuals in
California using data from the 2007 California
Health Interview Survey (CHIS).
Study Design: Using self-reported data from the
2007 CHIS, the most recent CHIS data available, this study uses logistic regression models to examine differences in breast, cervical, and colorectal cancer screening rates among HMO and enrollees. Additionally, analyses were performed on
African-American and Hispanic subpopulations to examine any differences in HMO rates among racial groups.
Population Studied: Adult respondents from the
CHIS 2007 (n=31.363). Excluded in the study are elderly patients (65 years of age and older) and those without any form of insurance.
Principal Findings: After controlling for demograhpic and other factors, individuals recommended for colorectal screening (aged 50-80 in accordance with HEDIS guidelines) and enrolled in an HMO were more likely to have had a screening in the past 5 years than their non-HMO counterparts (OR 1.19, p = 0.004). There was no significant difference in screening rates among
HMO and non-HMO enrollees for either breast
(ages 40-64) or cervical (ages 21-64) cancer screening. Cervical cancer screening rates were over 90% in both groups. Neither African-American or Hispanic-specific analyses revealed significant differences in cancer screenings between HMO and non-HMO counterparts.
Conclusion: Colorectal cancer screening was more likely to have occurred among HMO enrollees. Neither breast or cervical cancer screening rates differed significantly between those in HMO and non-HMO settings. Cervical cancer screening rates were over 90% in both HMO and non-HMO groups.
Implications for Policy, Delivery or Practice:
These findings suggest that the HMO setting may provide similar screening rates, at least for breast and cervical cancer, as non-HMO settings. This runs contrary to studies suggesting HMO increases screening rates. Screening for older adults
(colorectal) cancer may still be done at higher rates in an HMO setting.
Theme: Coverage and Access
● Quality of Care & Pediatric Nonurgent
Emergency Department Utilization: A Qualitative
Analysis
Mark Nimmer, B.A.; David, Brousseau, M.D., M.S.;
Nichole Yunk, B.A.
Presented by: Mark Nimmer, B.A., Clinical
Research Assistant II, Pediatrics, Medical College of Wisconsin, 999 North 92nd Street, Milwaukee,
WI 53226, Phone: (414) 337-7184; Email: mnimmer@mcw.edu
Research Objective: Previous research has shown that parents often bring their children to the emergency department for nonurgent care because of primary care referrals and perceived emergency department convenience. We sought to explore, through in-depth narratives, additional and underlying motivations for parents to bring their children to the emergency department for nonurgent conditions.
Study Design: Purposeful sampling was used to select participants with both high and low parent reported quality of primary care. Two domains of care were analyzed: family-centeredness and timeliness. Parent narratives were obtained through
semi-structured interviews for all events that preceded the decision to seek emergency department care. Parents were also asked to describe their child’s physician practice and past experiences with the physician’s office. Transcribed interviews were coded with the aid of qualitative software and analyzed for patterns related to quality of primary care.
Population Studied: Parents of children presenting to the emergency department who were triaged at the lowest of five acuity levels were sampled based on high versus low parent reported quality of primary care. Twenty-five eligible parents agreed to be interviewed at a later date.
Principal Findings: All but three parents whose primary physicians scored high on familycenteredness called a medical professional and were referred to the emergency department. Five of the referrals were physician referrals, five were nurse referrals, and one was an office staff referral.
Primary physician expertise was a decision-making factor for five parents who reported low familycenteredness, with four of those five stating there was no reason to call because they were coming to the emergency department anyway. Many parents reported that long delays in the availability of a primary care visit were responsible for their emergency department visits, including a lack of calling for advice prior to seeking emergency department care. Nine visits occurred outside of regular office hours, but only three of these visits were by parents who reported low quality familycenteredness.
Conclusion: Families with physicians having high family-centeredness ratings seem to have different reasons for seeking emergency department care than those with low physician ratings. Primary care referrals and after hours care most often explained visits among parents reporting high familycenteredness while poor physician expertise occurred more often than any other theme among parents reporting low family-centeredness. Further exploration of the results revealed that primary care referrals often include not only physician referrals, but also nurse and office staff referrals.
Implications for Policy, Delivery or Practice: The reasons for nonurgent pediatric emergency department utilization vary by perceived quality of primary care. Emergency department referrals might be reduced by educating office staff on appropriate criterion for referral or having physicians screen referrals made by office staff.
Extra efforts by physicians to build and maintain trust during well-child visits may also lead more parents to contact their physicians prior to visiting the emergency department. Blocking more time for acute care appointments may also reduce nonurgent emergency visits for parents who reported scheduling frustrations. Regarding after hours care, a better explanation from the primary physician of why a situation is nonurgent may make parents with high family-centeredness ratings more willing to wait and schedule an appointment.
Funding Source(s): Agency for Healthcare
Research and Quality
Theme: Child Health
● General Surgery - A Medical Specialty in
Crisis? An Exploratory Case Study
Nurit Nirel, M.A.; Shlomo Birkenfeld, M.D.;
Jochanan Benbassat, M.D.
Presented by: Nurit Nirel, M.A., Senior
Researcher, Smokler Center for Health Policy
Research, Myers-JDC-Brookdale institute, JDC Hill,
P.O. Box 3886, Jerusalem, 91037, IL, Phone:
(972) 265-7420; Email: nuritn@jdc.org.il
Research Objective: To identify criteria for clinical specialties in workforce crisis, measures for the assessment of its extent, and propose solutions.
Study Design: To identify criteria for a "medical specialty in crisis", define measures of its extent, and list proposed solutions, we examined the case of general surgery (GS) and performed a qualitative analysis of the content of: (a) written testimonies submitted to a public commission, (b) interviews of key personnel in the Israeli and other health systems, and (c) open, semi-structured interviews with hospital directors and heads of GS departments. To translate the criteria defining a clinical specialty in crisis into quantitative measures we performed an analysis of administrative data from national databases. To compare the job characteristics and self-reported quality of general surgeons' working life with those of other specialists, we performed a re-analysis of data from an earlier national survey by one of the authors.
Population Studied: (a) Key personnel in the health system (b) 13 hospital directors and 17 heads of GS departments (c) A sample of 50% of the board certified specialists in GS, cardiology,
Ob/Gyn, ENT, ophthalmology and dermatology.
Altogether, we contacted by mail 1,410 physicians; of these, 890 (63%) responded and completed a structured questionnaire.
Principal Findings: We identified five criteria of a clinical specialty in crisis: shortage of suitable applicants for residency (male graduates of Israeli medical schools with high cognitive achievements), difficulties in filling vacant positions, high clinical workload, perceived low remuneration and, perceived poor quality of professional working life.
These measures only partially support the claim that GS is a medical specialty in crisis. Although pass rates at Stage I examinations declined among
GS residents, they remain higher than those in several other specialties. The proportion of foreign graduates in GS increased; but remains lower than in other specialties. The proportion of women increased; but the percentage of men completing
GS residency is still higher than in many other specialties. However, between 2000 and 2005, the
clinical workload (measured by the annual number of hospital admissions per staffed position) had increased, there were fewer candidates for every vacant GS specialty position then in some other specialties, and the ability of GS to attract male residents, Israeli graduates and applicants with higher cognitive skills had declined. Furthermore, the perceived remuneration and self-reported quality of working life were less then in other specialties. Proposed solutions included changes at the national and hospital level, within the specialty and its residency programs and in the medical undergraduate curriculum.
Conclusion: The study shows that it is possible to define criteria for a medical specialty in crisis and to quantify the level of crisis. Even though the measures used have their limitations, they partially support the problems we identified.
Implications for Policy, Delivery or Practice: The study findings could constitute the basis for establishing criteria for additional specialties in crisis, clarifying the extent of the crisis, and even identifying possible ways of solving it.
Funding Source(s): Israel National Institute for
Health Policy and Health Services Research
Theme: Health Care Workforce
● A Systematic Review of Maternal & Child
Health Outcomes for Community Health
Workers Interventions
Brett Nishikawa, M.D.; Meera Viswanathan, Ph.D.;
Jennifer Kraschnewski, M.D.; Laura Morgan, M.A.;
Patricia Thieda, M.A.; Dan Jonas, M.D., M.P.H.
Presented by: Brett Nishikawa, M.D., Chief
Resident, Preventive Medicine, University of North
Carolina at Chapel Hill, MacNider, CB #7240,
Chapel Hill, NC 27599, Phone: (919) 843-8077;
Email: brett.nishikawa@unc.edu
Research Objective: Community health workers
(CHWs) can serve as a bridge to the health care system for maternal and child health care needs.
We aimed to conduct a systematic review of the effectiveness of CHW interventions in the areas of maternal and child health.
Study Design: We conducted a systematic review using standard Evidence-based Practice Center methods. We searched Medline, the Cochrane
Database, and CINAHL from their inception through
October 2008 using a variety of terms for CHWs.
We included studies with a comparison group that were conducted in the United States, published in
English, with at least 40 participants. Two reviewers independently assessed each abstract and full-text article for inclusion, resolving disagreements by consensus. One reviewer abstracted data onto a standard form; a second reviewer checked for completeness and accuracy.
Trained reviewers abstracted data and assessed the methodologic quality of studies using predefined criteria from international standards.
Population Studied: Studies that primarily involved interventions targeted toward pregnant women and/or children were selected for the maternal and child health subset of this systematic review. Of 992 abstracts we identified, 17 studies meeting our inclusion criteria involved primarily maternal and/or child health. Of these, 12 focused exclusively on potentially vulnerable populations (e.g., racial or ethnic minorities, recent immigrants, low income families, inner city residents). Another three targeted families identified as high risk for child maltreatment. Pregnant women were part of the target population for 8 studies. One study each addressed pregnant women with phenylketonuria
(PKU), children with failure to thrive, and children with “chronic disease” (not otherwise characterized).
Principal Findings: Outcome measures assessed in the studies included adequacy of prenatal care, birth outcomes including incidence of low birth weight and very low birth weight, breastfeeding initiation and continuation, and presence of neonatal health problems. Child development was frequently considered among outcomes, e.g., cognitive development, language development, and physical growth. Many of the measured outcomes involved factors that could contribute to childhood health and well-being. Among these outcomes were environmental tobacco exposure, incidence of child maltreatment, parent-child interaction, assessments of appropriateness of home environment for child development, and parental stress or psychiatric well-being. Preliminary results indicate a wide range of effectiveness to CHW interventions without a clearly discernable pattern emerging to determine which interventions or population types would be most effectively served by CHWs.
Conclusion: Most studies involving CHWs for maternal and child health have been concerned with high-risk populations. For maternal and child health, the evidence on CHW interventions appears to be most consistent when addressing existing health conditions instead of potential conditions, i.e., primary prevention. A minority of the studies evaluated reported statistically significant results, suggesting a paucity of evidence from which to draw further conclusions.
Implications for Policy, Delivery or Practice: The evidence for effectiveness of community health workers in improving key health outcomes relating to maternal and child health such as prematurity, low birth weight, sustained breastfeeding, or child maltreatment remains unclear. The lack of such findings suggests that further research is needed to determine true benefit.
Funding Source(s): AHRQ
Theme: Disparities
● Integrating Screening & Intervention for
Substance Use into Routine Prenatal Care
James Nocon, M.D., J.D.
Presented by: James Nocon, M.D., J.D., Director,
Prenatal Recovery Program, Obstetrics &
Gynecology, Indiana University School of Medicine,
1001 West 10th Street, F5102, Indianapolis, IN
46202, Phone: (317) 630-2663; Email: jnocon@iupui.edu
Research Objective: To implement simple motivational processes designed to assist obstetrical providers to integrate treatment for substance use into routine prenatal care.
Study Design: This is a prospective cohort study of pregnant patients referred to a high risk clinic for substance use disorders. They were counselled at each prenatal visit by a physician using a simple motivational technique designed to lower the patient's resistance to treatment. In addition, each patient received routine prenatal care, and dietary and social service consultations. A urine drug screen was performed at each visit.
Population Studied: 287 patients were referred to an obstetrical clinic that treats substance use disorders in pregnancy who recived simple motivational empowerment therapy. These women were compared to a control group of 145 pregnant substance users seen in the same clinic with providers who did not use motivational treatment.
The control group was evaluated by the same nurisng, dietary and social service personel. The care took place in a county hospital with 3200 deleiveries per year, 78% of which were Medicaid funded. The population is culturally diverse: 45%
Black; 25% White; 30% Hispanic.
Principal Findings: Of the 287 patients treated for substance use, 244, or 85%, were free of drugs at delivery. In contrast only 49% of the control group were drug free. Follow-up revealed that 79% of those negative were still abstinent six months postpartum. The most commonly used drugs were
"crack" cocaine, marijuana and opiates. A major finding in the study group revealed that crack cocaine smokers who abstained prior to 24 weeks gestation had normal weight newborns free of any cocaine sequelae. Another principal finding was that newborns (62) of mothers treated with methadone for opiate dependence had a mean
LOS 28 days in a special care nursery for NAS compared to a LOS of 3.5 days for newborns whose mothers were treated with buprenorphine.
Many other significant findings were observed.
Conclusion: Motivational Empowerment treatment is easy to learn and implement and has a profound impact on the outcome of pregnant patients with substance use disorders including greater abstinence, less complications and substantially shorter hospital LOS for newborns.
Implications for Policy, Delivery or Practice: The
Indiana State Commission on Prenatal Substance
Abuse has adopted the author's approach to integrating substance use treatment into routine prenatal care. A DVD with various demonstration videos, outlines and treatment protocols is being developed by the Indiana Perinatal Network to be distributed to all obstetrical care providers in
Indiana.
Theme: Behavioral Health
● Neonatal Outcome in Mothers Treated for
Opiate Dependence with Buprenorphine
Compared to Methadone
James Nocon, M.D., J.D.; David Haas, M.D.
Presented by: James Nocon, M.D., J.D., Associate
Clinical Professor, Obstetrics & Gynecology,
Indiana University School of Medicine, 1001 West
10th Street, F5102, Indianapolis, IN 46260, Phone:
(317) 630-2663; Email: jnocon@iupui.edu
Research Objective: To compare neonatal outcomes in opiate dependent patients managed with buprenorphine or methadone throughout their respective pregnancies
Study Design: A cohort study comparing the treatment of neonatal abstinence syndrome as reflected in length of stay (LOS) between newborns of mothers treated with buprenorphine (19) or methadone (54). In addition, preterm delivery, birth weight and other neonatal complications were evaluated
Population Studied: Opiate dependent women referred for prenatal care in a clinic designed for treating substance use in pregnancy. Over 95% of the patients were White, multigravidas who conceived while undergoing opiate maintenance treatment with either buprenorphine or methadone.
Principal Findings: Newborns of buprenorphine treated mothers had a significantly lower length of hospital stay (mean 3.5 days) compared to methadone newborns (28.53 days). In addition, buprenorphine newborns had significantly higher birthweights (3073 grams) compared to methadone newborns (2766 grams). Buprenorphine treated patients had significanty lower preterm delivery rates (16%) compared with the methadone mothers
(34%). There were no untoward effects in patients or newborns of mothers who used either buprenorphine or a buprenorphine/naloxone combination. In fact, the trends substantially pointed to better tolerance to the medication, less preterm delivery and higher birthweights in the buprenorphine/naloxone combination.
Conclusion: Newborns of buprenorphine treated mothers have substantially better outcomes when compared to methadone newborns. They have significantly shorter LOS, higher birthweight, less preterm delivery and less complications. In addition, buprenorphine and buprenorphine/naloxone combinations are equally effective and safe alternatives to methdone.
Implications for Policy, Delivery or Practice: The evidence is becoming increasingly clear for the safety and substantially improved outcomes for newborns of buprenorphine treated mothers. In terms of LOS, with special care nursery costs in
excess of $2,000.00 per day, each buprenorphine treated mother reduces hospital costs by at least
$50,000.00, compared to a methodone treated patient. Moreover, the reduction in preterm delivery from 34% in methodone patients to 16% in buprenorphine patients will significantly improve infant morbididty and mortality and result in a substantial reduction in hospital LOS and infant care costs.
Theme: Behavioral Health
● Developing a Model of Recovery in Mental
Health
Sylvie Noiseux, Ph.D.
Presented by: Sylvie Noiseux, Ph.D., Professor,
Nursing Sciences, University of Montreal, C.P.
6128, Succursale Centre-ville, Montreal, H3C 3J7,
CA, Phone: (514) 343-6111 ext. 0211; Email: s.noiseux@videotron.ca
Research Objective: The aim of this research project is to put forward a theoretical model of the recovery process for people with a mental health problem. Over the past decade, recovery has become the guiding principle of the mental health system in Canada (Everett et al., 2003). Currently, the process elicits great enthusiasm, although the term assumes many different meanings depending on the context in which it is ued. In addition, the literature regularly refers to a “model” of recovery, despite the fact that no such model has actually been defined or implemented in Canada (Everett et al., 2003). One may therefore ask, “How can one offer recovery-oriented services and care to people with a mental health problem when the nature of the process remains obscure or is often confused with the concepts of remission, cure, readjustment, or even rehabilitation?”
Study Design: Considering the objective of this project and the state of knowledge regarding the recovery process, a qualitative research design is called for. More specifically, the grounded theory approach is deemed most suitable for developing a substantive theoretical model of recovery. Such an approach makes it possible to push beyond descriptive analysis by defining the conditions that influence recovery and relating them to each other in a model that comprises a refined, detailed proposal (Strauss & Corbin, 1990). Grounded theory recommends the use of a variety of data sources to better understand, delineate and explain the phenomenon under study.
Population Studied: The participants in the study will represent people with a mental health problem, one of their family members and one of their care providers. In all, it is planned to select 108 participants divided into four comparison groups
(n=36) each representing one of the four types of mental health problem under investigation
(schizophrenic, affective, anxiety, and borderline personality disorders).Each comparison group will comprise 9 units (one unit = one person with a mental health problem, one family member, and one care provider). They will be recruited from specialized institutions, self-help groups and community organizations in three sociodemographic regions. Data will be collected in semi-structured interviews, which will be recorded and transcribed in full. Data analysis will follow the three steps of the Strauss and Corbin model (1990): namely, open, axial and selective coding.
Principal Findings: Seven categories emerged from the analysis and conceptualisation: perceiving a mental health problem as a ‘descent into hell’; igniting a spark of hope; developing insight; activating the instinct to fight back; discovering keys to well-being; maintaining a constant equilibrium between internal and external forces; and, finally, seeing light at the end of the tunnel. Comparison of these categories led to their being consolidated into a core category in which recovery is defined as a
‘process involving intrinsic, non-linear progress that is primarily generated by the role as actor that the individual adopts to rebuild his or her sense of self and to manage the imbalance between internal and external forces with the objective of charting a path through the social world and regaining a sense of well-being on all biopsychosocial levels.’
Conclusion: The theoretical explanation of recovery that we have proposed is based on the inner attributes and resources of the individual and on his or her ultimate potential to make a recovery.
Finally, like the members of any professional discipline, research nurses and health experts have a social responsibility for the actions they undertake with people with mental health and are expected to contribute to developing knowledge of care practices that will facilitate recovery.Healthcare practice should focus on developing an innovative macro-social intervention programme that goes beyond the objectives recommended in standard rehabilitation programmes. There should additionally be in-depth examinations of already existing intervention programmes with more individualistic orientations. Above all, this study is an initial step in directing attention towards the potential of the individuals rather than focussing on the deficits associated with the disease. It is therefore important to target clinical nursing interventions at the individuals with mental health so that they can identify their own keys to wellbeing; identify points of reference from their past that will give them a new sense of self; identify sources of motivation despite their anguish; and restore themselves to the state they were in before the disease.
Implications for Policy, Delivery or Practice: The recovery process highlights the way people use their personal potential and resources to rebuild themselves despite an illness (Brookes, 2006).
While recognition of these resources is of central importance in the discipline of nursing, it does not seem to play an influential role in the research and
practice of nurses in mental health problem and today has not progressed beyond the level of topic of discussion. At the theoretical level, the major contribution of this research has been to provide a holistic perspective on the recovery of people living with mental health problem. On one hand, approaching the subject from the point of view of human response entails connecting every dimension of the individual. In this sense, we concur with Lauzon and Pepin (2002) who maintain,
"that it is now important for nurses to integrate concepts in their research that are based on the complete and indivisible nature of individuals and their experiences’’ (p.12). On the other hand, our theory of recovery provides for in-depth scrutiny of the characteristics of this human response: in our opinion, these can only be biological, psychological,and social in nature, a view supported by the participants’ beliefs. Our results demonstrate that people living with mental health problem disease and people close to them are cognisant of aspects of well-being in each of the biopsychosocial dimensions of their life.
Researchers examining the role of nursing in the mental-health field have not really documented how these dimensions or resources interact or influence people’s responses to situations of illness. Equally, on the level of practice, there is limited research about how to take advantage of an individual’s potential or activate the resources that are available to him or her. Our midrange theory addresses these shortcomings and should have a significant impact on nursing practice. It should also result in changes to the way mental-health practitioners approach people living with mental health problem and channel their thinking away from the patients’inadequacies. In short, these findings are a useful basis for deliberation on the revitalisation of care practices in the mental-health field.
Funding Source(s): NIMH
Theme: Behavioral Health
● Assessing Hospital Capacity in a Major Urban
Area
Bonnie Norton; Bruce Siegel, M.D., M.P.H.; Janice
Blanchard, M.D., Ph.D.; Marsha Regenstein, Ph.D.
Presented by: Bonnie Norton, Senior Research
Assistant, Department of Health Policy, The George
Washington University, 2021 K Street, NW, Suite
800, Washington, DC 20006, Phone: (202) 994-
8670; Email: blnorton@gwu.edu
Research Objective: Urban hospitals provide care to millions of Americans and often serve as an entry-point to the health care system for low-income patients, resulting in overwhelmed emergency departments (EDs). However, data to measure hospital capacity community-wide is generally unavailable. While existing tools provide valuable information on factors that could potentially influence patient flow and crowding in EDs, none includes all of the elements necessary for a community-wide assessment of hospitals. Our objective was to assess hospital capacity, ED crowding, and patient flow in all District of Columbia
(DC) hospitals using the newly developed hospital capacity assessment tool (HCAT).
Study Design: The HCAT was developed to measure system capacity and demand and to help identify the root causes of hospital crowding community-wide. Eighteen of the 57 HCAT questions were adapted from existing hospital surveys, while the remaining 39 were constructed for use in this assessment. Instrument domains include throughput measures, hospital data collection, factors contributing to ED crowding and hospital closure/diversion, and human capital issues. The HCAT was distributed both in person and electronically to the eight acute care hospitals in DC with active EDs, all of which returned the survey. To the researchers’ knowledge, this is the first survey to measure hospital capacity and patient flow at all hospitals in a major metropolitan area.
Population Studied: General acute-care hospitals in Washington, DC.
Principal Findings: Overall, DC hospitals indicate that their capacity exceeds supply by 36%.The median wait time to see a physician in DC hospitals is 90 minutes and the median length of time from
ED triage to ED discharge is 3.7 hours. The median nurse vacancy rate within DC EDs is 10%, and the median hospital-wide nurse vacancy rate is
13.6%. Hospitals ranked lack of critical care beds, overcrowded ED, lack of general acute care beds, and a shortage of nursing staff the highest among factors causing ED closure/diversion. Although six of the hospitals reported that they regularly collect throughput times, only three use these times to monitor flow and forecast ED demand. In addition, the hospitals use different measures to evaluate ED crowding.
Conclusion: While median DC throughput times exceed the national average, the current lack of standard measurement and forecasting limits the implementation of effective interventions to improve timeliness and quality of care. Effective patient flow requires the ability to match resources, such as staff and available beds, to demand. In the absence of national ED measures, hospitals may forego measurement or are forced to rely on their own measures with varying specifications.
Implications for Policy, Delivery or Practice:
With no end in sight to hospital crowding, it is essential to effectively manage demand. The first step in that process is to measure hospital capacity and demand and identify the primary issues contributing to crowding. Successfully field-tested in
DC, the HCAT could also be useful to the many other US communities grappling with the need to improve emergency care.
Funding Source(s): District of Columbia
Theme: Quality and Efficiency: Measurement
● Disruption to Disaster: Determinant of
Dilemma for Mental Health Care Consumers in
Texas
Gul Nowshad, M.B.B.S., M.P.H.; Sartaj Alam, M.S;
Scott Hickey, Ph.D.;
Presented by: Gul Nowshad, M.B.B.S., M.P.H.,
Senior Graduate Research Assistant, Management,
Policy & Community Health, School of Public
Health, University of Texas, Houston, 1200
Hermann Pressler, Houston, TX 77030, Phone:
(713) 500-9489; Email: Gul.Nowshad@uth.tmc.edu
Research Objective: This study examined the risk of harm (suicidal ideation) among mentally ill homeless patients receiving treatment at the Mental
Health Mental Retardation Association of Harris
County Houston (MHMRA). The study aim was to determine the characteristics of this vulnerable population, determinant of suicidal ideation and their mental health service needs.
Study Design: The study cohort includes 2,631 homeless mental health consumers at MHMRA between January 2004 and December 2005, followed for 2 years to ascertain whether they have been assessed according to Texas Recommended
Assessment Guidelines (TRAG), for self harm and involvement with criminal justice system. 1,527 homeless mental health consumers out of 2,631 had a second assessment for suicidal ideation within the past 90 days. The identification of these consumers and their booking history was obtained by matching all consumers served by MHMRA with
Harris County Justice Information Management
Systems (JIMS) booking records.
Population Studied: These MHMRA mental health consumers were 17 years and older, residents of
Harris County, homeless mentally ill who had been involved with criminal justice system.
Principal Findings: Over the two years period,
1,527(58%) of the study cohort had at least one time, thought about harming themselves, and 2393
(91%) had involvement with criminal justice system. 40 % of the cohort was female, 47%
African American and 46 % had a dual diagnosis.
Age, gender, education, marital status, lack of social support, clinical and criminal history was associated with suicidal ideation. Women were 36
% more likely to harm themselves than men after controlling for all socio-economic and demographic factors (OR 1.36; 95% CI 1.18 – 1.98), while married patients were 2 times more likely to consider harming themselves compared to divorced patients. Compared to patient who completed 9th grade or lesser education, those with advanced degrees were 40% more likely to consider suicide within the past 90 days (OR 1.39; 95% CI: 1.04-
1.87). Those who had a criminal history were also
40% more susceptible to harm themselves. (OR
1.40; 95% CI: 1.13- 1.74). Younger and depressed patients were 30% less likely to think about suicide.
(OR 0.65; 95% CI: 0.50- 0.84). Compared to schizophrenics, those who suffered from with other primary diagnoses were 3.75 times more likely thought about suicide. Compared to those with abundant help, patients who have very few actual or potential natural or community support were 3 times more to harm themselves while those with no support at all were 16 times more likely to hurt themselves.
Conclusion: This study helps us in understanding the vital roles mental health and social support can play in the lives of those who are at the verge of disaster.
Implications for Policy, Delivery or Practice:
These findings will enable mental health care researchers, judicial personnel and policy makers to understand the factors responsible for channeling this vulnerable population from disruption to disaster.
Funding Source(s): Houston Endowment Fund
Theme: Behavioral Health
● Challenges in Modeling Disease Progression for Economic Evaluation: Choice Between
Randomized Trial & Observational Data
Katia Noyes, Ph.D., M.P.H.; Alina Bajorska, M.S.;
Andre Chappel, B.A.; Steven Schwid, M.D.; Lahar
Mehta, M.D.; Andrew Dick, Ph.D.
Presented by: Katia Noyes, Ph.D., M.P.H.,
Associate Professor, Community & Preventive
Medicine, University of Rochester, 601 Elmwood
Avenue, Box 644, Rochester, NY 14642, Phone:
(585) 275-8467; Email: katia_noyes@urmc.rochester.edu
Research Objective: Cost-effectiveness analyses
(CEA) require comparison of outcomes in treated and untreated populations. Data from randomized clinical trials do not provide progression rates representative of the general population while treatment effects in observational data may be biased due to lack of randomization. We developed a novel approach for estimating untreated progression rates (controls) by using data from a population-based longitudinal study, correcting for the effects of patients’ treatments as reported by pivotal trials.
Study Design: Disease progression was characterized by disability-based disease states and by the presence of relapses. We modeled probabilities of disease state transitions using a first-order annual Markov model that adjusted for age, gender, disease duration, recent relapse rates, prior states, and the specific disease-modifying therapy (DMT), accounting for clustering and complex survey design. To estimate transitional probabilities, we developed an iterative multinomial logistic regression algorithm that constrained the effects of DMT to match those reported by RCTs as follows. We selected initial annual treatment factors and estimated first progression probabilities for controls. For those probabilities, using a numerical
algorithm, we found new treatment factors that resulted in the same risk ratios of progression as reported by the trials. The new factors were used in the regression model to adjust for DMT effects and to re-estimate the probabilities for controls. We continued this process iteratively, until the identified factors for the final control probabilities matched published DMT effects from RCTs.
Population Studied: We used a subsample
(n=844) of patients with relapsing remitting and secondary progressive multiple sclerosis (MS) from the 2000-2005 Sonya Slifka nationally representative longitudinal cohort of MS patients.
Principal Findings: After correcting for the treatment effects of DMTs and other observable risk factors, the probability of disability progression was greater for estimates based on all MS patients compared to the estimates based on untreated individuals only. The 95% confidence intervals obtained by bootstrapping using the entire cohort
(including treated and untreated individuals) were narrower than the intervals based on the subsample of untreated patients only.
Conclusion: Our results indicate that the untreated patients in our study had lower estimates of disease progression than the treated patients in the study would have had if they remained untreated. This suggests that patients who forgo treatment are likely to have milder, slower progressing forms of
MS.
Implications for Policy, Delivery or Practice:
Correcting for treatment effects in a broader, more inclusive group of patients likely provides a more realistic estimate of the natural history of disease progression than simply characterizing disease progression in an untreated cohort. The use of a broader population-based cohort also substantially improves the precision of the estimates.
Funding Source(s): Multiple Sclerosis Society
Theme: Quality and Efficiency: Measurement
● Cost-Effectiveness of Disease-Modifying
Therapy for Multiple Sclerosis: A Population-
Based Evaluation
Katia Noyes, Ph.D., M.P.H.; Alina Bajorska, M.S.;
Andre Chappel; Steven Schwid, M.D.; Lahar Mehta,
M.D.; Andrew Dick, Ph.D.
Presented by: Katia Noyes, Ph.D., M.P.H.,
Associate Professor, Community & Preventive
Medicine, University of Rochester, 601 Elmwood
Avenue, Box 644, Rochester, NY 14642, Phone:
(585) 275-8467; Email: katia_noyes@urmc.rochester.edu
Research Objective: Highly expensive disease modifying therapies (DMTs) were introduced in the
1990s to reduce the frequency of relapse and to slow disease progression in patients with multiple sclerosis (MS). While there is some evidence that these drugs slow MS progression and reduce the frequency of relapses, these therapies are characterized by uncomfortable and persistent side effects and high costs, representing great economic burden to patients, as well as public and private payers. This study estimates the cost-effectiveness
(CE) of DMT use for the US population with MS.
Study Design: To evaluate the CE of the following
DMTs: interferon beta-1a either given weekly
(Avonex) or three times a week (Rebif), interferon beta-1b given every other day (Betaseron) and glatiramer acetate (Copaxone) compared to therapy without DMT, we generated 10-year disease progression paths for an MS cohort (n=1121) with varying individual characteristics by employing firstorder Markov models to estimate transitional probabilities and logistic models to estimate the presence of relapses. Published estimates of DMT treatment effects were used to modify progressions for treated cohorts. To estimate the costs, we examined utilization of hospital and outpatient care, emergency department, physical therapy, mental health care services, home health and personal care, and priced each utilization item using
Medicare rates. The outcomes were measured as gains in quality-adjusted life years (QALY, measured via SF-6D) and relapse-free years, differences in the number of disease progressions
(as measured by disability status), and gains in years spent in lower disability states (without a walking aid or wheelchair). Monte Carlo (n=50) simulations, bootstrap resampling (n=250) methods, and sensitivity analyses were conducted to estimate expected total costs and effects, to characterize the statistical uncertainty of the parameters and to evaluate the robustness of the results to the study assumptions.
Population Studied: 2000-2005 population-based survey of MS patients from the Sonya Slifka
Longitudinal Multiple Sclerosis Study
Principal Findings: Using MS DMT for 10 years resulted in significant health gains. The choice of the optimal therapy depends on the outcome, with
Avonex generating the highest QALY gain (0.187
QALY), fewer disease progressions (by 0.91) compared to no DMT, and fewer years spent with walking disability (by 0.81 year), and with Rebif leading to more relapse-free years (by 1.12 year).
The basecase CE of all DMTs exceeded
$1,000,000/QALY, with Copaxone being the most cost-ineffective ($5,209,524/QALY). As the cost of
DMT decreases, DMTs become more costeffective.
Conclusion: While the current practice of recommending DMT for any patient with relapsing or secondary progressive MS results in substantial health gains, these gains, due to the high price of the DMTs, come at a very high societal and patient costs, rendering the incremental cost-effectiveness ratios of each of the DMTs far above currently accepted standards.
Implications for Policy, Delivery or Practice:
This advocates for changes in the current treatment
guidelines and wider acceptance of economic evaluations for health policy decision making.
Funding Source(s): Multiple Sclerosis Society
Theme: Public Health
● How much does Continuity Matter for High-
Quality Diabetic Care?
David Nyweide, M.S.
Presented by: David Nyweide, M.S., Ph.D.
Candidate, The Center for Health Policy Research,
The Dartmouth Institute for Health Policy & Clinical
Practice, 5 North Park Street, Hanover, NH 03755,
Phone: (603) 643-4056; Email: nyweide@dartmouth.edu
Research Objective: To determine whether higher continuity is associated with higher technical quality of care for diabetic older adults.
Study Design: Each patient’s care pattern over the course of a year was used to calculate a Continuity of Care (COC) score, which measures the concentration of ambulatory care visits across physicians. Diabetic quality indicators endorsed by the National Committee for Quality Assurance were used as metrics of technical quality: hemoglobin
A1c test, lipid test, and eye exam as well as a comprehensive measure encompassing all three processes. Patient-level covariates were included in multiple logistic regression models to test the likelihood of receiving each quality indicator. The significance level was set at P < 0.01.
Population Studied: Fee-for-service Medicare beneficiaries 65 to 75 years old from the 2005 20 percent Part B sample with a diabetes diagnosis and at least 4 evaluation and management visits (n
= 373,537). Restricting to at least 4 visits was mathematically necessary to compute the COC score and avoided ascribing high COC scores to beneficiaries with only a few visits to a single physician.
Principal Findings: After adjustment for patient age, sex, race, Charlson comorbidity score, and mean ZIP Code retiree Social Security income, likelihood of receipt was significantly associated with higher continuity for each indicator (P < 0.001): hemoglobin A1c test (OR 1.05, 95% CI 1.02—1.08), lipid test (OR 1.14, 95% CI 1.11—1.17), eye exam
(OR 2.46, 95% CI 2.40—2.53), and comprehensive quality metric (OR 1.74, 95% CI 1.69—1.78).
Conclusion: Diabetic older adults receive higherquality care when their care patterns exhibit higher continuity.
Implications for Policy, Delivery or Practice:
Policies that reduce fragmented care patterns should be pursued as a means to improve quality for older diabetic patients. For example, the medical home may be a helpful way to ensure high continuity. The COC score may also be a useful performance metric for physicians or organizations intent on improving the delivery of care for the chronically ill.
Funding Source(s): NIA
Theme: Prevention and Treatment of Chronic
Illness
● The Use of Performance Based Dashboards to
Address Cancer Healthcare Disparities in the
NCI Community Cancer Centers Pilot Program
(NCCCP)
Donna O'Brien, M.H.A.; Maureen Johnson, Ph.D.;
Kenneth Chu, Ph.D.; Mary Fennell, Ph.D.; Deb
Hood, M.B.A.; Nora Katurakes, M.S.
Presented by: Donna O'Brien, M.H.A., President,
Community Healthcare Strategies, LLC, 300 Park
Avenue, Suite 1700, New York, NY 11530, Phone:
(212) 705-4268; Email: donnamobrien@aol.com
Research Objective: To examine the effectiveness of a performance-based dashboard as a management tool to improve the performance of community hospitals in providing integrated state of the art cancer care and research to underserved populations across the six focus areas of the
NCCCP program: clinical trials, quality of care, survivorship/palliative care, biospecimens, information technology/electronic medical records, and healthcare disparities.
Study Design: With forty percent of the NCCCP program funding allocated to healthcare disparities, a cross-cutting infrastructure and a “real time” communication and tracking system was needed to support the integration of the various program components as they relate to meeting the needs of underserved populations for the respective pilot sites and for the program overall. The sites represent great heterogeneity and serve different hard to reach subpopulations. The pilot program has utilized site specific workplans and customized goal setting but there was a need for standardized documentation for cross site comparisons. Through a joint planning effort of the NCI and NCCCP sites, a dashboard and disparities workplan was created with metrics for each focus area (e.g. minority patient accrual) in order to track progress.
Population Studied: The NCI Community Cancer
Centers Program is a public private partnership involving 16 hospital sites in 14 states representing approximately 27,000 new cancer cases each year. For the NCCCP, the populations affected by healthcare disparities are defined as racial and ethnic minorities, and other underserved populations, including residents of rural areas, women, children, the elderly, persons with disabilities, the uninsured, underinsured and those who are socioeconomically disadvantaged. The pilot sites serve different subpopulations which include African American, Hispanic, Asian, and
Native Americans.
Principal Findings: The dashboard, as a management tool, has enabled the NCCCP participants at the site level and at the NCI to: plan and manage a cross-cutting initiative to address
cancer healthcare disparities; build the skills of the community hospital sites to track efforts to improve their performance in addressing healthcare disparities; and enhance the understanding of the
NCI in its development of effective metrics to track healthcare disparities efforts in community-based settings.
Conclusion: The NCCCP program participants have a broader understanding of how to plan and implement a cross-cutting initiative for improving the quality of care and research for populations experiencing healthcare disparities, how to focus these efforts with more complete data collection and metrics, and how best to draw upon the expertise and technical assistance of the NCI, and the successful healthcare disparities programs of other network members.
Implications for Policy, Delivery or Practice: To advance cancer care and research in community settings where most cancer patients receive their care, cross cutting approaches for addressing areas such as disparities, quality of care and information technology are needed and require input from a range of NCI technical experts and leadership from the delivery system. The dashboard, as a management tool developed through a public private partnership of the NCI with a network of community-based hospitals, provides an evidence based approach to facilitate the integration of various program components relevant to improving health care and research within a community context.
Funding Source(s): NCI
● Understanding Racial & Ethnic Differences in the Prevalence of Overweight & Obesity in
Mississippi
Sule Ochai, Ph.D.
Presented by: Sule Ochai, Ph.D., Assistant
Professor of Health Policy & Management, Health
Policy & Management, Jackson State University,
350 West Woodrow Wilson Drive, Suite 2301-A,
Jackson, MS 39213, Phone: (601) 979-8830;
Email: sule.ochai@jsums.edu
Research Objective: The prevalence of overweight and obesity (OWO) has reached an epidemic proportion in the state of Mississippi. A recent CDC report has named the state as the most obese.
While the state’s 2007 prevalence rate stands at
32%, the rate between black and white is alarmingly different with 42.8% for black and only 27.6% for white. The epidemic presents the state with negative social, economic and public health consequences. Yet understanding the factors that explain racial and ethnic disparities in OWO prevalence in Mississippi is germane to the development of culturally competent micro-level policies and intervention programs to tame the epidemic. This study describes racial differences in
OWO in Mississippi, and identifies factors that explain those differences.
Study Design: Data came from 2007 Mississippi
Behavioral Risk Factor Surveillance System
(BRFSS). BRFSS is an annual telephone survey of people 18 years and older conducted by State health departments in collaboration with CDC. The
2007 survey collected self-reported information on height and weight, demographic characteristics, health risk behaviors, and other socioeconomics factors from 7,818 randomly selected respondents.
In this study, OWO is measured as Body Mass
Index (BMI) greater than 25.00 kg/m2. BMI is weight measured in kilograms divided by height measured in meters squared.
BMI was recoded as a binary variable, and serves as the dependent variable. Race/ethnic group was classified as black and white. Gender was coded as male or female. All analyses were conducted with
Stata version 10 and took into account the complex survey design and unequal probabilities of sample selection in BRFSS. Six logistic regression models were estimated. First, index of OWO was regressed on only race/ethnicity variable (uncontrolled model).
Subsequent model estimations were controlled for differences in demographics, SES, Food and
Lifestyle choices. In addition, separate models were estimated for black, white, male and female subsamples. The “complete” model is a logistic regression of BMI >=25 on demographics, socioeconomic, food, health and lifestyle choices of the respondents.
Population Studied: Randomly Selected adults, 18 years and older
Principal Findings: Controlling for demographics and SES, black has a highly significant higher odds of OWO relative to white (OR=2.28, P<.001).
Similarly, the probability of male being OWO is over
70% higher than female (OR=1.79, P<.001).
Married adults are more likely to be OWO
(OR=1.46, P<.001). Compared to those in West
Central Public Health District # 5, adults in Districts
# 1, 2, 3, 6, 7 and 8 are more likely to be OWO.
After controlling for Food, Health and Lifestyle choices, racial and ethnic differences attenuated
(OR=1.94, p<.oo1).
Conclusion: Study highlights the need for health policymakers to develop OWO intervention strategies that take cognizant of racial/ethnic differences in Mississippi.
Implications for Policy, Delivery or Practice:
Factors that significantly predict OWO vary by race and gender. This calls for a multi-faceted policy and intervention approach that takes into account these differences.
Theme: Obesity Prevention and Treatment
● Factors that Explain Racial Disparities in
Restrictive Placement of Children in Special
Education & Mental Health Services
Sule Ochai, Ph.D.
Presented by: Sule Ochai, Ph.D., Health Policy &
Management, Jackson State University, 350 West
Woodrow Wilson Drive, Suite 2301-A, Jackson, MS
39213, Phone: (601) 979-8830; Email: sule.ochai@jsums.edu
Research Objective: Two main objectives were central to this study--1) evaluate factors that predict children’s placement in restrictive settings of special education (SPED) and mental health services
(MHS), and 2) examine whether restrictive placement predictors in SPED and MHS differ by race.
Study Design: Multivariate logistic regression models were used to analyze a subset of data on
1,199 children collected as part of the Fast Track
Project, a longitudinal study of children at risk for emotional and/or behavioral problems conducted in four locations: Durham, North Carolina; Nashville,
Tennessee; rural Pennsylvania; and Seattle,
Washington. Two categorical dependent variables were constructed to indicate level of placements in
SPED and MHS. For SPED placement, students that received their services in regular classrooms assumed a categorical value of 0 (least restrictive) while those that had their placement in separate or resource rooms took the value of 1 (most restrictive). For MHS placement, children that obtained past-year MHS from any outpatient services sectors were given a categorical value of 0
(least restrictive) while those that secured their mental health services from any inpatient services sectors took the value of 1 (most restrictive).
Population Studied: 1,199 children at the risk of emotional and behavioral problems.
Principal Findings: The relative risk of placing children in a restrictive SPED setting is higher for children whose mothers have higher level of education (OR=1.37, p<0.001) and come from households with other adults present (OR=1.31, p=0.01). The presence of biological father decreases the relative risk of children’s placement in self-contained classrooms (OR=0.68, p=0.02).
The relative risk of black children receiving SPED in self-contained classrooms compared to white children is highly significant (OR=1. 90, p<0.001).
Teachers’ report of a child’s inattention problem, diagnoses of mental retardation, learning disabilities and emotional and behavioral problems are significant predictors of a child’s relative odds of placement in self-contained classrooms for SPED.
Students with higher level of intelligence are relatively less likely to be placed in the restrictive inpatient MHS setting as opposed to the outpatient setting (OR=0.93, p<0.001). Students exhibiting greater oppositional and aggressive behavior have a significant greater risk of experiencing inpatient admission (OR=3. 53, p<0.001). White children who are diagnosed with emotional and behavioral problems are most likely to be placed in an inpatient mental healthcare setting compared to the outpatient sector.
Conclusion: Racial and ethnic disparity persists in the “full” model in the context of restrictive SPED placement but remains insignificant in the context of inpatient MHS placement; suggesting that other variables not fully captured in the present study may yet underlie the observed differences in black and white children’s level of placement in the two service sectors examined. Furthermore, the variables that predict children’s placement in the restrictive settings of SPED and MHS vary by race and ethnicity.
Implications for Policy, Delivery or Practice: The understanding that children who are placed at certain levels in SPED are more likely to have access to and use MHS calls for inter-agency cooperation in service delivery to behaviorally and emotionally challenged children.
● Innovations in Interdisciplinary Work
Redesigns in the U.S. & Canada: Impacts on
Patient Safety & Quality of Worklife
Patricia O'Connor, M.S.N.; Maureen Bisognano,
M.S.N.; Ross Baker, Ph.D.; Judith Ritchie, Ph.D.;
Christine Covell
Presented by: Patricia O'Connor, M.S.N.,
Associate Director of Nursing, Neurosciences; CF
Harkness Fellow, Nursing, McGill University Health
Centre, 3801 University Street, Montreal, Quebec,
H3A 2B4, CA, Phone: (514) 398-1962; Email: patty.oconnor@muhc.mcgill.ca
Research Objective: Severe staffing shortages are significant barriers to system performance in both the U.S. and Canada, and have been clearly linked to decreased access to care, increased adverse events, increased morbidity and mortality. This study’s aims were to: Identify four cases of interdisciplinary work redesign innovations that were implemented in U.S. and Canada to address shortages in human resources and have resulted in improved patient outcomes and better working environments for staff. 1. Identify the key contextual elements that led to and supported the implementation processes. 2. Describe the strategic processes including policy changes, needed to create more nimble and responsive service delivery solutions. 3. Determine how long the innovation has been in place; if and to what extent spread of the innovation has occurred? 5. Identify the outcomes associated with the innovation.
Study Design: An explanatory case study design using qualitative methods (semi-structured individual and group interviews, and document and report reviews) was used. Four organizations known to have innovated were studied during site visits of 2.5 days. Detailed case study narratives provided rich details of the context or why the work redesign was begun, the change processes used, and the nature of the innovation itself (the what of the strategic change).
Population Studied: Four organizational cases of work redesign innovations were selected on the basis of a nomination process from experts in healthcare. Units of analysis selected for the study included persons directly or indirectly involved in the inception, implementation and evaluation of the innovation at the micro, meso and macro levels within or external to the organization. Key stakeholders included executive leaders, formal and informal leaders, clinicians/ disciplines engaged in the actual service delivery of the innovation, committee groups responsible for the innovation oversight, quality department, unions, human resources and finance.
Principal Findings: Interdisciplinary work redesign innovations were achieved in four organizations that varied in size and complexity, providing rich accounts of the key contextual and strategic change processes needed for successful adoption and sustainability. Buy-in and leadership from all levels within the organization, executive stewardship, use of champions, broad stakeholder engagement, regular audit and feedback on both processes and outcomes (clinical, worklife, organizational) were key. Multiple barriers to change were tackled, and constant, multi-level facilitation was evident. The multiple outcomes, positive and negative, on safety, quality, worklife will be described in detail, as will human resource, financial, and physical environment impacts.
Conclusion: Organizations who are highly focused, with clear alignment of goals at all levels within organization, and prepared to think outside the box, can successfully transition through severe staffing shortages. Multiple improvements were achieved.
Implications for Policy, Delivery or Practice:
This study focuses on how organizations “make it happen”, and can serve to guide health system executives, managers and policy makers. It provides useful information on: a) types of work redesign innovations that can address severe human resource issues; b) leadership and other contextual factors required for the emergence and sustainability of such innovations; c) policy changes needed to encourage re-design to create more nimble and responsive service delivery solutions; d) direction for a future research agenda.
Funding Source(s): CWF
Theme: Health Care Workforce
● Stroke Case-Fatality & Time of Admission:
Investigating the Influence of Weekend, Day of
Week, Shift & Hour of Admission
Uzor Ogbu, M.B.B.S., M.S.; Gert Westert, Ph.D.;
Laurentius Slobbe, M.S.; Karien Stronks, Ph.D.;
Onyebuchi Arah, Ph.D.
Presented by: Uzor Ogbu, M.B.B.S., M.S., Ph.D.
Researcher, Social Medicine, Academic Medical
Center, University of Amsterdam, Meibergdreef 9,
Amsterdam, 1105AZ, NL, Phone: +31
(0)205664511; Email: u.c.ogbu@amc.uva.nl
Research Objective: Hospital admissions occurring during the weekend, nights or holidays have been associated with higher short-term casefatalities of emergency conditions. The urgent nature of these admissions may contribute to but does not fully explain the higher case-fatality.
Ischemic stroke represents a treatable emergency condition in which the speed and appropriateness of treatment is predictive of survival and subsequent level of functioning. Using data from the
Netherlands, we examined the influence of the time of admission on short-term case-fatality among ischemic stroke patients.
Study Design: We analyzed datasets linking hospital admissions and discharge data to the population register obtained from Statistics
Netherlands. Using data from 114 hospitals, representing more than 90% of all Dutch hospitals, we identified patients discharged in the years 2000 to 2004 with a diagnosis of ischemic stroke, based on the International Classification of Diseases,
Ninth Revision, Clinical Modification codes. We used multilevel logistic regression to examine the associations between time of admission and sevenday case-fatality among ischemic stroke patients, after taking into account both patient-level and hospital-level covariates. These covariates included age, gender, Charlson comorbidity score, urgency of admission, socioeconomic status, ethnicity, specialty of attending physician, insurance type, and hospital location and teaching status. We defined time of admission in five different ways: weekend versus weekday; the day of admission; weekend/weekday shifts; shifts by specific day of admission (for example, Sunday night-shift versus
Monday day-shift); and hour of admission.
Population Studied: We studied 58,352 ischemic stroke patients aged 50 to 100, discharged from
114 hospitals in the Netherlands from 2000 to 2004.
Principal Findings: Compared to weekdays, weekend admissions were associated with higher seven-day case-fatality: odds ratio (OR) 1.33 (95%
CI: 1.25 – 1.41). Sunday admissions had the highest case-fatality of any day (OR 1.37, 95% CI:
1.24 – 1.52), when compared to Monday. Patients admitted during weekday night-shift had the highest case-fatality among all such shifts (OR 1.86, 95%
CI: 1.69 – 2.04), when they are compared with those admitted during the weekday day-shift.
Among the specific days and shifts, Wednesday night-shift admissions experienced the highest case fatality (OR 1.97, 95% CI: 1.60 – 2.44). Based on the hour of admission, those admissions between midnight and 6:00am had the highest odds of dying
(OR ranging from 1.49 at midnight to 2.31 at
5:00am) while those admitted between 3:00pm and
5:00pm had the least odds of dying with ORs of
0.75 (0.61 – 0.94), 0.73 (0.59 – 0.91), and 0.74
(0.59 – 0.93) at 3:00pm, 4:00pm and 5:00pm respectively.
Conclusion: Weekends represent a high-risk fatality period for ischemic stroke patients. However this weekend /weekday dichotomy hides the higher case-fatality occurring during weekday nights, in particular the Wednesday night-shift or between midnight and 6:00am. The decreased odds of dying observed around the doctors’ shift changeover period (3:00pm to 5:00pm) may represent the benefit of temporarily increased staff numbers.
Implications for Policy, Delivery or Practice:
Short-term in-hospital survival of ischemic stroke patients might be improved by examining possible staffing and other organizational shortcomings during weekday nights, weekends (in particular
Sundays), and early morning hours.
Funding Source(s): Dutch National Institute of
Public Health and the Environment (RIVM)
Theme: Organizational Performance and
Management
● Is Acculturation Associated with Use of
Foreign Medical Care?
Victoria Ojeda, Ph.D., M.P.H.
Presented by: Victoria Ojeda, Ph.D., M.P.H.,
Assistant Professor, Medicine, University of
California, San Diego School of Medicine, 1011
South Torrey Pines Road- MC 0507, Phone: (858)
822-6165; Email: vojeda@ucsd.edu
Research Objective: While outcomes for specific measures vary, disparities in health status and access to care are prevalent among racial/ethnic minorities (including Latinos), linguistically isolated individuals, and the uninsured. U.S. residents of states that border Mexico may opt to pursue care in
Mexico, also known as medical tourism. This study is guided by the following research questions: (1)
To what extent do California Latinos and their subgroups rely on foreign medical care (FMC)? and
(2) Are acculturation-related factors associated with
Latinos’ use of FMC?
Study Design: We analyze the 2001 random-digit dial survey, the California Health Interview Survey
(CHIS). The study follows a cross-sectional design.
Descriptive analyses provide a prevalence estimate of FMC use among Latinos by selected sociodemographic characteristics. We will construct multivariate logistic regressions to examine associations between the three dependent variables and the key independent variables of interest (i.e., language proficiency, citizenship status, years in the U.S. and generational status, and potentially others), while controlling for other covariates.
Population Studied: The adult survey samples persons ages =18 from major race/ethnic groups and subgroups. Participants are interviewed in
English, Spanish, and several Asian languages.
CHIS 2001 sampled 57,848 California adults ages
18+, including 11,840 self-identified Latinos (i.e.,
8304 Mexicans, 1019 Central Americans, 180
Puerto Ricans, 275 South Americans, 788 other
Latinos, and 1155 2+Latino types)
Principal Findings: Preliminary findings indicate that 3% of California adults ages 18-64 consumed foreign medical or dental care, and 3% purchased prescription drugs outside the U.S., translating to more than 605, 000 persons for medical/dental care and 632,000 persons for prescription drugs. Of those who obtained foreign medical/dental care,
72% were of Latino origin. Among those who purchased foreign prescription drugs, 47% were of
Latino descent and 38% were non-Latino whites.
Among those pursuing FMC, 26% spoke Spanish primarily at home and 12% spoke English; among those who consumed both medical and dental care abroad 32% spoke Spanish. Among those who bought pharmaceuticals outside the U.S., 35% spoke English and 16% spoke Spanish. Patterns of use of FMC or drugs by citizenship status varied: noncitizens were more likely to seek foreign medical or dental care (57% vs. 43% among citizens) while U.S.-born persons were more likely to buy prescription drugs outside the U.S. than noncitizens (51.3% vs. 31%). Regression analyses will allow us to determine whether these and other relationships persist after controlling for other socioeconomic and labor market factors (i.e., income, health insurance coverage).
Conclusion: Latinos, the uninsured, and English
Limited persons appear vulnerable to use of FMC.
Implications for Policy, Delivery or Practice:
This study will inform efforts to reduce cultural and linguistic disparities in health service use and efforts designed to expand health care coverage to uninsured populations, particularly in this time of exceedingly constrained financial resources at all levels of government. The consumption of foreign pharmaceuticals by privately insured persons may also signal the need to more closely examine policies around pharmaceutical co-pays and pharmacy-related consumer education. Findings also have implications for the continued development of binational health insurance products, since in California, 18.8% of the population or 6.5 million persons lacked health insurance coverage (average of 2003-2005 data).
Funding Source(s): RWJF
Theme: Disparities
● Price & the Marginal Propensity to Waive
Coverage
Edward Okeke, M.D., M.P.H.; Richard Hirth, Ph.D.;
Kyle Grazier, Ph.D.
Presented by: Edward Okeke, M.D., M.P.H.,
Health Management & Policy, University of
Michigan, 109 South Observatory, Ann Arbor, MI
48109, Phone: (734) 332-6288; Email: blkwrt@umich.edu
Research Objective: We revisit the question of price elasticity of ESI take-up by directly examining
changes in take-up of employer-offered insurance at a large public University in response to exogenous changes in premium contributions.
Study Design: In this study our price variation derives from a natural experiment. Between 2003 and 2004, the University changed how its health plan offerings were priced to employees by changing the employee contribution formula. This had the effect of raising employee out-of-pocket
(OOP) contributions for some employees and lowering it for others. On average, OOP contributions increased by about 54 percent. To identify the effect of the price shocks on the propensity to waive coverage, we estimated a variety of panel and cross-sectional probit models including a rich set of employee characteristics as controls.
Population Studied: The population studied included more than 20,000 active employees eligible for health insurance coverage between
2001 and 2004.
Principal Findings: We found that on average, a
10 percent increase in OOP premium costs increased the probability of dropping coverage by
0.7 percent. More importantly we found that there were heterogeneous impacts. Married workers were much more price-sensitive than single employees, in fact we found no statistically significant effect of price for unmarried employees. We also found evidence that lower-paid workers were disproportionately more likely to drop coverage relative to higher-paid workers. Elasticity estimates for employees below the 25th percentile of salary distribution in our sample were nearly twice the average.
Conclusion: Overall we find evidence that the decision to take-up ESI is responsive to price. On average we estimate the elasticity to be around -
0.07. This is a modest effect implying that within the range of premiums that we observe, doubling the
OOP premiums will only result in a 7 percent decline in take-up of ESI. We also show that there are heterogeneous effects across groups. In addition we provide some evidence that suggests that the employees dropping coverage are probably taking up coverage elsewhere and not becoming uninsured.
Implications for Policy, Delivery or Practice: Our results have several policy implications: they provide evidence that at least part of the recent increase in the percentage of workers declining ESI is as a result of price increases, thus complementing previous work. They also provide suggestive evidence that the marginal worker who drops coverage, at least in a large employer like the one we studied here, is probably taking it up somewhere else. Finally, we show that lower-paid workers are the most sensitive to increases in OOP premium contributions suggesting that particular attention be paid to this group of employees when changes in premium contributions are being contemplated. Future research could focus on the likelihood that lower paid workers within a large, diverse firm are more likely to drop coverage rather than obtaining coverage from another source.
Theme: Consumer Choices in Health Care
● Psychometric Evaluation of an Instrument to
Assess Patient-Reported “Psychosocial Care by
Physicians:" A Structural Equation Modelling
Approach
Oliver Ommen, M.D., M.P.H., Melanie Neumann,
Ph.D.; Nicole Ernstmann, Ph.D.; Markus Wirtz;
Christian Janssen, Ph.D.; Holger Pfaff
Presented by: Oliver Ommen, M.D., M.P.H.,
Scientific Coordinator, Center for Health Services
Research Cologne, University Hospital of Cologne,
Germany, Eupener Straße 129, Cologne DE
50933, Phone: +49(0) 221-478-97112 Email: oliver.ommen@uk-koeln.de
Research Objective: The objective of our study was to develop a theory-based and empiricallytested instrument for measuring patient-reported
“psychosocial care by physicians” (PCP index) for practical and economic application. We propose a model integrating patients’ perceptions regarding 1) supportive behavior of physicians, 2) physicians´ empathy, 3) physicians´ informational behavior, and
4) shared-decision-making behavior of physicians.
Study Design: Data were gathered during 2001 within a cross-sectional, retrospective mail survey.
Specific scales of the Cologne-Patient-
Questionnaire were used. A two-step structural equation model (SEM) procedure was applied. In the first SEM, all items were modelled as indicators of the intended underlying latent construct
“psychosocial care by physicians”. In the second
SEM criterion-related validity of the intended construct was tested in regard to patients´
"satisfaction", “trust in physicians,” and the "image of the hospital.”
Population Studied: 4,192 inpatients of 6 German hospitals.
Principal Findings: Results confirmed that the aspects of psychosocial care provided by physicians measured by the scale items are indeed indicators of the same construct. Furthermore indicator reliabilities and selectivities revealed that the content of all 13 items was highly representative of the underlying construct. The second SEM showed that "PCP" predicted patients´
“satisfaction," "trust in physicians," and "hospitalimage" in a significant and relevant manner.
Conclusion: To establish and maintain a trusting and satisfying physician-patient relationship, it is important for the physician to consider several points. First, the physician has to behave in an empathic way. Second, the physician has to provide patients with comprehensive and extensive information about their current state of health, the course of their therapy, and their treatment prospects. Third, the physician has to actively
involve patients in the treatment process. The psychosocial care provided by physicians is, therefore, first and foremost a communicative and interactive service that takes place on both the relationship and informational levels.
Implications for Policy, Delivery or Practice:
Based on our instrument’s reported psychometric characteristics and on initial validity indicators, it may be regarded as an adequate measure for further use in outcome and intervention research, and as a quality indicator for the physician-patient relationship.
Funding Source(s): Ernst- und Berta-Grimmke-
Foundation (Application-Number: 98130)
● Effectiveness of Smoking Cessation Advice by Primary Care Providers for Smokers with
Comorbid Alcohol, Drug, or Mental Disorders:
An Instrumental Variable Analysis
Michael Ong, M.D., Ph.D.; Qiong Zhou, M.S.; Hai-
Yen Sung, Ph.D.
Presented by: Michael Ong, M.D., Ph.D., Assistant
Professor, Medicine, University of California, Los
Angeles, 911 Broxton Avenue, 1st Floor, Los
Angeles, CA 90024, Phone: (310) 794-0154;
Email: michael.ong@ucla.edu
Research Objective: Individuals with comorbid alcohol, drug, or mental (ADM) disorders combined make up over 40% of all smokers in the U.S.
Primary care providers play an important role in smoking cessation counseling, but their effectiveness with this population is not well known.
The purpose of this study was to determine the effectiveness of smoking cessation counseling by primary care providers for smokers with ADM disorders.
Study Design: We conducted multivariate probit regressions that examined the relationship between past year primary care provider smoking cessation counseling and current smoking status in 2000-
2001. Covariates included age, gender, ethnicity, region of residence, education, household income, marital status, employment status, health insurance, body mass index, and physical activity.
Past year primary care provider exercise counseling was also used as an instrumental variable for past year primary care provider smoking cessation counseling to account for potential reverse causality between smoking status and receipt of smoking cessation counseling.
Population Studied: We examined 1,356 adults who reported that they were current smokers in the
1998-1999 Community Tracking Study survey, and who responded in the follow-up 2000-2001
Healthcare for Communities survey that they had seen a primary care provider in the past year. Past year ADM disorders were defined by self report of disorder and symptoms from the 2000-2001 survey.
Principal Findings: Among study individuals,
45.6% (n = 619) had ADM disorders. Those with and without ADM disorders were equally likely to receive smoking cessation counseling (72.9% and
69.8%, respectively). Analyses of both all individuals and only individuals with ADM disorders found that smoking cessation counseling by primary care providers had a positive significant association with current smoking status (coefficient = 1.04, p <
0.01 for all; coefficient = 0.93, p < 0.01 for those with ADM disorders only). Hausman´s specification tests could not reject simultaneity (p < 0.01). When exercise counseling was used as an instrumental variable for smoking cessation counseling, smoking cessation counseling by primary care providers had a negative significant association with current smoking status (coefficient = -0.88, p <0.01 for all; coefficient = -1.063, p < 0.01 for those with ADM disorders). Predicted probabilities of current smoking status without smoking cessation counseling were 90.8% for all smokers, 94.0% for smokers with ADM disorders and 89.5% for smokers without ADM disorders. Predicted probabilities of current smoking status with smoking cessation counseling were 67.3% for all smokers,
68.7% for smokers with ADM disorders and 65.1% for smokers without ADM disorders.
Conclusion: The use of an instrumental variable was able to identify that past year primary care provider counseling is associated with smoking cessation among all individuals and individuals with
ADM disorders who had been previously identified as current smokers.
Implications for Policy, Delivery or Practice:
Smokers with ADM disorders should continue to be targeted for smoking cessation counseling, and described in future editions of the smoking cessation clinical practice guideline. Future prospective studies should be conducted to verify a causal relationship between smoking cessation counseling and actual smoking cessation in this important population.
Funding Source(s): RWJF
Theme: Behavioral Health
● Is the Likelihood of a Patient Admission
Through the Emergency Room Reduced by
Participation in a Medicare HMO?
Elijah Onsomu, B.A., M.S., M.P.H., C.H.E.S.;
Elizabeth Tait, B.S., M.H.S.; Barbara Ferguson,
B.S.N., R.N., M.S.N., M.H.A., M.B.A.; Kathryn
Willits, B.S.N., R.N., M.S.N., F.N.P.-B.C.
Presented by: Elijah Onsomu, B.A., M.S., M.P.H.,
C.H.E.S., Research/Teaching Assistant, Public
Health Sciences, The University of North Carolina at Charlotte, 9201 University City Boulevard,
Charlotte, NC 28223-0001, Phone: (704) 649-6240;
Email: eoonsomu@uncc.edu
Research Objective: The research to date has not adequately addressed whether hospitalizations are reduced by participation in a Medicare managed care program (HMO) verses participation in the
regular, Fee-for-Service (FFS) Medicare plan. If the
HMO system functions optimally, patients are managed in the office or over the telephone rather than be referred to the Emergency Department
(ED). Furthermore, HMO provides higher preventive care as opposed to the traditional FFS plan. The researchers studied the use of inpatient hospital services by Medicare beneficiaries by plantype (HMO verses FFS). We adjusted for age, sex, race/ethnicity, and co-morbidities using the
Elixhauser methodology, since the elderly have multiple diagnoses that may affect hospitalizations.
Study Design: The researchers used a crosssectional data from the 2005 Florida Inpatient
Discharge Database which is collected from hospital inpatient data, ambulatory surgical data, emergency department data and surgical infection prevention measures. Univariate analysis and a manual forward stepwise method were performed for variable inclusion; bivariate and multivariate logistic regression analyses were also performed.
The model for admission type was fit using the variables: age, race, ethnicity and co-morbidity.
The model was tested using the Hosmer and
Lemeshow Goodness of Fit test.
Population Studied: The study sample
(n=867,098) was for individuals =65 years in the
2005 Florida Inpatient Discharge Database.
Principal Findings: Elderly with HMO insurance plans were significantly more likely than elderly with
FFS insurance plans to be admitted through the
Emergency Department [OR=1.72 (95CI, 1.69 -
1.74) and OR=0.58 (95CI, 0.57 - 0.59)] respectively. Although there were few elderly
Blacks (26%) admitted to the ED with HMO compared to those with FFS (74%). Blacks in an
HMO had greater odds [OR=1.43 (95CI, 1.36 -
1.49)] than other races to be admitted to the hospital through non-trauma ED and physician referral. In contrast, Blacks in FFS had lower odds
[OR= 0.70 (95CI, 0.67 - 0.73)] than other races to be admitted to the hospital through non-trauma ED and physician referral. Furthermore, those with
HMO were 72% more likely to be admitted to the hospital through the ED than physician referral. In contrast those with FFS were 42% less likely to be admitted to the hospital through the ED than physician referral. There were also few elderly
Hispanic (33%) who were admitted through the ED with HMO compared to FFS (67%). However, they also had higher odds [OR=1.90 (95CI, 1.82 - 1.99)] than other races to be admitted to the hospital.
With FFS, elderly Hispanics had lower odds
[OR=0.53 (95CI, 0.50 - 0.55)] as compared to other races to be admitted.
Conclusion: Participation in an HMO has a higher protective effect for co-morbidities relative to those participating in FFS. However, they have a higher admittance through the ER. To ensure that all elderly irrespective of disease or condition, have access to and are enrolled in a health insurance plan that is both cost efficient and high quality.
Implications for Policy, Delivery or Practice:
HMO participants based on co-morbidities are healthier than those with a FFS plan. However, more attention needs to be given to HMO practices based on the ED admissions.
● What Physician Characteristics are
Associated with Higher Quality Care?
Rachel Orler, B.A.; Mark Friedberg, M.D.; John
Adams, Ph.D.; Elizabeth McGlynn, Ph.D.; Ateev
Mehrotra, M.D.
Presented by: Rachel Orler, B.A., Medical Student,
University of Pittsburgh School of Medicine, 726
Copeland Street #1, Pittsburgh, PA 15232, Phone:
(724) 312-1828; Email: orler.rachel@medstudent.pitt.edu
Research Objective: As evidenced by pay-forperformance incentives and physician quality
“report cards,” there is a growing interest in identifying high quality physicians. However, the characteristics of physicians providing the best care are unknown. Previous studies of the relationship between physician characteristics and quality measure performance have been limited by low numbers of physicians, few available physician characteristics, and limited scope of quality measures. In this study we examined, in a large sample of Massachusetts physicians, the relationship between several individual physician characteristics and performance on a broad range of quality measures.
Study Design: We obtained insurance claims generated by over 1 million adult patients continuously enrolled in 4 of the largest commercial health plans in Massachusetts during 2004-2005.
These claims reflected clinical care provided by
12,781 physicians in 27 specialties. The RAND
Quality Assurance (QA) Toolkit, which contains 131 measures of acute, chronic, and preventive care processes, was used to generate claims-based quality measure performance scores for each physician on each measure. QA Toolkit measures consist of delivery of guideline-based care (e.g., screening for anemia at a first prenatal visit). On each measure, physician performance was calculated as a ratio: denominator is number of instances where patients eligible for indicated care and care was assigned to a physician, numerator is among these instances number of times patient received indicated care. We created 4 composite performance scores (overall, acute, chronic, and preventive) for each physician by summing the individual measure numerators and dividing by the sum of the corresponding individual measure denominators. The composite scores were adjusted for the degree of difficulty of each individual measure. Physician characteristics were obtained from publicly available data provided by the
Massachusetts Board of Registration: gender, medical school attended, years in practice, board
certification status, and disciplinary and malpractice information. Medical school rankings were obtained from U.S. News and World Report. We constructed multivariate linear regression models of performance on the 4 composites, using all available physician characteristics as predictors.
Population Studied: 10,776 (84.31%)
Massachusetts physicians with complete demographic and performance profiles.
Principal Findings: The only physician characteristics independently associated with significantly higher overall performance scores were gender (female 1.4% higher), board certification status (board certified 3.0% higher), and medical school location (domestically trained 0.9% higher).
The higher scores in these groups were primarily driven by performance on the preventive care composite. Differences on acute and chronic care measures were much smaller. For example, the effect of gender on performance was much greater for preventive care (female 3.8% higher) than for chronic care (male 0.2% higher) or acute care
(female 0.55% higher). Notably, characteristics such as malpractice claims, medical school ranking, or years of experience were not associated with higher or lower physician performance.
Conclusion: Few characteristics of individual physicians are associated with higher quality, and when present, the associations are small in magnitude.
Implications for Policy, Delivery or Practice:
Publicly available characteristics of individual physicians are poor predictors of the quality of delivered care. Reporting physician performance scores—or shifting the focus of reporting to higher levels of physician organization—may offer the best guidance to patients seeking high-quality care.
Funding Source(s): University of Pittsburgh
Medical School Dean's Summer Research Program
Theme: Health Care Workforce
● Assessing the Prevalence & Variation of Knee
Replacement Surgeries among a Commercially
Insured Population
John Orwat, Ph.D.; Dan Gunselman; John Mason;
Amin Serehali; Andrea Marks; John Orwat
Presented by: John Orwat, Ph.D., Assistant
Professor, Loyola University Chicago, 820 North
Michigan Avenue, 1229 Lewis Towers, Chicago, IL
60611, Phone: (312) 915-7494; Email: jorwat@luc.edu
Research Objective: To use the Blue Health
Intelligence (BHI) dataset to estimate national utilization rates and trends associated with knee replacement surgeries and complications, as well as to identify associated demographic, regional, and clinical factors.
Study Design: BHI healthcare claims data were retrospectively analyzed. A 30- and 90- day followup period was used to identify complications and associated factors. Bivariate and multivariate analyses were conducted. Factors for analysis included region of the country, age, gender, procedure type (single vs. bilateral), and prior surgery status (another knee replacement within 7 or 90 days). Complications were examined by individual complication, complication group, and region of the country.
Population Studied: Knee replacement surgeries that occurred between January 2005 and
December 2007 were identified for members between the ages of 18 and 65 (n=144,187). This study population was drawn from the BHI dataset which contains the healthcare claims experience of approximately 54 million members from all regions of the United States from BHI participating Blues
Plans.
Principal Findings: Initial findings show a national prevalence rate for knee replacements of 19.2 per
10,000 beneficiaries with an upward trend in utilization, with an 18.7% increase in the rate of knee replacements from 2005 through 2007. This trend is higher for males (at 23%) than for females
(at 16%). For both males and females, the age group with the largest utilization increase is the 45-
49 age group. Within this age group, males show a
25% increase in surgery and females show a 17% increase from 2005 through 2007. Age and sex adjusted utilization rates and trends vary by region.
In addition, 22% of knee replacements have a complication within 30 days, and this rate increases to 25.3% at 90 days. More than half of the complications at 30 days are classified as hemorrhage/hematoma or blood clots.
Conclusion: Knee replacement and complication rates are trending upward and appear to be related to non-clinical factors such as gender and region.
Further analysis will examine factors that influence the wide difference in utilization trends between regions, the upward utilization trend among younger members, and risk factors for complications and revisions.
Implications for Policy, Delivery or Practice:
Orthopedic conditions and procedures are some of the largest drivers of cost and utilization in commercially insured health plans. A better understanding of treatment variation and associated complications and risk factors has implications not only for patient safety and quality of care, but also for a significant proportion of commercial health plan and employer costs. Knee replacement is an expensive and increasingly prevalent procedure.
What is less clear is whether the utilization drivers are due to system-wide issues or local issues. A better understanding of the factors that influence utilization may improve the patient’s quality of care.
Funding Source(s): BlueCross BlueShield
Association
● A Preliminary Look at Consumer Views on
PHRs: Towards a Social Transformation of
Managing Health Information
John Orwat, Ph.D.; Mary Madison
Presented by: John Orwat, Ph.D., Assistant
Professor, Loyola University Chicago, 820 North
Michigan Avenue, 1229 Lewis Towers, Chicago, IL
60611, Phone: (312) 915-7494; Email: jorwat@luc.edu
Research Objective: Personal Health Records
(PHRs) appear to hold the promise to improve quality, safety, and efficiency of health care.
However, little is known about consumer perceptions, attitudes and behaviors related to PHR adoption and utilization in an age of concerns related to privacy and security. Even less is known about the impact of educational interventions regarding PHRs. This analysis examines the factors associated with PHR adoption and utilization of a sample of consumers who attended workshops on
PHRs.
Study Design: Analysis of surveys completed by participants of workshops to increase awareness of
PHR adoption and utilization. Workshop goals were to increase awareness of PHRs generally as well as encourage use of MyPHR.com, a product offered by the AHIMA Foundation of Research and
Education, a 501(c)(3) charitable entity of the
American Health Information Management
Association (AHIMA). Analysis will identify factors associated with PHR adoption and the type of PHR planned, such as consumer preferences/attitudes, health care experiences, and demographics.
Population Studied: Exit surveys (N=9,051) of participants who attended educational workshops sponsored regional associations of AHIMA.
Workshops were conducted in a variety of settings from 2005-2008 reaching diverse audiences.
Principal Findings: The sample was 72% female, a majority were between 26 and 50 years old
(54%), 33% self managed their chronic illness, and
66% reported a preference for accessing their health information online if appropriate security precautions are in place. Almost one quarter had a
PHR and 88% indicated they would create one post workshop. Paper was preferred by 57% who indicated they would create a PHR, 21% would use
EXCEL, 17% would use an insurance plan provided service, and 3% an internet service. Preliminary multivariable analyses indicate that those over 65 years had a higher odds of already having a PHR
(OR=3.05, CI = 2.40- 3.89) as well as those who manage their own chronic illness (OR = 1.95, 95%
CI = 1.74- 2.18). Older age was associated with a lower odds of adopting forms other than paper formats; and those who self-manage their chronic illness had a higher odds of managing their PHR with their own computer when compared to the paper form. (Analysis of all factors will be complete
3/2009.)
Conclusion: Preliminary analysis suggests that adoption and preferred type of PHR is related to gender and age. Although we find a preference for paper PHRs regardless of age, these analysis are very preliminary. Future analysis will focus on consumer level experience, health care experiences, and preferences.
Implications for Policy, Delivery or Practice: The social transformation of health information suggests a shift in which consumers maintain health records.
Educational interventions to encourage PHR adoption should take different forms for different age groups and levels of illness, crucial if those with the most need will access this benefit. Current analysis is exploring the role of perceived ease, security and privacy of internet and computer based
PHRs, which may differ by age preferences. This analysis offers valuable insight into consumer level factors not previously identified.
Funding Source(s): AHIMA FORE & Loyola
University Chicago
● Employee Generation & Workplace
Experience: Relationship to Workplace Climate
Perceptions
Katerine Osatuke, Ph.D.; Scott Moore, M.A.;
Jonathan Fishman, B.S.; Jill Draime, Ph.D.; Sue
Dyrenforth, Ph.D.
Presented by: Katerine Osatuke, Ph.D.,
Psychologist, Veterans Health Administration
National Center for Organization Development,
11500 Northlake Drive, Ste 230, Cincinnati, OH
45249, Phone: (513) 247-2255; Email:
Katerine.Osatuke@va.gov
Research Objective: Generational differences among employees have recently been a focus of attention, due to quickly changing demographics in healthcare workforce. Veterans Health
Administration (VHA) strives to account for these changes in workforce development planning. This study explored relative importance of differences among generations of employees, using qualitative data of VHA workplace interview assessments, and
VHA All Employee Survey (AES) quantitative ratings.
Study Design: VHA workplace assessments ask respondents to describe strengths, weaknesses, and desired changes in their organizations.
Interview data were analysed using grounded theory methods. Themes relevant to generational differences were then summarized. The AES, an annual voluntary anonymous census survey of VHA employees, asks respondents to rate their perceptions of workplace characteristics and job satisfaction. Previous exploratory and confirmatory factor analyses of AES scores indentified three reliable dimensions in employee ratings of their workgroup climate which are of particular interest to this study: Civility, Management for Achievement or
Leadership, and Customer Focus. In a series of univariate ANOVAs, we examined differences in scores on these dimensions among the groups of
AES respondents whose age placed them into the
categories of baby boomers, X generation, and
Millennials, while distinguishing between respondents who worked more than, or less than 2 years at their current workplace.
Population Studied: Veterans Health
Administration employees of 3 generations
Principal Findings: Qualitative data suggested that employees’ experience at the specific workplace was salient, and generational differences were not salient in respondents’ perspective within the context of discussing strengths and weaknesses of their organizations. In quantitative data, main effects of generation on perceptions of workgroup climate were significant, if small, for
Customer Focus only. Having spent more than versus less than 2 years at the current workplace had significant and large main effects for all the three dimensions examined.
Conclusion: These findings place generational differences in the contex of exployees’ experience and suggest that the latter may be more important than the former. That is, age at position rather than age as such may be an important aspect to study.
Further studies are needed to systematically assess the effects of experience on employees perceptions of workplace, related needs (e.g. supervisory support), related individual outcomes (e.g. job satisfaction), and organizational outcomes (e.g. work quality, productivity, customer satisfaction).
Implications for Policy, Delivery or Practice:
These findings may redefine the importance or possibly the meaning of generational differences within the context of workplace perceptions. Placing more emphasis on employees’ age within organization rather than on their generational differences would directly impact decisions about workforce development and succession planning within Veterans Healthcare.
Theme: Health Care Workforce
● Can National Survey Data Be Used to
Benchmark Quality Indicators for Lung Cancer?
Jean Owen, Ph.D.; Benjamin Movsas, M.D.; J.
Frank Wilson, M.D.
Presented by: Jean Owen, Ph.D., Senior Director,
Quality Research in Radiation Oncology, American
College of Radiology, 1818 Market Street, Suite
1600, Philadelphia, PA 19103, Phone: (215) 574-
3164; Email: jowen@phila.acr.org
Research Objective: To determine adequacy of an existing database as evidence for measuring proposed Quality Indicators (QI) for lung cancer treatment.
Study Design: QI were developed based on
National Comprehensive Cancer Network (NCCN) guidelines for non-small cell lung cancer (NSCLC) management and were compared to published results of the Patterns of Care Study (PCS) database of patients receiving radiation therapy
(RT) for lung cancer in 1998-99. The survey data came from a national survey with a two-stage stratified sample of radiation oncology practices that allowed calculation of national averages of defined measures for patients treated in all types of radiation oncology facilities. Data was extracted from on-site chart reviews of medical and radiation therapy records.
Population Studied: Eligible cases were diagnosed with Stage I - III NSCLC cancer, were treated with RT starting in 1998 or 1999, had a
Karnofsky Performance Status (KPS) >=60, and did not have any of the following ineligibility criteria: distant metastases or malignant pleural effusion, prior thoracic RT, or prior or concurrent malignancy within five years (except in-situ or non-melanoma skin cancer).
Principal Findings: Compliance results for QI derived from NCCN guidelines for workup are: performance status described (99%), weight loss assessed (66%), smoking status described (95%), chest CT (98%), bone scan in clinical stage (CS) III patients (66%), brain imaging in CS III patients
(48%), mediastinum pathologically staged in surgical patients (51%). QI for treatment include:
CT based treatment planning (49%), 3D conformal therapy applied (52%).
Conclusion: Benchmarking utilization patterns provides a foundation for assessing appropriateness of care in the future. Although not collected originally for the purpose of measuring QI, the PCS database is a potentially rich data source that can be used to validate these measures.
Identifying additional data fields for new survey data collection allows measurement of more detailed QI for more specific patient subgroups. New data collection will provide current benchmark data.
Implications for Policy, Delivery or Practice: As
QI are used for more purposes, including quality improvement programs and accountability, finding data sources to benchmark QI is important. The required level of detail for meaningful QI exceeds most current administrative data, while clinical trials provide data only for a limited set of specific and usually experimental treatments. Well designed surveys can provide a data source with sufficient detail for a broad range of QI and investigators can define additional data requirements to allow use of administrative data.
Funding Source(s): NCI, Pennsylvania
Department of Health, Tobacco Settlement Act 77-
2001, Commonwealth Universal Research
Enhancement for Fiscal Year 2005.
Theme: Quality and Efficiency: Measurement
● University of Nevada School of Medicine
Resident Exit Survey - 2004 to 2008
John Packham, Ph.D.; M. Tabor Griswold, M.S.;
Caroline Ford, M.P.H.
Presented by: John Packham, Ph.D., Director of
Health Policy Research, University of Nevada
School of Medicine, 411 West Second Street,
Reno, NV 89503, Phone: (775) 784-1235; Email: jpackham@medicine.nevada.edu
Research Objective: The University of Nevada
School of Medicine Resident Exit Survey (RES) is designed to shed light on the experiences and preferences of physicians who remain in Nevada to practice and the reasons given by exiting residents for their decisions concerning initial practice locations, residency program outcomes, level of educational debt, and the initial geographical location selection criteria. As such, the survey data provides a snapshot of physician employment opportunities in Nevada and the decision making preferences of surveyed residents.
Study Design: The RES is administered annually and original data is collected from all exiting residents and fellows currently from the twenty-four programs offered at the University of Nevada
School of Medicine. The study design is quantitative in nature primarily with emphasis on residents’ opinions about entering patient care practice or continuing their training.
Population Studied: All physicians completing
University of Nevada School of Medicine residency and fellowship programs (2004 to 2008).
Principal Findings: The proposed poster highlights
RES results and trends over the past five years, trends have changed as women achieved parity, international medical students became forty percent of the residents, more Nevadan and residents from other countries were accepted into the program, and a decreasing number of residents were entering patient care. There was a rise in hospitalist professionals, decreasing satisfaction with the expected first year earnings, and variation with instate retention rates by specialty. Most importantly, residents entering primary care practices carried increasing higher levels of debt and were concerned most about availability of free time, dealing with managed care payers, and educational debt than the ability to find a job or insufficient practice knowledge or medical knowledge.
Conclusion: Conclusions of the survey over the five year period include relatively high levels of residents who remain in Nevada to practice upon completing residencies and fellowships in Nevada.
Implications for Policy, Delivery or Practice:
Implications of these results point to financial difficulties of training primary care providers particularly with rural recruitment in Nevada.
Funding Source(s): Medical Education Council of
Nevada
● Effect of Pediatric Providers' Participation in a
Preventive Dental Program on Children’s
Access to Dental Care
Bhavna Pahel, B.D.S., M.P.H., Ph.D.; Sally
Stearns, Ph.D.; John Preisser, Ph.D.; R. Gary
Rozier, D.D.S., M.P.H.; Dennis Clements, M.D.,
M.P.H., Ph.D.; Michelle Mayer, R.N., M.P.H., Ph.D.
Presented by: Bhavna Pahel, B.D.S., M.P.H.,
Ph.D., Post-doctoral Fellow, Sheps Center for
Health Services Research, University of North
Carolina at Chapel Hill, 725 Martin Luther King Jr.
Boulevard, CB# 7590, Chapel Hill, NC 27599-7590,
Phone: (919) 724-2239; Email: bhavna_pahel@unc.edu
Research Objective: To expand access to preventive dental care, the North Carolina (NC)
Medicaid program began reimbursing pediatric practitioners in 2000 to provide a package of three services for Medicaid-enrolled preschool age children through an initiative called the Into the
Mouths of Babes (IMB) program. These services, usually occurring with a well-child visit, include oral health assessments and dentist referrals (if needed), topical fluoride for the child’s teeth and parent education about infant oral health in up to six
IMB visits before three years of age. This study examined whether IMB participating practices facilitated access to dental care for preschool age children receiving IMB services in those practices compared to non-IMB practices.
Study Design: The observational longitudinal analysis identifies IMB visits in medical offices and dental visits (preventive and restorative) in dental offices with child observations aggregated within medical practices by child age group. We classify medical practices into three groups (well child visit with no IMB services, and high and low volume IMB practices) based on the average number of IMB visits per well child visit (WCV) provided by the practices. We compare time to use of dental care from the child’s first WCV for the three practice groups using a piecewise exponential survival model to determine whether participation in the IMB program results in earlier use of dentist services
(e.g., from referrals for existing tooth decay) or later use (e.g., due to decreased need for restorative services), controlling for child’s county of residence
(rural versus urban).
Population Studied: All children (N=167,358) born on or after July 1, 2003 who were enrolled in NC
Medicaid during 2004-2006.
Principal Findings: Of 1,192 medical practices included in the study, 914 (76.6%) were WCV only,
215 (18.0%) were low volume IMB and 63 (5.2%) were classified as high volume IMB practices.
Overall, 16,634 (9.9%) of the 167,358 children in the sample had at least one dental visit during the study period, with a majority being for restorative
(vs. preventive) dental services. Children seen for well-child care in high volume IMB practices went significantly longer without a visit in a dental office compared to children with WCVs in non-IMB practices. Children seen in low volume IMB practices did not differ significantly from those seen in non-IMB practices in time to dental office visits.
Conclusion: The IMB program increased time to use of dental services for children seen in high volume IMB practices compared to non-IMB
practices. The referral effect might not have been detected because the preventive effect of fluoride on dental caries could have overshadowed any referral effect in our study population.
Implications for Policy, Delivery or Practice:
Expanding access to preventive dental care in medical offices appears to lead to a net decrease in the use of dentist services (mainly restorative care) among children younger than age three. Future studies should account for the oral health status of the study population and also explore whether the
IMB program leads to greater use of preventive dental services for children after age three.
Funding Source(s): NRSA Post-doctoral traineeship from AHRQ, National Institute for Dental and Craniofacial Research (R03 DE017350).
Theme: Medicaid, SCHIP and State Health Reform
● Factors Associated with Incidental Sickness
Absence among Employees in One Health Care
System
Chih-Wen Pai, Ph.D.; John Mullin, Ph.D.; Gina
Payne, Ph.D.; Jeaneeta Love, R.N., M.B.A.; Gayle
O'Connell, M.S.; Dee Edington, Ph.D.
Presented by: Chih-Wen Pai, Ph.D., Analyst,
Health Management Research Center, University of
Michigan, 1015 East Huron, Ann Arbor, MI 48104,
Phone: (734) 647-0519; Email: cwpai@umich.edu
Research Objective: To examine the association of health risks and frequency of health management program participation with incidental sickness absence among employees in one Midwest health care system.
Study Design: Cross-sectional, multivariate logistic regression was conducted to assess the association of taking any incidental sickness absence with health status, health behavior, and program participation. We performed the analysis for the entire study population as well as analyses for four subpopulations (male, female, ages 19-44, age 45 or over). Administrative absence tracking data of 2007 were used to define any incidental sickness absence as cumulative time away from work due to personal illness for more than one day in 12 months. Absence record from 2006 was also used to assess past absence. Health risk appraisal
(HRA) data from 2007 were the primary data source to measure health status and behaviors (13 health risks, 16 self-reported medical conditions), age and gender specific compliance with preventive service guidelines, and lost productivity. We included in the model demographic factors based on personnel data to control for differences in employment status and location. Using multi-year data, two variables measured program participation, years of HRA participation (4-7 years vs. 0-3 years) and years of wellness program participation (3-5 years vs. 0-2 year).
Population Studied: Individuals who were employed by one Midwest health care system for both 2006 and 2007 and had completed HRAs in
2007 (n=3,790).
Principal Findings: Among the overall study population, employees who were smokers had 1.43 times greater odds of taking absence, compared to non-smokers (p<0.05). Similarly, the odds of taking absence for overweight employees (body mass index =27.5) was 1.22 times higher than those with normal weight (p<0.05). Analyses of the four subpopulation did not generate congruent results.
For example, no consistency was observed in the association with health risks or health conditions.
However, one constant finding was that participation in wellness program three years or more had a negative association with taking absence (at least p<0.05).
Conclusion: Results from this study showed different patterns by gender and age group. One consistent association with taking incidental absence was observed in more years in wellness program participation.
Implications for Policy, Delivery or Practice: To combat lost productivity, employers may incorporate health risks in identifying employees at risk of incidental absence and implement early interventions to focus on preventable causes.
Specifically, wellness programs may target weight and smoking issues. The benefit of absencereducing interventions would be more far reaching since the same risk factors also contribute to high health care cost. Some consider taking incidental absence as one coping mechanism that prevents serious illness in the long term, rather than perceiving incidental absence as a direct manifest of poor health. Thus, certain degree of flexibility in work arrangement would facilitate work-life balance.
Funding Source(s): Saint Luke’s Health System
Theme: Health Care Workforce
● Risk of Stroke Associated with Antipsychotic
Use: A Retrospective Cohort Study
Jacob Painter, Pharm.D., M.B.A.; Jeffrey Talbert,
Ph.D.
Presented by: Jacob Painter, Pharm.D., M.B.A.,
Fellow, Research & Data Managment Center,
University of Kentucky, 1080 Export Street, Suite
280, Lexington, KY 40504, Email: jtpain2@uky.edu
Research Objective: In 2002, concerns regarding the association of stroke with atypical antipsychotics first arose from limited trial data. In
2003, the FDA issued a Dear Healthcare
Professional Letter warning about the use of risperidone in elderly patients suffering from dementia. Labeling changes for risperidone were instated followed by similar changes for olanzapine and aripiprazole. Significant gaps in the evidence still exist; however, regarding issues such as the baseline cardiovascular risk of patients on antipsychotics, the extent to which this association extends to patients without dementia, and the
differentiation of the risk of atypical versus typical antipsychotics. The most recent addition to the body of evidence regarding the relationship of stroke and antipsychotics was published in the
British Medical Journal in August 2008. The authors, Douglas & Smeeth, conducted a retrospective study examining the association of typical and atypical antipsychotics in patients with and without dementia. They employed a selfcontrolled case series study in an effort to minimize confounding and performed subgroup analyses to differentiate the risks seen in various types of drugs and patients. This article was heavily criticized due to the research design chosen to examine this relationship. We attempt to examine this relationship further by using a large patient database. Does a relationship exist between antipsychotic use and stroke risk? We hypothesize that a relationship does exist for both typical and atypical antipsychotics.
Study Design: We use a selected populationbased retrospective cohort study design to look at the outcome of stroke. Data is an extract from a health-claims data warehouse consisting of eligibility, pharmacy, institutional, and medical service information. The database represents the entire claim history for Kentucky Medicaid recipients during the calendar years 1998-2006. This database contains medical information such as diagnoses as indicated by International
Classification of Diseases (ICD) codes, and procedures as indicated by Current Procedural
Terminology (CPT) codes. It also contains information for prescription medications such as drug name, strength, day supply, and prescribing physician.
Population Studied: Using this information we scan the database for patients to include in our study. Our inclusion criteria were all adults 45 years of age and older, with continuous eligibility
(11 months) for at least one year. From this initial population we selected only patients who had either: A diagnosis within the Mental Health
Services Act (MHSA) codes 650, 651, 653, 657,
658, or 659 OR Seen a mental health provider
(Psychiatrist, Psychologist or Community Mental
Health Clinic). From this initial selection we chose only those patients who had 90 consecutive days of consistent antipsychotic use. If the required time period was on a typical antipsychotic the patient was entered into the Typical group; whereas, if the spell was on an atypical antipsychotic they were entered into the Atypical group. Then each patient in these groups were matched 1:1 with patients who met the inclusion criteria, but had no exposure to an antipsychotic. Control patients had a random 90 day spell chosen during the same year as their matched counterpart.
Principal Findings: Preliminary findings suggest the absence of a link between the use of antipsychotics and the occurrence of stroke. While atypical antipsychotic use shows negligible difference from the control group, the typical group has a non-significant protective effect with an odds ratio of 0.89. Factors that did affect stroke occurrence included prior occurrence of stroke, age, Charlson Index, and recent coronary angiography or angioplasty. The significance of these factors suggest the model is sound.
Conclusion: Although preliminary data shows no significant differences between the groups studied, there is still much clarification needed regarding the contributing factors to stroke in this patient population. While no significant findings have occurred to date evidence to warrant continued investigation does exist. The postulated protective properties of haloperidol in the literature may contribute to the 0.89 OR seen in our typical antipsychotic group. Subgroup analysis to examine individual antipsychotics is also necessary in order to clarify whether intra-class variation exists. Our future research will focus on clarification of factors contributing to stroke risk in an attempt to find the true contribution of antipsychotics to stroke risk.
Analysis of the data with inclusion of concurrent medications and exclusion of hemorrhagic stroke are already underway.
Implications for Policy, Delivery or Practice:
There are substantial limitations to these preliminary results. Although we know certain medications other than antipsychotics are contribute to stroke risk in negative and positive directions, concurrent medications are not considered in this analysis. Furthermore, while antipsychotics likely only play a role in ischemic strokes, hemorrhagic strokes are included in this data. Further clarification is underway in order to elucidate the role antipsychotics play in the stroke risk seen in this population and to clarify the implications this effect may have on practice.
● Does the Level of Background Check
Agencies make a Difference in Firearm Death
Rates?
Anantachai Panjamapirom, M.S., M.B.A.; Bisakha
Sen, Ph.D.
Presented by: Anantachai Panjamapirom, M.S.,
M.B.A., Ph.D. Student, Health Services
Administration, University of Alabama at
Birmingham, WEBB 564 1530 3rd Avenue South,
Birmingham, AL 35294-3361, Phone: (205) 934-
3113; Email: apirom@uab.edu
Research Objective: Limited studies focused on the effects of background check agencies on firearm deaths. The 50 states have utilized different departments and organizations as a background check agency for gun purchase. This study investigated whether more rigorous background check for a gun purchase led to reduction in firearm suicide and homicide deaths.
Study Design: Longitudinal data (1996-2005) on firearm and non-firearm suicide and homicide
deaths were retrieved from the Web-based Injury
Statistics Query and Reporting System
(WISQARS™) of the National Center for Injury
Prevention and Control. The death data were divided into three age groups: 15-20, 21-29, and 30 and above. The level of background check agencies were classified into local, state, and federal based on the reports from the Survey of
State Procedures Related to Firearm Sales.
Because the integer nature of homicide and suicide death data, we utilized negative binomial regression to examine the association between the level of background check agencies and firearm and nonfirearm suicide and homicide deaths while controlling for income, poverty rate, unemployment rate, divorce rate, per capital alcohol consumption, per capita officers, and per capita agencies. We also used historical data as a lagged dependent variable to control for unobserved confounding factors.
Population Studied: All firearm and non-firearm suicide and homicide deaths reported in
WISQARS™ from 1996-2005. Legal intervention homicide deaths were excluded from the analyses.
Principal Findings: Using federal-level agencies as the reference group, in the states where a statelevel agency was used to perform firearm background checks, firearm homicide deaths were higher among black in the 15-20 and 30 and above age groups regardless of the historical data control.
When the historical data was not controlled, the incidence rate ratios for black in both age groups are 1.35 (95% CI = 1.10-1.67) and 1.26 (95% CI =
1.07-1.48). After controlling for the historical data, the incidence rate ratios for black in both age groups are 1.29 (95% CI = 1.09-1.53) and 1.23
(95% CI = 1.08-1.42).
Conclusion: If the historical data was not controlled, the level of background check agency for firearm purchase appeared to be significantly associated with firearm suicide and homicide deaths for different races across age groups.
However, after controlling for the historical data, no association was generally presented. The similar results were also found with non-firearm suicide and homicide deaths.
Implications for Policy, Delivery or Practice:
Previous studies suggested that the use of local agency to perform background check for firearm purchase should be considered. However, based on the results from this study, we suggested cautions in employing this type of intervention because those states that have employed a local- or state-level agency are the ones that might be more rigorous in controlling the firearm deaths to begin with.
Theme: Behavioral Health
● Assessing the Magnitude & Implications of
Risky Sexual Behaviours Linked to Mental
Health in Adolescents in California, CHIS 2005
Christopher Pankonin, M.P.H., M.S.; May Aydin,
Ph.D.
Presented by: Christopher Pankonin, M.P.H., M.S.,
University of California, Los Angeles Center for
Health Policy Research, 10960 Wilshire Boulevard,
Suite 1550, Los Angeles, CA 90024, Phone: (310)
794-8362; Email: cpankonin@ucla.edu
Research Objective: The purpose of this paper is to assess the magnitude of the association between mental/emotional health, specifically likely depression, and risky sexual behaviors among adolescents living in California.
Study Design: The association between depression and risky sexual behaviors among adolescents has been previously described at national and local levels, although not at a statewide level. This paper will analyze data from the
2005 California Health Interview Survey (CHIS), a population-based random-digit dial telephone survey conducted every two years across all age groups living in California. A logistic regression model is created using a composite Risky Sexual
Behavior outcome variable. Analysis is conducted separately by gender.
Population Studied: A total of 4,029 adolescents aged 12 to 17 were sampled in CHIS 2005.
Principal Findings: Odds ratios are presented for the composite Risky Sexual Behavior variable as well as for individual sexual risk factors contained within the CHIS data set.
Implications for Policy, Delivery or Practice:
Understanding the specific association between mental health and risky sexual behaviours as well as the specific needs of “risky” adolescents will allow for more targeted and effective interventions aimed at improving mental health but also as a means of preventing risky sexual behaviours.
Possible policy implications and targeted interventions such as counseling and identifying those at risk of depression are discussed.
Theme: Behavioral Health
● Learning in Primary Care Teams &
Implementation of the Chronic Care Model
Michael Parchman, M.D., M.P.H.; Holly Lanham,
Ph.D.; Michelle Jordan, Ph.D.; Polly Noel, Ph.D.;
Nedal Arar, Ph.D.
Presented by: Michael Parchman, M.D., M.P.H.,
Investigator, South Texas Veterans Health Care
System, VERDICT (11C6), 7400 Merton Minter
Boulevard, San Antonio, TX 78229-4404, Phone:
(210) 617-5314; Email: parchman@uthscsa.edu
Research Objective: Efforts to date to understand the slowness of the health care system to adopt new evidence into practice, especially for chronic disease care, may have been hindered by an overreliance on the knowledge, decision-making, and actions of individuals and under-recognition of the
crucial role of these activities in teams. Primary care teams have been characterized as complex adaptive systems (CAS). A CAS is comprised of individuals who learn, inter-relate, self-organize, and co-evolve with their environment in non-linear ways, creating uncertainty and unpredictability.
Individuals deal with uncertainty by forming relationships that promote effective learning. We examine the relationship between learning in primary care teams and the elements of the chronic care model.(CCM)
Study Design: Clinicians and staff completed a survey containing the Assessment of Chronic
Illness Care (ACIC) survey, a validated measure of the CCM, and a learning survey. The ACIC measures six elements of the CCM: organizational support, community linkages, self-care support, decision support, delivery system design and clinical information systems. The 22 question learning survey was developed by members of our team to measure the degree to which learning occurs in a small health care team. Examples of items on the survey include: “I learn how to do things in this clinic by sharing knowledge with team members,” and “I learn a lot about how to do my job by talking with other people in this clinic,” and “This practice encourages everybody’s input for making changes.”
Population Studied: Clinicians and staff in 40 small, autonomous primary care clinics.
Principal Findings: 142 surveys have been collected in 22 clinics. Final results from all 40 clinics will be presented at the meeting. Cronbach’s alpha for the learning survey was 0.78. The learning score was significantly associated with the total
ACIC score (r=0.43, p<.001). Regarding sub-scale scores representing different components of the
CCM, the learning score was significantly associated with all 6 components. The two components most strongly associated with learning were organizational support and delivery system design. A principal component factor analysis
(PCA) with a promax rotation on all of the items in both instruments revealed no overlap in factor loadings between learning scale items and ACIC items, suggesting that the learning scale is measuring a different underlying construct than those measured by the ACIC scale.
Conclusion: Primary care teams with high levels of sharing and learning from each other are more likely to have implemented the CCM. From a CAS perspective, this may reflect the ability of a team to learn and adapt to patient needs in real time leading to higher CCM scores.
Implications for Policy, Delivery or Practice: In caring for patients with a chronic illness, team-level learning is critical. If one can take advantage of the adaptability of CAS, strategies for team-level learning can be encouraged. Emphasizing teamlevel construction of meaning within a local context and using knowledge in new ways will help primary care teams meet the needs of the particular patient at hand.
Funding Source(s): National Institue of Diabetes,
Digestive and Kidney Disorders
● Balancing Health Care Evidence & Art to Meet
Clinical Needs: Policy Makers’ Perspectives
Louise Parker, M.A., Ph.D.; Mona Ritchie, Ph.D.,
M.S.W.; JoAnn Kirchner, M.D.; Richard Owen, M.D.
Presented by: Louise Parker, M.A., Ph.D.,
Independent Consultant, Veterans Health
Administration, 1 Warwick Park, Unit 1, Cambridge,
MA 02140, Phone: (617) 497-4952; Email: parkerlouise@earthlink.net
Research Objective: Historically, health practitioners have considered their work as much an art as a science. Since the early 1990’s, however, practice based on scientific empirical evidence or evidence-based practice (EBP), has taken the health care industry by storm. Although many believe that EBP has great potential to improve health outcomes, critics have identified a number of limitations. Much of the concern focuses on EBP proponents’ assumption that randomized clinical trials (RCTs) should provide the gold standard for clinical evidence. Critics of this stance have argued that typical patients are often more complicated and less compliant than those included in RCTs. Further, critics have argued that what we observe on average in groups may not apply to a particular individual or context; by contrast, the art of medicine accommodates individual patients’, providers’, health care organizations’, communities’, and societies’ needs, preferences, and circumstances. Others have argued, however, that proper application of EBP involves reasoned consideration of a wide variety of information; thus the dichotomy between medical science and art may be false. We explored the views of executivelevel policy makers from Veterans Health
Administration (VHA), a leader in the EBP movement, regarding what constitutes evidence and the relative importance of evidence versus practical needs when determining clinical policy.
Study Design: We conducted 26 semi-structured qualitative interviews and performed a content analysis. Three highly experienced interviewers conducted all the interviews. Two interviewers participated in each interview with one serving as the lead interviewer and the other serving as a back-up interviewer taking notes and ensuring that the lead interviewer covered all crucial topics. The three interviewers took turns serving in these capacities. We were able to conduct 12 interviews face-to-face at informants’ offices; we conducted the other 14 via telephone. We audio-recorded all interviews and produced verbatim transcripts. Two investigators then each inductively coded one half of the transcripts, reviewed 100% of each other’s coding, and resolved any differences. Coding
proceeded iteratively over several rounds, as we refined our understanding of the data.
Population Studied: We identified a purposive sample of 28 current and former executive-level
VHA policy makers and key members of their staff.
Twenty-six of these agreed to participate in the study.
Principal Findings: Although informants generally believed in the value of EBP and the role of RCTs within it, they also valued other types of evidence.
Further, they had concerns that were sometimes antithetical with strict adherence to an evidencebased approach including: practical concerns, fit with organizational values and with local circumstances, resources, political pressures, and patient needs. They were especially concerned about how to address medical conditions that affect many individuals or high risk populations that have no evidence-based treatment.
Conclusion: When possible, practice should be evidence-based. When this is not possible, providers must turn to the health care art by using consensus-based best practices.
Implications for Policy, Delivery or Practice:
When considering research priorities and opportunities, researchers would benefit from increased understanding of policy makers’ perspectives on balancing the evidence base and patient needs.
Funding Source(s): VA
Theme: Organizational Performance and
Management
● Panning for Health Reform In Communities;
Does It Add Up?
Christopher Parker, M.D., M.P.H.; Glenn Landers,
M.B.A.; Naima Wong, Ph.D.
Presented by: Christopher Parker, M.D., M.P.H.,
Senior Research Associate, Andrew Young School of Policy Studies, Georgia Health Policy Center, 14
Marietta Street, Atlanta, GA 30303, Phone: (404)
413-0311; Email: chrisparker@gsu.edu
Research Objective: To determine the impact of community engagement and planning on the sustainability of chronic disease prevention and management programs in communities, and to estimate whether such impact might have national implications.
Study Design: This work utilizes comparative case studies of three Georgia communities that have been engaged in planning and implementing health improvement approaches in an effort to reduce the burden of chronic disease. The three studies examine the end products and outcomes of differing planning processes that varied by size of community, type of funder, nature of stakeholders engaged and health improvement approach or emphasis, i.e., prevention, care coordination, care management or some combination of all.
Quantitative and qualitative methods and reporting are used to draw inferences and support the conclusions about the usefulness of the process to broader health reform and the relative influence and importance of public/private, local, state and federal partnerships and engagement.
Population Studied: Three populations of
Georgians were studied in this project: (1) An
Atlanta based community of approximately 7,000 individuals, from whom a group of nearly 50 community members and partners worked with
Kaiser Permanente to develop policy, environmental changes and programs to improve the eating and physical activity habits of residents.
(2) A rural community-based volunteer clinic that focused on working with a for-profit software development company in a CDC sponsored project to develop utilizable and configurable chronic disease prevention and management software.
Thirty-two individuals participated in designing the software and nearly 200 participated in the pilot of this approach. (3) A collaboration of 25 mostly rural counties to plan for the enhancement of the regional safety net to focus on wellness and the regional coordination of care and wellness of care.
Nearly 60 members of the community participated in the 6 month state sponsored planning process.
Principal Findings: (1)In designing the health improvement process, community stakeholders commonly linked coordination of care with prevention efforts even where their original intention or focus was one or the other (2)Though longer planning processes resulted in fluctuating engagement, it often meant that the groups were better prepared for implementation (3)Planning progress and implementation were more likely to occur where the private sector ( not-for-profit/forprofit) was centrally engaged (4)The larger the community size, the greater the challenges to stakeholder expectations, success and sustainability.
Conclusion: Planning for health reform at the community level can by itself succeed in changing community norms and preparing resident groups to tackle the adverse impact of chronic disease. Best opportunities for successful initiation and implementation include the full engagement of public and private sector partners, the choice of a well defined community and an approach that focuses on both prevention and coordination of care and values policy and environmental change alongside program development.
Implications for Policy, Delivery or Practice:
While not negating the need national health reform, this study supports the reinforcement of community level planning processes that are efficient in stakeholder engagement, not over ambitious in scope, and relies on established private and public sector partnerships and leadership.
Funding Source(s): CDC, Kaiser Permanente and
Georgia Department of Community Health
Theme: Prevention and Treatment of Chronic
Illness
● Finding the Voice of Public Health in the
National Health Reform Dialogue: An Integrative
Model for Health System Transformation
Christopher Parker, M.P.H., M.B., B.S.; Karen
Minyard, Ph.D.; Holly Avey, Ph.D., M.P.H.; Mary
Ann Phillips, M.P.H.; Glenn Landers, M.H.A.,
M.B.A.; Angela Snyder, Ph.D., M.P.H., M.Phil.
Presented by: Christopher Parker, M.P.H., M.B.,
B.S., Senior Research Associate, Georgia Health
Policy Center, Georgia State University, P.O. Box
3992, Atlanta, GA 30302-3992, Phone: (404) 413-
0311; Email: alhcap@langate.gsu.edu
Research Objective: U.S. health reform policy discussions have routinely focused on expanding access to health care, containing costs, financing the system and delivering quality care. Within the
Public Health community, much of this conversation is perceived to be occurring without a full consideration of the value of prevention and population-based approaches to improved health.
The objective of this research project was to identify opportunities and establish a framework for Public
Health to become part of the health reform debate.
Study Design: Phase I: Literature and document reviews, focus groups, and key informant interviews were conducted to identify strategic health improvement priority areas and locate the priorities common to both the Public Health and non public health sectors. Phase II: Stakeholders were convened to discuss how to integrate health improvement priorities into health transformation at the local, state, and national levels. Stakeholders returned to their respective constituents to discuss specific strategies and policies, then reconvened to share and integrate what was learned. Nearly 20 sectors and over 500 individuals participated.
Population Studied: Public Health and non public health stakeholders from local, state, and national groups representing academic, government, business, philanthropic and broader health policy communities
Principal Findings: Phase I: Five health improvement priority areas were identified to be common to both sectors and align with stakeholder recommendations for disease prevention, cost/benefit, and upstream approaches: physical activity, obesity and nutrition, tobacco cessation, environment, and access.
Phase II: Stakeholder convenings identified the following strategies: coordination of effort, accountability and shared responsibility, focusing on children and the education sector, making the economic case for disease prevention and health promotion, linking primary care and Public Health, focusing on health disparities and targeting those most at risk, incorporating prevention and wellness benefits into the health care financing system, and using simple, positive messaging. A model was developed to define and visualize three levels of reform efforts. In the model, health care service delivery and financing is located within more comprehensive levels of reform for population/community-based strategies and health in all policies.
Conclusion: By broadening the health reform conversation to include the consideration of population health and the impact of all policies – social, economic and others – on health, an intersectoral process emerged to engage stakeholders at every level as instruments of reform.
Implications for Policy, Delivery or Practice: The margin between health care system reform and community-based reform offers the greatest opportunity for progress in advancing the health transformation dialogue. Health reform possibilities include the integration of disease prevention and primary care. Reform that targets health in all policies highlights the need to put challenges into context, to understand the relationship between various problems, and to think in advance about both the immediate and more distal consequences of policy decisions. It recognizes that there are multiple realities within a system and that all are valid. This moves the reform rhetoric away from
“either/or” or “right/wrong”, to a more inclusive
“both/and” view of proposed interventions. By broadening the health reform conversation to include health in all policies, an intersectoral process emerges that creates “seats at multiple tables for everyone.”
Funding Source(s): CDC
● The Effectiveness of Adolescent Depression
Screening in Primary Care
Helen Parsons, M.P.H.; Donna McAlpine, Ph.D.;
Karen Kuntz, Sc.D.
Presented by: Helen Parsons, M.P.H., Research
Assistant, Division of Health Policy & Management,
University of Minnesota, Mayo Mail Code 729, 420
Delaware Street, Southeast, Minneapolis, MN
55455, Phone: (612) 625-9135; Email: pars0100@umn.edu
Research Objective: Depressive disorders are common in younger persons and are associated with functional impairment, decreased academic performance and increased risk for suicide. Yet, for many adolescents, depression goes undetected and untreated. The objective of this research is to assess whether screening adolescents for depression in primary care settings is an effective strategy to address these problems. Through a systematic literature review we assess effectiveness according to five criteria: 1) the burden of disease on the population; 2) the presence of a detectable asymptomatic stage; 3) the availability of effective treatments for the condition; 4) the relative effectiveness of treatment
delivered early; and 5) the test performance of the screening instrument.
Study Design: We assessed effectiveness through a systematic literature review. We conducted a
MEDLINE and PsychInfo search of peer-reviewed literature published within the past 15 years to determine the most recent estimates of disease prevalence and treatment patterns, as well as recommendations for adolescent depression screening and treatment. Articles were selected for inclusion in prevalence, treatment and outcomes estimates if the study was based on a representative US sample published after 1993.
Estimates for treatment effects were only included if they originated from randomized placebo controlled trials. Approximately 300 articles were identified and 150 articles were selected for further review.
Population Studied: Adolescents ages 12-18 in the United States.
Principal Findings: Estimates of depressive symptoms and disorders were highly variable in the literature, ranging from 1% to 30% in the adolescent population. Current screening tools to detect depression have variable sensitivity and specificity, the most popular having a sensitivity ranging from
71 to 100 and specificity ranging from 57 to 93.
Once identified, current treatments showed varying short-term success, with the largest reduction in depressive symptoms seen in patients undergoing behavior therapy or a combination of behavior and drug therapy in up to 70% of patients. However, studies indicated that remission recurred in up to
31% of these patients. Additionally, studies have increasingly focused on the safety of prescribing depression medications to adolescents and young adults, with indications of potential increases in suicide and suicidal thoughts.
Conclusion: While the prevalence of the disorder and its comorbidities are indicative of the importance of depression in the adolescent population, detecting the disease at a set interval such as through screening programs may be costly, identify a large proportion of individuals who may not benefit from treatment, and place a great deal of stress on an overburdened health system. The cyclical nature of the disease presents the challenge of not identifying many adolescents through this process. Therefore, general medical screening for adolescent depression is not recommended at this time.
Implications for Policy, Delivery or Practice:
Despite their popularity, there is no evidence that implementing depression screening programs in primary care would be an effective strategy to reduce the burden of disease on the population.
Programming should instead be prioritized to improve the effectiveness of screening tools and disease management of depression in adolescents.
● Community Infrastructure that Promotes
Consumer Access & Exchange of Health
Information
Karen Paschal, D.P.T., M.S.; Kimberly Galt,
Pharm.D., Ph.D.; Sue Crawford, Ph.D.; Kevin Fuji,
Pharm.D.
Presented by: Karen Paschal, D.P.T., M.S.,
Associate Professor, Physical Therapy, Creighton
Health Services Research Program, Creighton
University, 2500 California Plaza, Omaha, NE
68178, Phone: (402) 280-5690; Email: kpaschal@creighton.edu
Research Objective: Access to health information is critical for consumers engaged in making healthcare decisions for themselves and family members. This project describes community infrastructure that facilitates interactions between health care professionals, organizations, and consumers to improve access to health-related knowledge.
Study Design: An exploratory sequential mixed methods design was used. The first phase of the study was a qualitative in-depth examination of consumer access to health information and exchange through interviews. Each was audiotaped, transcribed verbatim and coded. Thematic analysis was conducted across interviews. A quantitative phase followed with a survey using a modified Dillman technique to construct a broaderbased community inventory of access to health information. Data from both phases were integrated to describe access and availability.
Population Studied: Nine residents of two rural
Nebraska communities were identified for interviews through contacts of convenience within the state government, private health care organizations, health professions organizations, interfaith ministries offices in Nebraska, the state
Offices on Aging, and the consumer board of the state Quality Improvement Organization. The surveyed sample consisted of 4,256 consumers from non-metropolitan Nebraska counties drawn from the InfoUSA database.
Principal Findings: Consumers sought health information to increase their general knowledge base and for use in making health care decisions for themselves and/or their family members. A variety of means to access and exchange health information was described. All described person-toperson interactions with health care professionals, family members, friends and acquaintances, and at community programs and events (e.g., churches, senior center, health fairs, informal gatherings at a coffee shop or bakery). Printed materials (books, brochures, and newspapers), television, and the
Internet provided additional information. Of particular interest was the importance consumers placed on the knowledge they gained from healthrelated and non-health-related community programs, events, and gatherings. Initial survey results confirmed these qualitative findings.
Finding information on the Internet lagged behind more traditional sources of information for rural
Nebraska residents, but over half reported using the
Internet to some extent to obtain health information.
Conclusion: Community infrastructure, both formal and informal, is a key component in consumer access and exchange of health information.
Individuals access and share health information in a variety of health care and non-health care venues within their communities. These are well-known and respected by community members.
Implications for Policy, Delivery or Practice:
This work may result in understanding the importance of local community infrastructure in improving the quality of healthcare of its citizens through the development and dissemination of accurate and accessible health-related resources.
The value of person-to-person interaction and community involvement presents opportunities to develop strategies for increasing consumer engagement in healthcare. In addition, opportunities may emerge for community infrastructure to facilitate increased access to health information through the use of technology.
Funding Source(s): Nebraska Office of Rural
Health
Theme: Consumer Choices in Health Care
● Consumers’ Interest & Potential Usage of
PHRs: Findings from a Regional Health
Information Organization’s Consumer Survey
Vaishali Patel, Ph.D., M.P.H.; Rina Dhopeshwarkar,
M.P.H.; Alison Edwards, M.S.; Yolanda Barron-
Vaya, M.S.; Jeffrey Sparenborg, B.A.; Lisa Kern,
M.D., M.P.H.; Rainu Kaushal, M.D., M.P.H.
Presented by: Vaishali Patel, Ph.D., M.P.H.,
Assistant Professor, Pediatrics & Public Health,
Weill Cornell Medical College, 411 East 69th Street,
KB313, New York, NY 10021, Phone: (212) 746-
4253; Email: vap2005@med.cornell.edu
Research Objective: New York State is undertaking a major health information technology
(health IT) initiative to improve the adoption and usage of health IT by providers and consumers, including funding projects to support personal health records (PHRs). In its first phase, NY State provided $53 million to 26 grantees, each an alliance of several health care stakeholders.
Despite the heavy public investments in health IT and the recognition that consumers are a major stakeholder, limited work has focused on consumers’ perceptions and usage of health IT.
We conducted a study to describe consumers’ perceptions of health IT, and to explore factors associated with their interest and use of PHRs to access and manage their healthcare information.
Study Design: We conducted a cross-sectional survey of English-speaking adults in the waiting rooms of 5 primary care physician practices participating in a pilot program operated by the
Greater Rochester Regional Health Information
Organization (gR-RHIO), a NY State health IT grantee that provides an online platform for health information exchange to providers across nine counties. They have also conducted a consumer outreach program. Logistic regression models were used to determine factors associated with consumers’ potential usage of personal PHRs.
Population Studied: A total of 117 individuals responded to the survey. Over 73% (n=79) of respondents were women and 84% (n=90) were white. Fifty-two percent (n=56) of respondents were between 45-64 years old, 38% (n=41) between 18-
44, and 10% (n=11) were older than 65. About
50% (n=53) of the respondents were college graduates and 72% (n=77) were employed full time.
About 89% (n=103) had access to the Internet and
79% (n=90) used the Internet daily or more often.
Principal Findings: About 57% (n=61) of respondents reported being interested in accessing their personal health information online.
Respondents were interested in viewing their medical data, such as test results (n=67) and medical record (n=73), doing administrative tasks such as filling out paperwork (n=69), and using interactive features, such as receiving reminders for preventive care (n=71). About 75% (n=77) of respondents stated they would potentially use these
PHR-related features. Preliminary analyses identified a number of factors associated with consumers’ potential usage: increased usage of the
Internet, higher comfort with sharing personal information online, positive perceptions regarding the health benefits associated with managing their healthcare online, user-friendliness of a PHR website, higher education, being employed and having a child in the household.
Conclusion: A majority of consumers who may be affected by the RHIO’s health IT activities expressed interest in accessing their personal health information and using clinical and administrative PHR features. However consumers’ potential usage of a PHR would likely be limited unless concerns regarding the privacy and security of their data are addressed.
Implications for Policy, Delivery or Practice:
Initiatives that seek to engage consumers in the use of PHRs should consider addressing concerns regarding the privacy and security of their personal health information in addition to providing advanced clinical features and a user-friendly website.
Funding Source(s): Greater Rochester Regional
Health Information Organization
Theme: Health Information Technology
● Specialty Care Referral Patterns for the
Underserved: A Study of Community Health
Centers on the South Side of Chicago
Gail Patrick, M.D., M.P.P.; Tony Navarra, B.A.; Irma
Hasham, B.S.; John Hickner, M.D., M.S.
Presented by: Gail Patrick, M.D., M.P.P., Clinical
Assistant Professor, Family Medicine, University of
Chicago, 5841 South Maryland Avenue MC7110,
Chicago, IL 60645, Phone: (773) 834-9852; Email: gpatrick@uchicago.edu
Research Objective: Although community health centers (CHCs) have had some success meeting the primary medical care needs of underserved populations in the US, access to specialty medical care remains a problem. Few studies have examined specialty care referrals for underserved patients. The purpose of this pilot study was to describe the process and outcomes of specialty referrals from CHCs on Chicago’s South Side.
Study Design: We performed chart reviews and examined referral logs of patients referred for specialty medical care during 3 two-week periods in
2007-2008. We included only referrals to medical specialists that required a face-to-face visit.
Referrals to emergency departments were excluded. We gathered data about patient demographics and insurance status, reason for referral, specialty requested, dates of referral and specialty appointments, documentation of attendance at the appointment, and documentation of receipt of a report from the specialist.
Population Studied: Patient’s ages ranged from 1 week to 90 years; 119 (47%) patients were under
19 years old. Fifty-three percent of patients were women; 46% were African-American and 46% were
Latino. Insurance status was: 51% Medicaid, 15%
Medicare, 13% uninsured, and 17% private insurance.
Principal Findings: Two hundred fifty-six (256) medical specialty referrals were made during the study period. The main specialties requested for adults were gastroenterology (51(37%)), general surgery (15(11%)), and a four-way tie among dermatology, otolaryngology, orthopedics and neurology (9 each(7%)). The main specialties requested for children were otolaryngology
(21(18%)), dermatology(17(14%)) and orthopedics
(12(10%)). There was no documentation in the chart regarding whether the patient attended the specialty appointment for 126(49%) referrals, and
21(8%) patients were known not to have attended the visit. There were significant time lags for receiving an appointment regardless of insurance type, with a trend toward longer wait times for patients with Medicaid (42 days) or who are uninsured (39 days) compared to privately insured patients (30 days), p = 0.053. A referral letter from the specialist was documented in the patient’s health center chart for 34%(88) of referrals.
Conclusion: The top requested specialties are the same as those identified in other studies as difficult for underserved patients to access. A large proportion of patients do not show up for specialty referral appointments. Insurance status is an important predictor of timely access to specialty care. Reports from specialty visits often are not available in the health center chart.
Implications for Policy, Delivery or Practice: Our findings from this pilot study suggest that there are major safety and quality issues related to specialty medical referrals for underserved urban populations. Our conversations with providers from other US cities suggest this is a widespread problem. Understanding the referral process and knowing the specialties in high demand will allow us to target interventions and resources to areas of greatest need. Further research is needed to understand the primary-specialty care interface on a national level and to gather patient perspectives on this issue. Expanding access to specialty care will require creative solutions and a close examination of the roles of primary and specialty care providers.
Funding Source(s): Michael Reese Health Trust,
Chicago, IL
Theme: Coverage and Access
● Safety Culture in Emergency Medical
Services: A National Study
Daniel Patterson, Ph.D., M.P.H.; David Huang,
M.D., M.P.H.; Rollin Fairbanks, M.D., M.S.; Henry
Wang, M.D., M.S.; Scott Simone, B.S., N.R.E.M.T.-
P.; Matthew Weaver, N.R.E.M.T.P.
Presented by: Daniel Patterson, Ph.D., M.P.H.,
Research Assistant Professor, Emergency
Medicine, University of Pittsburgh School of
Medicine, 230 McKee Place, 4th Floor, Pittsburgh,
PA 15213, Phone: (412) 647-3183; Email: pattersond@upmc.edu
Research Objective: Safety culture refers to workers’ perceptions and collective beliefs regarding the conditions and safety of workplace operations. Some evidence suggests that safety culture is associated with patient outcomes. The
Joint Commission and other leading authorities recommend periodic assessment of safety culture as part of any quality improvement program. This study characterized perceptions of safety culture in a nationwide sample of Emergency Medical
Services (EMS) agencies.
Study Design: We used a cross-sectional study design and a modified version of the Safety
Attitudes Questionnaire (SAQ) to measure safety culture. The SAQ is a previously validated instrument that measures workplace safety culture across six domains Safety Climate, Teamwork
Climate, Perceptions of Management, Job
Satisfaction, Working Conditions, and Stress
Recognition. We evaluated both the mean score
(score 0-100) and proportion of respondents that had a positive perception of the domain measured
(% positive). We identified variations in domain scores and pct. of positive responses across agency and respondent characteristics using
ANOVA and Chi-Square tests while controlling for clustering at the agency level.
Population Studied: We received 1,715 completed surveys from a convenience sample of 61 EMS agencies spread across all 4 major census regions
(mean agency response rate = 47.3% +/-18.7%).
Agencies were predominately ground-based delivery models (80.3%), employed between 21 and
50 employees, and saw <2,500 patients annually.
Respondents were 68% male, 60% paramedic certified with 9.5 mean years of agency affiliation.
Principal Findings: There was wide variation in mean domain scores across EMS agencies
(mean±SD): Safety Climate (74.4±17.8), Teamwork
Climate (70.9±20.4), Perceptions of Management
(65.9±22.3), Job Satisfaction (75.1±20.2), Working
Conditions (66.5±21.7),Stress Recognition
(56.1±23.7). The % positive responses across all 6 domains were low and varied across all agencies
(%, 95%CI); Safety Climate (56.8%, 51.3-62.3),
Teamwork Climate (51.0%, 45.2-56.9), Perceptions
Of Management (42.4%, 35.6-49.1), Job
Satisfaction (61.4%, 55.8-67), Working Conditions
(44.7%, 39.9-49.5), Stress Recognition (28.1%,
24.7-31.5). Both the domain mean scores and % positive responses were correlated with agency and respondent demographics. Respondents working at air-medical EMS delivery models tended to score higher across all safety culture domains compared to ground only EMS delivery models. Increases in annual patient contacts was associated with a decrease in safety culture scores across all domains except stress recognition.
Conclusion: Safety culture in EMS varies widely between agencies across the United States.
Implications for Policy, Delivery or Practice:
This is the first study characterizing safety culture in the Emergency Medical Services (EMS) on a national level. Findings may be used by local EMS agency directors and officials at all levels of government as a benchmark for evaluating safety conditions in the EMS setting.
Funding Source(s): Foundation for Air-Medical
Research and Education (FARE) and Pittsburgh
Emergency Medicine Foundation (PEMF)
● Implementing Change in Nursing Unit Work
Environment
Marjorie Pearson, Ph.D., M.S.H.S.; Jack
Needleman, Ph.D.; Pat Parkerton, Ph.D., M.P.H.;
Valda Upenieks, Ph.D., M.P.H., R.N.; Lynn Soban,
Ph.D., M.P.H., R.N.; Tracy Yee, M.P.H.
Presented by: Marjorie Pearson, Ph.D., M.S.H.S.,
Social Scientist, Health, The RAND Corporation,
1776 Main Street, Santa Monica, CA 90401,
Phone: (310) 393-0411; Email: mpearson@rand.org
Research Objective: Midst widespread concern about the nursing shortage, the Transforming Care at the Bedside (TCAB) initiative was launched to help hospitals engage frontline staff in organized change processes to simultaneously improve the work environment and care on medical-surgical units. Our goal is to assess the hospitals’ implementation of these change processes and to analyze the relationship between their implementation and improved nurse vitality.
Study Design: We used multiple data sources
(project documentation, collaborative meeting observation, and 150 semi-structured hospital interviews) and mixed qualitative and quantitative methods to analyze the hospital-level implementation of the organized change processes promoted by TCAB (unit-based improvement teams, brainstorming sessions, nurse-led rapid tests of change, and use of measurement data in the improvement process). We developed indicators of organized change process performance, quantity of innovations tested, and change in workplace vitality. Descriptive statistics were used to assess the units’ performance of individual change processes and the intensity of their performance. Parsimonious regression analyses were used to examine the relationship of change process performance and quantity of innovations to staff vitality. All analyses were at the nursing unit level.
Population Studied: The study population included
17 pilot units in the 13 hospitals participating in the second phase of TCAB.
Principal Findings: Implementation of TCAB’s organized change processes varied but was high overall. In the majority of units, eighty percent or more of the nurses participated in tests of change.
Over 400 innovations were tested, and unit managers and staff from 71% of the units reported an increase in nursing staff vitality. Bivariate analyses revealed a positive association between
TCAB implementation and staff vitality, both for the indicator of implementation intensity (p<.01) and individual change process indicators (unit team decision making on tests of change (p<.05); early involvement of nurses in tests of change (p<.10); ongoing and frequent team meetings (p<.05), the centrality of the unit manager’s role (p<.01), the leadership of the nursing unit in TCAB activities
(p<.05), and the consistent provision of staff time for
TCAB activities (p<.05).) TCAB process implementation was not significantly associated with the number of innovations tested. The parsimonious regression analyses showed that the intensity of the unit’s implementation of TCAB processes remained strongly related to increased staff vitality when the number of innovations tested was held constant (p<.01).
Conclusion: These findings demonstrate that it is feasible to engineer nurse engagement in change.
Unit team decision making, frequent team meetings, brainstorming sessions, and use of measurement data are viable mechanisms for engaging nurses in change and are associated with improved staff vitality. This engagement is orchestrated through the central role of the nursing unit and unit manager in directing these processes.
While the study design does not allow causal inferences, the significant association between implementation of many of these mechanisms and
increased nurse vitality suggests that nurse engagement in improvement processes may be a critical intermediate variable leading to improved workplace vitality.
Implications for Policy, Delivery or Practice:
Future efforts to improve work environment on nursing units would do well to focus on implementation of these mechanisms for staff participation in change.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Organized
Processes
● Public Health Workforce Data Resources
Robin Pendley, M.P.H., B.S.; Tourette Jackson,
M.P.H., B.S.
Presented by: Robin Pendley, M.P.H., B.S.,
Research Assistant, College of Public Health,
University of Kentucky, 121 Washington Avenue,
Suite 107, Lexington, KY 40536-0003, Phone:
(859) 218-2193; Email: rppe222@uky.edu
Research Objective: To compile existing public health workforce data resources and make resources available to researchers through National
Library of Medicine’s Health Services and Sciences
Research Resources and Center of Public Health
Systems and Services Research (CPHSSR) websites.
Study Design: Extract existing public health workforce data resources from Association of State and Territorial Health Officials (ASTHO) affiliate organizations such as Council of State and
Territorial Epidemiologists (CSTE), Association of
Maternal and Child Health Programs (AMCHP),
Association of Public Health Laboratories (APHL),
Association of State and Territorial Dental Directors
(ASTDD), and Association of State and Territorial
Public Health Nutrition Directors (ASTPHND) on through own organization enumeration efforts.
Principal Findings: There are individual affiliate organization enumeration of public health workforce efforts. However, there appears to be a gap crosscommunication between organizations in enumeration efforts.
Conclusion: There is appears to be a gap in internal communication within affiliate organizations and in research efforts within public health workforce enumeration.
Implications for Policy, Delivery or Practice:
Researchers having knowledge and access to public health workforce data resources will close the gap and foster growth and interest in public health workforce and policy to assure a competent public health workforce. This will help support a national effort to evaluate public health workforce issues.
Theme: Health Care Workforce
● A Multimodality Outpatient Quality
Improvement Intervention Using Precision
Performance Measurement Embedded Within an
Electionic Health Record
Stephen Persell, M.D., M.P.H.; Darren Kasier, M.A.;
Nancy Dolan, M.D.; Elisha Friesema, B.A.; Jason
Thompson, B.A.; Beth Andrews, Ph.D.; David
Baker, M.D., M.P.H.
Presented by: Stephen Persell, M.D., M.P.H.,
Assistant Professor, Medicine, Northwestern
University, 750 South Lake Shore Drive, 10th Floor,
Chicago, IL 60611, Phone: (312) 503-6464; Email: spersell@nmff.org
Research Objective: Quality improvement (QI) strategies such as audit and feedback, reminders, and clinical decision support (CDS) have achieved modest improvements. A limiting factor is that many apparent quality failures are due to unmeasured contraindications or patient refusals (exceptions).
This limits the ability to identify patients with true quality deficits. We aimed to assess the impact of a multimodality QI intervention that relies on capturing exceptions to permit more precise quality measurement and more accurate CDS and outreach.
Study Design: We programmed 16 chronic disease and preventive care quality measures into the EHR CDS and implemented a QI intervention for all patients in the practice consisting of reminders with linked order sets, standard ways to document patient or medical exceptions, and performance feedback and monthly lists of patients not receiving essential medications to physicians.
We used recorded exceptions for focused QI activities (peer review and academic detailing, counseling about drug costs, and educational outreach to patients refusing services). The primary outcome was the proportion of patients eligible for each measure without a quality deficit (either met the measure or had an exception recorded).
Secondary outcomes were (1) the proportion that met the measure and (2) the proportion with exceptions. We conducted time series analyses using data extracted from the EHR to determine changes in quality over the 10 months of the intervention and the statistical significance of changes in the rate of improvement during the intervention period compared to the previous year.
Population Studied: The study was done at a large academic internal medicine practice (37 attendings, 43,900 patients per year) with a commercial electronic health record (EHR).
Principal Findings: For 9 measures, the proportion of patients without a quality deficit improved significantly over the 10 month implementation: antiplatelet in coronary heart disease (CHD)
(improved from 90.2% to 95.3%), lipid drug in CHD
(87.9% to 93.2%), ACE inhibitor in diabetes (DM) and CHD (83.7% to 88.9%), beta blocker after myocardial infarction (91.0% to 94.3%), beta blocker in heart failure (HF) (83.6% to 89.8%), anticoagulation in atrial fibrillation (AFib) and HF
(65.1% to 87.6%), LDL control in DM (52.8% to
58.2%), aspirin in DM (78.6% to 91.9%), and nephropathy screening or treatment in DM (81.2% to 86.8%) (p <0.05 for all changes in time trends).
Many physicians reached 100% on these measures. For 4 other measures, the proportion without a quality deficit improved significantly, but the rate of improvement was not significantly different from the previous year: pneumococcal vaccination (81.1% to 89.2%); cervical cancer screening (80.5% to 87.2%), osteoporosis screening or treatment (77.9% to 82.5%); and colon cancer screening (58.2% to 60.7%); p<0.001 for improvement, p>0.05 for difference in rate of improvement. ACE inhibitor in HF, HbA1c control in
DM, and mammography did not improve. Some measures improved largely due to an increase in patients receiving the service (e.g., aspirin in DM), while others (e.g. anticoagulation for AFib) changed largely from documentation of exceptions; most changes were a combination of both.
Conclusion: A multimodality QI intervention significantly improved quality for 9 of 16 measures.
Pre-existing improvement trends were sustained for
4 others.
Implications for Policy, Delivery or Practice: As the number of quality measures proliferates, this approach could help practices use EHRS to improve quality measurement, CDS and outreach, and achieve very high quality care for many measures simultaneously.
Funding Source(s): AHRQ
Theme: Health Information Technology
● Does Pay-for-Performance Generate Greater
Interest in Adhering to Performance Measures than Audit and Feedback Alone?
Laura Petersen, M.D., M.P.H.; Kate Simpson,
M.P.H.; Tracy Urech, M.P.H.; LeChauncy Woodard,
M.D., M.P.H.; Sylvia Hysong, Ph.D.; R. Adams
Dudley, M.D., M.B.A.
Presented by: Laura Petersen, M.D., M.P.H.,
Principal Investigator, Health Services Research,
Department of Medicine, Baylor College of
Medicine, VA Health Services Research and
Development Service Center of Excellence, Michael
E. DeBakey VAMC (152), 2002 Holcombe
Boulevard, Houston, TX 77030, Phone: (713) 794-
8623; Email: laurap@bcm.edu
Research Objective: Few data exist regarding the effectiveness of pay-for-performance programs, and some claim that financial incentives for quality are ineffective and/or unnecessary in motivating physicians to improve care. We evaluated whether participants who receive financial incentives in addition to audit and feedback for their performance in delivering guideline-recommended hypertension care demonstrate greater interest in their performance than those who receive audit and feedback alone.
Study Design: We used data from an ongoing 12site cluster randomized controlled trial designed to assess the effectiveness of financial incentives
(pay-for-performance) in improving adherence to guideline-recommended hypertension care, as measured by prescription of Seventh Report of the
Joint National Committee on Prevention, Detection,
Evaluation, and Treatment of High Blood Pressure
(JNC 7) guideline-recommended medications, achievement of guideline-recommended blood pressure goals, and appropriate clinical response to uncontrolled blood pressure. Participants were primary care providers and their team members who supported their care of hypertension patients
(98 participants in the intervention arms and 20 in the control arm). All study participants received a baseline educational session about JNC 7 hypertension management as well as audit and feedback on their performance at the end of each four-month study period. Intervention subjects additionally received monetary rewards at the end of a four-month study period based on their adherence to guideline-recommended care.
Participants accessed their feedback reports and viewed any payments earned by logging onto a secure study website, which also provided resources on the JNC 7 hypertension guidelines and information about the study measures and procedures. Having tracked participants’ activity on the study website, we assessed participant interest in viewing their performance results by comparing the following measures in the intervention versus control arms approximately two months after participants had received their website login information: ever logging into the website; ever visiting the page which displays a participant’s feedback report; and ever completing a survey which follows the feedback reports of each participant. A two-sided Fisher’s exact statistic evaluated the significance of each comparison.
Population Studied: Participants were primary care providers and their team members who supported their care of patients with hypertension.
Principal Findings: The percentage of participants having logged into the website was significantly greater in the intervention arms (64%) than in the control arm (30%; p=0.006) after the first incentive period. In addition, a significantly greater percentage of intervention arm participants (60%) viewed their performance feedback reports on the study website compared to control arm participants
(25%; p=0.006). Finally, the percentage of participants who completed the survey following their feedback report was significantly greater in the intervention arms (47%) than in the control arm
(15%; p=0.011).
Conclusion: Participants who receive financial incentives in addition to audit and feedback for their performance in delivering guideline-recommended hypertension care demonstrate greater interest in their performance than those who receive audit and feedback alone.
Implications for Policy, Delivery or Practice:
These preliminary results suggest that, if properly designed, financial incentives have the potential to increase awareness of the extent to which providers’ practice adheres to evidence-based guidelines.
Funding Source(s): VA
● Patient Satisfaction with Primary Care in
Armenia: Are Standard Measurements
Informative?
Varduhi Petrosyan, M.S., Ph.D.; Anahit
Demirchyan, M.D., M.P.H.; Michael Thompson,
M.S., Dr.P.H.; Tsovinar Harutyunyan, M.P.H.
Presented by: Varduhi Petrosyan, M.S., Ph.D.,
Assistant Professor, Center for Health Services
Research & Development, American University of
Armenia, 40 Marshal Baghramian, Yerevan, 0019,
Armenia, Phone: +37410 512564; Email: vpetrosi@aua.am
Research Objective: This study, conducted in
Summer 2006, assessed the level of patient satisfaction with services at select primary health care facilities in Lori and Shirak provinces of
Armenia within the scope of the Primary Health
Care Reform Project funded by the United States
Agency for International Development.
Study Design: Stratified random sampling was used to select participating facilities. Selfadministered questionnaires were distributed to 684 recent clients of 48 selected primary health care facilities. The survey instrument, based on tools used in similar surveys in Armenia and internationally, included a fifteen-item scale that assessed satisfaction with health care providers.
The following domains also were included in the questionnaire: 1) accessibility of care; 2) confidentiality of care; 3) respondent’s ability to obtain prescribed medicines; 4) facility conditions;
5) respondent’s willingness to recommend the provider and respondent’s overall assessment of the care received; 6) respondent’s suggestions for improving services at the clinic; and 7) brief demographic characteristics.
Population Studied: To be eligible, respondents had to be at least age 16 and having visited the selected facility in the last month.
Principal Findings: The majority of patients were satisfied with their provider (mean satisfaction score of 1.75 out of maximum 2). Most patients (89.0%) would visit the same provider again, and would recommend the provider to friends (85.6%).
Satisfaction with other aspects of care, including waiting time, accessibility of services, confidentiality, and cleanliness of the facility also was high (mean score of 1.70 out of 2). Seventyeight percent of respondents considered the care they received to be “excellent” or “good”. The less educated and those in rural areas were more likely to be satisfied with the provider’s quality. The most frequently mentioned suggestions for improving care included improving pharmaceutical supplies
(80.5%), followed by increasing the salary of providers (44.7%), and increasing the facility’s capacity (35.3%).
Conclusion: Clients report being highly satisfied with the care that they receive. This high client satisfaction level reported in this study, however, is in conflict with the current state of primary health care services in Armenia, which is characterized by substantial technical, equipment, and staffing constraints. In this environment, provider-client interactions likely dominate clients’ overall assessments of care in Armenia.
Implications for Policy, Delivery or Practice: The implication is therefore, that patient satisfaction surveys in Armenia –and in other transitional countries – might not provide a valid proxy for the broader concept of quality, of which patient satisfaction is but one component. Quantitative assessment tools for patient satisfaction need further refinement to ensure that they also provide the robust measures of quality needed by policymakers. Qualitative research that explores how people in Armenia assess quality and how they would like to report on it is recommended.
Funding Source(s): United States Agency for
International Development
Theme: Quality and Efficiency: Measurement
● TB Control System in Armenia [civilian and penitentiary sectors]
Varduhi Petrosyan, Ph.D., M.S.; Ruzanna
Grigoryan, M.D., M.P.H.; Nune Truzyan, D.V.M.,
M.P.H.; Hripsime Martirosyan, M.D., M.P.H.; Byron
Crape, M.S.P.H., Ph.D.
Presented by: Varduhi Petrosyan, Ph.D., M.S.,
Assistant Professor, Director, College of Health
Sciences, American University of Armenia, 40
Bagramian, Yerevan, 0019, AM, Phone: +37410
512564; Email: vpetrosi@aua.am
Research Objective: Armenia is facing a reemerging threat from tuberculosis (TB). New registered TB cases have almost tripled from 1995 to 2005. Armenia is ranked the 10th highest for TB incidence rate among the 53 WHO European
Region countries. This threat has increased significantly as the rates of new drug resistant strains of TB have grown. To respond to these increasing threats, the government of Armenia decentralized TB services to better control this disease. This study provides an assessment of stakeholders’ capacities in different TB control activities and an evaluation of potential capacities for support and cooperation with the Ministry of
Health and Ministry of Justice.
Study Design: Qualitative semi-structured in-depth interview and focus group discussion methodologies were used by the research team from the Center for Health Services Research and
Development at the American University of
Armenia.
Population Studied: Participants at all levels of TB control: experts and administrators; providers of primary and specialized care; auxiliary medical personnel; patients.
Principal Findings: Study participants provided a number of strengths in the TB control system in
Armenia. The establishment of the National TB
Program (NTP) provided a national infrastructure for TB control. The decentralization process brings services closer to where patients live. Free TB services and TB medication are provided to the entire population. Successful collaborations in TB control have been seen in the penitentiary system and DOTS+ pilot sites. Some of the leading challenges identified included lack of definition of roles and responsibilities in TB control in the civilian sector and problems in recruiting and retaining qualified professionals in TB control in both penitentiary and civilian sectors due to lower compensation and fewer benefits. Participants also indicated lack of modern standards for infection control, inadequate physical infrastructure and substandard equipment and supplies. Participants widely indicated that there was a lack of capacity to conduct rigorous DOTS in the civilian sector and to expand DOTS+ to both civilian and penitentiary sectors. Moreover, inappropriate management of
TB drugs by health providers [in pharmacies and dispensaries] was mentioned by participants as an important problem in TB control in Armenia. One of the important findings was stigma and fear of TB that exist among both the general population and health care workers. This stigma contributes to late utilization of TB services, negative social consequences and spread of disease.
Conclusion: Recommendations to meet these challenges include improving supervision, developing quality control systems, increasing resources in TB control and further developing protocols, procedures and regulations. Apart from improvement in health care system, participants recommended awareness campaigns as part of TB control to change the negative attitudes and stigma among the general public.
Implications for Policy, Delivery or Practice: The findings suggest that a transitional collaboration between the NTP with its national expertise and the
International Committee of the Red Cross (ICRC) and the Médecins Sans Frontières France (MSF) with their international experience may most effectively improve TB control and resolve the deficiencies in the newly decentralized TB control system.
Funding Source(s): International Committee of the
Red Cross
Theme: Global Health
● Systemic Hypertension & Risk of Obstructive
Sleep Apnea Syndrome: A Case- Control Study in Yerevan, Armenia
Varduhi Petrosyan, Ph.D., M.S.; Ruzan Udumyan,
M.P.H.; George Piligian, M.D., M.P.H.; Parounak
Zelveian, M.D., Ph.D., F.E.S.C.
Presented by: Varduhi Petrosyan, Ph.D., M.S.,
Assistant Professor, Director, College of Health
Sciences, American University of Armenia, 40
Bagramian, Yerevan, 0019, AM, Phone: +37410
512564; Email: vpetrosi@aua.am
Research Objective: Obstructive sleep apnea syndrome (OSAS) has been recognized in the
Western world as a public health burden, but there have been no OSAS related epidemiological studies conducted in Armenia. The aim of the study was to measure the independent association between risk of OSAS and systemic hypertension
(HTN) in adult people living in Yerevan, the capital of Armenia, considering all known confounders.
Study Design: The study utilized a case-control design. The control variables of interest were: age, gender, BMI, neck circumference, waist circumference, waist-to-height ratio, smoking status, weekly alcohol consumption, coffee consumption, weekly physical activity, and presence of diabetes or renal diseases. The study measured the OSAS risk using the Berlin
Questionnaire.
Population Studied: The study population (108 cases and 157 unmatched controls) came from the sample population (170 hypertensives and 578 normotensives) of Hypertension Extended Study in
Armenia (HESA) conducted in 2004 by the
Armenian Medical Association in Yerevan.
Principal Findings: The odds of HTN in people with high risk of OSAS was 2.17 times greater (95%
CI: 1.02, 4.63) than odds of HTN in people with low risk of OSAS after adjusting for age, neck circumference and co-morbidities (diabetes and/or renal disease).
Conclusion: This case-control study found an independent positive association between OSAS risk and HTN. The study identified fat deposition in the neck as an influential determinant of HTN and
OSAS risk, and indicated that development of diabetes and renal disease were important risk factors for HTN.
Implications for Policy, Delivery or Practice:
There is a need to increase population awareness of OSAS risk factors, symptoms and consequences, as well as to consider the OSAS when developing national clinical guidelines, particularly in those related to the management of
HTN.
Theme: Prevention and Treatment of Chronic
Illness
● Hospital Expenses in the United States: State
Level Analysis
Varduhi Petrosyan, Ph.D., M.S.; Gerard Anderson,
Ph.D.
Presented by: Varduhi Petrosyan, Ph.D., M.S.,
Assistant Professor, Director, College of Health
Sciences, American University of Armenia, 40
Bagramian, Yerevan, 0019, AM, Phone: +37410
512564; Email: vpetrosi@aua.am
Research Objective: Hospital spending has been a significant contributor to health care spending growth in the United States. Interest in the factors explaining hospital cost increases has intensified during the recent years among economists and policy makers. Better understanding of the factors behind hospital expenses could help develop longterm and sustainable policy solutions. The current study contributes to the literature by analyzing longrun state level hospital spending drivers. It explores the short- and long-run relationships between per capita state level hospital expenses and four factors commonly believed to be associated with increased spending: labor, availability of technology, medical education, and per capita aggregate income.
Study Design: During the last three decades, analysis of the determinants of hospital spending has focused on short run causes of hospital spending increase and has ignored the long-term relationships. This study conducts an individual state time-series analysis, applies macroeconometric analysis methods (Johansen’s cointegration modeling) to state-level data, and reports on long-run as well as short-run relationships. The augmented Dickey and Fuller
(ADF) and the Kwiatkowski-Phillips-Schmidt-Shin
(KPSS) tests offer robust evidence that the most state level data are non-stationary in levels and stationary in first or second differences in the period
1977 to 2002. Therefore, Johansen’s cointegration testing is applied to look at the relationships between the variables of interest.
Population Studied: Hospital data are extracted for the 50 US states and DC from the American
Hospital Association’s (AHA) Annual Survey of
Hospitals for each of the years 1977 through 2002.
This study focuses on all non-federal, short-term general and special hospitals with facilities and services available to the public.
Principal Findings: This testing provides evidence for a significant cointegrating relationship at the 5% significance level in 37 states. This means cointegrating vector is identified that reflects the long-run relationships between the variables of interest.
Conclusion: The study provides evidence that labor and economic development are major longrun drivers behind accelerating hospital costs. It also suggests that in majority of the US states medical technology does not significantly affect the pace of growth of hospital expenses in the long run, and similarly teaching commitment does not raise the cost of hospital care. These findings remain robust when adjusted for the patient severity.
Implications for Policy, Delivery or Practice:
This study shows to policy makers that labor cost is one area where they could intervene to limit cost growth.
● The Impact of ‘Social Capital’ on the ‘Overall
Perceptions of Safety’
Holger Pfaff, Ph.D.; Antje Hammer; Oliver Ommen,
M.D. M.P.H.; Nicole Ernstmann; Christian Günster;
Günther Heller, M.D.
Presented by: Holger Pfaff, Ph.D., Head of the
Center, Center for Health Services, University
Cologne, Eupener Straße 129, Cologne, 50933,
DE, Phone: +49 221 478 97101; Email: holger.paff@uk-koeln.de
Research Objective: Prior research has found that management and leadership factors have an impact on safety culture outcomes. However, little or no research exists on the influence of ‘Social Capital’ on safety culture outcomes. Therefore, the aim of our study was to analyse the relationship between
‘Social Capital’ and the top-managements perception of safety in a hospital.
Study Design: In 2008 we conducted a crosssectional representative study with 1224 medical directors from all German hospitals who reached the following inclusion criteria: at least one internal medicine and one surgery unit. The standardized, written questionnaire contained a reliable scale measuring ‘Social Capital’ and a measure for the
‘Overall Perceptions of Safety’ from the HSOPSquestionnaire. We use a multiple linear regression model to examine the relationship between the
‘Social Capital’ of the hospital and the ‘Overall
Perceptions of Safety’.
Population Studied: We received questionnaires of 551 medical directors. The response rate was
45.2 %.
Principal Findings: We found a significant association between ‘Overall Perceptions of Safety’ and the dimensions ‘Social Capital’. Explained variance for both variables is 21,0 %.
Conclusion: We analysed data from the German medical directors’ point of view. We found a significant association between the hospital ‘Social
Capital’ and the ‘Overall Perceptions of Safety’.
Implications for Policy, Delivery or Practice:
This analysis emphasizes the importance of hospital culture for patient safety issues. Improving safety culture should begin with investments into the ‘Social Capital’ of the hospitals.
Funding Source(s): Bundesärztekammer
Theme: Organizational Performance and
Management
● Systems of Care for New Moms: Integrating
Depression Treatment
Susan Pfefferle, Ph.D.; Ana Baumann, Ph.D.; Paul
Sterzing, M.S.W.; Ben Cooper, B.A.; Debbie
Layton, R.N.; Samantha Books, M.P.E.
Presented by: Susan Pfefferle, Ph.D., Research
Associate, George Warren Brown School of Social
Work, Washington University in Saint Louis, 1
Brookings Drive, Campus Box 1093, Saint Louis,
MO 63130, Phone: (314) 935-6376; Email: spfefferle@gwbmail.wustl.edu
Research Objective: Our study involves adaptation and co-location of a depression treatment, problem solving therapy, to fit low income mothers' preferences and beliefs and provision by nurses through home visitation. This study is a collaboration between a nurse home visitation agency and university researchers.
Study Design: In this mixed methods study we gathered contextual data on women's beliefs about depression and their preferences for depression treatment. We also gathered data from home visitation nurses and community providers in order to capture information about women's barriers to depression treatment and any service system characteristics that might impact our adaptation of problem solving therapy for use in nurse home visitation. We conducted three focus groups with women who scored 10 or above on the CES-D 10.
Focus groups were conducted at a local service agency with on site child care. We also provided transportation to the site. Each focus group lasted about 1 hour. We also conducted in person semistructed interviews with four nurses and telephone semi-structured interviews five community key informants who work with the home visitation agency and the depressed mothers. Finally, we administered a survey to nurses at the home visitation agency to explore if we needed to conduct an organizational intervention prior to adaptation of problem solving therapy. The survey included the psychological safety scale, the evidence-based practice attitudes scale, the relational coordination scale, and the organizational social context survey. 26 nurses completed the survey for a response rate of 68%. Thematic analysis of the qualitative data were performed by two researchers,who independently analyzed the data by type: focus group, nurses in person interviews, and telephone interviews with community providers. A theory driven immersioncrystalization approach was used for analysis of the qualitative data. Master reports were written once researchers came to consensus on themes. These reports then were compared with the existing literature on depression treatment for low income mothers and reports were written synthesizing our local and global data. These reports were then used to make treatment adaptation decisions. Survey data were used to gather a contextual picture of the work environment in the home visitation agency.
Scales were created and bivariate analyses completed in order to assess the need for an organizational intervention prior to implementation of the adapted problem solving therapy.
Population Studied: Low income depressed mothers served through nurse home visitation and community key informants in the St. Louis, MO. metropolitan area.
Principal Findings: Major themes from our qualitative data included the "myth of the strong woman", as well as religiosity and fatalism as ways these low income depressed women coped with depression. These women acknowledged feeling depressed but expressed the sentiment that
"whatever doesn't kill me makes me stronger." They also had extreme distrust of service providers other than their nurses. Trust with the nurses took time to develop. Nurses and community providers identified these themes as well, although they were more likely than the women to identify competing demands and barriers such as lack of insurance or transportation as well as poor coordination between providers. Our survey results showed that no organizational intervention was needed prior to implementing the adapted problem solving therapy.
Nurses reported high psychological safety and relational coordination. They also expressed positive attitudes toward the use of evidence based practices.
Conclusion: Prior to co-locating depression treatment in nurse home visitation it was necessary to gather contextual data to adapt the treatment.
Women focused on internal barriers to treatment while nurses and community providers identified internal and external barriers and competing demands that interferred with receipt of depression treatment.
Implications for Policy, Delivery or Practice: Our methodology and findings provide useful information for both providers and services researchers as they attempt to culturally adapt treatments for underserved populations. Our study also points to important service system gaps for low income women with mild to moderate depression.
Funding Source(s): NIMH
Theme: Behavioral Health
● Provider Reactions to Declining Medicare
Reimbursement for Separately Paid Outpatient
Drugs
Christine Pierce, R.N., B.S., M.S.N.; Michael
Ziskind, M.A., M.Sc.
Presented by: Christine Pierce, R.N., B.S., M.S.N.,
Partner, The Resource Group, P.O. Box 373,
Richfield, OH 44286, Phone: (330) 659-2324;
Email: cpierce@consultresourcegroup.com
Research Objective: To assess hospital providers regarding actual or planned operational changes in response to progressively decreasing Medicare reimbursement for separately paid outpatient drugs.
Study Design: Survey-based. A survey instrument consisting of both quantitative and qualitative queries was developed, evaluated and pre-tested.
Preparation input included review of legislation and
Medicare regulations relevant to outpatient drug payment. Questionnaires were mailed to the
Director of Pharmacy and Chief Financial Officer at
4,000 U.S. hospitals. Participation was voluntary and respondents received no remuneration.
Population Studied: For-profit and not-for-profit
U.S. hospitals, having a Director of Pharmacy,
Chief Financial Officer, or both. The following hospital categories were excluded by design: Kaiser
Permanente; Maryland hospitals; Department of
Defense or Veteran’s Hospitals, and Children’s
Hospitals.
Principal Findings: There were 364 valid questionnaires returned for a response rate of nearly ten percent. Both for-profit and not-for-profit hospitals were represented, ranging in size from less than 100 beds to over 500 beds. At least one response was received from each of 48 states and over thirty percent reported classification as disproportionate share hospitals. Greater than eighty six percent responded that the 2008
Medicare reimbursement for separately paid drugs, at average sales price plus five percent, was not adequate to cover hospital costs for both drug acquisition and costs associated with pharmacy services and handling. Twenty six percent reported having made changes to outpatient drug administration services during 2007-2008 as a result of the decreased Medicare rates and nearly one third reported having decreased program availability. Over twenty seven percent reported that if Medicare drug payment rates continued to decline in 2009, they would definitely make changes, with more than half of those reporting that they would decrease program availability.
Conclusion: Medicare reimbursement for separately paid drugs in the hospital outpatient setting decreased from average sales price plus six percent in 2007 to average sales price plus five percent in 2008, at which point hospital providers responded with changes to their drug administration services, including program elimination. Providers reported that further reimbursement decreases would result in additional program elimination.
Effective January 1, 2009, Medicare reduced outpatient drug payments to average sales price plus four percent.
Implications for Policy, Delivery or Practice:
Providers are responding to decreased Medicare drug reimbursement by implementing or planning interventions that may threaten patient access to care. This information may assist policymakers in designing and implementing strategies designed to preserve access.
Funding Source(s): Centocor Ortho Biotech
Services, LLC
● Promise Keeping: The Provision of Healthcare for Survivors of the Tuskegee Study of
Untreated Syphilis in Negro Males & their
Families
Lauretta Pinckney, Ph.D.; Bryan Lindsey, Ph.D.,
M.P.P.A.; Lehida Meléndez-Morales, M.P.H.,
C.P.H.; Norman Hayes, B.S.; Hazel Dean, Sc.D.,
M.P.H.
Presented by: Lauretta Pinckney, Ph.D., Senior
Public Health Analyst, National Center for
HIV/AIDS, Viral Hepatitis, STD & TB Prevention,
Centers for Disease Control and Prevention, 1600
Clifton Road, Northeast, Mailstop E07, Atlanta, GA
30333, Phone: (404) 639-8019; Email:
LPinckney@cdc.gov
Research Objective: The Tuskegee Health Benefit
Program (THBP) was established by the United
States government in 1973 to provide free medical and burial services to all living participants of the
Tuskegee Study of Untreated Syphilis in the Negro
Male, 1932-1972. The program expanded in 1975 to include medical expenses for wives, widows and offspring who became infected because their spouses or fathers who were not untreated. The program paid all direct medical costs incurred by participants that were not paid by other medical programs. The aim of this study was to describe the types of health care services, frequency of utilization, reimbursement practices and cost of health services provided to THBP participants.
Study Design: We abstracted administrative and program data from October 2000 to September
2007 and developed a database to capture client demographics, diagnostic and utilization information, and cost of medical services. Additional data were obtained from contracts, case files, trip reports, and field notes to assess program compliance and implementation. We developed two data abstraction tools which were linked to an
Access database. Data were analyzed using
Microsoft Excel.
Population Studied: Tuskegee Syphilis Study participants, their wives, widows and off-spring enrolled in the Health Benefit Program
Principal Findings: We reviewed 5570 invoices for health care services provided during October 2000 to September 2007. From October 2000 to
September 2001, THBP paid medical and healthcare-related expenses for 38 clients – 65% were female, 47% were offspring of Tuskegee
Study participants, and the average age was 76 years. From October 2006 to September 2007, medical and healthcare-related expenses were paid for 19 clients – 68% were female, 84% were offspring of Tuskegee Study participants, and the average age was 65 years. At the end of the study period, all Tuskegee Study participants were deceased. The total THBP expenditures, excluding contracts, declined 41% from $745,045 during
October 2000 to September 2001 to $473,444 during October 2006 to September 2007. Medical care and nursing home services accounted for most of the costs, 32% and 34% respectively. Medical care accounted for approximately half (46%) of the
services utilized, followed by pharmacy (22%) and home care (12%). Diabetes Mellitus and hypertension were the most frequently diagnosed conditions. Contractual service costs declined from
$1.5 million to $470,000 during the period, but still accounted for almost 45% of overall program expenditures. Contractors were employed to provide home health services, but the program lacked a mechanism to monitor the quality of those services.
Conclusion: THBP was financially stable and able to fulfill its fiduciary responsibility to provide free medical and health care for participants. The program supported an array of medical and healthcare-related services. The assessment highlighted the need for continued long-term planning to sustain the program and address identified weaknesses.
Implications for Policy, Delivery or Practice: The program assessment was undertaken to comprehensively examine the nature of program operations, characteristics of participants utilizing medical services, types of medical services provided to participants, costs associated with medical services, and characteristics of service providers. The study provided valuable information to strengthen the program and plan for the health needs of aging THBP participants.
● Attitudes, Susceptibility, & the Utilization of
HIV Testing Services among Women in the
United States
Crystal Piper, M.P.H., M.H.A., Ph.D.; Saundra
Glover, Ph.D.; Lucy Annang, Ph.D.; Shaniece
Charlemagne, M.P.H.; Dayna Campbell, M.S.;
Wilhemenia Lee
Presented by: Crystal Piper, M.P.H., M.H.A.,
Ph.D., Research Associate, Institute for Health
Disparities, University of South Carolina, 220
Stoneridge Drive, Suite 208, Columbia, SC 29210,
Phone: (803) 463-1739; Email: cnpiper@yahoo.com
Research Objective: The purpose of this study is to examine the correlates of utilization of HIV testing services among women age 18-64 in the
United States. This study will assess women’s attitudes, beliefs, characteristics, and the predictors of perception of risk and the relationship in HIV testing behaviors. The hypotheses are: (1) Minority women tested for HIV are more likely to believe they have no risks of getting HIV; (2) Low-income women tested for HIV are more likely to believe they have no risks of getting HIV; and (3) Women tested for HIV that believe they have no risks of getting HIV are more likely to not be tested for HIV in the future.
Study Design: This study is a retrospective study and secondary data analysis of the 2006 National
Health Interview Survey. Aday and Andersen’s
Access to Health Services is the framework used in this study to examine HIV testing service utilization among women. To address multi-stage data from
NHIS, SAS callable SUDAAN was utilized.
Parametric testing using
Univariate/Bivariate/Multivariate analysis was performed to HIV testing service utilization and perception of risk among women in the United
States. The partial t-test in each model was considered in each model. To construct efficient models, stepwise regression procedure was used to identify demographic predictors for each outcome and then removed if the p-value>.05.
Population Studied: The 2006 original weighted sample population consisted of approximately
24,114 women, and a subset of the data were used to account for the 10, 379 women that responded to the question (Have you ever been tested for HIV?).
The percentage of women that responded (Yes) to ever being tested for HIV was (47.74%; n=4955), and the women that responded (No) to ever being tested for HIV was (52.26%; n=5424). This study will report on women that have been tested for HIV.
Principal Findings: More than half of the HIV tested women believed they had no chance of ever getting HIV (69.48%). In multivariate analysis, women who had been tested for HIV that believed they had no chance of getting HIV were more likely to have a family income of $0 - $14,999 (OR=3.14, p<.0001). Women in this study that had been tested for HIV that believed they had no chance of getting HIV were less likely to report having an HIV test in the next twelve months (OR=.54, p=.0180).
Conclusion: The consistent and widespread utilization of HIV testing services among women is one aspect of the reduction of HIV infection in that population in the United States. Findings from this study can lend themselves to the development of more efficient and sustainable interventions to prevent HIV infection and decrease high-risk behaviors among more susceptible populations and for the development of HIV testing policy.
Implications for Policy, Delivery or Practice:
Increasing HIV testing utilization among women requires national guidance. National recommendations should be developed and followed by primary care providers in order to increase HIV testing among women. Federal regulations should also consider other nontraditional locations for required HIV testing.
Policies to address these issues should address fundamental health system changes.
● A Cross-Sectional Model to Assess U.S.
Fetal/Infant Health
Grace Poertner, R.N., M.S.W., Ph.D.
Presented by: Grace Poertner, R.N., M.S.W.,
Ph.D., President, Administration, Wellness
Research, P.O. Box 159, Saint Charles, MO 63302,
Phone: (636) 946-4555; Email:
MissouriWellnessResearch@gmail.com
Research Objective: This poster presents a model that assesses, for a one-year U.S. birth cohort, the health of the fetal/infant, which is defined as a child reaching the climax of two periods of life: birth (end of the fetal period) and a first birthday (end of infancy).
Study Design: Linked birth and infant death records provide the means to study: [1] the odds of birth at very-low-birthweight (VLBW: < 1500 gm) and [2] the odds of death by one-year of age.
Independent variables include: infant race, maternal age/education, parental marriage, and inadequate prenatal care. An 8% random sample of survivors
(290,843), appropriately weighted, is studied along with the population of infant deaths (38,292).
Logistic regression tests the odds of VLBW in the first equation and the odds of infant mortality, given birth at VLBW or Above-VLBW, in the next series of equations.
Population Studied: The 1984 U.S. birth cohort is the database available to this researcher; this includes the 50 states and Washington, D.C.
Current reports from this author's 1991 study
(Medicaid Policy and Infant Survivability) are one source of studies that provide a basis for on-going assessments of fetal/infant health across the U.S.
Principal Findings: For this birth cohort, VLBW accounts for 1.2% of births and 43.6% of infant deaths. [1] Risk of VLBW is significantly increased with inadequate prenatal care; single parent-hood; lower maternal education; and, for African-
Americans, older maternal age. [2a] Factors affecting death among VLBW infants are minimally significant and inconsistent across the races. [2b]
Significant factors increasing risk of Above-VLBW infant death for Caucasians include inadequate prenatal care, single-parenthood, lower maternal age and lower education; significant factors for
African-Americans include inadequate prenatal care, single parent-hood, lower maternal education.
Conclusion: VLBW most strongly predicts infant mortality; inadequate prenatal care most strongly predicts VLBW birth and above-VLBW mortality.
The non-association of parental factors with VLBW mortality most likely reflects similarly available medical technology.
Implications for Policy, Delivery or Practice:
U.S. health policy must include prenatal care that is minimally adequate and available across the states and in Washington, D.C.
Funding Source(s): AHRQ, Wellness Research
● Effects of smoking, alcohol, fruit and vegetable consumption on Oral Cancer
Efstathios Polychronopoulos, Ph.D.; Balas, Ph.D.;
Patricia Hentosh, Ph.D.; Stella Bondi, Ph.D.;
Angeliki Polychronopoulou, M.P.H.;
Presented by: Efstathios Polychronopoulos, Ph.D.,
Ph.D. Student, Health Servives Research, Old
Dominion University, 330 West Brambleton Avenue,
Apartment 1804, Norfolk, VA 23510, Email: stathypoly18@gmail.com
Research Objective: It is believed that tobacco use and alcohol consumption increase the risk of developing oral and pharyngeal cancer, while there is some controversy on the impact of fruit and vegetable consumption impact. The purpose of this study was to examine the effects of these factors.
Study Design: Several electronic databases including Medline, PubMed, Google Scholar and the SEER database were searched, supplemented by the screening of reference lists. 252 case-control studies examining the effect of tobacco use, alcohol consumption as well as fruit and vegetable intake on oral and pharyngeal cancer were investigated.
Population Studied: Adult general population in the United States.
Principal Findings: Overall, the effects of twenty five lifestyle risk factors were assessed. Ultimately, analysis reveals that tobacco use (OR ranging from
2.5 to 14.3) and alcohol consumption (OR ranging from 3.2 to 8.1) appear to be major risk factors in all age groups, with higher risk in males and in African
Americans, while fruit and vegetable consumption appears to be associated with a protective effect in all age groups (OR ranging from 0.08 to 0.7).
Conclusion: The pattern of the stereotype for an oral and pharyngeal cancer patient has changed, and it is no longer a male individual over 65 years from a low social class, who reports high tobacco use and alcohol consumption.Smoking cessation programs, alcohol reduction initiatives and awareness programs on the protective effect of fruit and vegetable consumption would reduce racial and gender differences among adults regarding oral and pharyngeal cancer.
Implications for Policy, Delivery or Practice: A main implication is the need for an aggressive tobacco cessation policy, moderating alcohol consumption programs, and an emphasis on the benefits of a diet rich in fresh fruits and vegetables.
Adolescents will benefit from a tobacco control policy because the results of these studies show an increased risk associated with smoking at an early age. This should also be matched with increasing awareness in young people on the risk factors of oral and pharyngeal cancer. Education of health professionals such as those involved in the dental practice, will help prevent a diagnostic delay.
Theme: Prevention and Treatment of Chronic
Illness
● Hydroxyurea Use, Perceived Trust, Care, &
Pain Management in the Emergency Department among Patients with Sickle Cell Disease
Jerlym Porter, Ph.D.; Paula Tanabe, Ph.D., M.P.H.,
R.N.
Presented by: Jerlym Porter, Ph.D., Postdoctoral
Fellow, Institute for Healthcare Studies,
Northwestern University, 750 North Lake Shore
Drive, 10th Floor, Chicago, IL 60611, Phone: (312)
503-5570; Email: j-porter@northwestern.edu
Research Objective: Despite evidence that the use of hydroxyurea to prevent acute painful crises for patients with sickle cell disease (SCD) is beneficial, use is limited --- only 30% of eligible patients take hydroxyurea. We provide a descriptive analysis of hydroxyurea use (HU) among patients with SCD. We also examined patients’ selfreported perception of being treated with trust and respect by emergency department physicians and nurses, as well as self report of the most important aspects of emergency department care (ED) and pain management efficacy.
Study Design: Structured medical record review of patients in a three-site prospective longitudinal cohort study in emergency departments (ED) and
10 patient interviews per site per quarter over 15 months. The larger study focus was to develop a quality improvement system for enhanced ED analgesic treatment of SCD patients. One interview per participant was analyzed for the purposes of this study. Participants were asked: 1) do they take hydroxyurea, 2) were they treated with trust and respect by the triage nurses, ED nurses, and ED physicians (yes, no/somewhat), 3) which aspect was the most important part of their ED visit, and 4) how well did the ED clinician try to manage their pain (1-10). Frequencies were tabulated and independent t-tests were used for continuous variables.
Population Studied: Adult ED patients with a chief complaint of sickle cell pain episode from three sites, comprising one rural and two urban academic medical centers.
Principal Findings: The sample consisted of 54 adults (mean age 35.4 years; 54% men). Less than one-third (27.8%) were taking HU. Participants reported being treated with less trust and respect by the ED physician (29.6%) compared to the triage
(24.1%) and ED nurse (22.2%). The most important aspect of the ED visit was: 1) clinician understanding SCD (50.0%), 2) amount of time until medication administration (27.8%), 3) being treated with trust and respect (14.8%), and 4) type of pain medication received (5.6%). The results indicated an association between perceived trust and respect from ED clinicians and pain management efficacy.
Specifically, patients who reported trust and respect from the ED nurse and the ED physician had higher ratings of pain management (M=7.55, SD=3.12;
M=8.13, SD=2.40, respectively) when compared with patients who reported not being treated with trust and respect [M=4.42, SD=2.78; t(52) = -3.14, p
< .01 and M=3.81, SD=3.19; t(52) = -5.47, p < .001, respectively].
Conclusion: Our findings are consistent with previous reports of low HU use among patients with
SCD. Patients reported that clinicians’ understanding of SCD was an important aspect in their ED visit, as well as clinicians’ trust and respect of the patient. Being trusted and respected was associated with pain management efficacy, such that those who reported being treated with trust and respect had higher ratings of pain management.
Implications for Policy, Delivery or Practice: The results underscore the importance of trust in patient-provider interactions in the ED for SCD patients. Future research is necessary to better understand the impact of trust on quality of care and explore the barriers to hydroxyurea use in patients with SCD.
Funding Source(s): Mayday Fund
● The Medical Home & Employment Change among Parents of Children with Special Health
Care Needs
Shirley Porterfield, Ph.D.; LeaAnne, DeRigne, Ph.D.
Presented by: Shirley Porterfield, Ph.D., Associate
Professor, Social Work, University of Missouri-Saint
Louis, One University Boulevard, Saint Louis, MO
63121, Phone: (314) 516-4617; Email: porterfields@umsl.edu
Research Objective: Recent estimates are that one in five U.S. households with children has at least one child with a special health care need
(USDHHS, 2004). Like most parents, those with children with special health care needs struggle to balance child-rearing responsibilities with employment demands. Child-rearing responsibilities can bleed into paid work time resulting in increased absenteeism, distraction, and exhaustion. The eventual impacts may be a reduction in work hours or a cessation of paid employment. This research examines factors affecting parents’ decisions to make no change in their work hours, cut their number of hours worked, or stop working all together due to their child’s special health care needs, focusing specifically on whether having a medical home influences this choice.
Study Design: The data are analyzed both descriptively and in a multivariate model. The model is specified as reduced form, and is estimated using multinomial logistic regression with the choice of a parent to maintain their current level of employment, reduce work hours, or stop working specified as a function of the needs of the child, the resources of the family, and the socio-demographic characteristics of the family rather than as a function of the characteristics of the parent’s job.
Components of the medical home variable include
1) whether or not the child has a usual source of care and/or personal provider; 2) whether or not the care provided is “family centered;” 3) whether or not care coordination services are received; and 4) whether or not the family has problems getting all the referrals they need.
Population Studied: This study includes 35,252 children with special health care needs ages 5-17
included in the 2005-2006 National Survey of
Children with Special Health Care Needs.
Principal Findings: Half of children in our sample met criteria in all four facets and thus received care in a medical home. If the child has a medical home, the relative risk (odds) of a parent choosing to cut hours rather than not change hours decreases by
51%. The relative risk (odds) of choosing to stop working rather than not change hours decreases by an estimated 64%. All else equal, care coordination services significantly reduce the odds of changing employment status. The probability that parents who report that they spend more than 7 hours per week providing home health care for their child have cut work hours over the past year because of the child’s medical condition is 10.4% if the child receives care within a medical home and 21.1% if the child does not receive care within a medical home.
Conclusion: The medical home is a moderating factor in parental decisions concerning change in employment status.
Implications for Policy, Delivery or Practice:
From a policy perspective, if parents reduce work hours to provide care for a child with special health care needs, the resulting economic impacts may be devastating to families, especially those families who may have substantial medical expenses and/or other children. If the medical home can alleviate some of these problems, policies can be adopted to help support introduction or expansion of this aspect of pediatric care.
Theme: Child Health
● Who are the Dentists that Employe Dental
Hygiensits?
Nadereh Pourat, Ph.D.
Presented by: Nadereh Pourat, Ph.D., Associate
Professor, Health Services, University of California,
Los Angeles Center for Health Policy Research,
10960 Wilshire Boulevard, Suite 1550, Los
Angeles, CA 90024, Phone: (310) 794-2201;
Email: pourat@ucla.edu
Research Objective: Dental hygienists can increase dentists’ productivity, yet national data indicate that a third of dentists do not employ these personnel. Information on characteristics of dentists who employ hygienists, as well as the reasons for not employing hygienists is needed.
Study Design: A 2003 survey of California dentists in private practice and logistic regression was used to assess factors that were independently associated with hygienist employment. These factors included dentists’ personal, practice, patient, population, productivity, and patient care characteristics. Characteristics of dentists not employing hygienist and reasons for not employing hygienists are also described.
Population Studied: The sample for this study included dentists with any percentage of time providing care in private practice and excluded retired, faculty, students, radiologists, pathologists, public health dentists, and surgeons who were highly unlikely to provide substantial amounts of direct patient care. Dentists practicing in more than one dentist-owned private practice reported on the location they identified as having the largest volume of patients.A total of 3,599 dentists were included in this study.
Principal Findings: Dentists who work full-time, employ more front office personnel, have more operatories, longer appointments, more income from private payers, and more elderly patients exhibited a higher likelihood of employing hygienists than dentists with the alternative characteristics.
Foreign graduates and those with a lower percentage of white patients in their practice were less likely to employ hygienists.
Conclusion: Differences between dentists who do and do not employ hygienists reflected factors such as practice income and patient demand as reasons for this decision. Dentists’ practice preferences appeared also to be important.
Implications for Policy, Delivery or Practice:
Employing hygienists increases capacity for providing care, yet not all dentists can or choose to do so. Understanding the characteristics of dentists who employ hygienists informs policy decisions to maintain or increase the capacity of the dental workforce.
Funding Source(s): California Dental Association
Foundation
Theme: Health Care Workforce
● Translating National Breast Cancer Screening
Guidelines into Practice
Nadereh Pourat, Ph.D.; Sara McMenamin, Ph.D.;
Helen Halpin, Ph.D.; Gerald Kominski, Ph.D.; Jay
Ripps, F.S.A., M.A.A.A.
Presented by: Nadereh Pourat, Ph.D., Associate
Professor, Health Services, University of California,
Los Angeles Center for Health Policy Research,
10960 Wilshire Boulevard, Suite 1550, Los
Angeles, CA 90024, Phone: (310) 794-2201;
Email: pourat@ucla.edu
Research Objective: In 2007, the American
Cancer Society (ACS) published guidelines recommending women at “high risk” for breast cancer to receive annual screening using breast magnetic resonance imaging (BMRI) as an adjunct to mammography. “High risk” as defined in the guideline included women with a 20% or higher lifetime risk of breast cancer including women with a family history of breast or ovarian cancer, those with the BRCA1 and BRCA2 genes, or women who were treated previously for Hodgkin’s disease. We aim to calculate the lifetime risk of breast cancer among non-elderly women in California and estimate the likelihood of BMRI screening among
“high risk” women by using mammogram rates and
controlling for predisposing, enabling, and need characteristics.
Study Design: We compared the general population and the “high risk” population by their characteristics to identify potential differences. We used logistic regression models to predict the odds of receiving a mammogram (BMRI) in the past year among those women categorized as “high risk”. All high risk women who received a mammogram were assumed to receive a BMRI.
Population Studied: We used data from the 2005
California Health Interview Survey (CHIS) and the
Gail model to calculate the lifetime risk of breast cancer among women age 30-64 (n=14,312).
Women with a lifetime risk of 20% or greater were classified as “high risk.” This definition of “high risk” did not include women with BRCA1 and BRCA2 genes or those exposed to Hodgkin’s disease since the Gail model did not include those risk factors.
Principal Findings: The percentage of women ages 30-64 without a prior history of breast cancer who were classified as high risk ranged from 0.9% to 1.2% depending on the assumptions made in the model. The “high risk” sample was more likely to be older, high-school educated, white, born in the U.S., employed, speak English fluently, and visit the doctor in the past year. The “high risk” sample was less likely to have public fee-for-service health insurance or earn 200% or below the poverty level.
Approximately 74% of the women in the “high risk” group received a mammogram in the past year. The significant factors predicting annual mammograms
(BMRI) in the past year among the high risk group were older age, being employed, having at least one doctor visit in the past year, and having more chronic conditions. Of the approximately eight million women aged 30-64 with no prior history of breast cancer, 14% were uninsured, 65% were in health insurance plans that would not cover breast
MRI screening of “high risk” women, and 21% were in health insurance plans that would cover breast
MRI screening of “high risk” women. If health insurers were to adopt the 2007 ACS guidelines, nearly 34,000 more BMRIs may be conducted in
California annually.
Conclusion: The average cost of a BMRI is estimated at $1,300, leading to a significant rise in overall health care expenditures. Given the high rates of false positive BMRI results and its clinical and emotional consequences, broad implementation of this guideline and its public health impact should be carefully examined.
Furthermore, the costs of implementing this guideline among group and publicly insured populations may be partially absorbed by the employers and the public payers. However, the costs of such screening among the individually insured and the uninsured remain the responsibility of such individuals. Disparities in BMRI screening are likely to exist similar to those currently in place for mammograms.
Implications for Policy, Delivery or Practice: A careful examination of the clinical impact of guidelines should be combined with an examination of the impact on health care delivery, expenditures, and outcomes of recommended practices.
Funding Source(s): California Health Benefits
Review Program
Theme: Coverage and Access
● Computer Modeling of a Hospital-Wide
Approach to Decreasing Emergency
Department Congestion
Emilie Powell, M.D., M.B.A.; Rahul Khare, M.D.;
Arjun Venkatesh, M.D.; Ben Van Roo, Ph.D.; Gilles
Reinhardt, Ph.D.
Presented by: Emilie Powell, M.D., M.B.A.,
Department of Emergency Medicine, Northwestern
University, 259 East Erie Street, Suite 100,
Chicago, IL 60611, Phone: (773) 317-3877; Email: emilie.powell@gmail.com
Research Objective: The harms of crowding and boarding in the Emergency Department (ED) are well published. ED crowding is multi-factorial and some significant influences come from outside the
ED itself, including a limited supply of inpatient beds causing ED admitted patient boarding. Several studies support hospital-wide efforts to address the issue of ED crowding. The objective of this work is to utilize computer modeling to analyze 3 unique inpatient discharge timing strategies to optimize the limited supply of inpatient beds to reduce total ED admitted patient boarding hours.
Study Design: A time series analysis was performed to demonstrate the relationship between demand for inpatient beds from patients admitted through the ED and elective surgery and the supply of inpatient beds made available through inpatient discharges. The patient flow streams were analyzed against the primary output of ED admitted patient boarding hours. An Excel solver model was applied to compare 3 inpatient discharge timing strategies- discharging 75% of patients by noon, uniformly discharging all patients noon to midnight, or matching discharge timing with time of admission requests- with the goal output to minimize total ED boarding hours.
Population Studied: This was an observational cohort study during August-September 2007 weekdays in an urban, tertiary care hospital with a
75,000 patient/year ED and 490 inpatient beds.
Patients were included if they were admitted to the hospital from the ED or elective surgery on a weekday and excluded if data was not available on
ED boarding time and type of inpatient bed requested. All patients discharged from the inpatient floors were included if they were discharged on a weekday and full data was available on the type of bed the patient was discharged from and the timing of discharge. Two pre-existing hospital databases were used to collect
this information. Data collected includes time of patient hospital-admission, boarding hours for ED admitted patients, and time of inpatient discharge.
Principal Findings: The study period included
2077 ED admitted patients (2077/2219; 93.6%),
779 surgical patients (779/803; 96.9%), and 4623 inpatient discharges (4623/4624; 100%). A mean of
33 ED patients and 18 surgical patients were admitted/weekday. 72.3% of ED admissions and
96.8% of surgical admissions occurred noonmidnight. Mean inpatient bed occupancy was 95% at 0800 and the peak time of discharge was 1500.
The study period mean ED boarding time was 2.4 hours/patient and total daily boarding hours was
225.9 hours. Model results: discharging 75% of inpatients by noon increased boarding hours to
446.9, discharging all patient uniformly noonmidnight decreased boarding hours to 143.7, matching discharge timing to admission demand decreased boarding hours to 0.1/day.
Conclusion: In this model, discharging 75% of patients by noon increased boarding hours.
Matching supply with demand minimized boarding hours. An operationally feasible solution uniformly discharges all patients noon-midnight to decrease boarding hours by 36.4%.
Implications for Policy, Delivery or Practice:
Hospital operational data may be used to optimize inpatient discharge timing strategies to decrease
ED admitted patient boarding hours and ED congestion.
Theme: Organizational Performance and
Management
● The Role of Retail Health Clinics in the
Delivery of Vaccinations & Primary Care
Services
Alyssa Pozniak, Ph.D.; Allison Kennedy, M.P.H.;
Sarah Shoemaker, Ph.D., Pharm.D.; Shannon
Stokley, M.P.H.; Garrett Asay, Ph.D.; Margaret
Coleman, Ph.D.
Presented by: Alyssa Pozniak, Ph.D., Associate,
Domestic Health, Abt Associates, Inc., 55 Wheeler
Street, Cambridge, MA 02138, Phone: (617) 520-
2455; Email: alyssa.pozniak@abtassoc.com
Research Objective: Retail health clinics (RCs) are characterized as having flexible hours, short waiting times, transparent pricing, and a narrow scope of well-defined healthcare services in a convenient setting. Most RCs offer vaccinations, which are among the most common services requested. RCs are seen by some as an affordable alternative for people who do not regularly see a doctor, but separating vaccinations from other preventive services may also reduce comprehensive primary care. As RCs go out of business or leave some local markets, there are further concerns about continuity of care. This study describes how fees, vaccinations, and other services vary across RCs, what factors drive these differences, and the feasibility of RCs to provide primary care services in the long-term.
Study Design: We will construct a database of over 70 variables for all RC companies identified through publicly-available data and literature review. We produce frequency counts and crosstabulations for variables of interest, including which vaccinations and preventive services are offered, their published fees, and descriptive information about the RCs (e.g., size, geographic presence, retail partners). We also will interview a purposive sample of 15 representatives from industry trade groups, RC companies, and retail partners to gain a deeper understanding about the services offered and their costs. We will conduct a thematic analysis of the interview findings related to the services provided and synthesize the findings across the different interviewees.
Population Studied: All RCs identified through methods outlined above and 15 representatives from industry trade groups, RC companies, and retail partners.
Principal Findings: We identified 49 RC companies that operate 1,250 clinics in 38 states.
The most commonly offered services include vaccinations, physical exams, and treatment for influenza, earaches, and rashes. Among the vaccinations examined, influenza was the most commonly offered (N=30) followed by pneumococcal (N=27) and hepatitis B (N=26).
Approximately 25% of RCs did not list vaccinations among the services they provide. Prices for vaccinations ranged from $20 for influenza to $525 for the 3 dose-HPV vaccination. Several RC websites offer discounts (e.g., coupons, lower prices when customers pay cash). Despite transparent pricing being a characteristic of RCs, many websites did not list any prices, or only listed price ranges. The interviews will provide additional insight regarding services offered and prices charged by RCs, including how RC companies determine which services to offer (e.g., state regulations, expense, clientele) and what prices to charge; what makes services more profitable than others (e.g., economies of scale, additional services ordered); how services offered differ at individual clinics within the same RC company; and how continuity concerns are addressed.
Conclusion: Preliminary findings suggest considerable differences exist between RCs regarding vaccination and other services offered and fees charged. These differences have important implications to consumers and their choice of settings for primary care services.
Implications for Policy, Delivery or Practice: By providing current information about vaccinations and services offered and fees charged by RCs, this research contributes to a better understanding of the understudied but burgeoning RC industry.
These differences may be linked to the long-term viability of the clinics, which, in turn, may affect continuity of patient care.
Funding Source(s): CDC
Theme: Consumer Choices in Health Care
● How Do Insurers & Population Management
Companies Determine Coverage for Multiple
Sclerosis & Other Complex, Chronic Illnesses?
Alyssa Pozniak, Ph.D.; Danna Mauch, Ph.D.; Lisa
LeRoy, Ph.D.
Presented by: Alyssa Pozniak, Ph.D., Associate,
Domestic Health, Abt Associates, Inc., 55 Wheeler
Street, Cambridge, MA 02138, Email: alyssa.pozniak@abtassoc.com
Research Objective: Approximately 400,000 people in the US have multiple sclerosis (MS), a chronic illness with considerable variability in symptoms, impairments, and outcomes. Some people have relatively few impairments with little need for medical care, whereas others have rapid disease progression, multiple impairments, severe disability, and require ongoing and intensive medical care. For these latter individuals, MS care involves a high degree of healthcare utilization of neurologic and other specialty services (e.g., nursing, mental health, neuropsychologic, urologic, rehabilitative) which, over the long term, generate substantial costs that increase over time. While most people with MS have health insurance, not all necessary services are covered adequately, or even covered at all. In particular, anecdotal evidence suggest that many services that are critical to providing high quality care for a complex, chronic illnesses such as MS (e.g., multidisciplinary care coordination, obtaining authorizations, completing forms, supporting appeals) are under/-unreimbursed. Furthermore, although population management is common for other chronic diseases (e.g., diabetes), little is known about its role in providing MS care. This study explores the processes by which health insurers and population management companies define standards of care (and thus reimbursable services); the challenges faced by payers in setting reimbursement rates for MS and other complex chronic illnesses across a broad spectrum of settings and providers; and the role of population management in providing MS care.
Study Design: We will interview key informants from health insurance companies and population management companies to address challenges in delivering appropriate levels of care to the MS population and trends in reimbursement levels of services for people with MS. Interviews will be conducted via the telephone and will take approximately one hour. We will conduct a thematic analysis and synthesize the findings across the different interviewees.
Population Studied: Six companies that provide health insurance and/or population management services. Our sample consists of a small sample of regional and national companies across the nation.
Principal Findings: The interviews will provide insight regarding insurers’ and population management companies’ decision making process regarding coverage for MS, including: benefit design for MS or other high cost, chronic diseases; formulary coverage and determination for MS drugs; extent and use of population management for MS care; and management of costs for enrollees with MS, including interventions used (e.g., stepped care approach, care manager call line, patient financial incentives). We will also examine how coverage decisions have changed in recent years with the advent of increasingly effective but expensive drugs used for treating MS symptoms.
Conclusion: The findings will have important implications for people with MS and their providers regarding coverage for MS services by health insurers and in particular, the role of population management for MS care.
Implications for Policy, Delivery or Practice: Our findings contribute to a better understanding of the decision process and cost containment strategies employed by insurers regarding coverage for MS
(which is also applicable to other chronic illnesses) and how the process and strategies contribute to gaps in coverage for people with MS and other chronic illnesses. The study also provides preliminary information about the role of population management for people with MS.
Funding Source(s): National MS Society
Theme: Prevention and Treatment of Chronic
Illness
● State Scorecard: Comparative Analysis of
Health Care System Performance in the 29
Largest States of the U.S., 2000-2006
Ashwin Prabhu, B.S.; Sherry, Glied, Ph.D.
Presented by: Ashwin Prabhu, B.S., Research
Assistant, Health Policy & Management, Mailman
School of Public Health, Columbia University, 600
West 168th Street, 6th Floor, MSPH-HPM, New
York, NY 10032, Phone: (212) 305-4031; Email: agp2014@columbia.edu
Research Objective: The objective of this study is to analyze the overall performance of the health care system in the 29 largest states by population in the United States over a period of 7 years, 2000-
2006
Study Design: This report uses data from the
Medical Expenditure Panel Survey (MEPS). Statelevel weights provided by application to the Agency for Healthcare Research and Quality (AHRQ) allow a state-by-state comparison for the 29 most populated states in the United States. In addition, results are also available for the remaining 21 states taken as a whole. Multiple questions in seven major areas are analyzed - Health Insurance
Coverage, Access to Care, Preventive Care, Health
Status, Disability Days, Health Service Utilization and Health Care Expenditures. The years studied in
the report are 2000-2006. The study provides data for each year in this period as well as a comparative analysis of the periods 2000-2003 and 2004-2006.
Results are broken down by age, race and insurance coverage status, among other population characteristics.
Population Studied: Respondents in the Medical
Expenditure Panel Survey (MEPS), 2000-2006
Principal Findings: Study findings are subject to clearance by the Agency for Healthcare Research and Quality (AHRQ) and the Center for Economic
Studies (CES) division of the Census Bureau. We are currently in the process of gaining this clearance and therefore, due to confidentiality agreements, cannot reveal state-level results at this point in time. However, a region-level analysis of the MEPS shows some interesting trends. For example, average total medical expenditures are highest in the Midwest, followed by the Northeast,
South and West. A similar trend is seen when looking at Out-Of-Pocket (OOP) Expenditures. This gap is closing, however, as costs grow at different rates in different regions. For example, comparing an average year in the periods 2000-2003 and
2004-2006, we find that the Western region of the
US has experienced the fastest growth in OOP expenditures, of about 30%, compared to 20% in the Midwest. Differences are also seen in access to care. Up to 87% of people have access to a Usual
Source of Care (USC) in the Northwest, compared to only 77% in the South. Regional differences in access to a USC mirror differences in the proportion of people who seek preventive care. The proportion of patients in different regions with access to a USC has remained stable over the 7 year period of analysis.
Implications for Policy, Delivery or Practice: The results of this study will be valuable to both policymakers and practitioners as a resource for evaluating the relative performance of health care systems in the 29 largest states of the United
States.
Funding Source(s): CWF
● Quality & Financial Performance of Rehab-
Focused Nursing Homes
Rohit Pradhan, M.B.B.S.; Alex Laberge, P.T.,
M.B.A.; Robert Weech-Maldonado, Ph.D.; Kathy
Hyer, Ph.D.
Presented by: Rohit Pradhan, M.B.B.S., Research
Assistant, Department of Health Services
Research, Management, & Policy, University of
Florida, College of Public Health and Health
Professions. 101 South Newell Drive, Room 4151,
Gainesville, FL 32611, Phone: (352) 273-5217;
Email: pradhan@phhp.ufl.edu
Research Objective: The galloping cost of rehabilitation therapy was one of the most important factors driving the introduction of Medicare’s
Prospective Payment System (PPS) for nursing homes in 1997. Concern continues to be expressed over the quality of care as well recent increase-after a period of stability--in cost of rehabilitation therapy. While most nursing homes provide rehabilitation therapy, some are more specialized in providing rehabilitation therapy. The purpose of this study is to determine if these rehab-focused facilities achieve better quality and financial performance
Study Design: Cross-sectional study that utilized the Center for Medicare and Medicaid Services
(CMS) Minimum Data Set (MDS), Online Survey
Certification and Reporting (OSCAR), Area
Resource File (ARF) and Medicare Cost Reports.
Dependent Variables: Quality outcomes (quality of care deficiencies, post acute walk improvement, 4point ADL decline); staffing measures (Registered
Nurses (RN), Licensed Practioner Nurse (LPN),
Certified Nursing Assistant (CNA) ratios, and RN percent, PT percent and OT percent for skill mix); and financial measures (patient cost per day, patient revenue per day, and operating margin).
Independent Variable: Dichotomous variable defining rehab-focused facilities: SNFs in top 10% tier of Medicare beds. Control variables: Nursing home ownership status, system membership, acuity index, payer mix, market competition and size.
Analysis: Ordinary Linear Squares regression
(OLS) with county fixed effects
Population Studied: All free standing, nongovernment facilities that had Medicare beds in the
United States in 2004 and submitted Medicare cost reports (N= 11,247)
Principal Findings: Rehab-focused nursing homes had 1.2 % higher operating margin (p= .05), $66 additional patient revenues (p=.001), and $80 additional patient costs (p=.001) compared to facilities not focused on rehab. Rehab-focused nursing homes had a higher proportion of physical therapists by 2.3% (p=.09) and a lower decline of activities of daily living (ADL) of .01 (p=.001). There were no statistical differences between rehabfocused nursing and other nursing homes in terms of nurse staffing ratios, RN and OT skill mix, quality of care deficiencies and walking improvement
Conclusion: Rehab-focused nursing homes had mixed findings in terms of both financial and quality of care performance. While rehab-focused facilities had higher patient revenues, they also had higher patient costs which may be a result of their higher
PT skill mix compared to other facilities. Despite their higher costs, rehab-focused nursing homes still exhibited better financial performance, although marginally significant. Although rehab-focused facilities had better outcomes in terms of ADL decline, they did not show any significant difference in walk improvement compared to other nursing homes.
Implications for Policy, Delivery or Practice:
Further research is needed to examine the paradoxical finding that rehab-focused homes do not achieve better outcomes in terms of walk
improvement, despite their higher patient costs and
PT skill mix. Considering their higher cost, it raises important policy questions.
Theme: Medicare
● Does Higher Quality Lead to Better Financial
Performance: The Case of Hospitals
Rohit Pradhan, M.B.B.S.; Michael Morris, M.P.H.,
M.P.A.; Kimberly Elliot, B.A.; Robert Weech-
Maldonado, Ph.D.
Presented by: Rohit Pradhan, M.B.B.S., Research
Assistant, Department of Health Services
Research, Management, & Policy, University of
Florida, College of Public Health and Health
Professions. 101 South Newell Drive, Room 4151,
Gainesville, FL 32610, Phone: (352) 273-5217;
Email: pradhan@phhp.ufl.edu
Research Objective: In recent years, hospitals in
America have faced pressure from regulators, payers and the general public to improve their quality of care. This increased emphasis on quality comes at a time when hospitals continue to face significant financial challenges. Therefore, hospitals face twin challenges: Improve quality in an era of increasingly negative balance sheets. This study examines the relationship between quality of care and financial performance in hospitals.
Study Design: The study is a cross-sectional study that utilized data from Florida Hospital Association
(FHA) survey and the American Hospital
Association (AHA) hospital survey Study Design:
Cross-sectional data source. Dependent Variable:
Financial performance was measured through both operating and total profit margins. Independent
Variable: Hospital quality was measured through
RN skill mix and nurse staffing intensity. Skill mix was operationalized as the number of RN full time equivalents (FTEs) divided by the total number of nursing staff FTEs while intensity was measured as overall availability of nursing staff per adjusted patient day. Control variables: hospital size, hospital ownership status, system affiliation status, whether the facility is a teaching hospital, payer mix, market competition, urban/rural, and patient case mix.
Analysis Design: Ordinary Least Squares regression (OLS).
Population Studied: Data were examined from
170 non federal acute care hospitals who participated in the 2005 FHA survey and the AHA hospital survey.
Principal Findings: Higher RN skill mix and nurse staffing intensity were not significantly associated with hospital financial performance. Further analysis showed that while hospitals with higher RN skill mix and higher nurse staffing intensity had higher patient care revenues (0.42; p=0.03 & 0.33; p=0.04, respectively), they also had significantly higher operating costs (0.34; p=0.01 & 0.28; p=0.02, respectively). The percent of revenue from
Medicare was positively associated with both operating and total margin (0.60; p=.03 & 0.26; p=0.04), while a higher proportion of Medicaid revenues was significantly associated with lower financial measures (-0.04; p=0.006 & -0.05; p=0.01, respectively)
Conclusion: Results suggest that increasing nurse staffing has no statistically significant detrimental effect on hospital financial performance. While increased staffing ratios were associated with significantly high costs this was counterbalanced by a corresponding increase in patient care revenues.
It was found that hospital payer mix was the strongest predictor of both operating and total margins in this sample.
Implications for Policy, Delivery or Practice:
From the perspective of the hospital administrator, our findings indicate that focusing on cost control through staffing reductions is not necessarily the best approach to improve financial margins.
Similarly, from a public policy perspective, it suggests that greater emphasis can be laid on quality without negatively impacting the hospital’s bottom line.
Theme: Quality and Efficiency: Policies and
Incentives
● Are VA Health Care Services Rationed? Using
VA Wait Times to Model Supply & Demand for
VA Services
Julia Prentice, Ph.D.; Steven Pizer, Ph.D.
Presented by: Julia Prentice, Ph.D., Health
Scientist, Health Care Financing & Economics, VA
Boston Health Care System, 150 South Huntington
Avenue (152H), Boston, MA 02130, Phone: (857)
364-6057; Email: Julia.Prentice@va.gov
Research Objective: The Department of Veterans
Affairs (VA) reconciles the imbalance between supply and demand for healthcare through wait times. In effect, waiting for care is the price veterans pay for using the VA health care system.
Over the last decade, the VA has invested significant resources into decreasing wait times for appointments through performance measures and the Advanced Clinic Access Initiative. These initiatives focus on developing a more efficient scheduling environment, which increases facilities’ capacity to provide appointments and decreases waits. However, policies aimed at decreasing wait times have not considered how demand for VA care changes as wait times change (e.g. price elasticity of demand). This research is the first to use wait times to model the price elasticity of supply and demand for VA services.
Study Design: This is a cross-sectional study that models supply and demand for primary care separately. Supply in a particular month is defined as the total number of primary care appointments scheduled to occur in that month. The main predictors of supply are the average wait times for new patients for primary care in the previous month,
VA medical center budget and the number of community based outpatient clinics a VA medical center has. Demand in a particular month is defined as the estimated number of primary care appointments that were entered into the schedule in that month, regardless of when they were scheduled to occur. The main predictors of demand are predicted primary care wait time for new patients, population need characteristics (e.g. number of veterans in the area, per capita income), and the availability of non-VA options in the area
(e.g. Medicare HMOs). Due to the endogenous relationship between wait times and demand, we instrument for current primary care wait times using lagged wait times.
Population Studied: All VA medical centers providing outpatient primary care services between
2002 and 2005.
Principal Findings: There is a large and significant negative relationship between predicted wait times and the estimated number of appointments demanded. In contrast, there is a small significant relationship between higher wait times in the previous month and a greater number of appointments supplied in the current month.
Conclusion: VA facilities are responsive to wait times as the number of scheduled appointments increases when wait times increase in the previous month. However, VA patients are even more sensitive to wait times, exiting the VA system as wait times increase and reducing the demand for
VA services.
Implications for Policy, Delivery or Practice:
Substantial progress has been made in reducing
VA wait times by focusing on how to increase capacity through changes in scheduling. However, policymakers have not considered how decreases in wait times will influence demand for VA services.
This research indicates many veterans adjust their demand for VA services in response to changes in waiting times. Consequently, resources committed to expanding capacity will have part of their effect on waiting times offset by the demand response.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Perceptions of Policy Changes Regarding
Asthma Inhalers among Chicago Minority
Communities
Valerie Press, M.D., M.P.H.; Amber Pincavage,
M.D.; Dustyn Baker, M.D.; Walter Conwell, M.D.;
Julie Kleczek, M.D.; Vineet Arora, M.D., M.A.
Presented by: Valerie Press, M.D., M.P.H., Clinical
Associate, Research Fellow, Medicine, University of
Chicago, 5841 South Maryland Avenue, Chicago, IL
60637, Phone: (773) 702-5170; Email: vpress@medicine.bsd.uchicago.edu
Research Objective: Asthma affects a disproportionate share of minority citizens in
Chicago. Patients are at increased risk with policy changes to replace the familiar Metered Dose
Inhalers (MDIs) with environmentally friendly but harder to use hydrofluroalkane (HFA) inhalers. It is imperative that internal medicine (IM) residents and community members are prepared for this transition. The research objectives were: to conduct a needs assessment of resident physician knowledge regarding inhaler use and the change to
HFA inhalers; and, to assess minority patient perceptions of inhaler use and the change to HFA inhalers.
Study Design: An anonymous 22-item survey to assess resident knowledge and familiarity with asthma management, inhaler technique, and HFA policy changes was administered to a convenience sample of resident physicians attending the 2008
Illinois American College of Physicians (ACP) meeting in October 2008. Focus groups were conducted with minority asthma patients at Booker
Access Health, serving African-Americans in the
South Side of Chicago and at the Cicero
Community Center, in partnership with Corazon
Community Services with Spanish-speaking moderators, serving Latino patients.
Population Studied: Internal medicine residents in the State of Illinois and community members from
South Side and Cicero communities participated in surveys and focus groups.
Principal Findings: Of the 94% (134/143) of surveys received at the regional ACP conference, respondents represented 15 Illinois residency programs. While 59% of residents felt confident in their ability to teach inhaler technique, only 49% could correctly identify the steps in using an inhaler.
Correctly identifying the steps for inhaler use was not associated with confidence in the ability to teach inhaler technique (49% confident vs. 49% not confident, p=0.97). Residents were more likely to have ever prescribed an MDI-CFC versus an HFA
(81% MDI-CFC vs. 52% HFA prescribed, p<0.001).
While 26% of residents stated that they knew the difference between HFA and MDI-CFC inhalers, only 5% were able to correctly describe differences when asked. Only 3% of residents were aware of the upcoming policy change to HFA inhalers. To date, focus group participants (n=5 at Booker; n=20 at Corazon) highlight that patients are not aware of how to use inhalers, or of the upcoming policy change to HFA inhalers. They confirmed that patients were inadequately trained on inhaler technique and do not understand the switch to HFA inhalers, reporting “they don’t work the same,” and wondering why they could not get their old inhalers and why the new inhaler was so expensive.
Specific barriers for Spanish-speaking patients included access to Spanish-speaking healthcare personnel to ask questions and understanding of medication use (rescue vs. preventive).
Conclusion: Across the state of Illinois, both residents and patients do not have sufficient knowledge about how to correctly use inhalers, nor
do they have information about the new HFA-type inhaler.
Implications for Policy, Delivery or Practice:
Educational interventions are imperative to improve care for minority patients with asthma in Chicago during this critical transition to HFA inhalers.
Funding Source(s): ACP Foundation
Theme: Prevention and Treatment of Chronic
Illness
● Gender & Income Disparities in Medical
Services Utilization of Chronic Noncancer Pain
Patients: The Opioid Utilization Study (OPUS)
Amy Puenpatom, Ph.D.; Timothy Victor, Ph.D.
Presented by: Amy Puenpatom, Ph.D.,
Pharmacoeconomics, Endo Pharmaceuticals, 100
Endo Boulevard, Chadds Ford, PA 19318, Phone:
(610) 459-6480; Email: puenpatom.amy@endo.com
Research Objective: Disparities between populations in the utilization of health care services have been described in the literature. We examined differences between gender and income groups in a sample of chronic noncancer pain patients taking opioids.
Study Design: Using data collected from the
Opioid Utilization Study (OPUS), an ongoing 1-year, multicenter, prospective, observational cohort study, negative binomial models were used to examine determinants of health services utilization.
Dependent variables were: number of medical visits and utilization of alternative/complementary therapy. Explanatory variables were: physical (PF) and mental function (MF) scores from the Short
Form 12 Health Survey (SF-12), mean cost of medical visits, net household income, insurance type, sex, race and age.
Population Studied: Adult, male and female patients being treated with an opioid for chronic (> 3 months) noncancer pain in 2008.
Principal Findings: Of 811 patients, 61.4% were female and 90.9% were white and 94.3% had experienced chronic pain for more than 1 year. The mean number of monthly medical visits was 2.9 (SD
= 3.6) for females and 2.6 (SD = 3.6) for males. We found no evidence of gender differences in health services utilization (P < 0.74). MF, net household income, and the mean cost per visit was statistically significant (p< 0.05).
Conclusion: Measures of MF and health services costs were predictors of medical care utilization.
Whereas no evidence of gender disparities was observed, household income was inversely associated with utilization.
Implications for Policy, Delivery or Practice: Our study suggested that resources targeted toward gender inequalities may be better used if they were reallocated to address differences between income groups.
Funding Source(s): Endo Pharmaceuticals
● Prescription Trends of High Risk Medications in Elderly Veterans FY03-FY06
Mary Jo Pugh, Ph.D., R.N.; Dan Berlowitz, M.D.,
M.P.H.; Eric Mortensen, M.D., M.S.; Laurel
Copeland, Ph.D.; John Zeber, Ph.D.; Joseph
Hanlon, Pharm.D.
Presented by: Mary Jo Pugh, Ph.D., R.N.,
Research Health Scientist/ Assistant Professor,
VERDICT Research Enhancement Award Program,
South Texas Veterans Health Care System/
UTHSCSA, 7400 Merton Minter (11C6), San
Antonio, TX 78229, Phone: (210) 617-5300 x
17193; Email: pughm@uthscsa.edu
Research Objective: Potentially inappropriate prescribing in the elderly, an important patient safety concern due to adverse events (e.g., falls, fractures), led to a new HEDIS quality measure which identified a list of High Risk Drugs for the
Elderly (HRDE). This study examined trends in
HRDE use within the VA, focusing on the prevalence and incidence in elderly outpatient veterans.
Study Design: This retrospective database analysis used VA administrative and pharmacy data
(FY03-06) to identify older veterans (=65 years) who were regular users of VA care (at least one visit in the year prior to exposure). The main outcome was prevalent and incident exposure to
HRDE. Longitudinal change in overall HDRE exposure and for individual drugs or classes of drugs was examined using chi-square analyses.
Population Studied: Older veterans (=65 years) who were regular users of VA care (at least one visit in the year prior to exposure).
Principal Findings: The population each year was approximately 2 million. Prevalent exposure to
HRDE overall decreased from 14.1% in FY04 to
12.6% in FY06, while incident HRDE exposure also significantly declined from 6.4% to 5.1% (p < .001).
Prevalence and incidence of HRDE decreased somewhat in most drugs. Propoxyphene-the single most commonly used HRDE-had the greatest change over time with prevalent exposure decreasing from 2.8% to 2.3% and incident exposure decreasing from 1.2% to 0.7% (p < .001).
Conversely, prevalent use of nitrofurantoin, an antibiotic commonly used for urinary tract infections, slightly increased over time from 0.3 to 0.4% (p <
.001) while incident use remained relatively stable
(0.3%). Examination of alternative drugs for specific
HRDE found significant decreases over time for some alternatives to propoxyphene (e.g., nonsteroidal anti-inflammatory drugs [17% vs14%; p<.001]), stable use for others (e.g. opioid drugs
13% throughout), and a slight increase in gabapentin (4% vs. 5%; p < .01). With regard to nitrofurantoin, use of other antibiotics for UTI remained stable at approximately 7%
Conclusion: Use of HRDE was significantly reduced between 2004 and 2006. Although prescriptions for propoxyphene dropped, it still remains the most commonly used and controversial
HRDE due to its questionable efficacy. Reductions in use of propoxyphene were associated primarily with a slight increase in use of gabapentin--an antiepileptic drug that was approved for use in neuropathic pain at the time of this study. The continued use of nitrofurantoin in older veterans may reflect the growing problem of antibiotic resistance. Such use, however, is problematic. This drug is less effective for treating urinary tract infections in patients with reduced renal function, potentially subjecting them to needless adverse effects.
Implications for Policy, Delivery or Practice:
Reduced exposure to HRDE may lead to fewer adverse drug events. Theoretically, this may translate into better overall patient health status, lower treatment costs, and perhaps improved patient satisfaction with VA care. However, prior outcomes studies have documented inconsistent findings. The next phase of this study will examine the association of HRDE exposure and patient outcomes to determine if interventions to further reduce or eliminate use of these drugs in VA are warranted.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Ethnic/Race Differences in Preferences for
Disclosure of Organ Donation Intentions
Tanjala Purnell, M.P.H.; Neil Powe, M.D., M.P.H.,
M.B.A.; Misty Troll, M.P.H.; Nae-Yuh Wang, Ph.D.;
L. Ebony Boulware, M.D., M.P.H.
Presented by: Tanjala Purnell, M.P.H., Ph.D.
Student, Health Services Research & Policy, Welch
Center for Prevention, Epidemiology, & Clinical
Research, Johns Hopkins Bloomberg School of
Public Health, 2024 East Monument Street, Suite 2-
600, Baltimore, MD 21205, Phone: (410) 502-3770;
Email: tpurnell@jhsph.edu
Research Objective: Low rates of organ donation, particularly among ethnic/racial minorities, contribute greatly to ethnic/race disparities in rates of transplantation, in which minorities are less likely to receive transplants despite their significantly greater need. Because many in the general public intend to donate their organs but have not adequately disclosed their intentions before death, outreach efforts encourage potential donors to privately and publicly disclose their donation intentions through discussions and through greater use of public organ donor registries. Yet, little is known about the most preferred methods of disclosure among persons of different ethnic/racial backgrounds. We studied ethnic/race differences in attitudes toward different disclosure methods among U.S. potential donors.
Study Design: We performed a national, crosssectional study of attitudes regarding deceased organ donation among the U.S. general public.
Households were identified using random digit selection of telephone numbers, and a telephone questionnaire was administered by trained interviewers. We assessed differences in attitudes toward public (e.g. registration through mail/telephone/computer, workplace registration, registration at a religious place of worship, and registration at a grocery store/bank/post office) and private disclosure of donation intentions (e.g. discussion with family about organ donation, with a doctor, and with a religious representative). In multivariable analyses, we assessed the independent association of ethnicity/race with perceptions after adjusting for age, sex, marital status, insurance status, and annual household income.
Population Studied: Among 845 total study participants (from 85% of eligible households), 302
(36%) stated that they had not yet declared themselves organ donors on a drivers’ license or organ donor card. Among non-donors, 18% were
African American, 62% were female, 53% were married, 88% had health insurance, 67% were employed, 57% were older than 40 years, and 52% had annual household incomes greater than
$40,000.
Principal Findings: Among African Americans, discussion with family members was the most preferred method of disclosure (70%), followed by discussion with a doctor (61%), a religious representative (54%), registration through mail/ telephone/ computer (37%), workplace registration
(34%), registration at a religious place of worship
(33%), and at a grocery store/bank/post office
(25%). Discussion with physicians was the most preferred method among non-African Americans
(66%), followed by registration through mail/telephone/computer (64%), discussion with family (62%), workplace registration (43%), registration at a grocery store/bank/post office
(38.4%), registration at a religious place of worship
(38.3%), and discussion with a religious representative (37%). After adjustment, African
Americans were statistically significantly less likely than non-African Americans to favor registration through mail/telephone/computer (adjusted OR,
0.35; 95% CI, 0.15-0.81; P= 0.014) and workplace registration (adjusted OR, 0.22; 95% CI, 0.09-0. 55;
P= 0.001).
Conclusion: African Americans strongly favored private methods of disclosing their organ donation preferences, but they were less likely than non-
African Americans to favor public methods of disclosure.
Implications for Policy, Delivery or Practice: The development and implementation of culturally targeted interventions to enhance private disclosure of donation preferences among African American families, health care workers, and religious
representatives and to relay preferences to organ procurement professionals may improve donation rates among ethnic/racial minorities and help to ameliorate ethnic/race transplant disparities.
Funding Source(s): National Center for Minority
Health and Health Disparities and the National
Institute of Diabetes and Digestive and Kidney
Diseases
Theme: Disparities
● ACPNet Chronic Obstructive Pulmonary
Disease Quality Improvement Project
Amir Qaseem, M.D., Ph.D., M.H.A., F.A.C.P.;
Meghan Gannon, M.S.P.H.; Qianna Snooks, B.A.
Presented by: Amir Qaseem, M.D., Ph.D., M.H.A.,
F.A.C.P., Clinical Programs & Quality of Care,
American College of Physicians, 190 North
Independence Mall West, Philadelphia, PA 19106,
Phone: (215) 351-2570; Email: aqaseem@acponline.org
Research Objective: The goal of this project is to develop an electronic resource for physicians and practice teams to improve the diagnosis and management of COPD, based on evidence-based guidelines and current performance measures. The practice measures are based on national guidelines and approved performance measures from the
Physician Consortium for Performance
Improvement and the National Quality Forum.
Study Design: This study evaluates the impact of an educational program utilizing pre-post design.
Practice data is collected twice during the study through a patient chart abstraction and practice survey, baseline measurement is performed at the start of the study, second measurement done approximately six months post-intervention. The intervention is the online educational intervention.
Each physician is asked to collect demographic and practice data on 35 patients from the practice who have COPD, are age 18 or older, and have been in the practice for at least one year. This includes diagnostic information such as the appropriate diagnosis of COPD with spirometry, and management indicators such as the use of appropriate COPD treatments, behavioral change counseling for smoking cessation, immunization status, and referral to rehabilitation programs. The practice pattern survey was designed to capture physician behavior in managing patients with
COPD. Confidential reports and feedback are provided to the physicians with information on quality indicators for management of COPD.
Physicians then use these data to develop improvement strategies in their practice using practice teams. There are scheduled conference calls with the practices to help them set goals using the Plan-Do-Study-Act process.
Population Studied: The sample for this project was drawn from the American College of
Physicians' Quality Improvement Program, ACPNet.
From this sample, 18 physicians agreed to participate to date.
The unit of analysis for this study is the individual primary care physician.
Principal Findings: Based on the pre-survey results, practice data reveals physicians did not meet current standards of care, such as appropriate use of spirometry. Conference calls with the practices revealed physicians felt they lacked important resources for managing patients with
COPD. There was a significant quality gap in process measure performance.
Conclusion: There is a quality gap between physicians' perceived care and their actual practice data. This awareness has motivated the physicians to work towards quality improvement goals. The focus of the project moving forward is to ask physicians to evaluate their performance measures and identify gaps in care. Physician feedback reports were reported as being strong motivators toward practice improvement.
Implications for Policy, Delivery or Practice:
This project may result in improved care and effective management of patients with COPD in meeting current performance measures. Improved management of COPD reduces hospitalizations, thus reducing health care costs.
Funding Source(s): Boehringer Ingelheim
● Raising Achievement: What We Know versus
What We Do for Alcohol Use Disorders
Amir Qaseem, M.D., Ph.D., M.H.A., F.A.C.P.;
Meghan Gannon, M.S.P.H.
Presented by: Amir Qaseem, M.D., Ph.D., M.H.A.,
F.A.C.P. Clinical Programs & Quality of Care,
American College of Physicians, 190 North
Independence Mall West, Philadelphia, PA 19106,
Phone: (215) 351-2570; Email: aqaseem@acponline.org
Research Objective: This project was funded by a collaborative RWJF grant with four leading medical societies that included the American College of
Physicians, Society of General Internal Medicine,
American Academy of Family Physicians, and
American Geriatric Society. The goal of this collaboration was to promote the integration of a patient’s alcohol use information into the routine management and prevention of common chronic diseases. This study examines the impact of an online educational program on physician practice patterns and behaviors related to their patients' atrisk drinking.
Study Design: This study utilized a pre-post intervention design and used a survey to capture data to evaluate physicians’ alcohol screening behaviors for their patients diagnosed with hypertension, sleep disorder or depression; learn about frustration and confidence levels of physicians when screening and/or treating their patients with alcohol use disorder; and measure
physician knowledge regarding association between alcohol use disorder and hypertension, sleep disorder, and depression.
The web-based educational program was designed by experts in primary care, alcohol use disorder, and addiction medicine. The goals of the webbased educational program were to: help physicians identify what constitutes at-risk drinking using an appropriate detection instrument, and educate physicians about regular evaluation of patients diagnosed with hypertension, sleep disorder, or depression for their alcohol use consumption.
Population Studied: The sample for this project was drawn from the American College of
Physicians’ Quality Improvement Program,
ACPNet. From this sample, 18 physicians agreed to participate. The majority were from small practices with 7 large practices. On an average they reported seeing 100 patients a week and represented 13 states. The primary specialty of all physicians was internal medicine, and most had been practicing medicine for more than 10 years at the time of the project.The unit of analysis for this study is the primary care physician.
Principal Findings: Our preliminary results show that most physicians were aware of the association of at-risk drinking with hypertension, sleep disorder, and depression at the beginning of the study.
However, their at-risk screening behavior did not reflect this knowledge in the pre-intervention survey. Results also indicated at-risk screening practices of physicians improved over the course of the study, specifically more physicians incorporated a valid screening instrument into their screening practice as compared to pre-intervention data.
Post-intervention data also indicated an improvement in the self-reported success physicians feel in treating patients for their at-risk drinking as well as a decrease in self-reported frustration levels. The most common perceived barriers include: lack of time, lack of patient interest, and lack of training. Physicians reported that the online feature of the project was very helpful, fast, concise, and digestible.
Conclusion: This project developed: an electronic resource for physicians to help them provide a systematic approach to addressing at-risk drinking, and a survey which identified several reasons for poor care. Both were pilot tested and validated successfully.
Implications for Policy, Delivery or Practice:
Future plans include expanding this project on a larger scale, improving care and reducing cost at a larger level.
Funding Source(s): RWJF
● Racial/Ethnic Disparities in Prescription Drug
Use among Older Adults in the U.S.
Dima Qato, Pharm.D., M.P.H.; G. Caleb Alexander,
M.D., M.S.; Rena Conti, Ph.D.; Phil Schumm, M.A.;
Stacy Lindau, M.D., M.A.P.P.
Presented by: Dima Qato, Pharm.D., M.P.H.,
Research Associate, University of Chicago, 5841
South Maryland Avenue, Chicago, IL 60637,
Phone: (443) 858-5276; Email: dqato@babies.bsd.uchicago.edu
Research Objective: We sought to examine racial and ethnic disparities in prescription drug use among the U.S. older adult population.
Study Design: A national population-based sample of older adults in the US. Household interviews were conducted between July, 2005 and March,
2006. All medications used daily or weekly on a regular basis were recorded by direct observation using a computerized log.
Population Studied: A national population-based sample of 3,005 community-dwelling U.S. adults ages 57-85 years.
Principal Findings: Overall, 82% of older adults use a prescription medication. In unadjusted analyses, in comparison to Whites, the use of at least one prescription medication was similar in
Blacks (odds ratio [OR] 1.15, 95% confidence intervals [CI] 0.80, 1.57), but significantly less among Hispanics (OR 0.49, CI 0.38, 0.73). After adjusting for demographic and health-related characteristics including diagnosed medical conditions, Whites were significantly more likely to be using a prescription medication than either
Blacks (OR 0.68, CI 0.49, 0.94) or Hispanics (OR
0.52, CI 0.37, 0.73). Racial/ethnic disparities for
Hispanics were significantly reduced (OR 0.73, CI
0.47, 1.14) after accounting for insurance status and usual source of care, and racial/ethnic disparities varied by therapeutic drug class.
Conclusion: These data provide updated, nationally representative estimates of racial and ethnic differences in prescription medication use among community dwelling elders. While differences in demographic and health characteristics did not explain the observed disparities in prescription drug use for blacks, socioeconomic and access to care disparities may.
However, differences in access to care were more important for Hispanics. In addition, differences in the use of over-the-counter drugs and dietary supplements is also evident.
Implications for Policy, Delivery or Practice:
Findings suggest potentially unmet health care needs in minority older adults in the U.S. More research is needed to identify unobserved factors associated with race/ethnicity responsible for differences in prescription medication use that may be associated with increased morbidity that may be modifiable in clinical practice.
Funding Source(s): NIA, University of Chicago
Program in Pharmaceutical Policy
Theme: Disparities
● Public Health Impact & NIH Grants for
Women’s Health
Nicole Quon, Ph.D.
Presented by: Nicole Quon, Ph.D., Assistant
Professor, School of Public & Environmental Affairs,
Indiana University, 1315 East Tenth Street,
Bloomington, IN 47405, Phone: (812) 855-0563;
Email: ncquon@indiana.edu
Research Objective: Women’s health advocates have actively campaigned for additional research resources for women’s health concerns. Breast cancer, for example, is a prominent disease on the women’s health agenda that has received substantial increases in National Institutes of Health
(NIH) research funding since the early 1990s.
However, some advocates and experts are concerned about the allocation of research resources. They suggest that higher profile diseases like breast cancer and HIV/AIDS receive a disproportionate share of NIH funding, especially relative to the public health impact. Do increases in
NIH resources for women’s health correspond with disease burden?
Study Design: This study utilizes a new dataset that links measures of the number and budget of
NIH research grants with key measures of disease burden. Previous research on NIH priority setting of women’s health concerns relied primarily on measures of prevalence and death rates. While these measures are useful, additional measures that capture aspects of the quality of life of these diseases are more informative. This study is the first to use a comprehensive set of disease burden measures to explore NIH priorities for a variety of women’s health concerns.
Population Studied: The dependent measures are the number of NIH grants awarded for key women’s health concerns from 1972-2008 and the NIH budget for key women’s health concerns from 2004-
2008.
Principal Findings: Results from this study suggest that priorities for NIH funding are shaped by higher profile diseases such as breast cancer that do not have the highest disease morbidity, mortality, or quality of life impact. The relationship between disease burden and NIH resource allocation also varies by diseases that only affect women (such as cervical cancer) and diseases that are important to women (such as stroke).
Conclusion: This paper explores the relationship between NIH research funding and public health need. The number and amount of NIH grants are loosely correlated with measures of disease burden, which suggests that other factors such as advocacy or social interest are influencing the women’s health priorities of the NIH.
Implications for Policy, Delivery or Practice:
Findings from this study provide insights about how the NIH sets priorities and the value of the NIH investment in women’s health research.
Theme: Gender and Health
● A Time-Series Analysis of the Association
Between Regional Post-Acute Care Supply & the Rate of Major Orthopedic Surgical
Procedures
David Radley, Ph.D., M.P.H.; Anna Tosteson,
Sc.D.; Elliott Fisher, M.D., M.P.H.; Doug Staiger,
Ph.D.; Vince Mor, Ph.D.
Presented by: David Radley, Ph.D., M.P.H.,
Associate, Health Policy, Abt Associates Inc., 55
Wheeler Street, Cambridge, MA 02138, Phone:
(617) 520-2651; Email: david_radley@abtassoc.com
Research Objective: The importance of care in transitional settings has surged for older adults.
The availability of inpatient post-acute care (PAC) and associated PAC spending increased dramatically as hospitals reduced patient length of stay in response Medicare payment policy. Our objective was to determine if increased availability of inpatient post-acute rehabilitation was associated with changes in orthopedic surgical rates.
Study Design: We performed a longitudinal panel analysis with difference-in-difference statistical modeling to examine the relationship between growth in PAC supply and changes in surgical rates for knee replacement and hip fracture repair. Our key analytic assumption was that the elective nature of knee replacement could make its frequency sensitive to PAC capacity, while nonelective treatment for hip fracture is not. Data were aggregated to the hospital referral region (HRR) level for analysis. Inpatient rehab (IRF) and skillednursing (SNF) supply estimates were derived from
Provider of Service (POS) and Online Survey
Certification and Reporting (OSCAR) data.
MedPAR claims files were used to identify the surgical cohorts. PAC supply estimates and surgical rates were adjusted for the demographic distribution of the region’s Medicare population. Our primary exposure was the annual change in regional PAC bed supply; the primary outcome was the annual difference in knee replacement rates less the annual difference in hip fracture rates.
Multivariate analyses were adjusted for regional acute bed supply, orthopedic surgical capacity,
Medicaid dual eligibility, residential income,
Medicare-HMO penetration, and home healthcare capacity. Data were considered annually from
1995-2004.
Population Studied: Fee-for-service Medicare beneficiaries aged 65-99 who experienced a knee replacement or surgical repair of a hip fracture.
Principal Findings: Post-acute capacity varied significantly across regions and grew over time (ptrend < 0.01). From 1995 to 2004, the number of
IRFs and SNFs each increased approximately 15%.
The number of beds available for skilled nursing care grew more substantially (130%); IRF bed supply grew modestly (11%). Knee replacement became more frequent, increasing from 4.01 to 5.83
per 1000 beneficiaries (p-trend < 0.01). Hip fracture decreased slightly, by about 1 per 1000 (ptrend < 0. 01). The correlation between annual change in PAC supply and annual change in knee replacement was modest (range: r = 0.05–0.21) but statistically significant (p<0.05) in most years; there was no correlation with annual change in hip fracture. For every 1% annual increase in PAC bed supply, there was a 5% (B=0.047, p<0.01) annual increase in the rate of knee replacement; no associated change in hip fracture; and about 3%
(B=0.032, p=0.03) higher growth in knee replacement relative to hip fracture.
Conclusion: Periods of substantial expansion in inpatient PAC capacity coincide neatly with increased incidence of knee replacement. However, this relationship likely has more to do with surgical advances and provider response to Medicare payment policy and than it does with the increased availability of PAC being able to accept postsurgical patients.
Implications for Policy, Delivery or Practice:
Medicare payment policies designed to counter financial incentives that shift care from acute inpatient to post-acute settings are associated with increased incidence of elective surgical procedures; our results suggest such strategies are not likely to produce net cost-savings.
Theme: Medicare
● Post-Acute Care & Surgical Throughput at
Acute Hospitals
David Radley, Ph.D., M.P.H.; Anna Tosteson,
Sc.D.; Elliott Fisher, M.D., M.P.H.; Doug Staiger,
Ph.D.; Vince Mor, Ph.D.
Presented by: David Radley, Ph.D., M.P.H.,
Associate, Health Policy, Abt Associates, Inc., 55
Wheeler Street, Cambridge, MA 02138, Phone:
(617)520-2651; Email: david_radley@abtassoc.com
Research Objective: To determine if patients who receive care from hospitals that more actively engage inpatient post-acute care (PAC) are more likely to experience a major orthopedic surgical procedure and/or shorter length of stay (LOS).
Study Design: Data for this study came from
MedPAR and Part-B claims for years 2000-2004.
Medicare beneficiaries were linked to their predominant ambulatory care physician and then to the acute hospital where that physician generated the plurality of their billing, resulting in hospitalspecific cohorts whose healthcare experience is attributable to the characteristics of their assigned hospital. Cohort members were followed to identify incident hospitalizations for knee replacement
(TKR) or surgical repair of hip fracture (HFX); controls had neither. For each hospital, we derived a measure of post-acute engagement based on its acute to post-acute discharge patterns among patients with neither surgical procedure. Our primary exposure was the hospital’s quartile of PAC engagement; the primary outcomes were receipt of a TKR or HFX repair, and LOS among cohort members who experienced a surgical procedure.
Multivariate analyses were stratified by surgical loyalty and were controlled for patient demographics and comorbidity; and acute hospital size, the number of affiliated orthopedic surgeons, and the presence of an on-site PAC facility.
Population Studied: We used a 20% sample of fee-for-service Medicare beneficiaries aged 65-99.
Principal Findings: Hospitals varied significantly across quartiles of PAC engagement. Mean acute to post-acute discharges were: q1(10%), q2(14%), q3(17%), q4(21%); p-trend <0.01. Hospitals with higher PAC engagement tended to be smaller (234 beds in q4 vs. 344 beds in q1), have fewer affiliated orthopedic surgeons (8 vs. 10), and were less often publicly owned (10% vs. 26%). Surgical Incidence was higher at hospitals with greater PAC engagement (TKR: 7.2 per 1000 in q4 vs. 5.6 per
1000 in q1; HFX: 8.8 per 1000 vs. 7.2 per 1000), and LOS tended to be shorter (TKR: 4.0 vs. 4.4 days; HFX: 5.9 vs. 7.0 days). The adjusted odds of receiving TKR were higher among patients assigned to hospitals with higher PAC engagement.
Odds ratios across quartiles were: q1(1.0, ref), q2(1.19, p<0.01), q3(1.23, p<0.01), q4(1.31, p<0.01). Hospitalization patterns for hip fracture followed a similar, albeit flatter, trend: q2(1.11, p<0.01), q3(1.13, p<0.01), q4(1.15, p<0.01). There was a differentially stronger association between
PAC engagement and TKR relative to HFX (q3:
OR=1.075, p=0.02; q4: OR=1.11, p<0.01).
Treatment patterns among patients who received care at hospitals with high orthopedic volume tend to be less sensitive to the hospital’s post-acute engagement than among patients treated at hospitals with lower orthopedic volume.
Conclusion: Hospitals with high post-acute engagement discharge patients more quickly and their patients have a higher likelihood of experiencing orthopedic surgery than hospitals that rely less on PAC.
Implications for Policy, Delivery or Practice: The differentially stronger association between PAC engagement and TKR relative to HFX suggests that elective surgical procedures may be more sensitive to post-surgical capacity created by transitions to
PAC. Elective surgical treatment can add cost without value if it is unwarranted or conflicts with patients’ preferences; increased churning between care settings adds complexity and may undermine efforts to improve healthcare quality.
Funding Source(s): NIA
● A Protocol for Capturing Daily Variability in
System Characteristics at the Individual Patient
Level
Laurel Radwin, R.N., Ph.D.; Howard Cabral, Ph.D.,
M.P.H.; Leslie Chen, B.S.; Bonnie Mowinski
Jennings, D.N.Sc., R.N., F.A.A.N.
Presented by: Laurel Radwin, R.N., Ph.D., Yvonne
L. Munn Nurse Researcher, Institute for Patient
Care, Massachusetts General Hospital, 275
Cambridge Street, P.O. Box 430, Boston, MA
02114, Phone: (617) 726-1018; Email: lradwin@partners.org
Research Objective: Studies linking predictors, such as nurse staffing, with outcomes, such as patient mortality, have garnered great interest in the popular and professional literature. These studies have informed practices and policies addressing patient safety and care quality. Researchers routinely examine nursing care and patient outcomes at the hospital or unit level of analysis, using information from large existing data sets (e.g.,
AHA). However, a detailed understanding is lacking of the effects of staffing, nursing experience or other nursing-focused system variables on individual patient outcomes. Understanding care variations from patient-to-patient and from day-today requires quantitatively analyzing system data at an individual patient level. The objective of this methodological paper is to report the strategies we used to collect and manage data for a patient level analysis.
Study Design: Our challenge was to devise data collection, entry, and management strategies that allowed linking a patient’s assessment of patientcentered nursing care (independent variable) and resultant HRQOL outcomes (dependent variables) with data reflecting nursing-focused health care system variables (independent variables such as nurse staffing, MD-nurse collaboration, nurse experience) specific to the patient’s length of stay.
We devised a database with scores provided by study nurse participants and administrative records.
We calculated a daily score reflecting responses provided by nurses on duty each day and daily administrative data. We then took each individual patient’s days of hospitalization and calculated a composite score of those days’ values for the nursing system variables. Measures of central tendency were calculated so that each patient had a summary score for each system variable that was specific to the nurses who provided care during that patient’s stay. This score for the system variables for the patient’s unique days of hospitalization was then linked with the patient’s independent and dependent variables’ values. The overall technical approach was to enter data into customized Access data sets, import the Access data into SAS, clean the data, and then serially merge SAS data sets.
Links for the merges included unique codes for patients and nurses, and specific calendar dates.
Population Studied: Patient participants were recruited over an 18-month period from an inpatient hematology-oncology unit at a large NE urban hospital (N=178). Male=51%; White=56%; African
American/Black=28%; Hispanic=11%. Mean age=58.4 (13.4). Nurse participants (N=41) were almost entirely women (92.7%). White=78.1%;
Mean age=40.2 years (9.1). Mean years in nursing=15 (9.0).
Principal Findings: We were able to analyze nursing-focused health care system characteristic variables specific to an individual patient’s hospitalization. As a result, more variability in key concepts was introduced into the analysis with our method than if we had used unit level data.
Conclusion: These strategies offer an alternative to investigators who wish to “drill down” to an individual patient’s nursing care and to answer questions that focus on individual patients, rather than units or hospitals. Strategies would be particularly useful for investigators studying heterogeneity on units where staffing, collaboration, etc. vary considerably.
Implications for Policy, Delivery or Practice:
Patients care varies from day-to-day and methodologies are needed that capture this variability. By understanding this variability, more precise practice changes can be made to affect care processes and resultant outcomes.
Funding Source(s): AHRQ
● Is Patient-Centered Cancer Nursing Care for
Racially Diverse Patients Related to Health &
Societal Outcomes?
Laurel Radwin, R.N., Ph.DHoward Cabral, Ph.D.,
M.P.H.; Gail Wilkes, R.N., M.S., A.O.C.N.
Presented by: Laurel Radwin, R.N., Ph.D., Yvonne
L. Munn Nurse Researcher, Institute for Patient
Care, Massachusetts General Hospital, 275
Cambridge Street, P.O. Box 430, Boston, MA
02114, Phone: (617) 726-1018; Email: lradwin@partners.org
Research Objective: Patient-centered care, although effective in achieving desired outcomes, is unequally distributed. This ongoing study’s purpose is to describe and analyze the relationships among race, patient-centered nursing care (PCC-N) for hospitalized cancer patients, health-related quality of life (HRQOL) and hospital readmissions within 30 days (READMIT-30).
Study Design: This is a secondary analysis of data from a parent study that examined relationships between PCC-N and HRQOL outcomes in the context of the health care system. Race effects were not examined in the parent study. Patients’ responses to valid, reliable scales operationalized four aspects of PCC-N: responsiveness, individualization, coordination, proficiency. While hospitalized and post-discharge, patients completed valid, reliable scales that operationalized HRQOL: trust in nurses, authentic self-representation, cancer optimism, cancer distress, sense of well-being, and post-discharge functional QOL. Patient demographic, disease, and other data were collected from medical records, as was the patient’s
READMIT-30 status. Our findings will result from
the following activities. (1) Regression analyses will be conducted to assess the relative risk of
READMIT-30 as affected by race and PCC-N.
Covariates such as age, gender, metastatic disease, treatment modality, index admission length of stay will be examined. (2) Descriptive statistics will provide an examination of the relationships between HRQOL, race, and PCC-N. (3) An expert panel will conduct a chart review of each readmitted patient to describe the reason for READMIT-30.
Agreement will be evaluated using kappa. Reasons for READMIT-30 will be qualitatively analyzed and categorized initially as guided by the literature and sequentially refined as befits the data.
Population Studied: Patients from a hematologyoncology unit at an urban safety net hospital participated; (N=220). Mean age = 59 years
(SD=13); 58% White, 29% AA/B; 11% Hispanic;
35% had metastatic disease. The most prevalent four types of cancer were GI (26%), hematologic
(22%), lung (14%), breast (14%)
Principal Findings: (1) Preliminary regressions included potential confounders and suggested that the likelihood of READMIT-30 decreased as PCC-N increased. More complete models will estimate the effects of PCC-N on READMIT-30 while accounting for race. (2) Relationships between PCC-N,
READMIT-30 and race will be described. (3) A typology of reasons for READMIT-30 will result from qualitative analyses.
Conclusion: We have a unique opportunity in this ongoing study to examine racial disparities in the quality of PCC-N and in important outcomes
(HRQOL and readmissions). Such examination will enhance understanding of the variance in quality care that is uniquely attributable to nursing.
Implications for Policy, Delivery or Practice:
Certain questions will be informed by the analyses:
(1) Are there meaningful racial differences in the relationships between PCC-N, HRQOL and
READMIT-30 among cancer patients? If so, eliminating racial barriers to receiving high-quality nursing care is elevated as a goal. (2) Does PCC-N mediate the effect of race on cancer patients’
HRQOL and READMIT-30? If a mediating effect is present, then strengthening the implementation of
PCC-N “best practices” can be expected to reduce racial disparities in these outcomes. (3) Does PCC-
N moderate the effect of race on HRQOL and
READMIT-30? If so, then learning more about how
PCC-N should be adapted for different groups becomes important.
Funding Source(s): AHRQ
● Disentangling the Relationship Between
Placement Instability & Behavioral Problems among Children in Child Welfare: An
Instrumental Variables Approach
Ramesh Raghavan, M.D., Ph.D.; Peichang Shi,
M.S.W.; Jillian Waid, M.S.W.
Presented by: Ramesh Raghavan, M.D., Ph.D.,
Assistant Professor, Washington University in St
Louis, Campus Box 1196, Saint Louis, MO 63130,
Phone: (314) 935-4469; Email: raghavan@wustl.edu
Research Objective: Assuring stable placements for children in foster care is of great concern to child welfare policymakers. However, children who exhibit behavioral problems often disrupt placements, and disruptive placements can in turn affect the emotional well-being of children. Prior research has not controlled for the endogeneity between placement change and behavioral problems, and the unobserved heterogeneity associated with observational samples. In this study we disentangle the relationship between placement change and behavioral problems among children in child welfare in order to inform policymaking by child welfare agencies.
Study Design: Data are from 3 years of the nation’s first representative panel study of children coming into contact with child welfare agencies, linked to the Area Resource File. We observed children with at least one movement into foster care. We used race/ethnicity as an instrument for placement, and maltreatment history as an instrument for behavioral problems, in lagged models that controlled for the child’s demographic characteristics, caseworker experience and education, and population adjusted ratios of health service providers in the county of the child’s residence. All models used an additional indicator variable of urbanicity of the county. Weighted instrumental variables estimations were performed in Stata.
Population Studied: 496 children followed up for 3 years in the National Survey of Child and
Adolescent Well-Being.
Principal Findings: These children displayed a mean of 1.5 placements (range 1-8) over the 3 years of follow-up. Children had a mean score of 58 on the externalizing subscale of the Child Behavior
Checklist (range 30-94; scores above 63 indicate probable disorder). Instrumental variables regression revealed that behavior problems at year
1 increased the total number of placement changes between years 1 and 2 (b=0.06; SE: 0.01; p<0.0001), but did not have any effects on placement changes between years 2 and 3.
Placement changes, however, only worsened behavioral outcomes between years 2 and 3
(b=13.2; SE: 5.2; p=0.01). Children living in counties with 1 more psychiatrist per 10,000 individuals displayed lower levels of behavioral distress (b=-7.3; SE: 2.1; p<0.001). Caseworker characteristics did not seem to play a role in the relationships between placement change and behavioral change.
Conclusion: The relationship between placement change and behavioral disturbances are cyclical, and characterized by differences in magnitude and
temporality. These findings reveal relatively strong, distal effects of placement change on behavioral distress; in contrast, the effects of behavioral disturbances do not seem to have a similar magnitude, and are more proximal.
Implications for Policy, Delivery or Practice:
Child welfare agencies can break this cycle of behavior problems and placement change by identifying, screening, and treating children with large numbers of placements. Screenings for emotional distress should occur not just in the short-term but also in the long-term. Treatment strategies, when appropriately deployed, may reduce future placement instability fairly quickly, and may offer a way to protect the emotional wellbeing of these highly vulnerable children.
Funding Source(s): NIMH
Theme: Behavioral Health
● Medical Tourism: A Cost-Benefit Analysis
Zo Ramamonjiarivelo, M.B.A.; Gouri Gupte, M.H.A.
Presented by: Zo Ramamonjiarivelo, M.B.A.,
Research Assistant, Department of Health Services
Administration, University of Alabama at
Birmingham, 1705 University Boulevard,
Birmingham, AL 35292-1212, Phone: (205) 934-
3113; Email: zoramam@uab.edu
Research Objective: The rising healthcare costs in industrialized nations along with long waiting lists are the protagonists towards a trend we call
“medical tourism.” Medical tourism can be divided into three major categories: outbound medical tourism, when US citizens travel overseas for treatment; inbound medical tourism, when foreigners seek treatment in the US; and intrabound medical tourism, when patients travel to another state for treatment. This paper focuses on outbound medical tourism. Outbound medical tourism is a large and growing industry with annual gross income of $60 billion worldwide in 2006 (Herrick,
2007). The availability of the latest technology, skills and reduced transportation, and accommodation costs are acting as the catalysts towards this movement of global healthcare. However, the existing research does not provide an adequate cost-benefit analysis. This study performs a cost benefit analysis of some of the most common medical procedures such as bypass heart surgery, kidney transplant, hip replacement, and knee surgery. These procedures are performed in India,
Singapore, Thailand, or Eastern Europe rather than in the U.S.
Study Design: The study reviews available information from journals, and research websites to analyze the difference in the costs of the most common surgeries performed in the United States and overseas. It thus evaluates the financial gains or losses patients and third party payers could achieve or suffer through medical tourism.
Population Studied: The study focuses on the literature and data that targets medical tourism in
Asia and Eastern Europe comparing the costs in
United States.
Principal Findings: On average, most of the medical procedures performed abroad cost approximately 40% less than if they are performed in the United States. About 38% of the uninsured and 25% of the insured would travel overseas if their savings exceed $10,000. Roughly 25% of the uninsured and 10% of the insured would travel abroad if they can save between $1,000 and
$2,400. This is consistent with other studies like
Herrick (2007). Deloitte Development LLC (2006) estimates that spending on US outbound medical tourism approximately amounts to $2.1 billion; which results in $15.9 billion of lost revenue to the
US health care industry.
Conclusion: Medical tourism could generate substantial cost savings for patients and third party payers and loss for health care providers. However, this practice involves some risks such as lack of legal protection for patients in case of medical malpractice and complications, and long hour flights before and after medical procedure. In addition,
American physicians are reluctant to provide medical follow-up on patients that had procedures performed overseas.
Implications for Policy, Delivery or Practice:
Major challenges for health care exist due to the rising costs, decreased patient satisfaction and quality issues. Medical tourism is not the complete solution to the problem but can guarantee quality care within limited resources for a number of diseases. However, the growth of medical tourism should motivate American healthcare providers to be more effective and efficient. Moreover, international law and regulations on medical tourism should be established to facilitate the practice and protect all parties involved.
Theme: Global Health
● The Effect of the 2003 EMTALA Update On
Trauma Care
Mitesh Rao, M.D.; Erika Martin, M.S.; Andrew
Epstein, Ph.D.; Linda Degutis, Dr.P.H.; Cary Gross,
M.D.
Presented by: Mitesh Rao, M.D., Clinical Scholar,
Robert Wood Johnson Clinical Scholars Program,
Yale School of Medicine, IE-61 SHM P.O. Box
208088, New Haven, CT 06520-8088, Phone:
(203) 785-6499; Email: mitesh.rao@yale.edu
Research Objective: The Emergency Medical
Treatment and Labor Act (EMTALA) was updated in
2003 relaxed the requirements for hospitals to provide specialist coverage in their Emergency
Departments. There has been concern that this updated regulation and its subsequent effect on oncall coverage has led to lead to increased wait times and transfers for trauma patients. The primary
objective of this study is to examine the effect of the
2003 EMTALA update on probability of transfer and time to transfer for trauma conditions, as well as the association between patient, hospital, and encounter characteristics and the probability of a trauma patient being transferred.
Study Design: We analyzed the National Hospital
Ambulatory Medical Care Survey from 2000-2006.
Outcomes of interest were patient transfer to another facility and time to transfer. Predictors included year, day, and time of visit, sex, ethnicity/race, payment source, geography, and hospital ownership. Probability of transfer was estimated using logistic regression. Generalized estimating equations were used to adjust for hospital-level and year-level clustering. Time-totransfer was analyzed using a random coefficients model. Differences in transfer probability and timeto-transfer pre (2000-2003) and post (2004-2006)
EMTALA update were assessed by testing for differences in time coefficients.
Population Studied: The NHAMCS is a national probability sample of patient visits to Emergency
Departments, excluding Federal, military, and
Veteran's hospitals in the US. We limited our sample to adults with a primary ICD-9 code indicating trauma.
Principal Findings: The adjusted probability of transfer for trauma patients dropped in 2005 to 59% of the adjusted probability from 2004, and then in
2006 down to 50% of the adjusted probability from
2004. Categorical test of trend on the probability of transfer were significant with a p<0.01. Patients who were older (p<0.01), males (p<0.01), African-
Americans (p=0.04), and who arrived to the ER on weekends (p=0.04) were significantly more likely to be transferred. Hospitals located in the Western
(p=0.03) and Midwestern (p=0.02) regions of the country, those outside of MSA regions (p<0.01), and that are designated government non-federal
(p<0.01) were more likely to transfer their trauma patients. Hispanic ethnicity and payer-status were not found to be significant predictors of transfer for trauma patients.
Conclusion: Probability of transfer of trauma patients significantly decreased since the 2003 update. Males, African-Americans, elderly, and arrivals on weekends are more likely to be transferred. Hospitals in the Midwest/West, outside
MSAs, or government non-federal owned are more likely to transfer trauma patients. Payer status, ethnicity, and arrival on a weeknight did not significantly affect transfer probability of trauma patients
Implications for Policy, Delivery or Practice: The
2003 EMTALA update has not been detrimental to trauma care access. Significant disparities in transfer probability do exist based on examination of patient-level characteristics such as age, race, and gender. As trauma and surgical specialty coverage is anecdotally more difficult to obtain on weekends, in the Midwest and Western areas of the country, and in more rural areas, it is not surprising to see increased odds of transfer associated with these characteristics. It is, however surprising to see an increased odds of transfer for government non-federal hospitals, which typically have acceptable trauma and specialty-surgeon coverage.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Policies and
Incentives
● Access to a Medical Home Reduces
Disparities in Emergency Care Utilization by
Children with Special Health Care Needs
Jean Raphael, M.D., M.P.H.; Yiqun Zhang, B.S.;
Hao Liu, Ph.D.; Carl Tapia, M.D., M.P.H.; Angelo
Giardino, M.D., Ph.D., M.P.H.
Presented by: Jean Raphael, M.D., M.P.H.,
Assistant Professor of Pediatrics, Pediatrics, Baylor
College of Medicine, 6621 Fannin Street, Suite
1540.00, Houston, TX 77030, Phone: (832) 822-
1791; Email: raphael@bcm.edu
Research Objective: The concept of the medical home has been expanded beyond simply having a usual source of care. The effect of the medical home in reducing racial/ethnic disparities has not been well studied, nor which aspects of the medical home (comprehensive, accessible, family centered, culturally sensitive, coordinated, or continuous) are important. Children with special health care needs
(CSHCN) are of particular interest, as they represent a resource-intense and medically fragile population. Objectives: 1) To analyze racial/ethnic disparities in emergency department (ER) utilization for CSHCN. 2) To determine whether having a medical home was associated with reduced disparities in ER utilization in comparison to having just a usual source of care.
Study Design: We analyzed data on children ages
0-17, for whom data regarding medical home were available (n=36,864) from the 2005 National Survey of Children with Special Health Care Needs. The main dependent variable was emergency care utilization (2 or more visits versus none) in the last
12 months. The primary independent variables were race/ethnicity and type of care (medical home, usual source of care, no medical home or usual source of care). Bi-variate analysis was conducted to investigate associations between race/ethnicity, medical home, and other socio-demographic characteristics and emergency care utilization.
Multivariate logistic regression analysis was utilized to further assess these associations while adjusting for multiple socio-demographic variables and their potential interactions.
Population Studied: Children ages 0-17 from the
National Survey of Children with Special Health
Care Needs, 2005-2006
Principal Findings: The odds of having 2 or more
ER encounters versus none were higher for Black,
Hispanic, and Other children compared to White
children. These differences persisted for Blacks
(Adj OR 1.56, confidence interval 1.35-1.81) and
Others (Adj OR 1.29, confidence interval 1.06-
1.56), after controlling for socio-demographic variables. Having just a usual source of care was associated with a higher odds of emergency encounters compared to having a medical home
(Adj OR 1.32, confidence interval 1.17-1.47).
Racial/ethnic disparities in ER utilization disappeared when children had medical homes compared to just a usual source of care.
Conclusion: Emergency care utilization for
CSHCN differed significantly according to race/ethnicity. Having a medical home, and not just a usual source of care, appears to reduce disparities in emergency care utilization.
Implications for Policy, Delivery or Practice: The patient-centered medical home has been increasingly promoted as a potentially transformative health care delivery innovation in primary care and physician reimbursement. Our study demonstrates that it has important implications in reducing racial/ethnic disparities in pediatric primary care.
Theme: Child Health
● Parenting Stress in U.S. Families is
Associated with Pediatric Healthcare Utilization
Jean Raphael, M.D., M.P.H.; Yiqun Zhang, B.S.;
Hao Liu, Ph.D.; Angelo Giardino, M.D., Ph.D.,
M.P.H.
Presented by: Jean Raphael, M.D., M.P.H.,
Assistant Professor of Pediatrics, Pediatrics, Baylor
College of Medicine, 6621 Fannin Street, Suite
1540.00, Houston, TX 77030, Phone: (832) 822-
1791; Email: raphael@bcm.edu
Research Objective: Parenting stress has been defined as the accessibility and perceived availability of resources for parenting relative to the demands of parenting. According to family stress models, parental responses to stress can disrupt interactions between parent and child, leading to parental inability to seek timely medical care for their children. To date, no studies have used national data to examine relationships between parenting stress and healthcare utilization. The objective of this study was to quantitatively assess the relationship between parenting stress and child healthcare utilization.
Study Design: We conducted a cross-sectional analysis of 86,895 children, ages 0-17, using data from the 2003-2004 National Survey of Children s
Health. The primary independent variable was parenting stress, as defined by the Aggravation in
Parenting Scale and Parenting Stress Index. The main dependent variables were use of preventive and emergency care in the past 12 months. Multiple logistic regression was conducted to investigate associations between parenting stress and healthcare utilization, controlling for multiple sociodemographic variables and their potential interactions.
Population Studied: Children, 0-17 from the
National Survey of Children's Health, 2003-2004
Principal Findings: Parenting stress was greatest in households with the following parental characteristics: less than high school education, non-English primary language, and fair/poor mental health. Parents of Hispanic (23.7%) and Black
(17.1%) children reported the highest proportions of high parenting stress compared to those of White
(8.3%) children (P<.0001). For child preventive care, there were no differences in utilization between parents with high versus low parenting stress. For emergency care utilization, parents with high parenting stress had a higher odds (Adjusted odds ratio: 1.47, 95% confidence interval: 1.34-
1.60) of seeking emergency care for their children compared to parents with low parenting stress, controlling for multiple socio-demographic variables and parental mental and physical health.
Conclusion: Having a parent with high parenting stress is associated with greater utilization of pediatric emergency care. This pattern of increased use of costly resources may impact a child's receipt of timely medical care and contribute to the financial burden on the health care system.
Implications for Policy, Delivery or Practice:
Previous studies have assessed relationships between parental mental health and healthcare utilization. To our knowledge, this is the first national population based study to identify an association between parenting stress and healthcare utilization. Parenting stress may provide a new area for focus and intervention to provide families with needed support in the primary care setting and reduce unnecessary emergency care utilization and costs.
Theme: Child Health
● An Intervention to Improve Care & Reduce
Costs for Medicaid Patients with Frequent
Hospital Admissions
Maria Raven, M.D., M.P.H., M.Sc.; Brian Elbel,
Ph.D.; Shannon Kostrowski, M.P.A.; Colleen
Gillespie, Ph.D.; Marc Gourevitch, M.D., M.P.H.;
John Billings, J.D.
Presented by: Maria Raven, M.D., M.P.H., M.Sc.,
Assistant Professor, Emergency Medicine, New
York University School of Medicine, 462 1st
Avenue, New York, NY 10016, Phone: (917) 499-
5608; Email: maria.raven@med.nyu.edu
Research Objective: For a subset of fee-forservice Medicaid patients with frequent hospital admissions, contact with the health care system remains acute and episodic at high cost to
Medicaid, while less costly outpatient primary and preventive care services are underutilized.
Previous work validated the accuracy of a predictive case-finding algorithm to identify complex Medicaid
patients at risk for future high costs who might benefit from more intensive services, and identified remediable risk factors such as substance use, homelessness, and lack of social support associated with frequent hospital admissions. We aimed to pilot an intervention for a limited number of high-cost patients to address unmet health and social needs in both the hospital and community, to improve care while reducing hospital admissions and associated costs in this population. Our intention was to expand the program based on pilot success.
Study Design: Community and hospital-based care management intervention with process and implementation evaluation, and pre-post cost analysis. Eligible patients were offered intervention enrollment during an admission to an urban public hospital. Patients underwent in-depth psychosocial interviews by study social workers to identify immediate and long-term needs such as housing, primary care, transportation to and advocacy during appointments, medication management, entitlements enrollment, improved connections to psychiatric and substance use treatment, and home visits. Patients who met criteria for chronic homelessness were evaluated in-hospital by a community-housing partner who initiated housing applications based on a housing first model. Prepaid cell phones were provided to patients when needed to maintain close contact with study staff for reminder calls and crisis management. Study staff worked closely with inpatient providers to facilitate appropriate discharge planning and follow-up.
Population Studied: Consecutive Englishspeaking Medicaid fee-for-service patients aged 18-
64 identified as high-cost and high-risk for readmission in the following 12 months by a validated predictive case-finding algorithm.
Principal Findings: Over the past year, 19 patients have enrolled. 100% are male. 17/19 were chronic substance users at enrollment. 5/19 were lost to follow-up. Of the remaining patients, 8 met criteria for chronic homelessness that would facilitate expedited placement into permanent housing. Of these 8, 2 were placed in nursing homes and 2 died. The remaining 4 chronically homeless patients are now in transitional or permanent housing. Hospitalizations and ED visits have decreased, while establishment of an outpatient medical home has increased. Comparing the 9-12 months after the intervention to the 12 months before intervention revealed a decrease in average monthly inpatient Medicaid costs per patient ranging from $1205-$2881. This resulted in an average annual inpatient cost reduction from
$14,464 to $34,568.52. Prior research indicates without intervention, Medicaid costs for these patients in the following 12 months will increase.
Conclusion: A pilot intervention to improve care for medically, socially complex high-cost Medicaid patients shows savings to Medicaid and decreased hospitalizations and ED visits by addressing issues that are challenging for the traditional health care system to manage.
Implications for Policy, Delivery or Practice: Our model will be expanded to serve a greater number of patients across additional hospitals to determine if the success of our pilot can be replicated, and will include a more detailed cost analysis.
Funding Source(s): The United Hospital Fund
Theme: Medicaid, SCHIP and State Health Reform
● Community Reception Center Integrating All
Hazard Public Health Emergencies
Michael Rea, M.P.H.; Robert Gougelet, M.D.;
Roberto Nicolalde, M.S., Ph.D.
Presented by: Michael Rea, M.P.H., Research
Associate, New England Center for Emergency
Preparedness, Dartmouth Medical School, 1
Medical Center Drive, Lebanon, NH 03756, Phone:
(603) 653-1726; Email: michael.rea@dartmouth.edu
Research Objective: For public health emergencies affecting populations, a variety of models exist for community-based centers to address the needs of ambulatory and non-urgent care patients. Whether the "Points of Dispensing"
(POD) clinic designed by the Centers for Disease
Control and Prevention (CDC) for mass prophylaxis, the CDC Vaccination clinic (VC) to address diseases such as smallpox or pandemic flu, the Neighborhood Emergency Help Center
(NEHC) developed for bioterrorism events by the
Department of Defense under the Modular
Emergency Medical System (MEMS) concept, or the modified NEHC for radiological events proposed by the CDC; the concept of providing medical screening and treatment or intervention remains constant through these systems. We seek to analyze these systems in terms of staffing, resources and patient throughput characteristics to design an all-hazards generic community reception center flexible enough to accommodate a variety of public health emergencies.
Study Design: We analyzed hazard specific reception centers according to staffing in terms of clinical training and area of expertise, necessary facilities, hours of operation and number of shifts, processing time required for the specific hazard medical intervention, patient throughput in terms of flow and capacity, and clinical resources and supplies. Planning software tools developed by
AHRQ for mass prophylaxis and the CDC for vaccination clinics were compared in terms of output for the same characteristics and then validated using queuing times for each station or output in the proposed model.
Population Studied: We studied reception center plans developed for specific hazards to include the
CDC POD, CDC VC, NEHC, the CDC radiation reception center, and California and New Jersey state mass prophylaxis clinics.
Principal Findings: We found that a generic reception center is capable in an all-hazards approach to public health emergencies.
Characteristics of a generic reception center to accommodate 500 patients per hour includes a core of 150 clinically and non-clinically trained personnel, a stand-along structure of approximately 15,000 square feet operating for 24 hours in two 12 hour shifts, and using standard medical supplies augmented as necessary for the specific hazard.
Just in time training and defined responsibilities reduces the burden upon staff and may decrease requirements for clinically trained personnel in high demand in public health emergencies such as clinicians and nurses.
Conclusion: Given the common aspects to the various models and systems for specific hazards, a core generic center with a standard staffing approach is capable of addressing public health emergencies for ambulatory and non-urgent care patients in an all-hazards approach.
Implications for Policy, Delivery or Practice:
This all-hazards generic community reception center better enables a community-based response and fits into larger systems for medical surge and mass care such as the MEMS model.
Funding Source(s): Northern New England
Metropolitan Medical Response System (NNE
MMRS)
● Health Savings Accounts: Whose Skin Is in the Game & How Does it Affect Care-Seeking
Behavior?
Mary Reed, Dr.P.H.; Huihui Wang, Ph.D.; Vicki
Fung, Ph.D.; William Dow, Ph.D.; Joseph
Newhouse, Ph.D.; John Hsu, M.D., M.B.A.,
M.S.C.E.
Presented by: Mary Reed, Dr.P.H., Staff Scientist,
Kaiser Permanente Division of Research, 2000
Broadway, Oakland, CA 94612, Phone: (510) 891-
3808; Email: mary.e.reed@kp.org
Research Objective: Consumer directed health plans with Health Savings Accounts (HSAs) potentially could engage consumers more in their health care, while also decreasing unnecessary care; however, there is limited information on contributions or subsequent effects on patient careseeking choices. We examined HSA contributions, cost-related changes in care-seeking, and reports of financial burden.
Study Design: We conducted telephone interviews with a random sample of adult members of an integrated delivery system who had HSA-eligible plans in 2007. Patients reported the amount and source of HSA contributions, i.e., employee vs. employer; cost-related changes in use of the emergency department (ED), office visits, preventive medical tests/screenings (exempt from the deductible), and prescription drugs; and health care cost-related financial burdens, i.e., borrowing money or going without necessities to pay for care.
We used multivariate logistic regression to assess the association between patient characteristics
(including age, race, household income, and education) and contributions, and to assess the association between contributions and care-seeking behavior.
Population Studied: Among 488 participants (78% response rate), 51% were <45 years old, 49% were female, 60% had “very good” or “excellent” selfrated health, 84% had some college education, and
19% had incomes <$40,000.
Principal Findings: Overall, 81.5% of participants knew that they had an HSA-eligible plan; among whom 70% received some employer contribution;
54% made some contribution themselves; 23% had contributions from both sources; and 39% had HSA contributions totaling >100% of their deductible amount. In multivariate analyses, participants with employer HSA contributions were significantly less like to make a contribution themselves (OR=0.61,
95%CI=0.38-0.97), as were those age <35 (OR compared with age 55+=0.29, 95%CI=0.14-0.61), or of non-White race/ethnicity (OR=0.54,
95%CI=0.34-0.87). Among all participants, 38% reported cost related changes in ED care; 53% for office visits; 20% for preventive medical tests; and
55% for prescription drug use. Participants with employer HSA contributions only (and no employee contribution) were significantly less likely to change their ED (OR=0.53, 95%CI=0.30-0.96), and office visit (OR=0.57, 95%CI=0.32-1.00) care-seeking, compared with those with contributions from both sources. Overall, 21.4% reported that their health care costs created financial burdens (borrowed money or went without a necessity); participants who had only employer contributions were significantly less likely to report financial burdens
(OR=0.47, 95%CI=0.23-0.96).
Conclusion: While most consumers with HSAs have employer contributions to these accounts, most have unfunded portions of their annual deductible. One in five reported avoiding preventive tests exempt from the deductible, and one in five experienced financial burdens associated with paying for care. Consumers with employer HSA contributions were less likely to contribute themselves, reported fewer changes in their careseeking behavior, and were less likely to report experiencing financial burdens.
Implications for Policy, Delivery or Practice:
Consumers often report changing their careseeking behavior in response to costs, but the source of HSA contributions significantly impacts the frequency of these responses and the financial burden of health care costs. Some of these behavioral responses also may be clinically inappropriate.
Funding Source(s): Kaiser Foundation Research
Institute
Theme: Consumer Choices in Health Care
● Why do Tuberculosis Patients Default?
Tariq Rehman, M.D.; Anis, Rehman, M.D.; Shamim
Ur Rehman, M.B.B.S., D.T.M.H.S.,M.R.C.P. U.K.;
Rehman, Tariq
Presented by: Tariq Rehman, M.D., Physicians in
Internal Medicine, Internal Medicine, Nephrology
Section, Marshall University medical Center, 1249
15th Street, Suite 3000, Huntington, WV 25701,
Phone: (304) 634-0479; Email: anisramay@hotmail.com
Research Objective: Why do Tuberculosis
Patients default?
Study Design: A total of 105 adult patients were recruited from January 2005 to December 2007 in a
TB clinic of a teaching hospital. All the patients had active TB according to WHO criteria and history of default after they had been on therapy for 2-5 months. 56 patients had history of default on one occasion, 31 on two occasions, 15 on three occasions, and 3 patients took several therapies. A face to face interview using a structured questionnaire was conducted. Interviewers were physicians and they requested a verbal consent from the patients in their first language. The questionnaire included reasons for stopping the therapy in the past, duration of therapy before default, drugs with their dosages, physicians’ recommendations, and compliance during the previous therapy. The study was approved by the local institutional review board (IRB) of Mohammad
Medical College, Mirpurkhas.
Principal Findings: The factors affecting adherence were divided into broad categories
Structural Factors: includes inability to afford medication and to pay for transportation to the clinic.
Personal Factors: includes knowledge, beliefs, and attitude towards treatment, interpretation of illness, wellness and cure. Health Service Factors: comprised of health care provider, health care system, disease progression, and adverse effects of drugs. We found that 47% of patients defaulted due to personal factors. Out of these, 23% felt no improvement despite on therapy and 13% stopped therapy as they thought they are cured because their symptoms had resolved. 8% defaulted as they were tired of being on therapy and remaining 3% thought they did not have TB. 39% of our patients fell under category of structural factors, with 34% could not afford the cost of the therapy and the rest
5% were unable to afford the transportation related expenses to the health care facility. 15% of patients fell under the category of health services factor. Out of group, 12% claimed that their health care provider had told them that they are cured once their symptoms resolve and instructed to stop the treatment. Remaining 3% in this group stopped the therapy due to development of adverse effects to drugs without consulting their physician.
Conclusion: Our study is an effort to answer one of the most difficult questions of why patients default.
Default is a product of several barriers including low socio-economic status, cultural beliefs, inadequate health care system, complicated disease process, and long term therapy. We think, patients usually take their medication under difficult circumstances without much financial, medical and social support.
Implications for Policy, Delivery or Practice:
Adherence to long term therapy with side effects is a complex and challenging task. However, a system is desperately needed that pays more attention to structural, personal and health care barriers to help a poor sick man who is fighting the battle against
TB.
Theme: Prevention and Treatment of Chronic
Illness
● Strengthening State & Local Roles in the
National Food Safety System
Adam Reichardt, M.P.A.; Michael, Taylor, J.D.;
Stephanie David, J.D.; Subha Chandar, M.P.H.;
Adam Reichardt, M.P.A.; Joe Corby; Mike Batz
Presented by: Adam Reichardt, M.P.A., Director,
Environmental Health, Environmental Health,
Association of State & Territorial Health Officials,
2231 Crystal Drive Suite 450, Arlington, VA 22202,
Phone: (571) 527-3170; Email: areichardt@astho.org
Research Objective: This project aimed to develop an agenda for strengthening state and local roles in the national food safety system. The objectives of the project were to 1. Formulate and express a modern vision of the role of state and local government in an integrated, prevention-oriented food safety system; 2. Identify gaps or constraints in current law, policy and practice at the federal, state and local levels that inhibit fulfillment of that vision;
3. Recommend changes in law, policy, and practice that are needed to enhance the effectiveness of state and local agencies in addressing food safety problems at the local, state and national level; 4.
Identify specific opportunities to improve collaboration among state, local, and federal agencies; and 5. Describe current funding patterns and resource needs at the state and local level.
Study Design: The project held workshops that brought together state and local officials, their federal counterparts, and diverse stakeholders of the food safety system, including the food industry and consumer groups to develop recommendations on how to improve state and local public health’s role in the context of an integrated food safety system. The workshops focused on specific issues such as: national policy, regulatory and inspection activities, surveillance and outbreak response, and integrating the national food safety system.
Recommendations from the workshops were summarized and compiled into a final project report.
Population Studied: State and local governmental public health and food safety practitioners were the primary population queried for recommendations.
Food industry representatives, consumer advocacy groups, and federal government representatives also provided input into the project.
Principal Findings: The project produced 27 findings and 19 key recommendations. The findings describe the principal role state and local agencies play in the national food safety system as well as the gaps and challenges that currently exist. The recommendations express specific national policy changes that are needed to enhance and better integrate the national food safety system. Principal recommendations include giving the Secretary of
HHS a legislative mandate to lead the development of an integrated, national food safety system that incorporates and enhances the food safety capacity of state and local agencies; directing HHS to organizationally unify FDA’s food safety functions under a single official with clear responsibility and accountability for leading the development of an integrated, national food safety system; establishing a National Food Safety Leadership Council (FSLC) to serve as an ongoing vehicle for federal-statelocal collaboration to build an integrated national food safety system; and providing federal funding and establish a matching grant program, backed up by a line item in the HHS appropriation.
Conclusion: The recommendations in this report address many of the system changes needed to fulfill the vision of an integrated national food safety system.
Implications for Policy, Delivery or Practice: The goal of this project was to create a platform for a better integrated national food safety system. It is the goal of the project team that these recommendations will be seriously considered by national policymakers and implemented to strengthen public health protection through these programs.
Funding Source(s): RWJF
Theme: Public Health
● Applying the Chronic Care Model Across
Multiple Conditions: A Planned Care Quality
Improvement Initiative in the Indian Health
System
Ty Reidhead
Presented by: Ty Reidhead, National Chief Clinical
Consultant in Internal Medicine, Indian Health
System, P.O. Box 860, Whiteriver, AZ 85941,
Phone: (928) 338-4911; Email: charles.reidhead@ihs.gov
Population Studied: American Indian/Alaska
Native population
Principal Findings: Title: Applying the Chronic
Care Model across multiple conditions: A planned care quality improvement initiative in the Indian
Health System. Suggested Author List: Cindy
Hupke, Ty Reidhead, Bruce Finke, Pat Lundgren,
Lisa Dolan, Jerry Langley, Tracy Jacobs, Lindsay
Hunt, Kedar Mate, Don Goldmann. Chronic and preventable conditions result in a high burden of illness in American Indian and Alaska Native peoples. In 2007, the Indian Health Service (IHS) launched the Chronic Care Initiative (CCI) with the aim of improvement in clinical prevention and the management of chronic conditions using the framework of the Chronic Care Model. The
Innovations in Planned Care (IPC) collaborative, a component of the CCI, is based on the
Breakthrough Series Collaborative model and focuses on strengthening the relationship between the prepared, proactive care team and the patient, family, and community. Improvement is guided by measurement in four domains: preventive care, management of chronic conditions, patient experience of care, and cost of care. We report here on selected clinical indicators from the first 18 months of this comprehensive care redesign effort.
Methods: Fourteen pilot Indian health facilities participated in the IPC Collaborative. Within each facility, a group of providers and their patients were identified as a clinical “microsystem.” (Clinical microsystems are defined as the front-line units that provide most health care to most people.) From
March 2007 to August 2008, each microsystem utilized organizational and community assessment tools, process flow diagrams, and rapid cycle improvement methods (plan-do-study-act cycles) to identify, test, and implement changes to improve chronic care, preventive processes and patient experience, as well as to identify and mitigate waste in their systems. Data reflecting these parameters were tracked using web-based monthly reporting tools and analyzed using weighted averages in
Microsoft Excel. Results: Thirty improvement concepts and related change ideas were tested by the 14 microsystem participants in the IPC
Collaborative. Intake screening rates (a composite measure including screening for alcohol use, depression, body mass index, blood pressure, domestic violence and tobacco use) showed a mean improvement from 41.2% to 55.1% across all microsystems with 93% (n=13) improving and 43%
(n=6) showing more than 50% improvement.
Aggregate cancer screening rates (a composite measure including screening for colorectal, cervical and breast cancer) showed a mean improvement from 50.5% to 53.9% across all microsystems with
86% (n=12) showing improvement and 21% (n=3) showing more than 25% improvement. Diabetes comprehensive measures (a composite measure including key processes of care) showed a mean improvement from 17.2% to 31.1% across all microsystems with 79% of sites (n=11) showing improvement and 65% (n=9) showing more than
25% improvement. Improvements in the cost of care, patient experience and waste reduction have also been seen.
Conclusion: Indian Health teams working in a collaborative setting have demonstrated marked improvement in clinical prevention, management of chronic conditions, patient experience of care, and waste reduction using change concepts derived from the Chronic Care Model. Further follow-up is planned to identify the optimal set and sequence of changes to ensure sustainability and spread of these improvements.
Funding Source(s): Health Service
Theme: Prevention and Treatment of Chronic
Illness
● Improved Colorectal Cancer Screening in the
Indian Health System
Ty Reidhead
Presented by: Ty Reidhead, National Chief Clinical
Consultant in Internal Medicine, P.O. Box 860,
Whiteriver, AZ 85941, Phone: (928) 338-4911;
Email: charles.reidhead@ihs.gov
Study Design: Colorectal cancer (CRC) kills an estimated 50,000 Americans annually. The US
Preventive Services Task Force (USPSTF) has estimated that attainment of population-level goals for CRC screening could save 18,800 lives annually. Yet, despite the widespread availability of these screening techniques, CRC screening remains woefully inadequate with only 52% of
Americans screened annually. In the Indian Health
System (federal, tribal, and urban Indian health programs) the screening rate for CRC is 29%.
Methods: In 2007, the IHS partnered with the
Institute for Healthcare Improvement (IHI) to launch the Innovations in Planned Care (IPC)
Collaborative. Fourteen pilot Indian health facilities including federal, tribal, and urban sites from diverse geographical areas were enrolled in this
Breakthrough Series Collaborative (a short-term learning system that brings together teams from hospitals or clinics to improve a focused topic area) from March 2007 to August 2008. Within each facility, a group of providers and their patients were identified as a clinical “microsystem.” (Clinical microsystems are defined as the front-line units that provide most health care.) The microsystem teams used rapid cycle improvement methods (plan-dostudy-act cycles) to identify, test and implement strategies to improve CRC screening. Individual microsystem screening rates and aggregate collaborative-wide screening rates were tracked using web-based reporting tools.
Population Studied: American Indian/Alaska
Native population
Principal Findings: CRC screening rates across all outpatient visits to the 14 participating clinical microsystems increased from a weighted average of 31.6% (min 7.7%, max 83.6%) to 52.4% (min
28.6%, max 89.8%) over the 18-month period.
Initially, only two sites had rates of CRC screening above 50%. By August 2008, six sites had screening rates above 50%. Specific changes that led to these improvements in CRC screening rates included empanelment of patients, development of care teams, optimization of the roles of and communication between with members of the care team, standardization of protocols and orders for testing, referral, and follow-up, and utilization of the clinical information system for planning, feedback, and identification of opportunities for improvement.
Conclusion: The use of improvement methods to implement specific changes listed above have successfully improved CRC screening rates among selected microsystems within Indian Health sites.
Further follow-up will be needed to ascertain the durability and spread potential of these improvements.
Funding Source(s): Indian Health Service
Theme: Prevention and Treatment of Chronic
Illness
● Clinician Patterns in Treatment of Mental &
Substance Use Disorders in the Private Sector
Sharon Reif, Ph.D.; Constance Horgan, Sc.D.;
Maria Torres, M.A.; Elizabeth Merrick, Ph.D.,
M.S.W.
Presented by: Sharon Reif, Ph.D., Research
Scientist, Heller School, Institute for Behavioral
Health, Brandeis University, 415 South Street,
MS035, Waltham, MA 02454, Phone: (781) 736-
3924; Email: reif@brandeis.edu
Research Objective: A substantial proportion of mental health (MH) and substance abuse (SA) treatment occurs within the private sector, with private insurance accounting for about 22% of combined MH and SA expenditures in 2003. MH/SA treatment in the private sector is often carved out from health insurance to specialty managed behavioral healthcare organizations (MBHOs).
MBHOs maintain a network of approved or credentialed treatment providers, but little is known about the clinicians actually delivering MH/SA care in the private sector. Understanding clinician characteristics and treatment patterns will address a common belief that managed care relies on less experienced providers, and will also provide an essential baseline understanding as pay-forperformance initiatives enter behavioral healthcare.
This paper examines the degrees, experience, and demographics of clinicians in one large MBHO network, and considers the patterns of providers seen within outpatient episodes of care.
Study Design: Secondary data analysis of outpatient non-EAP claims, enrollment, and clinician credentialing data for 2004, from a national
MBHO. Univariate analyses describe the clinicians; bivariate analyses consider how clinician characteristics relate to client demographics,
MH/SA status, and outpatient episodes of care.
Population Studied: Clinicians enrolled in a private
MBHO's provider network, and MBHO clients served by those clinicians.
Principal Findings: The MBHO network includes physicians (18% of clinicians), psychologists (31%), social workers (32%), counselors (19%), and nurses (2%).The years of experience range widely
(median=13, range 0-51). One-third of clinicians report specialized training in alcohol or drug disorders; half in major psychiatric disorders; 40% in children and 59% in adolescents. On average, physicians see a higher number of patients than other types of providers, but fewer visits per patient.
Overall, 69% of clients see only one provider in an outpatient treatment episode. Of those with multiple providers, the vast majority see a physician plus another type of provider. Clients who have SA only or both MH and SA diagnoses are more likely to see multiple clinicians within an outpatient treatment episode, and more likely to see a clinician with specialized alcohol or drug training. Clients with both MH and SA diagnoses are more likely to see a physician than clients with only one diagnosis, and more likely to see a clinician with specialized training in major psychiatric disorders.
Conclusion: The clinician network of this MBHO has a significant representation of each provider type, with some differences across clinicians by number and types of clients, and distinctly different patterns for clients with both MH and SA. The finding that most episodes with multiple clinicians include physicians suggests that they are playing a medication management role, which further analyses will examine.
Implications for Policy, Delivery or Practice:
These findings allow a better understanding of how
MH and SA care are provided within managed behavioral healthcare, and provide enlightenment as to the potential challenges of clinician-level incentives in behavioral healthcare.
Funding Source(s): NIDA
Theme: Behavioral Health
● Predicting Member Compliance with Dilated
Retinal Eye Exams, LDL Screening & HbA1c
Testing
David Reisman, B.S.; Stephen Jones, M.S.; Soyal
Momin, M.S., M.B.A.; Sandy Bunting, R.N., M.B.A.;
Barbara Koslow, M.P.H.
Presented by: David Reisman, B.S., Associate
Accreditation Research Analyst, Medical
Informatics, Blue Cross Blue Shield of Tennessee,
801 Pine Street, Building 2G, Chattanooga, TN
37402, Phone: (423) 535-4238; Email: david_reisman@bcbst.com
Research Objective: To determine drivers of compliance for dilated retinal eye exam (DRE),
LDL-C screening, and hemoglobin A1c (HbA1c) measures for both commercial and Medicaid populations in an effort to improve Healthcare
Effectiveness Data and Information Set (HEDIS®) rates.
Study Design: Two years (Jan. 2006 – Dec. 2007) of administrative medical claims data including census information, and plan structure data were extracted for study. Criteria for member compliance with study measures were based upon HEDIS guidelines. Study population consisted of members that were non-compliant for at least three consecutive months (Assessment Period: June –
Aug. 2007). Study members were then observed for
4 months (Observation Period: Sept. – Dec 2007) to determine whether a compliant event occurred.
Member compliance was modeled against benefit structure, demographic factors, socio-economic factors, risk scores (Acute and Chronic impact scores) and compliance with additional evidence based guidelines using stepwise logistic regressions for each variable and line of business
(LOB). We evaluated model fit using the Hosmer and Lemeshow test for each of the six models.
Population Studied: Continuously enrolled
(January 2006 – December 2007) diabetic members aged 18-75 from a southeastern
Managed Care Organization (MCO) with at least 3 continuous months of non-compliance for one of the study measures (commercial N = 15,100; Medicaid
N = 2,751).
Principal Findings: Compliance with any one diabetes measure was associated with increased compliance for the other two measures. Stepwise logistic regression revealed Commercial DRE compliance to be affected by: age (OR: 1.02), diabetes type (OR: 0.7 for Type II vs. I), income
(OR: 1.07), primary care physician (PCP) office visits (OR: 1.03), and an Endocrinologist office visit within the 12 month period prior to a compliant event (OR: 1.87). Also, commercial members with vision benefits were 2.2 times more likely to be compliant. Medicaid DRE compliance was affected by: age (OR: 1.01), diabetes type (OR: 0.7 for Type
II vs. I), and an Endocrinologist office visit within the
12 month period (OR: 2.09). Commercial LDL rates were influenced by age (OR: 1.01) and an
Endocrinologist office visit within the 12 month period (OR: 1.51). Medicaid LDL rate analysis failed to reveal any factors affecting compliance outside of the chronic impact score (OR: 1.07) and acute impact score (OR: 0.99). Commercial HbA1c rate analysis was inconclusive in that days since PCP visits was found to be significant, but OR = 1.
Medicaid HbA1c rates were affected by the number of PCP visits (OR: 1.07).
Conclusion: Results consistently suggest that provider interactions and member age are correlated with increased compliance. Previous internal studies have shown that the additional costs associated with increased utilization in the short term are offset by future savings. Member education regarding covered services is needed as retinal eye exams are a medical procedure and covered under medical benefits.
Implications for Policy, Delivery or Practice:
Information gained from this study allows for efficient targeting of non-compliant members in either Line of Business (LOB). Findings found to be significant in both LOB can be incorporated in planwide outreach efforts. Factors unique to each population (e.g. benefit design) should be considered when developing HEDIS® improvement activities.
Funding Source(s): Blue Cross Blue Shield of
Tennessee
Theme: Prevention and Treatment of Chronic
Illness
● Challenges & Strategies for Health Finance of
Universal Coverage in China
Ran Ren, M.A.
Presented by: Ran Ren, M.A., Professor, Heath
Economics, Dalian Medical University, Fu Guo
Street 81# 3-4-2, Dalian, 116023, CN, Phone: +86
13604267830; Email: renran99@163.com
Research Objective: The objective of this study is to analysis the challenges for health finance on reaching UC in China, and find feasible solution, and provide valuable suggestions for the reform and system designing so as to making right choice in right time.
Study Design: There are facing a lot of significant reform for China at present but Universal
Coverage(UC) is most priority in China. It is not only a key for set up and modify the health security system but also push reform of health care system.
It will be significant for improving equity in health finance and access of essential health care as well as system performance, especially ultimately promoting the goal of improving health of the nation and reaching the goal of harmonious society.
However, there are challenges and obstacle, such as how to finance by tax versus social health insurance, and how to cover rapid population moving toward universal, how to reduce the gap on health finance level between urban and rural and so on.Based on definition of UC by WHO, “everyone has access to appropriate care when they need it and at affordable cost”, literature review, field survey, and questionnaire and the focus group adopt, interviewed policy makers and scholars, and collecting and exploring the point of view on UC in
China.
Population Studied: 120 of key policy makers and scholars interviewed, and collected and explored the point of view on UC in China.
Principal Findings: It showed the opinion on reaching UC by quesionnaire, 51% of respondent approve of tax-based, and 31% think contribution do well. Main obstacle of health finance to UC include: government budget limited lack of coordinated and efficient administrated system, and inconsistent of delivery system and no effective regulation for provider behavior. We advanced the three stages and objective on health finance for reaching UC of China: access of basic health care for all before 2010, and reduce the gap between regions and people in health finance before 2020 and improving equity in health finance after 2020. It is more feasible choice for consist of a mix of social health finance system, including social health insurance(SHI), and tax finance system, and community health finance and other system.
Political commitment and government role is key factors in reform and priority financed for the poor.
The strategies including extending the source of health finance and adjusting the allocation structures, and effective health care purchasing and payment for the sustainability of health care financing. Also, it emphasis on prevention and health promotion by enhancing the community health care and regulating, regulated for providers.
Conclusion: It concluded SHI not right choice for now but it is feasible in future, and a mixed system of health finance more feasible for universal coverage to Chinese. There are needed identify the different objective and development stage of UC especially urgent to explore “China Models” of UC and based on learning international experiences.
Implications for Policy, Delivery or Practice:
System Designing and policy suggestions on health finance for reaching the goal Universal Coverage for the policy makers for China government, and provide case for other countries.
Theme: Coverage and Access
● Comorbidity Measures for Risk-Adjusting
Outpatient Rehabilitation Models
Linda Resnik, P.T., Ph.D.; Pedro Gozalo, Ph.D.
Presented by: Linda Resnik, P.T., Ph.D., Assistant
Professor, Community Health, Brown University/
Providence VA Medical Center, 121 South Main
Street, Providence, RI 02912, Phone: (401) 863-
9214; Email: Linda_Resnik@Brown.edu
Research Objective: The Functional Comorbidity
Index (FCI) offers the potential to improve risk adjustment for rehabilitation. Although validity of the
FCI has been demonstrated for use with patients with back impairments, further testing is needed before it can be recommended for patients with other types of conditions. The purposes of this study were to: 1) Examine the association between the FCI and functional status (FS) at discharge among patients seeking outpatient rehabilitation 2)
Examine the impact of FCI on discharge FS when added to comprehensive predictive models of FS, and 3) Compare the current (additive) FCI to a proposed Weighted FCI.
Study Design: This was a prospective, longitudinal, cohort study. Discharge FS was collected using computer adaptive tests (CATs) for nine patient categories: cervical, lumbar, shoulder, elbow, wrist/hand, hip, knee, foot/ankle, neurological. Linear regression (controlling for censoring) was used to examine the association
between FCI and discharge FS for each category.
We compared three methods of including functional comorbidities (FC): the existing additively scored
FCI (FCI); a (sub-group specific) FCI weighted index (weighted FCI); and the ungrouped list of 18 condition variables. Comprehensive models of discharge FS compared these three FC measures while controlling for intake functional status, age, gender, payer type, exercise history, number of surgeries, and onset of condition
Population Studied: Study population was drawn from the Focus On Therapeutic Outcomes, Inc.
(FOTO) database. Individuals with a rehabilitation intake date between January 1, 2006 and
December 31, 2007 were included.
Principal Findings: The relationship between FCI and FS varied by patient group. The R2 of models ranged from .02 (elbow group) to .09 (hip group).
The weighted index or separate condition variables models produced almost identical R2 measures, ranging from.03 (shoulder) to.14 (neurological).
Weighted FCI or list of separate condition variables
(compared to additively scored FCI ) increased the
R2 of crude models by < .01 for cervical, shoulder, and lumbar groups; .01 wrist/hand, knee and foot/ankle; by.02 hip; 03 elbow; and .08 neurological group. In comprehensive models, addition of the FCI added <.01 to the R2 of all groups; whereas addition of the weighted FCI increased the R2 by <.01 for cervical, lumbar and shoulder,.01 wrist/hand, hip, knee, foot and ankle groups, .02 elbow , and .04 neurological group; and addition of the 18 condition variables increased the
R2 of comprehensive models by < .01 cervical, shoulder and lumbar groups, .01 knee and foot/ankle, .02 elbow, wrist/hand and hip and .05 neurological group.
Conclusion: Functional comorbidity measures can be useful in risk-adjustment models of rehabilitation, especially when relevant adjustors such as intake
FS are not available. Overall, using the full list of comorbidities or our proposed weighted FCI is preferable to using the existing additive FCI.
Implications for Policy, Delivery or Practice:
Many believe that it is time to develop alternative payment systems that reward effectiveness and efficiency of services. However before we can do this we need to develop accurate risk-adjustment methods that can help make valid patient comparisons.
Funding Source(s): NICHD
Theme: Quality and Efficiency: Measurement
● Psychotropic Medication Use among Youth with Acute Psychiatric Crisis
YongJoo Rhee, Dr.P.H.; Neil Jordan, Ph.D.
Presented by: YongJoo Rhee, Dr.P.H., Research
Assistant Professor, Psychiatric & Behavioral
Sciences, Northwestern University, 710 N. Lake
Shore Drive, Chicago, IL 60611, Phone: (312) 503-
1130; Email: y-rhee@northwestern.edu
Research Objective: Acute psychiatric services remain one of the most important components of a behavioral health system of care for children and youth. Traditionally, the behavioral health system had two services: restrictive, and high cost inpatient services, and less restrictive, less costly outpatient services. Psychotropic medications are also a common treatment modality for youth with acute psychiatric needs. The Illinois’ Screening,
Assessment, and Supportive Service (SASS) program is a state-funded program designed to more appropriately match youth to mental health crisis services. SASS clinicians and social workers are required to assess the severity of the youth’s mental crisis and make an appropriate treatment referral. Youth typically receive SASS services for
90 days. This study examines psychotropic medication use during a SASS treatment episode by children and adolescents with acute psychiatric crisis symptoms.
Study Design: This retrospective cohort study uses
Medicaid administrative data to assess mental health service use for youth served by the SASS program during 2006. We used the pharmacy claims to examine psychotropic medication use among youth who received any prescribed psychotropic medication.
Population Studied: The study population includes children and youth ages 3 to 21 with an acute psychiatric crisis who received inpatient or outpatient stabilization services as part of the SASS program. Using Medicaid administrative data
(n=30,154), we identified 3,998 children and adolescents those received psychiatric stabilization services and at least any psychotropic medications during the study period.
Principal Findings: Among youth in SASS, 48.3% were female, 45.1% were white, and 32.7% were black. The average age was 13 years. The most frequent primary diagnoses using the International
Classification of Disease, Ninth Revision (ICD 9 code) among youth were: attention deficit disorder
(8.0%), bipolar disorder (6.1%), posttraumatic stress disorder (4.2%), and oppositional defiant disorder (4.1%). We regrouped them into broader four diagnostic categories: mood disorder, disruptive behavior disorder, psychotic disorder, and developmental disorder. Youth treated for acute psychiatric symptoms in SASS were diagnosed for mood disorder (73.3%), disruptive behavior disorder (45.5%), psychotic disorder
(21.9%) and developmental disorder (13.2%). One third of youth had both behavior and mood disorder
(28.2%). About 66.3% of youth were diagnosed either for behavior or mood disorders.
Among those with any psychotropic medication claims, youth had an average of 7.5 pharmacy claims for medication use during SASS episode.
The most frequently prescribed classes were atypical antipsychotics (23.5%) and selective serotonin reuptake inhibitors (SSRIs)(15.3%).
SSRIs (34.8%) were most frequently prescribed for those diagnosed solely with a mood disorder or for those diagnosed with a behavior, mood, and psychotic disorder. Half of youth (49.0%) had more than two different classes of psychotropic medications.
Conclusion: Children and adolescents with acute psychiatric crisis symptoms that received psychiatric stabilization services through the SASS program commonly received polypharmacy as part of their treatment regimen.
Implications for Policy, Delivery or Practice:
Given ongoing concerns about the safety and efficacy of psychotropic medications for youth, these findings will help policymakers better understand the magnitude of psychotropic medication use among youth with a psychiatric crisis.
● Importing Medicine: A Look at Citizenship &
Immigration Status for Graduating Residents in
New York State from 1998-2007
Michael Richards, M.D.-M.P.H.; Chiu-Fang Chou,
Dr.P.H.; Anthony LoSasso, Ph.D.
Presented by: Michael Richards, M.D.-M.P.H.,
Student, Health Policy & Administration, University of Illinois-Chicago College of Medicine & School of
Public Health, 300 West Hill Street, Apartment 926,
Chicago, IL 60610, Phone: (618) 910-8790; Email: mricha1@uic.edu
Research Objective: Our work aims to illustrate the continued presence of International Medical
Graduates (IMGs) in the ongoing discussions regarding the US health care labor market, and ultimately, the impact of physician supply on the health of the population. Additionally, we hope to offer a more careful examination of IMG trends by including citizenship and immigration status in our analysis. This unique source of data allows us to focus on graduating residents in New York State, historically one of the most popular final destinations for IMGs, to compile a detailed description of the heterogeneity of IMGs over the previous ten years. Our more nuanced assessment of IMGs is likely more instructive for US policy makers and also offers a new angle on the global discourse regarding poor country physicians migrating to rich country practices.
Study Design: We use data from the New York
State Survey of Residents Completing Training conducted by the Center for Health Workforce
Studies of the State University of New York at
Albany. The survey was originally conducted annually each May and June beginning in 1998.
The survey was not fielded in 2004 but was restarted in 2005 on a biennial basis. The aggregate response rate for 1998-2007 of surveys is 64% (23,679/37,126).
Population Studied: Graduating resident physicians in the state of New York for the past decade.
Principal Findings: Despite the regular commentary of a continually increasing IMG work force in the US, our data demonstrate some starkly different trends. For example, there is a pronounced shift toward US citizens claiming residency slots, from 56% in 1998 to 73% in 2007.
The growing numbers of USIMGs (US citizens from foreign medical schools) accounts for approximately
40% of this climb. Also, J visa holders fell from
20% to 6%, and Permanent Residents dropped from 18% to 11% over the same time period. The
H visa holders (the smallest group) modestly increased. Moreover, the J visa holders, with the greater likelihood of Health Professional Shortage
Area (HPSA) service obligations, are not only dissipating in numbers, but are also seeking more specialized training instead of “generalist” patient care.
Conclusion: We believe a variety of factors may be placing downward pressure on IMG immigration and elements of substitution may be redrawing the landscape of physician immigration to the US.
More specifically, the rise in IMG levels documented by others may be an artifact of the increasing popularity of international medical schools for US citizens.
Implications for Policy, Delivery or Practice: If our results are representative of broad US trends, then it will significantly alter US policy makers’ strategies for dealing with physician shortages in underserved areas. Additionally, the infamous
“brain drain” takes on a new face in this light.
Although it is still occurring, the magnitude of the pull may be diminishing over time. Instead of the rich world “poaching” physicians from the developing world, perhaps the rich world is merely herding its own flock.
Theme: Health Care Workforce
● Risk Factors for Unintentional Injuries Among
Adolescents in California
Erin Richardson, M.S.; David Hemenway, Ph.D.;
Ninez Ponce, Ph.D.
Presented by: Erin Richardson, M.S., Student /
Teaching Fellow, Health Services, University of
California, Los Angeles, 1527 Corinth Avenue, # 1,
Los Angeles, CA 90025, Phone: (781) 704-4223;
Email: erin.richardson@gmail.com
Research Objective: This analysis is tries to discern whether there is a relationship between unintentional injuries among adolescents and factors from three areas: individual (e.g., sex), family (e.g., parental knowledge of adolescent’s free time activities), and society (e.g., health services variables).
Study Design: This is a retrospective, crosssectional analysis of the 2001 California Health
Interview Survey (CHIS). Weighted bivariate and multivariate survey logit regression models were used to assess the effect of demographic variables and risk variables on the injury status of adolescents. Uninjured adolescents (n=5,409) were compared with adolescents who were unintentionally injured (n=418).
Population Studied: Adolescents aged 12 through
17 years of age, who lived in California at some point between November, 2000 and October, 2001.
Principal Findings: Bivariate associations show significant relationships between being unintentionally injured and being male, having ever ridden in a car with a drunk driver, ever having held a gun, ever having had an alcoholic drink, living in an urban environment, and between races/ethnicities. The weighted multivariate logistic regression found that having ever held a gun
(p<0.0001), living in an urban area (p=0.02), being
Latino (p<0.01), and being Asian (p<0.0001) remained significant while holding all other variables constant. The odds of being unintentionally injured versus uninjured for those adolescents who had held a gun are 1.67 times higher than for those who had not held a gun
(OR=1.67, p<0.0001). Living in an urban area reduced the odds of being unintentionally injured
(OR=0.70, p=0.02). The odds of being unintentionally injured for Latino adolescents was
0.57 times the odds of Caucasian adolescents
(OR=0.57, p<0.01) and Asian adolescents had 0.27 times the odds when compared with Caucasian adolescents (OR=0.27, p<0.0001). After stratifying the data for urban and rural adolescents, the odds of being unintentionally injured remained significant for adolescents who had held a gun for both urban
(OR=1.62, p=0.001) and rural (OR=1.95, p=0.02) adolescents.
Conclusion: These preliminary analyses indicate that adolescents are at greater risk of an unintentional injury if they participated in certain risky behaviors, particularly having ever held a gun.
Implications for Policy, Delivery or Practice:
Targeting high risk adolescents and intervening will help to address unintentional injuries, some of which are chalked up to being simply accidents or a natural part of childhood. Clearly, those who have held a gun are at particular risk for being unintentionally injured and there is room for intervention among these high risk teens, which will not only address the issue of intentional injuries but will also help to reduce unintentional injuries. An important implication for practice is screening adolescents for gun exposure. It is important for physicians to screen adolescents for having access to guns not only as a marker for susceptibility to intentional injury but as a marker for susceptibility to all injuries, both intentional and unintentional.
Physicians can use a screener to discern which adolescents are most at risk and can then target these teens to provide anticipatory guidance and other educational measures.
● Variation in Home Health Care Heart Failure
Patient Service Use Across the Rural-Urban
Continuum
Jennifer Riggs, R.N., M.S.N.; Tanya Scharpf, Ph.D.;
Elizabeth Madigan, Ph.D., R.N., F.A.A.N.
Presented by: Jennifer Riggs, R.N., M.S.N., Ph.D.
Student, Frances Payne Bolton School of Nursing,
Case Western Reserve University, 10600 Euclid
Avenue, Cleveland, OH 44106-4904, Phone: (216)
368-0593; Email: jennifer.riggs@case.edu
Research Objective: Heart failure (HF) affects more than 5 million Americans at an estimated total cost of $33.2 billion. Incidence increases with age, thus HF is a high-cost high-volume condition in home health care, where the majority of patients are elders. Little is known about the service use patterns of heart failure patients in the home health care setting. The rural-urban variability in service use for home health care patients with HF has not been described. The purpose of the present study is to describe variation in service use across the rural-urban continuum for home health agency patients with a primary diagnosis of heart failure.
Study Design: The Outcome and Assessment
Information Set (OASIS) data set was merged with the Standard Analytic File for home health and the
Area Resource File to provide patient level clinical data, home health care service utilization data and rural-urban continuum codes. Home health care visit intensity was examined across four categories on the rural-urban continuum – urban counties, suburban counties, counties with towns and rural counties. The percentage of patients receiving home health care visits from each discipline is described in terms of the same four categories, highlighting differences in provision of services based on the degree of rurality of the county.
Population Studied: The 2005 national population of Medicare-certified home health agency patients with a primary diagnosis of heart failure.
Principal Findings: The population of home health care patients with heart failure (N=106,918) is mostly Caucasian and female (83% and 62%, respectively), with an average age of 81. Median visit intensity for the population ranges from a low of
3.04 in rural counties, to 3.18 in both urban counties and counties with towns, to a high of 3.20 in suburban counties. Almost all home health care heart failure patients (98%) receive nursing visits;
50% receive physical therapy visits, 22% receive home health aide visits and 17% and 14% receive occupational therapy and social work visits, respectively. While there is little or no variance in nursing or nurse aide visits by type of county, differences exist for physical and occupational therapy, and social work whereby 62% of patients residing in urban counties receive physical therapy visits, compared to 48% in counties with towns,
51% in suburban and 52% in rural counties; 24% of
patients in urban counties receive occupational therapy visits, compared to 19% in suburban and rural counties, and 17% in counties with towns;
20% of patients in urban counties receive social work visits, compared to only 11% in rural, 14% in counties with towns, and 16% in suburban counties.
Conclusion: A visit intensity measure that includes service provided by all disciplines masks differences in provision of services by specific home health care disciplines.
Implications for Policy, Delivery or Practice:
Results from the present study indicate that patients outside of urban areas received substantially fewer of these specialized services. Policies may need to be developed to address an imbalance in availability of physical and occupational therapy, and social work for patients who reside in rural or suburban counties or counties with towns, compared to urban counties.
Funding Source(s): NHLBI
● Access to Care Among Safety Net Patients with Comorbid Chronic Physical & Mental
Illness
Deborah Rinehart, M.A.; Josh Durfee, M.S.P.H.;
Susan Moore, M.S.P.H.; Sheri Eisert, Ph.D.
Presented by: Deborah Rinehart, M.A., Research
Project Coordinator, Health Services Research,
Denver Health & Hospital Authority, 777 Bannock
Street, Denver, CO 80204, Phone: (303) 436-3828;
Email: deborah.rinehart@dhha.org
Research Objective: Published studies have described high rates of mental illness among individuals who have chronic physical illness. To date, there have been few studies that have analyzed the differences in access to care for those with chronic physical conditions, with and without mental health illness. The safety net health care system, which provides care to vulnerable populations, offers an opportunity to study access to care for this comorbid population. This study explores differences in access and medical service utilization between patients with comorbid chronic physical and mental illness as compared to those with chronic physical conditions only.
Study Design: Denver Health (DH), an integrated safety net health care delivery system, also has an integrated information system that allows for extraction and analyses of health care utilization data throughout the continuum of care. A dataset of de-identified electronic medical records was created for all patients 18 and older who received medical services at DH between 2006 and 2007. ICD-9 codes for 2006 visits were used to identify two groups of patients: 1) patients with comorbid chronic physical and mental illness and 2) patients with chronic physical illness only. Chronic illness was defined as having had an ICD-9 code for diabetes, hypertension or cardiovascular disease and mental illness was defined as having had an
ICD-9 code for any mental disorder, excluding organic conditions, childhood disorders and mental retardation. Data from 2007 were used to explore service utilization patterns across the two groups.
Population Studied: The study sample consisted of 105,057 adult patients who had at least one visit to DH in 2006 and who were diagnosed with one of any of the three following chronic conditions: hypertension, diabetes, cardiovascular disease.
Principal Findings: Preliminary results indicate statistically significantly demographic differences.
Those who had comorbid conditions were younger
(53 vs. 55) and women (28%) had significantly higher proportions of comorbid physical and mental illness as compared to men (23%). White patients
(35%) had significantly higher proportions of comorbidity as compared to Hispanic (21%), African
American (24%), Asian (14%) and patients from other race/ethnic groups (19%). Preliminary results also suggest an association between comorbidity and access to care where patients with comorbid illnesses utilized less primary care (35%) and more specialty services (54%) as compared to those with chronic illness only (42% and 48%, respectively).
Analyses will also be presented for each of the three chronic conditions as well as regression models investigating service utilization that will control for demographics, insurance status and illness severity.
Conclusion: Individuals with comorbid physical and mental illness appear to have unique needs that influence service access and utilization.
Implications for Policy, Delivery or Practice: The prevalence of comorbid chronic physical and mental illness is high in an urban safety net population.
The fragmentation of systems for mental illness and medical care create a gap in access to care for this vulnerable population. Better understanding the factors related to access and utilization of services is helpful in targeting and providing effective care to this population.
● Public Deliberation & Preferences for Health
System Reform
Jill Rissi, Ph.D., R.N.
Presented by: Jill Rissi, Ph.D., R.N., Associate
Director, Research & Policy, Saint Luke's Health
Initiatives, 2929 South Central Avenue, Suite 1550,
Phoenix, AZ 85012, Phone: (602) 385-6503;
Email: jrissi@slhi.org
Research Objective: Health system reform continues to be at the forefront of policy issues at the federal and state levels. However, despite repeated attempts to enact significant reforms, in the face of ideological differences and entrenched stakeholder interests, little real progress has been made. Deliberative forums have offered the opportunity to identify and reconcile diverse perspectives on reform and reach consensus on a plan of action, even in the absence of agreement on
underlying personal values. This research explores how reform option preferences are re-structured through deliberation, considers the relative salience of factual data in that process, and identifies solutions the public, business and civic leaders will support when given the opportunity to consider value-based choices and trade-offs.
Study Design: Mixed-method study of individual and group preferences for health system reform.
Three, day-long facilitated public dialogues were conducted utilizing scenarios that incorporated value-based choices on how insurance would be provided and at what level of benefit. Pre- and post-dialogue questionnaires are complemented by narrative and rhetorical analysis of audio and videotapes to explore how and why preferences shifted.
Population Studied: 120 randomly selected, demographically representative adults residing in the Southwest.
Principal Findings: Process, content and participant identity were identified as primary mediating factors for deliberative outcomes.
Citizens’ initial support for a private system of comprehensive coverage shifted to preference for a system of quasi-governmental universal coverage through a basic insurance plan, administered by a single public insurance agency and funded through a combination of personal and business income taxes, sin and consumption taxes. Citizens shifted from focusing on their personal situation to considering the most workable and equitable system overall, and from a focus on ‘what’ was covered to ‘how’ insurance would be provided for all. On a 1 – 10 scale, a comprehensive private insurance plan initially rated 6.7, but fell to 4.0 in the post-dialogue ranking, while a comprehensive public plan increased from an initial score of 6.1 to a final rating of 6.7, and a limited-benefit public plan increased from 3.9 to 6.1.
Conclusion: Engaging citizens in dialogue revealed thoughtful consideration of the tradeoffs inherent in health system reform and strong support for moving away from employer-sponsored insurance to a single, quasi-governmental insurance plan. Although the dialogues revealed public perceptions, misperceptions and assumptions which were quite different from those of experts, citizens clearly understood tradeoffs between security, choice, cost and access inherent in health system design, and suggests that support for a publicly funded insurance plan may be stronger and more stable than public opinion polling indicates.
Implications for Policy, Delivery or Practice:
Devolution of health policy and discretionary spending decisions to states requires that citizens be engaged in decision-making to gain both a better understanding of public preferences and greater stability in public judgment. Citizen engagement through facilitated dialogue provides a basis for anticipating how the broader public will resolve issues when given the opportunity to come to grips with them, and insight on how deliberation enables persons with diverse perspectives to reach broader consensus on complex policy issues.
Funding Source(s): St. Luke's Health Initiatives
Theme: Consumer Choices in Health Care
● Optimizing Design of a Health Informatics
Intervention for Improving Safe NSAID Use:
Patient & Physician Perspectives
Douglas Roblin, Ph.D.; Jeroan Allison, M.D., M.S.;
Richard Shewchuk, Ph.D.; Suzanne Baker, R.N.,
M.P.H.; Renny Varghese, M.P.H.; Catarina Kiefe,
M.D., Ph.D.
Presented by: Douglas Roblin, Ph.D., Senior
Research Scientist, The Center for Health
Research / Southeast, Kaiser Permanente Georgia,
3495 Piedmont Road, Northeast, Building 9,
Atlanta, GA 30305, Phone: (404) 364-4805; Email:
Douglas.Roblin@KP.Org
Research Objective: Non-Steroidal Anti-
Inflammatory Drugs (NSAIDs) are a commonly used and mis-used prescription and over the counter (OTC) medication. Yet, OTC NSAID use is frequently not ascertained at the time of a patient visit or documented in the medical record. We are designing a health informatics intervention intended to facilitate ascertainment and documentation of
OTC NSAID use and to direct physician attention to instances of risky OTC NSAID use. Using established qualitative methods, we collected information on both patient and physician perspectives on intervention usability, benefits, and concerns.
Study Design: We conducted four patient focus groups and two primary care physician (PCP) nominal groups. Patients and PCPs were asked to comment on an intervention design that would: 1) deliver an OTC NSAID survey to patients through
KP.Org (the secure Internet portal in Kaiser
Permanente's electronic medical record [EMR] system for patient-physician messaging), then 2) deliver an e-mail message to the patient's PCP summarizing survey responses. Patients were also asked to comment on OTC NSAID recognition and use. After initial review of the group transcripts, we developed a concept map and classification scheme for structured analysis of the texts. Two trained coders labeled the transcripts using
ATLASti.
Population Studied: 48 Kaiser Permanente
Georgia (KPG) adults 25-70 years old who had a medical condition for which NSAIDs should be used cautiously or had a recent prescription for NSAIDs.
7 KPG adult medicine PCPs.
Principal Findings: 47 patients indicated current or recent use of OTC NSAIDs. Patients' use of OTC
NSAIDs raised concerns about safe use: lack of recognition of some OTC NSAIDs, combining OTC
NSAIDs, or increasing OTC NSAID doses. Both
patients and physicians acknowledged that the intervention would improve documentation of OTC
NSAID use in medical records but also noted challenges of keeping OTC NSAID use up-to-date.
Patients expected timely feedback on potentially risky OTC NSAID use – either at the point of completing the survey or upon review of that information by their PCPs. They were anxious about the ambiguous meaning of a no or slow response to the message delivered to their PCPs
("lost information", "no risk"). PCPs suggested integrating the survey with patients’ on-line appointment scheduling as an alternative to a survey delivered via a KP.Org message. They also endorsed a tighter, automated linkage of survey responses with medical history in the EMR (“best practice alert”) as an alternative to the e-mail message summarizing their patients' OTC NSAID use.
Conclusion: Patients and providers affirmed the value of the intervention (more complete medication history, potential to reduce risky NSAID use) and offered several suggestions to modify the original design and improve its usability.
Implications for Policy, Delivery or Practice:
Design of health informatics interventions to improve patient-provider interactions should build on their relevant experience and perspectives. In the short term, not all expectations may be satisfied by technical capabilities of an EMR; however, unmet expectations may guide future expansion of
EMR capabilities.
Funding Source(s): Kaiser Permanente
HealthConnect Research Initiative
Theme: Health Information Technology
● Psychosocial Stress & Depression among
Employed Middle-Age Adults
Douglas Roblin, Ph.D.; Peter Joski, M.S.P.H.;
Edmund Becker, Ph.D.
Presented by: Douglas Roblin, Ph.D., Senior
Research Scientist, The Center for Health
Research / Southeast, Kaiser Permanente Georgia,
3495 Piedmont Road, Northeast, Building 9,
Atlanta, GA 30305, Phone: (404) 364-4805; Email:
Douglas.Roblin@KP.Org
Research Objective: The Chronic Care Model
(CCM) posits that social resources are an important component in organizing health care interventions for achieving good health outcomes. Social resources can include the health benefits that can be obtained from supportive interpersonal relationships, both instrumental and emotional. Our objective was to estimate the potential for mitigating risk of depression among employed middle-age adults by reducing stress in work or social circumstances.
Study Design: Data were collected on a mixedmode survey (written or Internet) in 2005 (2,224 respondents of 5,309 sampled, 42% response rate) and again in 2007 (1,098 respondents of 1,939 sampled, 58% response rate). The survey included scales scored from 0 (most stressful) to 100 (most supportive) for work climate (decision latitude, job demands, coworker support, supervisor support) and social climate (friend/family support and strain) using items adapted from MIDUS I and the SF-12 items. Depression was defined as MCS12<43. In the 2005 cross-section, we estimated the likelihood of depression in 2005 as a function of work climate and social climate, adjusted for cohort, age, gender, race, marital status, and education. For the
2005/2007 respondents, we estimated likelihood of depression in 2007 as a function of 2005 depression, 2005 work and social climates, and
2005/2007 changes in work and social climates, adjusted for patient covariates.
Population Studied: 2,057 respondents to the
2005 survey with complete information on study variables; and, 979 of these who remained enrolled with Kaiser Permanente Georgia (KPG) and responded to the same survey in 2007.
Respondents were 25-59 year old KPG enrollees employed by large public and private employers in the Atlanta area, randomly sampled from 3 condition cohorts (diabetes, elevated lipids without acute CAD, and "low risk").
Principal Findings: In 2005, 12.3% of respondents met the criterion for depression. 11.2% met the criterion for depression in 2007. 38.3% of the 120 depressed in 2005 were also depressed in 2007.
Work and social climate subscales all had good to excellent reliability (Cronbach's alphas ranging between 0.73 and 0.89 in the 2005 cross-section).
Likelihood of depression in 2005 was significantly
(p<0.05) associated with job demands (adjusted
OR=0.981 per point change in job demands) and with social climate (adjusted OR=0.955) in 2005.
Likelihood of depression in 2007 was significantly associated with depression in 2005 (adjusted
OR=4.545), job demands and social climate in 2005
(adjusted ORs=0.961 and 0.958, respectively), and change from 2005 to 2007 in job demands and social climate (ORs=0.976 and 0.974, respectively).
Conclusion: In this MCO population generally representative of middle-class Atlanta, psychosocial circumstances were associated with depression.
Risk of depression was greater among adults with more demanding jobs and less supportive relationships with family and friends. Over 2-years, improvements in work and social circumstances
(reduced job demands, increased support from family and friends) decreased the risk of depression.
Implications for Policy, Delivery or Practice:
Behavioral health interventions that include strategies to address the mediating effects of adverse work or social climates can achieve incremental improvements in mitigating depression in employed adults, independent of individual factors predisposing adults to depression.
Theme: Behavioral Health
● Understanding Racial Disparities in Physician
Advice & Patient Actions to Control Blood
Pressure in a MCO
Douglas Roblin, Ph.D.; Shifan Dai, Ph.D.; Nora
Keenan, Ph.D.; Robert Merritt, M.A.; Peter Joski,
M.S.P.H.; Edmund Becker, Ph.D.
Presented by: Douglas Roblin, Ph.D., Senior
Research Scientist, The Center for Health
Research / Southeast, Kaiser Permanente Georgia,
3495 Piedmont Road, Northeast, Building 9,
Atlanta, GA 30305, Phone: (404) 364-4805; Email:
Douglas.Roblin@KP.Org
Research Objective: The Chronic Care Model
(CCM) links proactive practice teams and activated patients with better chronic care outcomes. We studied the associations of blood pressure (BP) control with 1) physician advice and patient actions on lifestyle behaviors (e.g. exercise and diet) to control BP and 2) medication adherence in a cohort of adults with hypertension (HTN) in a managed care organization (MCO). Our primary analysis focused on the apparent paradox that, in this MCO,
African Americans reported higher rates of advice and actions to control BP yet had worse BP control.
Study Design: Two independent samples of 3,000
18-74 year old Kaiser Permanente Georgia (KPG) adults with HTN (750 per JNC-VII level) were randomly selected for telephone survey in October
2007 and March 2008. The survey included
BRFSS items on advice and actions to control BP and self-reported race and education. Medication adherence was measured as proportion of days covered (PDC) with any HTN-related medication in the 12-month period preceding the survey. Mean
SBP (systolic BP) and DBP (diastolic BP) were measured from computerized data in the same period. Associations among receipt of advice, taking actions, PDC, and BP control were estimated using multivariate path analysis (controlling for age, gender, and education).
Population Studied: 1,330 survey respondents who were African American (N=751) or white
(N=579).
Principal Findings: Compared to whites, African
Americans had significantly (p<0.05) higher SBP
(standardized beta of 0.137). Better medication adherence was associated with lower SBP (-0.229); however, African Americans had lower medication adherence (-0.074). Patients taking action to control BP were more likely to have received physician advice to take action (0.232); and, African
Americans were more likely to receive advice
(0.187) and to take action (0.056). Poor SBP was associated with greater likelihood of taking action
(0.075). A similar pattern of associations was observed for DBP.
Conclusion: African Americans in this MCO were less adherent with antihypertensive medications than whites; and, this contributed to worse BP control. Worse BP control, however, increased the likelihood that physicians advised African
Americans to take actions to control BP. Consistent with the CCM, physicians appear to be directing advice on actions to control BP to the subset of patients most likely to benefit from lifestyle change.
Implications for Policy, Delivery or Practice: The disconnect between physician advice and patient actions for achieving BP control, yet lower BP control and medication adherence, among African
Americans raises questions about effectiveness of patient-provider interactions in some patient subgroups. More effective means of supporting the African American patient with hypertension need to be evaluated and implemented.
Funding Source(s): CDC
Theme: Disparities
● Estimating the Supply of Healthcare
Interpreters in Los Angeles County
Dylan Roby, Ph.D.
Presented by: Dylan Roby, Ph.D., Assistant
Professor, Health Services / Center for Health
Policy Research, University of California, Los
Angeles School of Public Health, 10960 Wilshire
Boulevard, Suite 1550, Los Angeles, CA 90024,
Phone: (310) 794-3953; Email: droby@ucla.edu
Research Objective: To determine the demand and supply for health care interpreters in Los
Angeles County, California. Explore the mismatch between languages provided and those spoken, and potential need for additional health care interpreters in the county.
Study Design: To estimate the actual need for health care interpreters by language type, we used data from the 2000 U.S. Census and a report from the Asian Pacific American Legal Center (APALC) entitled "L.A. Speaks." To approximate the growth in the limited English proficiency by language spoken, we used 2001 to 2005 California Health
Interview Survey data. This data allowed us to estimate the actual current level of demand for health care interpretation for Los Angeles County residents with limited English profiency.
A provider survey was also conducted to collect data on the scope of health care interpretation services by provider type, area of the county, and language. A total of 205 representative providers were in the sample, and 42 responded to the survey. A response rate of 20.5%.
Population Studied: Residents of Los Angeles
County over the age of 5 who spoke a language other than English at home.
Principal Findings: There appears to be a shortfall of health care interpreters when compared to the overall limited English proficient (LEP) population in
Los Angeles county. Smaller providers, especially medical groups and private physician offices, report using untrained bilingual staff and family members as interpreters, rather than using trained
interpreters, telephonic language lines, or outsourced health care interpreting agencies when dealing with LEP patients. In Los Angeles, we estimate over 2.8 million (almost one-third) of the population are LEP. Although bilingual Spanishspeaking staff may provide acceptable interpretation services to the almost 2 million spanish-speakers who are LEP, there are an additional 800,000 LEP residents who speak asian languages, armenian, farsi, and others. The supply of trained health care interpreters appears to be concentrated among hospitals and primary care clinics, but not available in most specialty clinics and medical groups/private physician offices. Most providers without in-house interpreters reported using telephonic language lines or bilingual staff to communicate. A minority reported using outsourced health care interpreter agencies that charge an average of $75 per visit to communicate in certain languages.In certain areas of the county, fewer providers reported patients needing interpretation services. The Downtown/Metro (50%), South Bay
(33.8%), San Gabriel (37.7%), and San Fernando
(37.5%) regions of Los Angeles reported higher levels of interpreter need. However, the West, East, and Antelope Valley areas reported less than 5% interpreter need.
Conclusion: Although Los Angeles County is a very diverse area, patients still have trouble receiving services in their native language in certain areas. Providers report barriers to communicating with patients and in hiring, retaining, and paying for health care interpreter services.
Implications for Policy, Delivery or Practice:
More effort needs to be made in training health care interpreters to meet current demands in diverse communities with high numbers of LEP individuals.
Providers should be provided with tools to meet the needs of their LEP patients, including reimbursement for provision of health care interpretation services. Attention needs to be paid to the geographic areas of need in the county, in order to focus on barriers to communication found in specific areas of need.
Funding Source(s): Asian Pacific American Legal
Center
Theme: Health Care Workforce
● Loss of Medicaid Eligibility Increases
Likelihood of Emergency Room Visits in
Chronically Ill Beneficiaries
Dylan Roby, Ph.D.; Gerald Kominski, Ph.D.;
Nadereh Pourat, Ph.D.; Ying-Ying Meng, Dr.P.H.;
Allison Diamant, M.D., M.S.H.S.; Gina Nicholson,
M.P.H.
Presented by: Dylan Roby, Ph.D., Assistant
Professor, Health Services & Center for Health
Policy Research, University of California, Los
Angeles School of Public Health, 10960 Wilshire
Boulevard, Suite 1550, Los Angeles, CA 90024,
Phone: (310) 794-3953; Email: droby@ucla.edu
Research Objective: To determine if loss of eligibility in Medicaid would result in increased use of the Emergency Room by Medicaid beneficiaries
Study Design: Cross-sectional California Medicaid claims data from September 2006 through August
2007 were analyzed. We examined total ER visits and loss of eligibility over time for Medicaid beneficiaries. We then examined differences in historical eligibility using a logistic regression model predicting having an ER visit or not, controlling for specific chronic condition, whether a comorbid condition was present, severity of disease, and other demographic and utilization covariates.
Population Studied: California Medicaid enrollees ages 22 and older enrolled in fee-for-service care with one or more of the following six chronic illnesses: asthma, atherosclerotic disease syndrome (ADS), congestive heart failure (CHF), diabetes, coronary artery disease (CAD), and chronic obstructive pulmonary disease (COPD).
Fee-for-service beneficiaries in aged, disabled, and blind aid codes who had full scope coverage without share of cost, who did not receive Medicare, and resided in two pilot counties were included.
Principal Findings: Medicaid beneficiaries who experienced a loss in eligibility over the prior three years of Medicaid membership were less likely (OR
= 0.64, p<.001) to experience an Emergency Room visit in the third year when compared to beneficiaries who did not lose eligibility over the last three years. Predictors of ER visits included being disabled (OR = 1.25, p<.01), African-American (OR
= 1.5, p<.001), female (OR = 1.27, p<.001), and having a co-morbid condition (OR = 1.16, p<.01).
Having lower disease severity levels and being
Armenian or Asian-American/Pacific Islander were associated with lower levels of ER visits in comparison to non-armenian whites with higher levels of disease severity.
Conclusion: After controlling for disease severity, race/ethnicity, age, disability, gender, and other covariates the likelihood of needing an ER visit is decreased in Medicaid beneficiaries who have lost
Medicaid for a period of time. Only 15% of the
Medicaid eligibles included in this analysis never lost eligibility, while 83.3% lost eligiibility for at least one month over the past three years. 82.7% of beneficiaries had Medicaid coverage for 35 of 36 months, indicating that the length of time of coverage loss was usually one month or less. This finding indicates that Medicaid beneficiaries who experience low levels of eligibility "churn" may not be at increased risk of ER visits once they return to
Medicaid. However, we do not have information on their health while the beneficiary was not enrolled in
Medicaid. Because of retroactive eligibility and emergency Medicaid coverage, it is likely that ER visits would be captured by Medicaid claims in this population.
Implications for Policy, Delivery or Practice:
Although eligibility churn did not appear to increase
ER visits for this population, it is still an important topic for Medicaid beneficiaries. This population was made up of the blind/disabled/aged, who are more likely to have long period of continuous
Medicaid eligibility and no share of cost. This was shown by the relatively low number (less than 15%) of beneficiaries who went without coverage for more than 1 month in the past 36 months. Attention should be paid to preserving eligibility for Medicaid beneficiaries, rather than letting eligibility lapse due to administrative issues like recertification and changing eligibility criteria. Based on these findings, it appears that losing Medicaid eligibility could potentially harm chronically ill beneficiaries with higher levels of disease severity, co-morbidities, and disabilities, or from african-american and female groups. If state budget cuts due to the current economic crisis result in changes to administrative rules around eligibility and recertification, those groups would be at the highest risk of decreased health status due to lack of
Medicaid coverage and access to care.
Theme: Prevention and Treatment of Chronic
Illness
● Determinants of Nurse Fluidity: A Theoretical
Model to Examine Processes to Support
Foreign Educated Nurse Integration to the
United States Health Care System
Monica Rochman, B.S.N., R.N.
Presented by: Monica Rochman, B.S.N., R.N.,
Ph.D. Student, School of Nursing, University of
Michigan, 400 N. Ingalls Room 4170, Ann Arbor, MI
48103, Phone: (734) 846-6077; Email: mrocwal@umich.edu
Research Objective: As a result of the nursing workforce shortage and concern for quality patient care, health care organizations have increased recruitment and retention efforts for nurses. One strategy has been aggressive recruitment of foreign educated nurses or FENs. Recent data suggest that employment for nurses who are trained outside the
United States is increasing as the general population ages and current nurses begin to reach retirement age. Researchers have identified specific processes that support FEN integration into the United States health care workforce. These processes include policies regarding visas and credentialing, anti-discriminatory practices, training and orientation programs, and supportive systems for acculturation. At the national level, visa regulation and credentialing are concrete processes that allow for FEN entry into healthcare systems.
However, the remaining processes; supportive systems for acculturation, anti-discrimination practices, training and orientation, are not consistently applied throughout healthcare institutions. Although these processes are identified as necessary for FEN integration, it is uncertain how they relate to each other and whether one is required to precede another for successful integration. The main objective is to develop a theoretical model that explicates the interrelationships of these processes. The model entitled Determinants of Nurse Fluidity, describes how training and orientation influence both risk of discrimination and professional acculturation.
Study Design: A literature review of relevant studies that address topics of foreign educated nurses, nurse migration, acculturation, and international nurse recruitment was conducted to allow for conceptualization of a theoretical model to examine processes to support FEN integration to the United States health care system.
Population Studied: Literature review of foreign educated nurses.
Principal Findings: The United States health care system may lack coherent, supportive tools needed to facilitate FEN integration.
Conclusion: Nurse Fluidity is defined as the degree to which a registered nurse integrates or adapts to a new work environment. The determinants refer to elements that define the nature of nurse fluidity. The relationship between the systemic variables, training and orientation, are assumed to be dependent and can support or impede nurse fluidity. In addition, systemic variables are assumed to support nurse fluidity and affect experience with discrimination risk and professional acculturation. Finally, discrimination risk and professional acculturation are assumed to have an inverse relationship; when professional acculturation is strong discrimination risk is minimal.
Implications for Policy, Delivery or Practice:
Consequent implications of insufficient support from the U.S. health care system include: inadequate adjustment of FENs to the U.S. health care system, lowered job satisfaction, poorer working relationships with colleagues and patients, and diminished quality of services. The model provides a beginning framework for future health services research that examines facilitators or barriers to
FEN fluidity, and stimulates a discussion of potential implications of FEN fluidity on the nurse practice environment in the United States.
● Heroin Purchasing is Income, Price & Arrest
Rate Sensitive
Juliette Roddy, Ph.D.; C.L. Steinmiller, Ph.D.; M.K.
Greenwald, Ph.D.
Presented by: Juliette Roddy, Ph.D., Assistant
Professor, Social Sciences, University of Michigan
Dearborn, 4901 Evergreen, 2230 SSB, Dearborn,
MI 48128, Phone: (248) 703-5628; Email: jroddy@umd.umich.edu
Research Objective: This updated study assesses behavioral economic drug demand in heroin dependent research volunteers. Findings on drug price, competing purchases, and past 30-day
income and consumption, established in a previous study, are replicated and extended.
Study Design: This study sample is comprised of
109 volunteers who completed the first screening visit from August 2005 – January 2008 at a downtown Detroit outpatient clinical research program. This included written informed consent, providing demographic information, a complete medical and drug use history, and a semi-structured interview lasting about 20 minutes that was designed to obtain specific information about factors influencing drug price, purchasing, and use.
All volunteers reported current daily heroin use and/or provided a urine sample that had to test positive for opioids (>300 ng/ml). Participants were paid U.S. $25 for completing the first screening visit. Participants are asked to indicate to what extent hypothetical environmental changes
(described in separate scenarios) would alter their heroin purchasing behavior. The structured interview uses an updated version of the Drug
Purchasing and Use Questionnaire (DPUQ II) developed by Roddy and Greenwald.
Population Studied: This cohort was predominantly African-American (n=65), male
(n=72) and about 45 years old with a high school or equivalent education. Participants reported using heroin for about 22 years. Of the 109 respondents,
72 individuals reported they primarily injected heroin and 37 reported they did not inject.
Principal Findings: In the overall sample (n=109), participants report they would significantly (p<.005) decrease their heroin daily purchasing amounts
(DPA) from their past 30-day levels
(mean=$60/day) if: (1) they encountered a 33% decrease in income (DPA=$34), (2) family/friends no longer paid their living expenses (DPA=$32), or
(3) they faced four-fold greater likelihood of police arrest at their purchasing location (DPA=$42).
Statistically significant gender differences are apparent within the hypothetical scenarios of (1) a
33% decrease in income (DPAf = $38, DPAm =
$32), and (2) less likely and same as arrest rate scenarios, where females react less than male counterparts. Injection and non-injection users also showed significantly different reactions (DPAi =$25,
DPAni=$19) to a 33% reduction in income.
Regression analysis uncovers income and dealer availability influence unit purchase amount; number of purchases per week is influenced by purchase time, dealer availability and income; and number of bags consumed per day is influenced by unit cost and income. Analysis of income quartiles reveals that higher income levels lead to increased purchase amounts of heroin among these daily users but has little to no effect on other drug use.
Conclusion: Reactions to potent income and arrest changes will vary between groups
Implications for Policy, Delivery or Practice: and policy implications are present. Policy variables such as arrest rates and income supplement reductions must be dramatic to influence heroin purchases.
Funding Source(s): NIDA, R01 DA15462
Theme: Behavioral Health
● National Health Expenditures & Health
Professions Employment & Compensation
Charles Roehrig, Ph.D.; Ani Turner, B.A.
Presented by: Charles Roehrig, Ph.D., Vice
President, Altarum Institute, 3520 Green Court,
Suite 300, Ann Arbor, MI 48105, Phone: (734) 302-
4646; Email: charles.roehrig@altarum.org
Research Objective: The National Health Statistics
Group within the Centers for Medicare and
Medicaid Services (CMS) produces the National
Health Expenditure Accounts (NHEA) which track annual national health expenditures (NHE). Each
February, they also release ten-year NHE forecasts. Much of NHE goes to compensate workers in the health professions. Our research objective is to estimate the relationship between
NHE and health workforce employment and compensation by NHE service type (hospital, physician, nursing home, etc.) and the 68 health workforce occupations defined by the Bureau of
Labor Statistics (BLS). The results will allow researchers and policy makers to better understand the drivers of health sector expenditures over time and the health workforce requirements implied by
NHE forecasts.
Study Design: Using the CMS mapping of NAICS industry codes into the major NHE spending categories, we mapped BLS employment and wage data by industry to the NHEA, creating estimates of both the number of health and non-health workers and the associated labor expense in each NHEA category. We examined and applied additional industry and occupation-specific data sources to compare and supplement these data. We used the
Bureau of Economic Analysis (BEA) input-output tables as another estimate of labor expenditures for comparison and to estimate the portion of expenditures going to non-labor inputs in each healthcare setting. Finally, we developed a time series to observe shifts in employment and expenditure patterns over time.
Population Studied: All workers in the 68 BLS health occupations as well as workers in non-health occupational categories who are employed in the health services industries (hospitals, physician offices, dentist offices, nursing homes, and home health).
Principal Findings: For each of the major NHEA
Personal Health Care spending categories, we produced estimates of employment and compensation by detailed health occupations as well as spending on non-health labor and major non-labor inputs. We produced this employment and expenditure detail for the NHEA from 1996 through 2006 to observe trends over time. Finally,
we produced forecasts of the workforce requirements implicit in the most recent CMS national health expenditure forecasts to 2018 and compared them to existing supply projections found in the literature.
Conclusion: We have demonstrated the feasibility of linking the NHEA, BLS employment data, and
BEA input-output data to provide a unique resource from which it will be possible to investigate a multitude of broad health workforce and spending issues.
Implications for Policy, Delivery or Practice: As we have begun here, researchers and policy makers can examine and apply this work to: (1) identify the occupations that are driving costs in each NHEA service category; (2) identify trends in the percent of healthcare expenditures going to non-health occupations and inputs; (3) estimate workforce requirements associated with NHE forecasts; (4) identify potential constraints on spending based on health workforce shortages by occupation; and (5) identify the potential impact of proposed approaches to controlling health care cost growth on health profession employment and compensation.
Funding Source(s): Internal Research and
Development Program
Theme: Health Care Workforce
● Modeling the Primary Care/Specialty Mix of
New Physicians in the U.S.
Charles Roehrig, Ph.D.; Ani Turner, B.A.; Russell
Robertson, M.D.
Presented by: Charles Roehrig, Ph.D., Vice
President, Altarum Institute, 3520 Green Court,
Suite 300, Ann Arbor, MI 48105, Phone: (734) 302-
4646; Email: charles.roehrig@altarum.org
Research Objective: With the aging of the US population, the potential for expanded insurance coverage under health care reform, and the aging of the physician workforce, there is growing concern about physician shortages. Concerns about primary care are especially strong because primary care incomes have not kept pace with those of other specialties and there has been a steady decline in the number of US medical school graduates who are interested in practicing as generalists. Our research objective is to model the process by which the specialty mix of new physicians is determined and identify factors that influence the number entering primary care.
Findings will support the Council on Graduate
Medical Education (COGME) which is addressing this critical issue in its latest report.
Study Design: We developed a model of the primary care/specialist mix of new physicians and examined forecasts under baseline and alternative policy scenarios. Our model captures physicians’ preferences for primary care and non-primary care, graduate medical education (GME) training positions offered, and the reconciliation of specialty preferences and training positions through the
National Resident Match Program.
Population Studied: Applicants and institutions participating in US GME, including US allopathic and osteopathic medical school graduates, graduates of international medical schools applying for US residency training, and US teaching institutions offering GME.
Principal Findings: A number of physicians enter training in primary care as a second choice because of constraints on training positions in their first choice specialty. Yet there remain unfilled
GME positions in primary care. One implication is that simply increasing GME positions offered in primary care will not increase the number trained but will lead to more unfilled positions. Secondly, moderate increases in physician preferences for primary care will have little effect on the number trained. Instead, the main effect will be to reduce the portion training in primary care as a second choice. Planned increases in US medical school enrollments have the potential to increase primary care supply by forcing additional physicians into previously unfilled primary care positions. Once unfilled positions are filled, expansion of primary care GME positions can increase supply further, but again, will mean more physicians in primary care as a second choice. With no expansion in GME, the increase in US graduates will eventually push out international graduates without further increasing the primary care or non-primary care supply.
Conclusion: Moderate increases in physician preferences for primary care may not increase the number trained but will increase the number there by first choice. Increases in the number of US medical school graduates recommended by AAMC have the potential to increase the number in primary care but mainly as a second choice.
Significant increases in primary care that reflect first choices will require achievement and sustainment of the kind of dramatic increase in preferences observed briefly in the 1990s.
Implications for Policy, Delivery or Practice:
This work supports COGME as they consider recommendations for their upcoming 20th Report, and other policymakers as they develop strategies to address concerns about primary care shortages.
Funding Source(s): HRSA
Theme: Health Care Workforce
● The Medical Home & Receipt of Health
Screenings & Anticipatory Guidance for
Children
Melissa Romaire, M.P.H.; Janice Bell, Ph.D.,
M.P.H.
Presented by: Melissa Romaire, M.P.H., Graduate
Student, Health Services, University of Washington,
School of Public Health & Community Medicine,
1959 Northeast Pacific Street, H-664; Box 357660,
Seattle, WA 98195, Phone: (206) 568-3415; Email: mromaire@u.washington.edu
Research Objective: The medical home for children is a system of family-centered, communitybased healthcare intended to deliver quality acute and preventive care. Receipt of certain components of preventive care for children, including advice about particular health topics, is low. The purpose of this study was to estimate the number of children in the U.S. who have a medical home and to examine the association between having a medical home and receipt of preventive care.
Study Design: We conducted a cross-sectional analysis of data from the 2004 and 2005 Medical
Expenditure Panel Survey (MEPS). We developed two binary indicators of having a medical home from 22 questions in MEPS, reflecting 4 of the 7
American Academy of Pediatrics’ recommended components of the medical home, including the child’s usual source of care. The first indicator considers a person (e.g. doctor) as the usual source of care, while the second considers a person or facility (e.g. health center) as the usual source of care. Using logistic regression, we examined the association between having a medical home and receipt of specific health screenings and anticipatory guidance during the past 12 months, controlling for age, sex, race, insurance, family income, health status, and region.
In all models, standard errors account for MEPS' complex survey design.
Population Studied: Children ages 0-17 years with a visit to a healthcare provider within the past year were included (N=14,096).
Principal Findings: Approximately 20% of children have a medical home if the child’s usual source of care is defined as a person; whereas, 50% have a medical home if the definition of usual source of care is expanded to include a facility. Receipt of screenings or guidance in the past year ranges from 26% of children receiving advice on seat belt use to 93% of children having their weight measured. In fully adjusted multivariable models, having a medical home is significantly associated with increased odds of having weight, height, and blood pressure measured and of receiving advice about dental check-ups, diet, exercise, and use of a seatbelt when usual source of care is defined as a person —odds ratios (ORs) ranging from 1.18 –
1.59. Using the more expansive definition of usual source of care that includes facility, having a medical home is significantly associated with same outcomes; however, the magnitude of the associations is greater (ORs from 1.32 – 1.71) and four additional outcomes (advice about bicycle helmet use, use of safety seat, use of booster seat, and secondhand smoke) were significant. The study findings were generally robust to multiple comparisons.
Conclusion: The medical home is associated with increased likelihood of children receiving health screenings and anticipatory guidance. MEPS data can be effectively used to examine the concept of the medical home for children; however, excluding facilities from the usual source of care component of the medical home may underestimate the prevalence and impact of the medical home.
Implications for Policy, Delivery or Practice:
Receipt of health-related screenings, advice, and education is an integral component of well-child care, and the medical home provides an opportunity to improve the delivery of these vital services for children.
Funding Source(s): AHRQ
Theme: Child Health
● Comparing Methods of Measuring Treatment
Intensification in Hypertension Care
Adam Rose, M.D., M.Sc.; Dan Berlowitz, M.D.,
M.P.H.; Meredith Manze, M.P.H.; Michelle Orner,
M.P.H.; Nancy Kressin, Ph.D.
Presented by: Adam Rose, M.D., M.Sc., Center for
Health Quality, Outcomes, & Economic Research,
Bedford VA Medical Center, 200 Springs Road,
Building 70, Bedford, MA 01730, Phone: (781) 687-
3198; Email: adamrose@bu.edu
Research Objective: Insufficient treatment intensification (TI) is an important barrier to blood pressure (BP) control. A valid measure of TI could be useful in quality improvement efforts. However, there is uncertainty in the literature regarding how best to measure TI, and a variety of approaches have been used. Our objective was to compare the ability of different measures of TI to predict BP control (predictive criterion validity).
Study Design: Patients were enrolled in this prospective cohort study starting in August 2004; follow-up was completed in January, 2008. The outcome was the final BP measurement prior to study completion. Other data collected included demographics, comorbid conditions, BP values at each clinic visit, and prescriptions. We used three different methods to calculate TI. The any/none method differentiates patients who had any medication increases from patients who had none.
The Norm-Based Method (NBM) models the chance of a medication increase at each visit based on patient and visit characteristics, then scores each patient based on whether they received more or fewer medication increases than predicted by the model. The Standard-Based Method (SBM) is similar to NBM but expects a medication increase whenever the BP is uncontrolled. We compared the ability of these three measures of TI to predict each patient’s final systolic blood pressure (SBP), controlling for covariates, using linear regression.
Dichotomous BP control (final SBP < 140 mm/Hg) was similarly examined using multivariable logistic regression.
Population Studied: 819 hypertensive patients managed in primary care at an academic, urban safety-net hospital. The mean follow-up time was
24 months. The mean age was 60, 66% of patients were female, and 58% were of Black race. At baseline, 25% of patients received 1 medication,
37% received 2, 25% received 3, and 13% received
4 or more. The mean baseline BP was 134/80 mm/Hg.
Principal Findings: The any/none method produced a paradoxical result: patients with any therapy increases had a higher final SBP than those who had none (134.8 mm/Hg vs. 128.4 mm/Hg, p < 0.001). NBM was not a significant predictor of BP (p = 0.28). A categorical analysis demonstrated a U-shaped relationship between the
NBM score and the final SBP (final SBP in quartiles of increasing TI: 138, 128, 131, and 134 mm/Hg).
The SBM score was the best predictor of BP control; the patient’s final SBP was 2.1 mm/Hg lower for each additional therapy increase per ten visits (p < 0.001). Controlling for patient-level covariates did not alter this result. Patients with the most intensive management (highest quartile) had a mean final SBP 16 mm/Hg lower than patients with the least intensive management (lowest quartile).
The SBM score also predicted greater odds of a final SBP < 140 mm/Hg (OR 1.30 for each additional therapy increase per ten visits, c-statistic
0.70, p < 0.001).
Conclusion: The SBM score was a powerful predictor of BP control, while the NBM score and the any/none score were not valid measures of TI.
The validity of SBM was robust to examining BP control as a continuous vs. a dichotomous outcome, as well as controlling for patient-level covariates.
Implications for Policy, Delivery or Practice: Our results suggest that the standard-based method is the preferred measure of treatment intensification for hypertension care. Given its strong link to BP control in our study and prior studies, treatment intensification, as measured by the standard-based method, could serve as a quality indicator for hypertension care.
Funding Source(s): National Heart, Lung, and
Blood Institute
Theme: Prevention and Treatment of Chronic
Illness
● Can Readmission be Used as a Measure of
Hospital Quality? Implications from the
Veterans Health Administration (VA)
Amy Rosen, Ph.D.; Susan Loveland, M.A.T.;
Amresh Hanchate, Ph.D.; Haytham Kaafarani,
M.D., M.P.H.
Presented by: Amy Rosen, Ph.D., Senior
Research Scientist, Center for Health Quality,
Outcomes & Economic Research, 200 Springs
Road, Bedford, MA 01730, Phone: (781) 687-2960;
Email: akrosen@bu.edu
Research Objective: Readmissions are costly, common, and may be potentially preventable if they result from poor quality of care. Because readmissions may reflect both quality and efficiency, they provide more information on healthcare delivery than can a single measure alone, such as mortality. Risk-adjusted readmission rates have gained acceptance as valid quality measures by organizations such as the
Center for Medicare and Medicaid Studies and the
National Quality Forum. Given the current focus on hospital reporting of outcome measures, it is critical that the validity of readmissions as quality measures be tested. This paper examines one aspect of validity, “construct validity,” by evaluating the association of readmissions with the Patient
Safety Indicators (PSIs), a set of potentially preventable outcomes. Since the PSIs are related to longer inpatient stays, higher mortality and increased costs, we hypothesized that PSI events would also be related to higher readmission rates
(i.e., PSI events would be higher in hospitalizations with readmissions than in hospitalizations without readmissions).
Study Design: We combined VA 2003-2007 inpatient administrative data files and merged them with dates of death obtained from the VA Vital
Status file. We ran the AHRQ PSI software, obtaining observed PSI rates. Logistic regression models estimated the likelihood of “all-cause readmission” within 30 days of discharge, controlling for “any PSI event” (any of 13 selected
PSIs) and the comorbidities obtained by AHRQ’s
Comorbidity Software. For hospitalizations with readmissions, we examined type of readmission
(emergent or elective) and whether the readmission occurred in the same hospital.
Population Studied: All acute-care hospitalizations in the VA between 2003 and 2007.
Principal Findings: Of 1,915,387 hospitalizations,
15.6% resulted in readmissions within 30 days of discharge; 81.7% were single admissions without death; and 2.8% ended in death. Patients who were readmitted were generally older and had longer lengths of stay and higher comorbidity rates
(e.g., diabetic rates were 25.1% vs. 22.7%, respectively) (p<0.05) than patients who were not readmitted. Although PSI rates were relatively low, rates for 8 of the 13 PSIs were higher in hospitalizations with readmissions than in those without readmissions (e.g., rates of postoperative respiratory failure were 2.4% and 0.9%, respectively) (p<0.001). The probability of readmission, controlling for comorbidities, was higher for hospitalizations with any PSI event than those without any PSI event; OR= 1.20 (95% CI,
1.16-1.23).
Conclusion: As hypothesized, readmission rates were higher for hospitalizations with PSI events compared to hospitalizations without PSI events.
These results suggest that the construct validity of readmissions is strong, although additional
research is needed to test this finding in other settings and healthcare systems.
Implications for Policy, Delivery or Practice:
Risk of readmission can be modified by quality and type of care provided. Hospitals can better assess readiness of patients for discharge, improve discharge planning, and carefully transition patients to other care venues. Inclusion of readmissions as a quality indicator may flag additional opportunities for quality improvements missed by mortality and prevention of adverse events.
Funding Source(s): VA
Theme: Quality and Efficiency: Measurement
● Do Magnet-Credentialed Hospitals Provide
Better Care?
Marie-Claire Rosenberg, Ph.D., R.N; Ann Barry
Flood, Ph.D.
Presented by: Marie-Claire Rosenberg, Ph.D.,
R.N, College of Nursing, New York University, 246
Greene Street, Room 712W, New York, NY 10010,
Phone: (212) 992-9387; Email: mcr11@nyu.edu
Research Objective: To investigate whether hospitals with Magnet status have better patient outcomes.
Study Design: In the early 1980s, during a nursing shortage in the US, a group of hospitals was noted for their ability to consistently attract and retain a high-quality nursing workforce. The original research of these 41 hospitals became the foundation for the Magnet recognition process.
Each year since 1994, the American Nurses
Credentialing Center (ANCC) evaluates tens of applicant US hospitals to determine whether they will become newly designated or retain Magnet status for another four years
Earlier research has supported the claim that
Magnet status hospitals provide better care. We investigated whether patients in Magnet designated hospitals had better nursing-sensitive outcomes and general outcomes compared to non-Magnet hospitals, using two patient cohorts where nursing care is important.
Population Studied: Using a cross-sectional, multi-level design of fee-for-service Medicare beneficiaries 65 years or older who were hospitalized for initial treatment of Hip Fracture (HF) or Acute Myocardial Infarction (AMI) in all US nonfederal hospitals in 2002-2004, we merged data from the ANCC on Magnet status, from the
American Hospital Association on hospital characteristics, and Medicare (using Part A, enrollment and death records). In 2004 there were
101 healthcare organizations in 45 states credentialed as Magnet among 4522 non federal hospitals. 969,905 HF and 657,814 AMI patients were included in our study. Using multi-level regression models, we examined the association between receiving care in a Magnet hospital and better outcomes, using four nursing-sensitive outcomes (e.g., presence of ducubitus ulcers or infections from intravenous cannulas) and two general outcomes (in-hospital mortality and length of stay). Models adjusted for patient and other hospital characteristics.
Principal Findings: For HF patients, only one outcome was statistically significant at p<.001; they were less likely (Odds Ratio 0.72, 95% Confidence
Interval: 0.54, 0.96) to experience a decubitus ulcer in Magnet hospitals. None of the tests based on nursing or general outcomes showed care in
Magnet hospitals was statistically significantly different for AMI patients.
Conclusion: This study did not provide much support for the hypothesis that patients receive better care in Magnet hospitals. There are several limitations that may have limited our ability to test this association, in particular how to measure and evaluate nursing care in hospitals, e.g., what cohorts to use and what outcomes to measure.
Although our cohorts were large, few patients were reported to have experienced adverse outcomes in the nursing-sensitive measures, e.g., the percent of
HF patients experiencing a decubitus ulcer in their initial hospitalization was 0.20%.
Implications for Policy, Delivery or Practice:
Further research is needed to comprehensively measure how nursing care in general and more specifically nursing care provided within hospitals recognized for their excellent nursing care, impacts patient outcomes.
Theme: Organizational
● Management of Obesity & Overweight: Self
Reported Attitudes, Barriers, Skills &
Educational Needs among South Carolina
Health Care Professionals
Shelia Roundtree, M.D., M.P.H.; Lori Malvern, M.D.;
Khosrow Heidari, M.S.; Ava Dean, M.P.H.
Presented by: Shelia Roundtree, M.D., M.P.H.,
Clinician, Department of Veterans' Affairs, Minority
Faculty Development, Harvard University, 109
Shefford Court, Greer, SC 29650, Phone: (864)
525-8482; Email: tuunee01@yahoo.com
Research Objective: The primary aim of this study was to evaluate among South Carolina primary care providers their attitudes, barriers, skill level and educational needs in the management of obesity and overweight.
Study Design: A two page questionnaire was mailed and also electronically sent to primary care providers in South Carolina. The providers were drawn from two practice groups.
Population Studied: The first group of practitioners selected for participation were nurse practitioners and physicians who practice in community health outpatient clinics in South Carolina (n=250). The physicians were either Family Practice, Internal
Medicine or Pediatric specialists. These clinics are member institutions of the South Carolina Primary
Care Association. One hundred and sixty-eight
(67%) of these providers responded. The second group consisted of Family Practice, Internal
Medicine, Pediatric and Obstetrician/ Gynecologists whom comprise the Regional HealthPlus (n=135).
Regional HealthPlus is a combined physician/ hospital owned entity associated with Spartanburg
Regional Hospital in Spartanburg, South Carolina.
Sixty providers (44%) responded.
Principal Findings: Practice Setting. When asked in which area they spend most of their clinical time caring for patients, the majority of respondents reported that they work in a Family Medicine
(50.6%). Others reported : Pediatrics (26.19%),
Internal Medicine (10.12%) and Obstetrics/
Gynecology (7.74%).
Attitudes: The majority of respondents in all groups felt that overweight children will not outgrow their overweight (93.9%), health care providers have an important role in addressing healthy weight issues with their patients (93.4%) and health care providers should support policy changes related to nutrition and physical activity (90.5%). Barriers:
Both nurse practitioners and physicians cited lack of patient motivation (85%), lack of parent involvement (86.3%), lack of support services such as nutrition and counseling services (73.1%) and lack of resources for weight management in the community (71.9%) as the most important barriers to treatment of overweight and obese patients. Skill
Level: Most nurse practitioners and physicians reported low to moderate skill level in the use of behavioral management strategies concerning nutrition/ physical activity.
Conclusion: Primary Care providers queried overwhelmingly view obesity and overweight as critical areas of clinical practice. They felt that children usually do not outgrow overwieght, and they cited several barriers to effective overweight/obesity prevention and management.
Further, providers indicated the educational vehicles they prefer for learning more about these topics.
Implications for Policy, Delivery or Practice:
This data highlights the necessity for comprehensive dietary and physical activity training in basic medical education and post graduate curriculae.
The medical community is challenged to ensure that providers possess neccessary skills to effectively combat this most daunting of public health issues. From an economic standpoint, insurers may provide incentives for providers to perform preventive care. The current model is based on disease management rather than prevention. Continuing education organizations should consider the results of this study for future planning purposes.
Theme: Obesity Prevention and Treatment
● Personnel Economics Approach to Strategic
Human Capital Management & Planning:
Insights from the Centers for Disease Control &
Prevention
Kakoli Roy, M.A., Ph.D.; Zhuo Adam Chen, M.S.,
Ph.D.; Carol Gotway Crawford, M.S., Ph.D.
Presented by: Kakoli Roy, M.A., Ph.D., Senior
Economist, Office of the Director, Office of
Workforce & Career Development, Centers for
Disease Control & Prevention, 1600 Clifton Road,
Atlanta, GA 30306, Phone: (404) 498-6298; Email: kjr3@cdc.gov
Research Objective: An organization’s workforce
— or human capital — is its most valuable asset.
The 2002 President’s Management Agenda emphasizes the importance of strategic human capital management by requiring all federal agencies to improve performance by enhancing personnel and compensation systems. In response to these directives, the Centers for Disease Control and Prevention (CDC) recently drafted its human capital management plan to ensure that it is strategically aligned to effectively support the agency’s mission and Health Protection goals. CDC is the lead federal agency responsible for public health promotion, prevention, and preparedness. In this presentation, we explore the personnel economics literature to identify topics that are of practical importance and empirical relevance to
CDC’s internal workforce and personnel needs and examine how the agency might be managing careers in terms of entry, exit, and career progression, including variation in career paths by occupational and other categories.
Study Design: We conducted a comprehensive review of the personnel economics literature to identify relevant topics with an emphasis on identifying promising theoretical and methodological approaches that are then employed to explore
CDC’s internal labor market, using personnel data for 1995–2006. We conducted descriptive statistical and multivariate regression analyses to examine changes in the hierarchical organization structure across time, timing of promotions and variation in promotion and exit rates by tenure, and differences in career progression by occupational and other clusters.
Population Studied: Data were extracted from administrative personnel records of federal civilian employees employed at CDC during 1995–2006.
CDC has a blended workforce of >9,000 government employees, comprising Civil Service full-time employees (FTEs), Commissioned Corps officers of the U.S. Public Health Service, and other nonpermanent FTEs. The study sample is restricted to approximately 7,000 white-collar federal employees, which excludes a limited number of federal blue-collar workers and the Commissioned
Corps officers, who comprise 10% of the workforce, because they have a compensation and merit system that differs from civilian employees.
Principal Findings: Preliminary findings indicate that employees follow somewhat defined paths within the organization that are stable across time, resulting in long-term worker-organization attachments. However, the workforce is aging and the proportion of employees eligible for retirement is increasing. The proportion of employees in higher grades as well as the proportion in professional and technical categories is also increasing with time.
Significant variation in career progression by occupational clusters and an indication of grade compression exists, implying relatively flat careers, among employees holding advanced degrees and serving in professional categories.
Conclusion: Our study demonstrates relevance of existing organizational and personnel economics theories by empirically testing different hypotheses, using agency-specific personnel data. Our findings are also consistent with similar analyses conducted on other federal agencies.
Implications for Policy, Delivery or Practice: The personnel economics literature offers theoretically sound and empirically rigorous approaches that can help shape an evidence-based approach to human capital planning and improve incentives to attract, retain, and motivate a talented federal public health workforce and promote the culture of highperformance government. Disclosure statement:
The findings and conclusions in this presentation have not been formally disseminated by the
Centers for Disease Control and Prevention and should not be construed to represent any agency determination or policy.
Theme: Organizational Performance and
Management
● DoD Patient Safety Success Story: Efforts to
Reduce Harm Events in Military Treatment
Facilities
Rajasri Roy, Ph.D., M.P.H.
Presented by: Rajasri Roy, Ph.D., M.P.H.,
Epidemiologist, Patient Safety Center, Department of Defense, 1335 East West Highway, Suite 6 -
100, Silver Spring, MD 20910, Phone: (301) 295-
8112; Email: rajasriroy@hotmail.com
Research Objective: The root causes of potential patient safety incidents need to be addressed, as they can lead to serious injuries or death. In response to demands for patient safety efforts in the Military Health System (MHS), the DoD Patient
Safety Center (PSC) was formed in 2002 and action plans were developed to implement coordinated activities to promote outreach, safety education, and reduction of adverse events. Using logic models, the PSC created a data infrastructure that allows Military Treatment Facilities (MTFs) to regularly collect and submit data by using uniform sets of definitions for medical adverse events and comparing error rates with benchmarks derived from the data submitted by all participating MTFs.
Study Design: MTFs used the following methods and activities to report and evaluate patient safety and to improve health care quality: For nonmedication adverse events – a real-time, Webbased reporting tool; For medication errors –
United States Pharmacopia’s MEDMARX, a webbased anonymous medication error reporting program that is the largest medication error database in the nation; and For sentinel events – the Patient Safety Center developed in-house databases for analysis of RCAs (Root Cause
Analysis) and FMEAs (Failure Mood Analysis) of
JACHO (Joint Commission on the Accreditation of
Healthcare Organizations) reviewable sentinel events. Other activities included: Continuous patient safety training for providers, pharmacists, nurse attendants, and other medical staff; Secured websites for focused reviews on relevant patient safety topics; and Annual and semiannual analysis and summary reports of submitted error data, providing feedback to patient safety managers.
Population Studied: The military personnel and their dependents treated as of approximately 180
Military Treatment Facilities (ambulatory clinics, community centers, and medical centers).
Principal Findings: Several positive outcomes were revealed as a result of the system of standardized reporting and thorough analysis of patient safety events: The yearly analysis showed overall reporting of error events gradually increased by 90% from 2003 to 2007. With almost 73% of incidents reported by medical facilities being near miss events, DoD medical facilities realized that there were opportunities for preventing an error from reaching the patient by using proactive interventions. Results showed definite improvement in the primary measure of success for patient safety: preventing harm associated with healthcare. Errors causing significant harm to patients declined by 25% from 2003 to 2007, as measured in medication error reports(MEDMARX), non-medication events (Monthly Summary
Reports), and incidents resulting in Root Cause
Analyses (RCAs).
Conclusion: The DoD patient safety system successfully encourages an awareness of patient safety issues and has reduced patient harm in
Military Treatment Facilities.
Implications for Policy, Delivery or Practice:
Through continuous education, leadership support, and safe working culture, it has been proven that
“events of harm” can be decreased.
Funding Source(s): Department of Defense
Theme: Quality and Efficiency: Measurement
● An Analysis of Marriage-Lock: The Effect of
Health Insurance Options on Divorce Rates
Jamie Rubenstein, A.B.
Presented by: Jamie Rubenstein, A.B., Ph.D.
Student, Economics, Cornell University, 4th Floor,
Uris Hall, Ithaca, NY 14853, Phone: (607) 244-
4061; Email: jcr19@cornell.edu
Research Objective: The relationship between marital status and health insurance has remained largely unstudied. Given the powerful effects that marriage and divorce have on the well-being, it is important to understand how health insurance and health policy may impact the decision to divorce.
This paper aims to describe the relationship between health insurance and divorce and to measure the causal impact of health insurance options on the decision to divorce. The theory draws on the job-lock literature. Much as job-lock occurs when an individual stays with an inefficient job match in order to retain health insurance, individuals may stay in marriages that they would otherwise leave in order to retain health insurance, resulting in “marriage-lock”.
Study Design: I measure the causal effect of health insurance on divorce using two difference-indifference methods, as well as a regression discontinuity approach. Both DD methods compare differences in divorce rates between couples where the husband does and does not have his own employer sponsored health insurance (ESI) offer.
Since there may be unobserved differences when the husband can and cannot provide ESI, I use women who place a low value on a spouse’s ESI as a control group for women who place a higher value on their spouse’s ESI. I subtract the difference in divorce rates for women with a low value of health insurance from the difference in divorce rates for women with a high value of health insurance. The first DD method measures the value of health insurance to the woman by whether she has her own health insurance offer, since women with their own offer will place a lower value on a spouse’s insurance. The second DD method measures the value of health insurance to the woman by whether she has health problems, since women with health problems will place a higher value on their husbands’ insurance. Both DD methods control for a number of demographic and socioeconomic characteristics, job characteristics, and marital history. The first method also controls for health status. The RD method compares differences in divorce rates before and after a woman becomes eligible for Medicare, since a spouse’s ESI offer becomes less valuable when the wife turns 65.
Control variables are similar to the DD models with the addition of controls for benefits such as social security.
Population Studied: I use the 1996, 2001 and
2004 panels of the Survey of Income and Program
Participation to study married couples.
Principal Findings: Of women aged 18-64 who are married in wave 1, 3 percent divorce by the end of the panel. Of the women who divorce during the panel and are insured prior to the divorce, 9 percent lose their health insurance immediately after divorce. The estimation of the DD and RD models is not yet complete.
Conclusion:
Implications for Policy, Delivery or Practice: If access to health insurance is tied to marital status through a spouse's employment, this may result in inefficiencies in the same way that inefficiencies result from job-lock. These inefficiencies could be solved either with either a single payer system or with a stronger non-group health insurance market.
Theme: Coverage and Access
● Using Zones of the AUDIT-C to Screen for
Risk of Alcohol Dependence Among Patients
With & Without Prior Alcohol Treatment
Anna Rubinsky, M.S.; Daniel Kivlahan, Ph.D.;
Katharine Bradley, M.D., M.P.H.
Presented by: Anna Rubinsky, M.S., Research
Health Science Specialist, Health Services
Research & Development, VA Puget Sound Health
Care System, 1100 Olive Way, Suite 1400, Seattle,
WA 98101, Phone: (202) 277-4156; Email:
Anna.DeBenedetti@va.gov
Research Objective: Fewer than 10% of patients who meet criteria for DSM-IV alcohol dependence are identified and offered appropriate treatment.
Alcohol screening in primary care settings is recommended for identification of alcohol misuse, but additional practical methods are needed to identify the minority of screen positive patients who have alcohol dependence that warrants more intensive intervention (e.g. specialty referral or medications for alcohol dependence). Ranges of
AUDIT-C alcohol screening scores (“risk zones”) have been suggested to help identify patients likely to have alcohol dependence among those who screen positive for alcohol misuse. However, prior alcohol treatment may influence the diagnostic performance of alcohol screening questionnaires.
This study evaluated whether AUDIT-C risk zones perform similarly among individuals with and without a history of prior alcohol treatment.
Study Design: This cross-sectional study compared the diagnostic properties of previously identified risk zones of the derived AUDIT-C between individuals who reported ever seeking help for drinking (“prior treatment”) and the remainder of the screen positive general population sample.
Likelihood ratios for past year alcohol dependence were computed for each risk zone and used to estimate post-screening probabilities.
Population Studied: U.S. adults from the nationally representative 2001-2002 National
Epidemiologic Survey on Alcohol and Related
Conditions (n= 43,093) who screened positive for alcohol misuse.
Principal Findings: Eight percent of men
(n=1,335) and 3% of women (n=575) reported prior alcohol treatment. Among men, likelihood ratios for alcohol dependence were significantly different
between those who did and did not report prior treatment for 3 of the 6 AUDIT-C risk zones, demonstrating differential performance of the risk zones in these populations. Using the prevalence of alcohol dependence in men who did vs. did not report prior alcohol treatment (20% vs. 5%, respectively) and corresponding likelihood ratios, the post-screening probabilities of alcohol dependence across risk zones were 13% vs. 2%
(score 4), 22% vs. 6% (score 5), 26% vs. 12%*
(scores 6-7), 40% vs. 24%* (scores 8-9), 59% vs.
34%* (scores 10-11), and 89% vs. 49% (score 12).
Among women, likelihood ratios were significantly different between those who did and did not report prior treatment for 4 of the 7 risk zones. Using the prevalence of alcohol dependence in women who did vs. did not report prior alcohol treatment (18% vs. 2%, respectively), the post-screening probabilities of alcohol dependence across risk zones were similar to those for men.
Conclusion: Among individuals who screen positive for alcohol misuse, those who report prior alcohol treatment have a 10-40% higher postscreening probability of alcohol dependence across
AUDIT-C risk zones compared to those who do not report prior treatment.
Implications for Policy, Delivery or Practice:
Appropriate likelihood ratios of AUDIT-C risk zones can be incorporated into nomograms or electronic medical records so that health care providers can quickly derive a patient’s post-screening probability of alcohol dependence. To most accurately identify patients likely to have alcohol dependence and best inform clinical care, separate likelihood ratios should be used for patients with and without prior alcohol treatment.
Funding Source(s): VA
● Marked Mortality Reduction Over Time in
Patients Hospitalized for Community-Acquired
Pneumonia
Gregory Ruhnke, M.D., M.S., M.P.H.
Presented by: Gregory Ruhnke, M.D., M.S.,
M.P.H., Instructor, Hospital Medicine, University of
Chicago, 5841 South Maryland Avenue, MC 2007,
Chicago, IL 60637, Phone: (773) 834-8350; Email: ruhnke@stanfordalumni.org
Research Objective: Community-acquired pneumonia (CAP) is the most common infectious cause of death in the United States, and has thus been the target of numerous quality initiatives.
Several preventive and therapeutic interventions have been shown to decrease mortality. However, the impact of their greater use on the outcomes of
CAP patients over time has not been demonstrated.
We examined the trend in mortality of patients hospitalized with CAP.
Study Design: Our analysis used hospitalizations from a 10% subsample of the HCUP National
Inpatient Sample from 1993 – 2005. Cases of CAP were identified by a principle diagnosis of pneumonia or respiratory distress/failure with a secondary diagnosis of pneumonia (shown to have sensitivity 84% and specificity 86%). Using the
AHRQ Comorbidity Software, 29 binary comorbidity indicator variables (CIVs) were defined based on the secondary diagnoses coded at discharge.
Univariate logistic regressions of inpatient mortality on each CIV were performed separately. A multivariate logistic regression was then performed, regressing mortality on the following independent variables: dummy variables for each year
(reference year 1993), age by 5-year category, gender, and 25 of the 29 CIVs that predicted an increased risk of mortality, because a decreased risk would likely reflect coding bias. The primary outcome measure was the odds ratio (OR) of death in each year relative to 1993. To adjust for changing admission and discharge practices, the regression was repeated, excluding patients admitted from long-term care (LTC) facilities and, separately, including dummy variables for discharge to a skilled nursing facility (SNF), intermediate care facility
(ICF), or with home health care (HHC). To assess the sensitivity of our results to increases over time in the number of fields available for coding (and a consequent increase in the proportion of patients with CIVs), we repeated the multivariate regression after restricting all patients to a maximum of five discharge diagnoses.
Population Studied: Patients >= age 18 hospitalized for CAP. 389,639 patients met the coding criteria, most of whom were in their 50s
(11%), 60s (16%), 70s (24%), and 80s (23%).
Principal Findings: Over the period studied, mean length of stay (LOS) declined from 8 to 5.5 days, while the percent of patients discharged to a
SNF/ICF and with HHC increased from 15 to 22 and from 5 to 12, respectively. The frequency of many comorbidities increased substantially.
Mortality adjusted only for age and gender declined from 9.0% to 5.7%. The OR of multivariate-adjusted mortality declined progressively in each year
(relative to 1993), with an OR of .56 (CI .51–.60) in
2005. With exclusion of LTC patients and adjustment for discharge status, the ORs of mortality in 2005 were .60 and .62, respectively.
After restricting all patients to a maximum of five diagnoses, the OR of mortality was .59.
Conclusion: The prevalence of comorbidities in hospitalized CAP patients has increased, while LOS and mortality have greatly reduced, suggesting an improvement in both the quality and efficiency of care.
Implications for Policy, Delivery or Practice:
Policies directed at increasing early administration of guideline-concordant antibiotics, as well as influenza and pneumococcal vaccination may further reduce mortality due to CAP.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Measurement
● Health Information Technology:
Characteristics that Support VA Clinical
Workflow
Alissa Russ, Ph.D.; Jason Saleem, Ph.D.; Connie
Justice, C.I.S.S.P.; Heather Woodward-Hagg, M.S.;
Peter Woodbridge, M.D., M.B.A.; Bradley
Doebbeling, M.D., M.Sc.
Presented by: Alissa Russ, Ph.D., Health Services
Research & Development, Roudebush VA Medical
Center, 1481 West 10th Street, 11-H, Indianapolis,
IN 46202, Phone: (317) 988-3132; Email: alruss@iupui.edu
Research Objective: The aim of this investigation was to identify characteristics of health information technology (HIT) that support VA clinical workflow.
The VA is a successful leader in HIT development and implementation. However, the characteristics of
HIT needed to support clinical workflow have not been well studied.
Study Design: In this investigation, we performed a secondary analysis of interview data from healthcare employees and identified electronic information characteristics that support clinical workflow. We had previously conducted 20 keyinformant, semi-structured interviews at a large VA
Medical Center and asked healthcare employees why paper use persists despite a fully implemented electronic health record (EHR). Employee responses naturally shed light on positive and problematic aspects of the EHR and other IT and revealed key HIT characteristics needed to support clinical workflow. Two researchers independently analyzed the interviews for emergent themes; researchers then discussed each coding discrepancy until reaching consensus.
Population Studied: The VA’s EHR is designed to be used by clinical staff as well as administrative and support staff who aid clinicians’ efforts to deliver quality patient care. Therefore, the following participants were recruited: 3 administrators; 3 local
IT specialists, including 2 clinical application coordinators (CACs) with clinical backgrounds; 2 pharmacists; 3 physicians; 2 nurse practitioners; 4 registered nurses; 2 health technicians; and 1 dietician. (CACs receive clinical software requests from employees and resolve software issues.)
Both men and women were recruited, and participants represented a wide range of work experience, varying from 1 to 31 years with a mean value of 12 yrs of experience. This range of experience is important because the facility transitioned to an EHR approximately 10 years ago.
Principal Findings: Participants provided 199 examples of how electronic information affects VA healthcare workflow. From these examples, we identified 17 HIT characteristics needed to support clinical workflow. Some characteristics, like information "availability" and "security" were strengths of VA HIT. However, many other characteristics were inadequately supported. For example, data demonstrated a need for the following: information, such as abnormal lab results, that is better "prioritized" and more readily
"locatable"; "customizable" settings for patient lists and quantitative data; and the "ability to trend" patient data. Altogether, the 17 characteristics were spread across four primary domains: trustworthy and reliable; ubiquitous; effectively displayed; and adaptable to work demands.
Conclusion: Healthcare employees revealed key information characteristics needed to support clinical workflow and provided insights on how to improve HIT. For instance, redesigns can incorporate: integrated spreadsheets for trending data; customizable settings for viewing and sorting data; and enhanced visual displays to prioritize critical information. Study findings provide a more comprehensive set of HIT characteristics needed to support clinical workflow.
Implications for Policy, Delivery or Practice:
Each of the 17 characteristics can be used to enhance HIT design and may promote HIT adoption and sustainability. Furthermore, at least 17 corresponding evaluation questions can be developed from these findings to help assess EHRs pre- and post-implementation. These results are central to understanding how to improve HIT to better support clinical workflow and healthcare delivery.
Funding Source(s): VA
Theme: Health Information Technology
● Use of Text Mining to Identify Follow-up
Appointments in Hospital Dismissal Summaries
Kari Ruud, M.Ed.; Matthew Johnson, M.P.H.; Carrie
Grafft, M.D.; Juliette Liesinger, B.A.; James
Naessens, Sc.D.
Presented by: Kari Ruud, M.Ed., Health Services
Analyst, Health Sciences Research, Mayo Clinic,
200 1st Street, Southwest, Rochester, MN 55905,
Phone: (507) 266-9822; Email: ruud.kari@mayo.edu
Research Objective: To determine the accuracy of text mining in identifying whether specific follow-up appointment arrangements were documented in hospital dismissal summaries
Study Design: A dataset consisting of the free-text entry of dismissal summaries for 6481 hospital patients was manually reviewed to determine whether the discharge records contained specific follow-up appointment arrangements. To be considered complete, documentation needed to provide specific date, time, and either physician or location for the appointment. Appointments with primary care providers and specialty consults were considered to be follow-up appointments, whereas appointments for procedures or therapy were not.
The dataset was also evaluated for the same criteria using the text parsing capabilities of SAS
Text Miner software. The two assessments were
compared to determine the accuracy of text mining for detecting records containing follow-up appointment arrangements.
Population Studied: Electronic medical records were extracted for all patients dismissed in 2006 from inpatient general medicine services at Mayo
Clinic in Rochester, Minnesota. Records of patients transferring to another inpatient service, those going to hospice care, and those without research authorization were excluded from the analysis.
Principal Findings: Of the 6481 dismissal summaries reviewed, 3576 (55.2%) contained all criteria for follow-up appointment arrangements according to manual review, 113 (3.2%) of which were missed through text mining. Text mining incorrectly identified 107 (3.7%) follow-up appointments that were not considered valid through manual review. Therefore, the text mining analysis concurred with the manual review in 96.6% of the appointment findings. Sensitivity (text mining finding appointments identified through manual review) and specificity (text mining agreeing with
“no appointment” results of manual review) were
96.8% and 96.3%, respectively; and positive predictive value and negative predictive value were
97.0% and 96.1%, respectively. Analysis of specific appointment criteria resulted in accuracy rates of
93.5% for date, 97.4% for time, 97.5% for physician, and 82.9% for location. Further investigation of discrepancies showed location to be more difficult to identify because the full sentence context is unknown when selecting terms with the
Text Miner tool. False positives through text mining were primarily explained by the identification of procedural appointments, which were not considered to be valid follow-up appointments.
General instructions to patients containing date, time, physician name, and/or location were also falsely flagged as follow-up appointments.
Conclusion: Text mining of medical records can accurately detect whether follow-up appointment arrangement elements are documented in hospital dismissal notes. Although the SAS Text Miner tool has the ability to identify parts of speech, some limitations exist without the ability to interpret full sentence context, which may be feasible with natural language processing. Additional research is needed to determine the extent to which text mining can identify other relevant discharge information from medical record documentation.
Implications for Policy, Delivery or Practice:
Results suggest that text mining software can be used to identify specific appointment criteria in a large number of textual medical records, thus saving considerable resources required for manual abstraction in quality-related research and performance assessment.
Theme: Health Information Technology
● Effectiveness of Quality Improvement Training at Mayo Clinic
Kari Ruud, M.Ed.; Jeffrey Leland, M.A.; Valerie
Halling, M.Ed.; Matthew Johnson, M.P.H.; Andrew
Majka, M.D.; James Naessens, Sc.D.
Presented by: Kari Ruud, M.Ed., Health Services
Analyst, Health Services Research, Mayo Clinic,
200 1st Street, Southwest, Rochester, MN 55905,
Phone: (507) 266-9822; Email: ruud.kari@mayo.edu
Research Objective: To evaluate the effectiveness of a quality/ process improvement training program at Mayo Clinic
Study Design: This study assessed the Quality
Academy Teams Training at levels two (learning) and three (behavior) of Kirkpatrick’s framework for evaluating training programs. Pretests and identical posttests were used to assess learning by attendees of this applied course for physicians, administrators, practitioners, and team members of institutional process improvement initiatives who wish to improve quality, service capacity, and/or cost for either clinical or non-clinical processes.
The training program is administered in three sessions spanning five, three, and two days, respectively. Course content is designed to equip individuals and teams with process improvement tools and techniques within Six Sigma and Lean methodologies, so they can conduct improvement projects to improve value. Gain score analysis was conducted using paired t-test procedures to determine statistical significance of the change in test scores for each session. Only tests with a corresponding pre/posttest taken by the same individual were used for the analysis. Electronic surveys were sent to participants 90 days following completion of the course to assess application of newly learned tools and techniques to the work setting. Survey results were consolidated and summarized.
Population Studied: The tests and surveys were administered to three cohorts of Mayo Clinic,
Rochester (Minnesota) employees attending the process improvement training in 2008, comprising a total of 107 participants. Overall, 104 employees took both the pre- and posttest for at least one session, with 91, 79, and 66 completing both tests for sessions A, B, and C, respectively. Thus far, 37 course participants have responded to the survey
(the third cohort has not received the survey as of the writing of this abstract).
Principal Findings: Mean pre-test/post-test scores for sessions A, B, and C were 47% vs. 67%, 46% vs. 62%, and 62% vs. 84%, respectively. Test scores improved for 90.1% of participants in session A, 86.1% in session B, and 83.3% in session C. Test improvements for all three sessions were statistically significant at the .01 level, and no cohort effect was found. Survey results showed that respondents increased their use of 34 out of 36 process improvement tools on the job after attending the training, and a majority of
respondents reported meeting 11 of 19 skill-based course objectives. In addition, 83% of respondents indicated they shared knowledge of process improvement tools with co-workers, and 57% reportedly mentored others who did not attend the training.
Conclusion: This study demonstrated the effectiveness of a process improvement training program at Mayo Clinic. Attendees learned the material taught in the course, as evidenced by a significant improvement in post-training test scores from pretest scores. Furthermore, employees who attended the training applied the tools and techniques on the job within 90 days of completing the course and shared knowledge and mentored co-workers, according to self-report.
Implications for Policy, Delivery or Practice:
Healthcare institutions may benefit from the implementation of quality-related training programs that teach employees to use process improvement tools and techniques.
Theme: Quality and Efficiency: Organized
Processes
● State-Wide Child Maltreatment Prevention
Programming: Results of a Structured
Evaluation of One Component of Arkansas Act
703 of 2007
Kevin Ryan, J.D., M.A.; Rebekah, Craig, M.P.H.;
Kevin Ryan, J.D., M.A.
Presented by: Kevin Ryan, J.D., M.A., Assistant
Professor, Department of Health Policy &
Management, University of Arkansas for Medical
Sciences College of Public Health, 1401 West
Capitol Avenue, Suite 300 Victory Building, Little
Rock, AR 72201, Phone: (501) 526-2244; Email: ryankevinw@uams.edu
Research Objective: To determine the risk, effectiveness, system costs, burden, and fairness of a state statute requiring implementation of a schoolbased child maltreatment program through application of a systematic and weighted evaluation process and whether states have an affirmative duty to implement child protection programs.
Study Design: Arkansas Act 703 of 2007 (codified as ACA 12-1-501 et. seq.) was passed and signed into law "to protect child victims of abuse and neglect through new and expanded programs." A key component of the Act requires development and implementation of an education platform targeting school aged children with a goal to prevent and / or reduce child abuse through empowerment. The authors evaluate the effectiveness of this statute through application of a public health law evaluation tool and discuss whether the state has an affirmative duty to intervene to reduce child maltreatment.
Population Studied: Adult policy makers and educators and school aged children in Arkansas.
Principal Findings: When a public health law oriented statutory evaluation tool is applied, the potential benefit of implementing a child maltreatment prevention curriculum in Arkansas public schools appears to outweigh the potential burden. State authority supports mandating these types of programs. However, despite well documented need and demonstrated effectiveness, the decision by states to exercise that authority remains largely discretionary.
Conclusion: Child maltreatment impacts over i million children in the United States each year.
Evidence exists to support merit of implementation of school-based programs designed to empower children to prevent and reduce child maltreatment.
Evaluation of the law mandating such programs in
Arkansas demonstrate that the potential benefit of the mandate will outweigh the potential regulatory burden. However, despite documented need to intervene, and existence of effective programs, exercise of state authority remains largely discretionary.
Implications for Policy, Delivery or Practice:
Determining the potential cost-benefit of a state intervention to reduce child maltreatment is informed through use of a public health law evaluation tool. Policy makers should consider application of such tools to inform more efficient and effective development and implementation of laws, statutes and regulations.
Theme: Child Health
● Care Coordination for Senior Patients with
Multiple Chronic Diseases: Examining the
Association Between Organizational &
Relational Factors & Patient Outcomes
Marian Ryan, Ph.D., M.P.H., M.A., C.H.E.S.
Presented by: Marian Ryan, Ph.D., M.P.H., M.A.,
C.H.E.S., Research Associate, Heller School,
Brandeis, 415 South Street, Mail Stop 035,
Waltham, MA 02454, Phone: (949) 290-7697;
Email: mryan@brandeis.edu
Research Objective: The Chronic Care Model
(CCM) provides a conceptual framework to guide organizations in making a system change to improve chronic illness care. However, the pathway(s) from the six organizational elements of the CCM to the improved physician-patient interactions leading to improved patient outcomes is currently underdeveloped in the model. Relational coordination (RC) and trust may constitute the critical relational factors that facilitate the productive interactions between prepared physicians and activated patients envisioned by the CCM. This study sought to identify the amount of variation in patient outcomes and patient adherence that could be explained by CCM, physician care coordination behavior, the physician-patient relationship, and individual characteristics.
Study Design: Two analytic models are being examined to evaluate quantitatively the association between practice characteristics (the CCM elements and relational factors) and outcomes
(patient outcomes and adherence). Both models employ the same dependent variables: recognized quality process measures, intermediate outcome measures such as A1c and LDL levels, and patient adherence as observed through prescription medication claims. Hierarchical Linear Modeling
(HLM) techniques are employed to permit the formulation of explicit structural models for processes that occur within each level of a hierarchy (patient, PCP and clinic), the specification of the unique random effects of each level and the estimation of the variances and covariances of these random effects. Model I uses four years of longitudinal data to estimate the impact of two elements of the Chronic Care Model (selfmanagement resources and electronic medical records) measured at the clinic level, three relational domains (access, communication and coordination) measured at the physician level, and individual characteristics on patient outcomes and adherence over time. Model II is examining the direct association between the CCM score, relational coordination and trust, and patient outcomes and adherence. Interviews were conducted with key clinic team members to complete the Assessment of Chronic Illness Care instrument for each site. Surveys were mailed and completed by 831 patients that assessed Relational
Coordination and trust with their respective primary care physician. 94 primary care physicians completed web-based surveys assessing care coordination behaviors.
Population Studied: 12,031 senior patients having diabetes and at least one additional comorbidity identified by a large multispeciality medical group in southern California between January 2004 and
December 31, 2007, and all primary care physicians and group model clinics associated with care provided to these patients constitute the study population.
Principal Findings: Analytic models are in final construction following data input and verification.
Results will be completed within 8 weeks.
Implications for Policy, Delivery or Practice: The findings of this study have the potential to elucidate the pathway to the productive interactions envisioned in the Chronic Care Model as related to senior patients with multi-morbidity. It is also testing relational coordination in a new context, the physician-patient team. This work has the potential to influence policies to support payments to primary care physicians for care coordination and identify the critical factors within the physician-patient relationship that promote better outcomes.
Funding Source(s): JHF/Schneider/self
Theme: Quality and Efficiency: Organized
Processes
● Associations Between Personal Wellbeing,
Sense of Coherence, Health Service Utilization
& Perceived Health Care Quality in a Rural
Region of Sweden: Implications for Service
Provision, Where is Our Help Most Needed?
Ingrid Rystedt, M.D., Ph.D.; Gun Nordström, Ph.D.,
R.N.; Bodil Wilde Larsson, Ph.D., R.N.
Presented by: Ingrid Rystedt, M.D., Ph.D.,
Biträdande Lektor, Department of Nursing, Karlstad
University, Universitetsgatan 2, Karlstad, 65188,
SE, Phone: +011-54-7001267; Email: ingrid.rystedt@kau.se
Research Objective: To explore associations between personal wellbeing, sense of coherence, health service utilization and perceived health care quality.
Study Design: The study utilized data from a cross-sectional questionnaire, the Life and Health
Survey. In 2008, the questionnaire was mailed to
11,900 randomly selected inhabitants of Värmland,
Sweden (54.4% response rate). Survey domains included subjective health, formal encounters/individual experiences with health services, lifestyle habits and life circumstances.
Rich data on attributes of quality of life enabled exploration of associations between life characteristics and health service utilization and/or satisfaction.
Population Studied: The survey population was
18-84 years, all covered by the national health insurance system and living in the mostly rural region of Värmland in midwestern Sweden. The county council has 5 hospitals and approximately
25 outpatient clinics. Based on several indicators, the regional health status is below the national average, e.g., higher sick leave and early retirement, higher prevalence of obesity and much higher expenditures for pharmaceuticals.
Furthermore, long-term unemployment is high.
Principal Findings: During the past three months, survey participants visited on average two types of service providers. Persons with lower scores on the sense of coherence scale (targeting comprehensibility, manageability, and meaningfulness in life) visited more providers.
Similarly, persons with lower psychological wellbeing saw more providers (contacts with psychologists and social workers did not account for this difference). Approximately half the individuals who contacted outpatient clinics by phone reported difficulty in reaching staff that could answer their questions. Similarly, 44% experienced such difficulties when contacting hospitals by phone. Persons who experienced difficulty contacting providers by phone more frequently reported a lower sense of coherence and/or lower psychological wellbeing, compared to persons who did not experience difficulty accessing providers by phone. Furthermore, persons who experienced difficulty reaching providers by phone more
frequently rated the care they received as lower quality. Additional factors associated with reports of lower quality of care included having a chronic illness, and, having repeatedly been disappointed by someone trusted. In regards to lifestyle factors, women were less likely to report being asked questions regarding diet, exercise and alcohol habits in outpatient settings. Gender differences in the receipt of questions regarding lifestyle were also observed in hospital settings. Among women 35-49 years of age, 16 % reported contact with alternative health service providers.
Conclusion: Attributes of quality of life were associated with health service related disparities.
Therefore, identification and attention to patients that perceive life, and health care encounters, as complex and burdensome should be an essential component of effective health services provision.
Patient gender was associated with whether patients reported that providers targeted certain lifestyle related habits. This circumstance brings to question whether health systems adequately meet lifestyle related needs of women. It may be important to investigate how women are most effectively supported regarding lifestyle in health care encounters.
Implications for Policy, Delivery or Practice:
Patients with complex needs likely benefit from integrated services and a broader-than-diseaseoriented focus. Equipping providers and systems to identify and comprehensively care for patients with complex and chronic needs should improve overall health system performance. Levering community resources may be helpful in this regard.
Funding Source(s): Värmland County Council and
Karlstad University
Theme: Disparities
● Trends in Care for Uninsured Adults &
Disparities in Care by Insurance Status
Lindsay Sabik
Presented by: Lindsay Sabik, Doctoral Candidate in Health Policy, Harvard University, 180 Longwood
Avenue, Suite 113, Boston, MA 02115, Phone:
(301) 908-2353; Email: lsabik@fas.harvard.edu
Research Objective: This study investigates how care for the uninsured is changing over time and how the differences in care received by the privately insured and the uninsured are changing.
Study Design: Trends in outcomes for the uninsured and differences in outcomes between the uninsured and privately insured in access to care, management of chronic disease, and invasive heart attack treatment are estimated using data from the
Community Tracking Study for 1996-2003, the
National Health and Nutrition Examination Survey for 1999-2006, and the National Hospital Discharge
Survey for 1996-2006, respectively. All trends are adjusted for individual characteristics and AMI-care is adjusted for hospital characteristics.
Population Studied: Uninsured and privately insured adults between the ages of 18 and 64.
Principal Findings: Between 1996 and 2003 access problems neither increased nor decreased among the uninsured, while the gap between the uninsured and privately insured increased. The adjusted predicted probability of privately insured individuals reporting having to forgo or delay necessary care decreased from approximately 9% to 6%, while the probability of uninsured individuals reporting these access problems stayed constant at
10%. Between 1999 and 2006 blood pressure control among uninsured hypertensives improved, while glycemic control among uninsured diabetics and cholesterol control among uninsured with coronary heart disease, diabetes, or stroke did not show significant change. The privately insured had better blood pressure control than the uninsured at baseline and the gap between the uninsured and privately insured increased over this period.
Differences in other measures of disease management by insurance status did not change significantly over the period. Over the 1996-2006 period, use of percutaneous transluminal coronary angioplasty (PTCA) or coronary artery bypass grafting (CABG) post-AMI was consistently higher among privately insured patients than among uninsured patients, and the difference did not increase or decrease consistently over the period.
On average across the period the privately insured had a 4.1 percentage point (9.1%) higher probability of receiving invasive care post-AMI than did the uninsured.
Conclusion: Over the decade starting in the mid-
1990’s, care for the uninsured showed little improvement on measures of access, chronic disease management, or invasive care post-AMI.
Uninsured adults are more likely than the privately insured to report problems accessing care, are less likely to successfully control chronic diseases along most measures, and are less likely to receive invasive care after heart attacks. Over the study period, the disparity between the uninsured and privately insured along most of these measures did not improve. The gap between the two groups in access to care widened, while the gap in other measures did not increase or decrease significantly.
Implications for Policy, Delivery or Practice:
Along certain measures, care for the privately insured has improved while care for the uninsured has not. While care for the insured is improving along some measures, increasing costs and numbers of uninsured may be increasing strain on safety net providers. The growing gap between the uninsured and privately insured suggests that improvements in care for the privately insured are not similarly accruing to safety net patients.
Funding Source(s): National Science Foundation
Graduate Research Fellowship
● Understanding Health Care Disparities:
Perspectives of Black Middle-Class Women
Tina Sacks, A.M.
Presented by: Tina Sacks, A.M., Student, School of Social Service Administration, University of
Chicago, 1755 E. 55th Street, Unit 505, Chicago, IL
60615, Phone: (773) 314-7402; Email: tsacks@uchicago.edu
Research Objective: To explore how middle-class
African-American women perceive health care institutions and their health care providers.
To investigate the extent to which middle-class
African-Americans use class resources/selfpresentation strategies to increase the likelihood that they receive “appropriate treatment” or avoid differential medical treatment based on race?
Study Design: The study uses a qualitative method, in-depth interviews, to explore the experiences of black middle-class women in a health care setting.
Population Studied: Black women between the ages of 30 and 60 in a large urban area. The women are middle-class based on their education, income, or other normative measure of social position.
Principal Findings: African-American middle-class women report some negative experiences with health care providers in which they felt they were evaluated based on negative stereotypes of
African-Americans. They also reported actively managing their appearance and language choice to emphasize their social position. The women described the process of adopting self-presentation strategies as exhausting. They did not report a uniform preference for race concordance with their provider. Finally, the women's health care experience varied by age (younger women reported fewer negative experiences); by type of provider
(women reported more negative experiences with
OB/GYN's compared to primary care or other specialty provider); and the environmental context of care (Women in a city with a high proportion of successful African-Americans did not report being as affected by negative stereotypes of African-
Americans as women in cities with more socioeconomically depressed populations).
Conclusion: The study provides important information about the health care experiences of a large, but understudied, segment of the population.
African-American middle-class women have many resources but are not immune to bias in health care. African-American women have their own strategies for mitigating bias through emphasizing their education, profession, or social class.
However, the process of self-presentation can be exhausting to the individual, negatively affect the woman's ability to listen and engage with the provider, and may reduce the woman's trust in the provider.
Implications for Policy, Delivery or Practice:
Overall, the health care experiences of African-
American women remind us that while providers may be unaware of patient-side strategies like selfpresentation management they are likely to affect the patient-provider encounter. The combination of clinical uncertainty and self-presentation strategies must be considered when developing interventions to ameliorate differences in treatment. Finally, because black middle-class women have multiple social identities, interventions based solely on increasing race-concordance are unlikely to completely eradicate racial disparities in health care.
● How do U.S. Children Fare in Kinship Care?
An Analysis of a National Sample of Children in the Child Welfare System
Christina Sakai, M.D.; Jane Brotanek, M.D., M.P.H.;
Hua Lin, Ph.D.; Glenn Flores, M.D., F.A.A.P.
Presented by: Christina Sakai, M.D., Fellow,
General Pediatrics, University of Texas
Southwestern Medical Center, 5323 Harry Hines
Boulevard, Dallas, TX 75390-9063, Phone: (214)
648-5040; Email:
Christina.Sakai@Utsouthwestern.edu
Research Objective: Current child welfare policy encourages placement of maltreated minors with relatives in kinship care. Questions continue, however, regarding supervision and support in kinship care, and little is known about physical, mental, and other outcomes of children in kinship versus general foster care. The main objective of this research is to comprehensively assess placement patterns and physical, mental, and other health outcomes for US children in kinship care versus general foster care.
Study Design: Retrospective cohort study using data collected from the National Survey of Child and Adolescent Well-Being, October 1999 to March
2004, of a nationally representative sample of children whom possible abuse or neglect was investigated by child welfare agencies.
Multivariable analyses adjusting for baseline covariates were performed to examine the independent association between placement setting and physical, mental and behavioral health outcomes 36 months after initial placement.
Population Studied: The study population consisted of children initially placed in out-of-home care kinship care (n=572) or general foster care
(n=736) following a maltreatment report.
Principal Findings: Compared with general foster care, significantly (P<0.01) lower proportions of children in kinship care have selected special healthcare needs (anemia, cerebral palsy, persistent bowel problems, and physical deformities, mental retardation, visual impairment) and baseline abnormal behavior (31% vs. 52%;
P<0.01). Follow-up revealed lower proportions of children in kinship care (versus general foster care) had continuing behavior problems (23% vs. 53%;
P<0.01), low social skills (24% vs. 40%; P<0.01)
and prescription medication use (14% vs. 28%;
P=0.02), but no significant differences in education, depression, or insurance coverage measures. In multivariable analyses, kinship care children had significantly lower adjusted odds versus general foster care children of continuing behavior problems
[OR, 0.3; 95% CI, 0.1-0.5], low social skills [OR,
0.3; 95% CI 0.1-0.6], and medication use [OR, 0.3;
95% CI 0.1-0.9], but were significantly less likely to be up-to-date on immunizations [OR, 0.02; 95% CI,
0.003-0.2].
Conclusion: Children with certain special health care needs and behavior problems are significantly more likely to be placed in general foster care than kinship care. Compared with general foster care, kinship care is associated with lower adjusted odds of continuing behavior problems, low social skills, and prescription medication use, but higher odds of delayed immunizations.
Implications for Policy, Delivery or Practice: The findings of this study indicate greater insight is needed into why children with special health care needs and behavioral problems are more often placed in general foster care, and why kinship care children fare better with behavioral problems, social skills, and medication use, but worse with immunizations. The findings also suggest general foster care children may benefit from intensive services for special healthcare needs and behavioral problems, and kinship care children from efforts to enhance immunization coverage.
Theme: Child Health
● Development, Implementation & Evaluation of a Web-Based Curriculum on Pharmaceutical
Marketing & Prescribing
April Salisbury, M.B.A.; Margaret Gunter, Ph.D.;
Robert White, M.D.
Presented by: April Salisbury, M.B.A., Director,
Continuing Professional Education, Lovelace Clinic
Foundation, 2309 Renard Place, Southeast, Suite
103, Albuquerque, NM 87106, Phone: (505) 938-
9925; Email: april.salisbury@lcfresearch.org
Research Objective: To develop, implement, and evaluate a web-based curriculum on drug marketing and prescribing with funding from off-label marketing lawsuit settlement funds.
Study Design: Content experts developed a course entitled “Optimal Prescribing” in two stages: a live pilot and a web-based course consisting of modules with slides, video clips, narration, cases, interviews, and interactive elements. The course was presented to physicians and midlevel providers from the Lovelace Health System, in Albuquerque,
New Mexico. Feedback was collected at the live pilot and participants were surveyed later, as well as a control group of providers who did not attend the pilot course, to assess the impact of the education on attitudes about relationships with industry representatives, the amount and source of information that guides prescribing decisions, and self-reported changes in prescribing patterns. Data were analyzed and summarized to revise the material to produce web-based modules for use by prescribers in the public domain. The final webbased, 3-credit course includes a pretest, posttest, and a 90-day follow up survey, which are currently being collected.
Population Studied: In the first two months the course was available, 173 people accessed the web-based course and 40 successfully completed it, earning credit (16 more took the posttest, but did not pass). Of the 40 who accessed the course, 21 were physicians (3 psychiatrists and 1 surgeon), 2 physician assistants, 6 nurse practitioners, 6 nurses, 1 pharmacist, and 4 were either not specified or non-clinicians.
Principal Findings: Initial reports on the learners’ performance show improvements on all posttest questions. Of the 20 questions 17 showed >10% improvement, with a maximum 37.12% improvement. The greatest improvements were in the learner’s ability to: 1) acknowledge the fact that even very brief sales calls can result in a 16% increase in drug sales, 2) distinguish “absolute” from “relative” risk reduction, and 3) identify acceptable evidence for off-label choices. Of the total users, 78% would recommend this course to colleagues (29.35% said “yes, enthusiastically”).
The first few participants have received the 90-day post course survey, but data has yet to come in as of the writing of this abstract.
Conclusion: Optimal Prescribing has been piloted, revised, and launched as a web-based course, and is being accessed by a number of people with positive results; however, less than 25% have fully utilized the course and earned continuing education credit. Data collection and analysis will continue over the remainder of the 3-year life of the course and may lead to improvements in course design.
Implications for Policy, Delivery or Practice: The health care policy landscape is changing rapidly.
Compliance policies, regulations, laws, and voluntary codes are becoming more restrictive of traditional marketing strategies. Increased awareness of the effectiveness of these strategies to influence prescribing decisions can help prescribers understand and support these changes and develop what the course calls a “Healthy
Skepticism” when evaluating information about drugs.
Funding Source(s): State Attorney General
Consumer and Prescriber Education Grant Program
Theme: Quality and Efficiency: Policies and
Incentives
● Multiple Chronic Physical Illnesses among
Individuals with Depression: An Application of
Clustering with Correspondence Analysis
Usha Sambamoorthi; Nethra Sambamoorthi, Ph.D.
Presented by: Usha Sambamoorthi, Professor, 11
Bartram Road, Englishtown, NJ 07726, Phone:
(732) 972-6232; Email: ushasambamoorthi@gmail.com
Research Objective: In the published literature multiple methods have been used to identify cooccurring chronic conditions (i.e. multi-morbidity) including summary scores. An aggregated multimorbidity index does not adequately capture the complexity because conditions contribute differentially to mortality, services use and functional status. Therefore, it is essential to identify clusters of chronic conditions that will facilitate the planning, management and treatment of individuals with similar clusters. We use correspondence analysis, a statistical tool that assorts different physical illness categories (in combination with depression) into groups in a way that the degree of association between physical illnesses among individuals with depression is maximal if they belong to the same group and minimal otherwise. This analytical approach is particularly suited for handling variables that are discrete (i.e. presence or absence of a condition).
The objective of the study is to identify clusters of chronic physical illnesses that tend to co-occur with depression using correspondence analysis and profile individuals and healthcare expenditures based on the clusters.
Study Design: Cross-sectional of adults over age
18. Condition codes and ICD9 codes were used to identify 23 different chronic physical conditions
(anemia, arthritis, asthma, cancer, COPD, diabetes, diverticulitis, eye problems, gall bladder, GERD, heart disease, HIV, hypertension, IBS, liver problems, enlarged prostate, peptic ulcer, PVD, chronic renal failure, stroke, and thyroid disorders) and depression. The impact of the physical illness clusters on healthcare expenditures was evaluated using multiple regression methods. All analyses accounted for the complex design of the survey.
Population Studied: Data are from the Medical
Expenditure Panel Survey, a nationally representative survey of healthcare by US households. Cross-sectional analysis of 9,932 adults over age 18 with depression and a chronic physical condition in 2005.
Principal Findings: Overall, 14% (N = 1,498) representing 14,220,169 individuals reported having depression and a chronic physical illness; there were 1,301 combinations of chronic physical illnesses. 26% individuals had 2 physical conditions; 37% had 3 or more physical conditions.
Correspondence analyses revealed 5 different distinct clusters. The hierarchical clusters related to infectious diseases (HIV), respiratory (Asthma and
COPD), pain (arthritis, cancer), cardiovascular
(diabetes, heart disease, hypertension), thyroid, and miscellaneous disorders. In 2005, the most expensive cluster was the one with cancer
($12,656), followed by HIV ($9,631) and respiratory diseases ($9,288), cardiovascular ($8,986). The lowest expenditures were among clusters with arthritis, hypertension and thyroid disorders
($4,610). There were distinct demographic and socio-economic characteristics associated with each of the clusters.
Conclusion: We demonstrated the feasibility of correspondence analysis to identify patterns of physical illness clusters in individuals with depression. Profile and healthcare expenditures of the clusters revealed distinct patterns.
Implications for Policy, Delivery or Practice: The clustering information could be used by healthcare providers to better assess physical illness clusters and customize healthcare management and service delivery for individuals with co-occurring physical illnesses and depression.
Theme: Behavioral Health
● Co-occurring Mental & Physical Illness among
Adults
Usha Sambamoorthi, Ph.D.; Rosalie, Torres-Stone,
Ph.D.; Marie Hobart, Ph.D.
Presented by: Usha Sambamoorthi, Ph.D.,
Professor, Department of Psychiatry, University of
Massachusetts Medical School, 11 Bartram Road,
Englishtown, NJ 07726, Phone: (732) 972-6232;
Email: ushasambamoorthi@gmail.com
Research Objective: Research in the areas of cooccurring physical illnesses in schizophrenia and depression has been well researched. However, a comprehensive account of physical disorders among those with other types of mental illnesses is scarce. Therefore our objective is to estimate the prevalence of common chronic physical disorders among a nationally representative sample of adults under 65 years of age with mental illness.
Study Design: Cross-sectional analysis of the
Medical Expenditure Panel Survey (MEPS. Mental illness consisted of: attention deficit, adjustment, anxiety, mood, personality, schizophrenia and other psychotic and miscellaneous mental disorders.
Physical disorders included Alzheimer's, arthritis, asthma, cancer, chronic obstructive pulmonary disease, diabetes, heart disease, hypertension,
HIV, osteoporosis, stroke, and thyroid disorders.
Chi-square and multinomial logistic regression were used to profile individuals with co-occurring mental and physical disorders. All analyses accounted for the complex survey design of the MEPS.
Population Studied: Cross-sectional analysis of
16,914 adults under age 65 from the 2005 household component of the MEPS.
Principal Findings: Overall 22% (N = 2,980) reported presence of mental illness. Among those with mental illness, depressive (61%) and anxiety
(48%) disorders were highly prevalent and the most common physical disorder was hypertension
(23.2%) followed by arthritis (17.4%). A comparison of physical disorders among individuals
with and without mental illness revealed that except for Alzheimer's, osteoporosis, HIV, and cancer all other disorders had elevated rates in individuals with mental illness (p < .01). Women, Caucasians, current smokers and those who were obese were more likely to report co-occurring mental and physical disorders.
Conclusion: Many types of physical illnesses occur at higher rates in individuals with mental illness.
Some subgroups are more likely to have cooccurring mental and physical disorders.
Implications for Policy, Delivery or Practice:
Subgroups of individuals with co-occurring physical and mental illness need special attention and pose challenges to healthcare management. Our findings highlight the need for integration of physical and mental illness care and point to the need for reverse integration (i.e providing primary care in mental health settings).
Theme: Prevention and Treatment of Chronic
Illness
● A Natural History Model to Estimate the
Potential Impact of Vaccination to Reduce
Cervical Cancer Disparities Between U.S.
Females With & Without Health Insurance
Myrlene Sanon, M.P.H.; Douglas Taylor, M.B.A.;
Denise Kruzikas, Ph.D., M.P.H.; Vivek Pawar,
Ph.D.; Kristen Gilmore; Milton Weinstein, Ph.D.
Presented by: Myrlene Sanon, M.P.H., Research
Analyst, Health Economics & Outcomes Research,
I3 Innovus, 10 Cabot Road, Suite 304, Medford, MA
02155, Phone: (781) 518-4042; Email: myrlene.sanon@i3innovus.com
Research Objective: Although health insurance coverage is a significant factor in accessing healthcare, over 17 million American females do not have health insurance. Pre-cancerous cervical lesions and cervical cancer (CC) are common in US females, and CC screening has been shown to be effective in reducing CC incidence, but access to screening may be more limited for females not covered by health insurance. Increasing screening for uninsured females may be impractical, because it would require regular contact with the healthcare system. Vaccination programs, on the other hand, have the potential to substantially reduce CC incidence without having to change screening practices. The objective of this study was to assess the clinical benefits of CC screening and vaccination in cohorts of insured and uninsured 19year-old females.
Study Design: A lifetime Markov model was developed to simulate the natural history of HPV infection and subsequent cervical disease in a cohort of 19-year-old females with and without health insurance. The National Health Interview
Survey 2005 data were used to estimate annual
Papanicolaou test screening rates among females with and without health insurance. Medicare screening rates were applied to females aged 65 years and older. The screening rates for females with and without insurance were then used as model inputs to estimate cumulative lifetime CC incidence and mortality. Published literature and clinical trial data were used to model HPV natural history, vaccine efficacy, practice patterns, and clinical outcomes. Lifetime impact of vaccination on
CC incidence and mortality was compared with CC screening only, for both the insured and uninsured female cohorts. The vaccine was assumed to provide 95% protection against HPV types 16/18 and 27% protection against all other high-risk HPV types.
Population Studied: Cohort of 2,020,693 19-yearold US females with and without health insurance
Principal Findings: Average annual screening rates in females aged 20-49 were 71% with health insurance and 54% without health insurance. In the absence of vaccination, the model predicted lifetime
CC incidence of 220.8 and mortality of 41.4 per
100,000 females, in the uninsured cohort.
Corresponding results were 144.3 and 23.0 per
100,000 for insured females. At 100% coverage, vaccination reduced CC incidence by 72% (61.5 per
100,000) and CC mortality by 62% (15.8 per
100,000) in uninsured females; CC incidence and mortality were reduced by 74% (37.7 per 100,000) and 62% (8.8 per 100,000), respectively, for females with health insurance. Vaccination reduced the disparity in CC incidence between insured and uninsured females by 69% and the disparity in CC mortality by 62%.
Conclusion: Females without health insurance on average have lower screening rates and a higher overall risk of developing CC compared with females who are insured. Vaccinating 19-year-old females, regardless of insurance status, has the potential to reduce the lifetime disparity in CC incidence and mortality between insured and uninsured females by more than 60%.
Implications for Policy, Delivery or Practice:
Instituting HPV vaccination programs for 19-yearold females who are unlikely to be covered by health insurance will result in lower rates of cervical cancer. Moreover, such vaccination program is predicted to reduce disparities between females with and without health insurance.
Funding Source(s): GlaxoSmithKline
Theme: Disparities
● HPV DNA Testing & Impact on Cervical
Cancer Screening Intervals
Mona Saraiya, M.D., M.P.H.; Zahava Berkowitz,
M.S.P.H.; Robin Yabroff, Ph.D.; Vicki Benard,
Ph.D.; Louise Wideroff, Ph.D.; Sarah Kobrin, Ph.D.
Presented by: Mona Saraiya, M.D., M.P.H.,
Medical Epidemiologist, Division of Cancer
Prevention & Control, Centers for Disease Control
& Prevention, 4770 Buford Highway, Atlanta, GA
30341, Phone: (770) 488-4293; Email: msaraiya@cdc.gov
Research Objective: To assess recommendations by U.S. primary care physicians’ (PCPs) for extending screening intervals that incorporate adjunct HPV DNA tests compared to Pap testing alone.
Study Design: We surveyed a nationally representative sample of practicing PCPs in 2006-
2007 (cooperation rate = 73.4%). We examined
Pap test recommendations using 2 clinical vignettes of a 35-year old female with: 1) no new sexual partners and 3 consecutive negative Pap tests or 2) a negative Pap and negative HPV DNA test.
Differences in recommendations were assessed by physician specialty.
Population Studied: 950 PCPs who perform Pap tests and routinely recommended HPV test for screening or management
Principal Findings: Approximately one third of
PCPs recommended HPV DNA testing as an adjunct test for cervical cancer screening, with little variation noted by specialty. For a 35-year old with three negative Pap tests, 32.4% would conduct the next Pap test in 3 or more years. For a 35-year old with a negative HPV DNA test and a normal Pap test, 19.6% would conduct the next Pap test in 3 or more years. HPV DNA test recommendations varied, with the majority of responses split among no HPV testing, HPV testing at the same frequency as the Pap test, or unknown HPV testing frequency.
Conclusion: Over 80% of PCPs do not recommend extending the screening interval to 3 years with adjunct HPV DNA testing.
Implications for Policy, Delivery or Practice:
While at least a third of the providers are recommending the HPV DNA testing as part of cervical cancer screening, there is apparently resistance to extending the screening intervals to 3 or more years. The cervical cancer screening paradigm is shifting to use emerging technologies such as the HPV DNA test and the HPV vaccine.
These changes allow for stronger support of policies against annual cervical cancer screening.
However, providers, policymakers, and the public will need more education and systems interventions to adapt the extended interval. Implementation of effective interventions that focus on physician familiarity with extended screening intervals, combined with not reimbursing annual screening in women with negative HPV DNA and negative Pap tests, or with several consecutive negative Pap tests, will be important for improving efficient and cost-effective screening practices.
Funding Source(s): NCI, Centers for Disease
Control and Prevention
Theme: Prevention and Treatment of Chronic
Illness
● Does Patient Origin Influence Care? An
Analysis of International Medical Travelers
Siriporn Satjapot, B.A.; Tricia Johnson, Ph.D.;
Andrew Garman, Psy.D., M.S; J. Robert Clapp Jr.,
M.S.A., F.A.C.H.E.; Samuel Hohmann, Ph.D., M.S.;
Frank Phillips, J.D.
Presented by: Siriporn Satjapot, B.A., Health
Systems Management, Rush University, 3852 North
Francisco, Chicago, IL 60618, Phone: (773) 510-
2984; Email: siriporn_satjapot@rush.edu
Research Objective: Although medical tourism has received considerable attention by the mainstream news media, there have been few credible efforts to assess the size and impact of the phenomenon generally, and even fewer examining inbound international medical travel to the United States.
Although travelers may benefit from access to higher quality care, it is conceivable that the health risks associated with travel, culture distance, and loss of social support may mitigate these benefits and warrant further investigation.
Study Design: A retrospective case control study of international patients coming to the United States for medical care. A linear regression model was used to test the relationship between LOS and patient geographic origin (international or domestic), controlling for age, sex, race, academic medical center, admission status, severity of illness, and MS-DRG.
Population Studied: The study included 5,729 patients coded with international payers, including international self-pay and international private/commercial, and 1,467,783 domestic patients who were admitted to academic medical centers that are members of the University
HealthSystem Consortium (UHC) between October
2005 and September 2008. Because of the large group size differences between international and domestic patients, further analysis of the study will be case-controlled as international patient LOS is matched with domestic patient LOS.
Principal Findings: The average LOS for international patients was 6.47 days (std dev =
11.33), compared to an average LOS of 5.37 days
(std dev = 8.33) for domestic patients. The p-value for LOS and patients (international and domestic) is less than .0001.
Conclusion: Discrepancies in LOS between international and domestic patients may be an indication that different treatment protocols exist based on the patient’s geographic origin. Further research is needed to understand whether payment source further explains differences in LOS among international patients.
Implications for Policy, Delivery or Practice:
Discrepancies in LOS between international and domestic patients raises the question as to who is driving these differences. Closer scrutiny on the system may help to pinpoint opportunities to better utilize resources for international patient care.
Academic medical centers should consider what factors are contributing to the differences in LOS
and how they relate to the characteristics of international patients. Such characteristics that may contribute to LOS differences that are unique to international patients include long distance travel, patient expectations and cultural differences, and severity of illness upon admission. In addition, similar to domestic patients, the primary payer source may also contribute to LOS differences.
These considerations will facilitate continuing advancement in patient care and improve efficiency in international patient case management.
● The Importance of Comorbidity Measures & other Individual Factors in Assessing Health
Disparities & Quality of Care
Barry Saver, M.D., M.P.H.; Sharon Dobie, M.C.P.,
M.D.; Pamela Green, M.P.H., Ph.D.; Ching-Yun
Wang, Ph.D.; Laura-Mae Baldwin, M.D., M.P.H.
Presented by: Barry Saver, M.D., M.P.H.,
Associate Professor, Family Medicine & Community
Health, University of Massachusetts Medical
School, 55 Lake Avenue North, Benedict Building
A3-146, Worcester, MA 01655, Phone: (508) 856-
3458; Email: Barry.Saver@umassmed.edu
Research Objective: Quality measures may be used with limited or no individual adjustment. We explored how individual factors, including different methods of comorbidity adjustment, affected assessments of differences and time trends in ambulatory care sensitive hospitalizations (ACSHs) associated with race and ethnicity.
Study Design: Retrospective, longitudinal cohort study
Population Studied: SEER-Medicare control subjects aged 66+ from 1993-1999; ~125,000 subjects/year
Principal Findings: We modeled occurrence of at least 1 acute or chronic ACSH in a year, using the
AHRQ Prevention Quality Indicator definitions, with logistic regression using GEE. We used the previous year’s claims to derive the Romano-
Charlson comorbidity index and individual flags for the Romano-Charlson conditions plus hypertension.
Additional predictors included age, gender, disability status, residence, geocoded income and education surrogates, county-level service availability, and continuity of care. For acute
ACSHs, unadjusted, all groups showed substantial upward trends: whites increased 31%, blacks increased similarly but had rates ~30% higher than whites, Hispanics started out similar to whites but increased over 60%, and Asians/PIs started out
52% lower than whites and rose over 100%. In full models using the Charlson index, all increases were attenuated, with whites showing an 18% rise by 1999 and blacks and Hispanics having rates very similar to those of whites. Asians/PIs in this model had 38% lower odds than whites in 1993; these increased 69% by 1999, no longer significantly lower than whites. Models utilizing the comorbidity flags eliminated the secular increases for whites, blacks, and Hispanics and further attenuated differences between Asians/PIs and whites – Asians/PIs started out only 28% lower in
1993 and increased only 35%, ending up almost identical to whites in 1999. For chronic ACSHs, changes were even more dramatic. Unadjusted, whites showed a nonsignificant, 6% decline, while other groups all had substantial increases: rates for blacks rose from 78% (1993) to 124% (1999) higher than whites, Hispanics from 6% to 52% higher, and
Asians/PIs from 54% lower to 6% lower. Adjustment including Charlson substantially changed differences and trends: whites had a 17% decline, blacks had steady rates about 25% higher than the white rate in 1993, Hispanics had no trend and were intermediate between blacks and whites, and
Asians/PIs had a significant, attenuated, upward trend, starting out 35% lower than whites and ending 4% higher. Adjustment including comorbidity flags markedly depressed rate estimates for later years, with whites, blacks, and Hispanics all showing downward trends - an approximately 45% decline for whites and blacks, with the black-white gap widening similarly to what was seen with
Charlson, from about 30% to almost 50%, and
Hispanic-white differences wider vs. Charlson and now similar to the black-white differences. For
Asians/PIs, the upward trend vanished, with rates16% lower than whites in 1993 and 26% higher than whites in 1999.
Conclusion: For ACSHs, findings for time trends and racial/ethnic differences varied substantially depending on adjustment, including the method used to control for individual comorbidity.
Implications for Policy, Delivery or Practice:
Quality measures assessing disparities and time trends need to be evaluated for sensitivity to individual factors including comorbidity; separate condition indicators may be preferable to aggregate indices.
Funding Source(s): NCI
Theme: Disparities
● Impact of Intervention Type on
Antihypertensive Medication Adherence Effects
Vernon Schabert, Ph.D.; Leon Dupclay, Jr.,
Pharm.D., Ph.D.; Siddhartha Roy; Chakkarin
Burudpakdee, Pharm.D.; Erica Goldberg
Presented by: Vernon Schabert, Ph.D., Senior
Director, Health Economics & Outcomes Research,
IMS Health, 300 South Washington Street, #303,
Falls Church, VA 22046, Phone: (703) 286-2873;
Email: vschabert@us.imshealth.com
Research Objective: Poor medication adherence may result in morbidity, mortality, and increased healthcare expenditures in many diseases, including hypertension. Medication adherence publications, including evaluations of adherence interventions, have increased substantially since
1997. However, a recent Cochrane review found little support for the effectiveness of any adherence intervention within or across diseases. We conducted a meta-analysis of published antihypertensive adherence interventions to explore factors associated with the size of intervention effects.
Study Design: A PubMed and grey literature search identified 25 studies, published 1975present, for review. Three were excluded for incomplete reporting of design or results; a fourth was excluded for an outlier effect size. Adherence measures were categorized by type (raw adherence measure, percentage achieving adherence threshold, ordinal adherence report) and converted into standard effect sizes (standard deviation units
[SD]). Adherence intervention types were categorized by primary intervention tactic, then quantified by duration (months) and number of patient contacts.
Population Studied: The 21 studies included interventions of 17,280 total hypertensive patients
(mean N 785.4, median 285.5, range 100-6,813).
Of these, 16 were from North America, 3 from
France, and 3 from other countries. Across studies, mean age was 57.9 years and 49.7% were male.
Subject recruitment occurred at clinic, work/community, HMO, and pharmacy sites for 11,
4, 3, and 3 studies, respectively.
Principal Findings: The mean effect of interventions was 0.26 SD (median 0.22, range -
0.01 to 0.87), conventionally considered a “small” effect. When studies were grouped by type of adherence measure, regardless of intervention type, mean effect sizes were comparable (ordinal rating = 0.21, raw adherence = 0.29, adherence threshold = 0.26). Seven studies measured effects of reduced daily dosing (TID to BID or BID to QD, mean effect = 0.27, median 0.13, range 0.09-0.87) and 4 measured effects of unit-dose packaging
(mean effect = 0.17, median = 0.19, range 0.04-
0.24); other types of interventions were used too infrequently to estimate pooled effects. Eleven interventions included increased patient contact to communicate adherence messages. These interventions included a mean of 2.26 additional inperson, 3.43 mail, and 2.64 telephone contacts compared to control groups. Intervention duration and the log of study sample size showed small negative correlations with effect size (r = -0.16 and -
0.21, respectively), while total patient contacts showed a small positive correlation (r = 0.22).
However, a fixed effects regression showed that none of these were a significant predictor of adherence effect (R2 = 0.16, F = 1.00, p = 0.42).
Conclusion: Identified antihypertensive intervention studies showed relatively little variability of intervention effects, regardless of adherence measure or intervention type. No intervention strategies (e.g., reduced daily dosing, fixed-dose packaging, increased patient contact) demonstrated superior effects for improving adherence. These findings suggest that medication non-adherence is a complex, multidimensional problem.
Implications for Policy, Delivery or Practice:
Policymakers and providers should note that many antihypertensive interventions achieve small adherence effects whose clinical significance has not been fully delineated. Larger effects and reproducible results may be required to establish the true benefit of specific interventions. Future antihypertensive adherence interventions would benefit from a tailored, patient-centric approach so that interventions more directly target root causes of non-adherence.
Funding Source(s): Novartis Pharmaceuticals
USA
Theme: Quality and Efficiency: Policies and
Incentives
● Reduced Rehospitalization & Increasing
Length of Service in Home Care Patients
Charles Schade, M.D., M.P.H.
Presented by: Charles Schade, M.D., M.P.H.,
Medical Epidemiologist, West Virginia Medical
Institute, 3001 Chesterfield Place, Charleston, WV
25304, Phone: (304) 346-9864 x 2243; Email: cschade@wvmi.org
Research Objective: To identify characteristics of home health agencies predicting reductions in acute care hospitalization (ACH) during home care episodes during a time when average ACH rates declined significantly.
Study Design: Retrospective observational study of Medicare home health claims using a linear regression model to predicting ACH, and a confirmatory stochastic model of acute care hospitalization from home health care.
Population Studied: 2,344 U.S. home health agencies with at least 10 episodes of care completed each month during a study period of
January 2004 and December 2007.
Principal Findings: Average risk adjusted ACH rates (adjusted for patient characteristics including severity of disability requiring home care) declined by 0.01% per month (p<0.0001) while average length of service (ALOS) increased by 0.087 days per month (p<0.0001) over the study period. The patient-level stochastic model assumed that home health care had no impact on ACH rate, and simulated outcomes in agencies based on a target length of stay and probability of ACH derived from
Medicare claims data. It predicted that increasing
ALOS would result in increased ACH. In the linear regression model of actual HHA data, ALOS was the most important predictor, explaining 21% of the variance in ACH (the remaining variables explained
4% collectively). The actual data and simulation agreed on the size of the effect of ALOS on ACH.
When we modeled ACH in single years, seven factors were significant predictors of monthly ACH
rate: ALOS, ALOS-squared, a regional dummy, rural location, percent of weekdays in the month, and sine and cosine terms for seasonal variation.
The multiple correlation coefficients were nearly identical all four years (multiple r2 between 0.249 and 0.273), as were most of the predictors’ coefficients. However, the intercept term declined from 0.277 to 0.147 while the coefficient for the percent of weekdays term increased from -0.22 to -
0.04.
Conclusion: Length of service was the major predictor of ACH, but changes in ALOS between
2004 and 2007 did not explain the reduced ACH rate during that time period, and in fact, predicted an average increase of 2%. Agency care practices and other factors that are not captured in the model would explain most of the variance in risk adjusted
ACH rates. Changes in agency practice over time, such as improved education of patient and caregiver in disease management, may be responsible for improvement in ACH rates. We speculate that more consistent management of patients on weekends may have contributed to the improvement.
Implications for Policy, Delivery or Practice:
Home health agencies can make significant reductions in ACH during home care. We do not know if these reductions persist after discharge from the service.
Funding Source(s): CMS
Theme: Long Term Care
● Functional Status Decline among Home
Health Care Patients with Heart Failure
Tanya Scharpf, Ph.D., M.S.; Jennifer Riggs, R.N.,
M.S.N.; Elizabeth Madigan, Ph.D., R.N., F.A.A.N.
Presented by: Tanya Scharpf, Ph.D., M.S., Data
Analyst, Frances Payne Bolton School of Nursing,
Case Western Reserve University, 10900 Euclid
Avenue, Cleveland, OH 44106, Phone: (216) 368-
0593; Email: tanyascharpf@hotmail.com
Research Objective: Patients with heart failure
(HF) face substantial and progressive decline in functional status. Post-acute follow up in the home is one strategy to reduce the burden of HF on the
US health care system. Home health care is the most common type of post-acute care for patients with HF. There is little evidence on the most effective ways to reduce functional status decline for HF patients receiving home health care. The purpose of the present study is to describe functional status decline among patients with HF receiving home health care and determine the explanatory factors associated with functional decline among these patients.
Study Design: Using the population of patients with HF receiving home health care in 2005, the following data sets were merged: Outcome and
Assessment Information Set (OASIS) providing demographic and clinical data, home health standard analytic file for home health care claims information, Provider of Service file to describe home health care agency characteristics, and Area
Resource File to describe the geographic area factors. There were 83,748 patients in the analytic data set with mean age of 81, 62% female, 83% white, 13% black, 1% Hispanic and 3% other racial and ethnic minorities or unknown. At baseline, 90% had dyspnea, 30% hypertension, 38% urinary incontinence, 36% cognitive functioning impairment,
18% depressive symptoms, while over 90% had 10 or more comorbidities. Functional status dependency at baseline increased with increasing age for all four racial categories (white, black,
Hispanic, and other minority or unknown) with
Hispanics having the highest baseline dependency for all four age groups.
Population Studied: The 2005 national population of Medicare-certified home health agency patients with a primary diagnosis of heart failure.
Principal Findings: There were 11,540 (14%) HF patients that experienced functional decline while receiving home health care during 2005. Of all functional status declines, 4% occurred among young patients (<65 years old); 13% among middle aged patients (65-74 years of age); 37% among old patients (75-84 years of age); 46% among the very old (85 and older). Functional decline by race was comparable to the patient population distribution
(e.g. 84% of declines among whites, 13% among blacks, <1% among Hispanics, and 3% other minority or unknown). Using logistic regression to predict functional decline, there were a number of factors significantly associated with increased probability of functional status decline: being female
(OR=1.06), increasing age(OR=1.02), increasing urinary incontinence(OR=1.16), poorer cognitive functioning(OR=1.09), increased levels of IADL dependency at baseline(1.32), and poorer ability to administer oral medications at baseline(2.16).
Conclusion: There are a substantial number of HF patients who have a functional status decline, likely related to disease progression. The most important factors for functional decline are increased inability to self-administer oral medications and increased levels of IADL dependency at baseline.
Implications for Policy, Delivery or Practice:
Much of home health care practice focuses on ADL performance where the present results suggest that focus on IADLs and medication self-management are also important for patients with heart failure.
Funding Source(s): NHLBI
Theme: Prevention and Treatment of Chronic
Illness
● Hiring & Retaining PCAs in a Consumer-
Directed Program
Karen Schneider, Ph.D.; James Maxwell, Ph.D.;
Jaya Mathur, B.A.; Tom Mangione, Ph.D.; Jack
Boesen, Esq.; Christine Bishop, Ph.D.
Presented by: Karen Schneider, Ph.D., Senior
Research Scientist, Health Care Policy &
Management Research, JSI Research & Training
Institute, Inc., 44 Farnsworth Street, Boston, MA
02210, Phone: (617) 482-9485; Email: kschneider@jsi.com
Research Objective: The Massachusetts Personal
Care Attendant or PCA program funds home-based care for elderly and disabled consumers. Given that it is a consumer-directed program, consumers are responsible for hiring and training workers, and they can hire anyone to serve as their PCA, regardless of prior training or level of experience. Anecdotal evidence suggests that consumers have difficulty finding and retaining PCA workers. The purpose of this project is to evaluate the extent to which hiring and retaining workers are issues in the consumerdirected PCA program.
Study Design: Questions were taken from evaluations of other state PCA programs and prior research on home-based care. A random sample of consumers was drawn from the PCA program consumer population. For consumers less than 18 years of age or those incapable of completing the survey on their own, a proxy respondent completed the interview. Survey data were analyzed in SAS.
Chi-square tests for categorical variables and ttests for continuous variables were used to test for significant differences, p less than 0.05.
Population Studied: The final sample consisted of
502 elderly and disabled consumers in the
Massachusetts PCA program.
Principal Findings: Thirty-eight percent of consumers reported having at least one PCA leave in the past year. The overall annual turnover rate was 15 percent. Turnover rates for varied by: length of time in the program, LTE 2 years, 11 percent vs.
3-5 yrs, 15 percent vs. GT 5 years, 20 percent; number of PCAs working for consumer, 1 PCA, 12 percent vs. 2 PCAs, 17 percent vs, 3 or more
PCAs, 22 percent; number of hours of PCA time per week, LT 40 hours, 14 percent vs. GTE 40 hours,
20 percent; and consumer-PCA relationship, family member, 12 percent vs. friend/neighbor, 12 percent vs. stranger/acquaintance, 20 percent. Additionally,
44 percent reported ever having a PCA leave the job and 36 percent reported ever having to fire a
PCA. Seventy-two percent reported that it took LT one month to find a new PCA; more specifically, nearly half reported that it took LT one week. Fortysix percent used family/friends to help recruit PCA workers. One-quarter reported any difficulty the last time they had to hire a PCA.
Conclusion: Results from this survey documented that, even though recruitment and retention problems are still critical issues requiring policy attention, they are not as severe as other segments of the long-term care workforce. Hiring family members or friends improves worker retention.
However, those with more complex health needs, as measured by number of PCAs providing care and number of hours of PCA time per week, reported greater problems with turnover, which is problematic since a stable workforce is more critical to these people.
Implications for Policy, Delivery or Practice:
Enhanced functional skills training should be provided to consumers so they can better organize and train their workers, which could then impact retention. Recent negotiations between the state and PCA worker union have resulted in a wage increase. Other benefits for workers should be negotiated, such as training and health insurance, which should also help create a more stable workforce.
Funding Source(s): PCA Quality Home Care
Workforce Council
Theme: Health Care Workforce
● Establishing All-Payer Claims Databases:
Case Studies of Eight States
Craig Schneider, Ph.D.
Presented by: Craig Schneider, Ph.D., Director of
Healthcare Policy, Massachusetts Health Data
Consortium, 460 Totten Pond Road, Suite 690,
Waltham, MA 02451, Phone: (781) 419-7810;
Email: cschneider@mahealthdata.org
Research Objective: This paper studies the experiences of eight states that have implemented all-payer claims databases, or that are in the process of doing so. The analysis examines the states on the dimensions of history of implementation, financing, legal framework, and promotion and stakeholder relations.
Study Design: This paper is adapted from a series of presentations given at the National Association of
Health Data Organizations (NAHDO) annual conference in October 2007, the National
Conference on All-Payer Claims Databases in April
2008, and the NAHDO annual conference in
October 2008. The author reorganized the presentations into the history/financing/legal/promotion case study framework, and then gave the state officials the opportunity to review and edit the information.
Population Studied: The population studied is the eight states addressed in the case studies:
Kansas, Louisiana, Maine, Maryland,
Massachusetts, New Hampshire, Utah, and
Vermont.
Principal Findings: The experiences of these eight states indicate that there are four key decisions that states wishing to establish all-payer claims databases have to make: (1) the legal authority for the implementing agency or organization; (2) the governance structure for this entity; (3) how to finance the program; and (4) the promotional strategy to get the program implemented. Of course, this premise assumes that the state accepts the rationale for establishing an APCD – the health data organization needs to have a clear rationale
for going forward with the effort of establishing an all-payer claims database, since, as the case studies describe, this is not an easy process.
Conclusion: Based upon the case study narratives and analysis of the four key decisions, our conclusions take the form of recommendations to states that are contemplating establishing all-payer claims databases. The legal authority should be designed to fit the needs and culture of the particular state, but the simplest implementation would be to give the authority to an existing state agency that handles healthcare data. The state agency that wishes to establish the all-payer claims database should draft model legislation for the state’s legislature to consider, including the desired governance and policy. Creating Websites for public reporting of the data should be thought of as a requirement, and cost and quality data should be as transparent as possible. The financing question is likely dependent on the political environment in each state, but if politically feasible, the most stable source of revenue would be a per-claim assessment. The promotional strategy is important and cannot be neglected. The recommendations are to involve stakeholders early, give them an opportunity to air their views, and once the design of the program is clear, prepare concise fact sheets and other materials that communicate the benefits of establishing an all-payer claims database.
Implications for Policy, Delivery or Practice:
Many states are launching health reform efforts and determining that establishing an all-payer claims database is a critical component of this initiative.
The lessons learned and best practices identified from studying these eight states will help inform the implementation efforts of states that are contemplating establishing such databases, and will ultimately contribute to cross-state coordination and harmonization to facilitate multi-state analyses of claims data.
Funding Source(s): CWF
Theme: Medicaid, SCHIP and State Health Reform
● Uninsured but Eligible for Medicaid; Using a
Robust Statewide Survey to Determine
Outreach Needs
Joseph Schuchter, M.P.H.; Gerry Fairbrother, Ph.D.
Presented by: Joseph Schuchter, M.P.H.,
Epidemiologist, HPCE, Cincinnati Children's
Hospital Medical Center, 3333 Burnet Avenue, MLC
7014, Cincinnati, OH 45229-3039, Phone: (513)
636-2206; Email: joseph.schuchter@cchmc.org
Research Objective: Public health insurance programs consistently fail to enroll all who are eligible. This includes the Medicaid and SCHIP programs for children. Despite a streamlined and simplified State Medicaid program, in Ohio in 2007 over 7% of all children were still uninsured. To maximize effectiveness of current programs, and for further expansions of eligibility to be realized, both legislators and county-level enrollment offices need to understand the characteristics of those children who are eligible yet uninsured. As well, new SCHIP legislation provides incentives for States and counties to enroll uninsured but eligible children.
Hence, a demand for such information is anticipated. The reasons for being uninsured are unclear, as is the degree to which uninsured but eligible children have unmet needs. Our objective was to assess child and household-level demographic and health status characteristics associated with being uninsured but eligible for
Medicaid.
Study Design: This study utilized data from the
Ohio Family Health Survey (OFHS), which was a random digit dialed telephone survey conducted from July to Dec 2008. The full dataset contains child and household-level information from over
15,000 surveys collected on children. Using 85% of the records in this full dataset for which quality checking was complete, we examined child-level demographic and household characteristics associated with participation in Medicaid. We also examined county-level differences as a potential planning tool and performance measure for counties. Non-participation was defined as being uninsured at the time of survey of having household income below 200% FPL, the upper limit for coverage in Ohio. Logistic regression was used to calculate univariate and multivariate odds ratios.
Population Studied: Children eligible for Medicaid who were uninsured.
Principal Findings: Approximately one-third of all uninsured children in Ohio are eligible for Medicaid.
In univariate analysis at the child level, white race and not having a special healthcare need were the only factors significantly associated with nonparticipation. In univariate analysis at the household level, having two adults in the household, income above 100% of poverty level, highest level of education of 4 years of college or beyond, and parent having Medicare, Job-based coverage or being uninsured were all significantly associated with non-participation. In multivariate analysis, only household income (OR 1.24, 95% CL 1.01-1.51) and adult insurance status (OR 1.60, 95% CL 1.46-
1.75) remained significantly associated with nonparticipation. The proportion of uninsured children who were eligible varied dramatically at the county level. Among those eligible but not-participating in
Medicaid, only 8% gave a reason indicating it was their choice. Still, children not-participating experienced much greater unmet needs, ranging from forgoing an annual well-child visit (OR 3.34,
95%CL 2.32-4.80) to having any unmet need (OR
7.29, 95% CL 4.12-12.90).
Conclusion: Childrens' participation is associated with household income and parents insurance status.
Implications for Policy, Delivery or Practice:
Statewide surveys examining children’s health insurance provide valuable information for both
legislators and county enrollment offices. The results indicate that outreach programs need to be county-specific, and should focus on reducing stigma and helping parents obtain coverage.
Funding Source(s): ODJFS
Theme: Medicaid, SCHIP and State Health Reform
● Increasing Financial Burden in Patients with
Metastatic Colorectal Cancer
Kathy Schulman, M.A.; Gregory Lenhart, M.S.;
Amelito Torres, M.B.A.; Daniel Huse, M.A.;
Presented by: Kathy Schulman, M.A., Lead
Researcher, Healthcare, Thomson Reuters, 150
Cambridge Park Drive, Cambrige, MA 02140,
Phone: (978) 779-2981; Email: kathy.schulman@thomsonreuters.com
Research Objective: Patients are experiencing increased financial burden as health care costs continue to rise faster than incomes and employers shift health care costs to employees and retirees.
Out of pocket expenditures among non-elderly patients with employer sponsored insurance rose
14.6% from $1181 in 2001 to $1354 in 2004 and are projected to increase annually by 6-7% through
2016. While these studies have focused on the general patient population, few studies have examined the healthcare financial burden in patients with life threatening illness. The objective of the current study was to measure out of pocket costs in elderly and non-elderly patients newly diagnosed with metastatic colorectal cancer
(mCRC).
Study Design: Patients diagnosed with local and widespread mCRC between 2000 and 2006 were extracted from the Marketscan research databases.
Patients were required to be continuously enrolled in Marketscan for a period of at least 6 months prior and one year after diagnosis except in those cases where death was documented on a hospital claim.
Patient and health plan expenditure in the one year post index was calculated and reported in 2007 dollars. Out of pocket expenditure was defined as the patient share of payments for healthcare services provided. It did not include patient share of healthcare premium, over-the-counter drugs and supplies, or the value of the patient’s or caregiver’s time. Generalized linear models were developed to adjust for differences over time in patient demographics, degree of metastases, comorbidity and death.
Population Studied: Active employees and dependents with employer sponsored private insurance and Medicare eligible retirees with employer sponsored supplemental coverage.
Principal Findings: We identified 5,806 patients with mCRC; 3,165 patients were commercially insured and 2,541 were Medicare eligible retirees with supplemental employer insurance. Average age in commercially insured patients was 54.2 (SD
7.7); 45.5% were female; 77.9% had colon cancer and 56.2% had widespread metastases. Out of pocket expenditure in commercially insured patients more than doubled between 2000 ($2,086) and
2006 ($4,387), growing at an annual adjusted rate of 14.6% per year (p<.0001). Meanwhile the percentage of patients with a traditional indemnity or PPO plan design grew from 47.2% in 2000 to
66.9% in 2006. Average age in Medicare eligible patients was 75.7 (SD 6.8); 51.4% were female;
81.8% had colon cancer and 54.8% had widespread metastases. Out of pocket expenditure in these patients increased 74.4% between 2000
($1,288) and 2006 ($2,247), growing at an annual adjusted rate of 13.3% per year (p<.0001). No significant linear trend was observed in the proportion of expenditure incurred by patients regardless of payer.
Conclusion: Out of pocket expenditure for healthcare services in patients with mCRC is 3 times that observed in the general patient population and is increasing at twice the observed annual rate
Implications for Policy, Delivery or Practice:
One in five Americans were in families with problems paying medical bills in 2007. Patients with life threatening illness are at increased risk for such health related financial burden with potentially adverse implications for access, adherence and overall quality of life.
● Universal Coverage of Children: Benefits for
Children & Tradeoffs for Adults
Nathaniel Schuster; Matthew Davis, M.D., M.A.P.P.
Presented by: Nathaniel Schuster, Child Health
Evaluation & Research Unit, Divisions of General
Pediatrics & General Internal Medicine, University of Michigan Medical School, 507 Glen Court,
Apartment 4, Ann Arbor, MI 48104, Phone: (952)
237-8648; Email: nschu@med.umich.edu
Research Objective: In his campaign for president,
Barack Obama proposed mandating insurance coverage for all children in the United States. This policy goal is likely to receive attention from lawmakers and policymakers in the near future.
Although coverage is likely to benefit previously uninsured children, consequences for adults are not frequently discussed as part of the policy dialogue.
The purpose of this study is to anticipate tradeoffs involved with universal coverage of children through a mixed system of public and private coverage similar to the system currently in place in the US.
Study Design: We searched PubMed, JSTOR, the
National Bureau of Economic Research’s (NBER) working papers, the Kaiser Family Foundation’s reports, and the reference lists of articles for papers published between May 1996 and August 2008.
We reviewed studies of responses to public health insurance expansions for children and synthesized the findings to project the effects of a universal insurance mandate for children. Specifically, we
focused on studies regarding: decisions by employers about whether to offer employer sponsored insurance (ESI); decisions by employees about whether to purchase ESI for themselves and their children; and insurance cost-sharing arrangements between employers and employees.
Among studies of national public insurance expansions (Medicaid expansions and the State
Children’s Health Insurance Program), we included studies in which econometric microsimulation models were reported.
Population Studied: We reviewed studies that used national survey data or microsimulation models regarding the cost and accessibility of health insurance for children, parents, and employers.
Principal Findings: Based on empiric data from
Medicaid expansions and SCHIP, universal children’s coverage would be expected to improve children’s health outcomes and health care access and affordability, without causing much direct substitution crowd-out (ie, switching from private coverage to public coverage). However, we expect that mandating insurance for children will be necessary to ensure that all children have coverage, because otherwise children in families at higher income levels may not seek coverage. Such a mandate would be expected to lead to “spillover effects” in which parents would decline ESI and become uninsured as their children were enrolled in public plans – when, in the absence of the new child plans, the parents would have remained insured. The published literature on public programs also suggests that covering children without mandating coverage of adults will cause increased premiums and cost-sharing for employersponsored family coverage and general erosion of
ESI through decreased employer offer rates.
Consequently, covering children through a universal mandate would be expected to lead to higher rates of uninsurance among adults.
Conclusion: While covering all children appears to be a worthwhile policy goal, such a strategy may expand coverage of generally low-risk, healthy children while reducing coverage among generally higher-risk adults.
Implications for Policy, Delivery or Practice:
Policymakers must be aware of these expected tradeoffs of universal mandated coverage for children and design policy to prevent these consequences or to compensate for them prior to implementing such initiatives.
Funding Source(s): University of Michigan Medical
School’s Summer Biomedical Research Program
● Evaluation of Pay for Performance & a Multi-
Disciplinary Model for Coordinating Primary
Care with Disease Management on Incentivized
& Non-Incentivized Quality Measures & Costs
Alyson Schuster, M.B.A., M.P.H.; Peter, Fagan,
Ph.D.; Jill Marsteller, Ph.D., M.P.P.; Cynthia Boyd,
M.D., M.P.H.; Michael Griswold, Ph.D.; Shannon
Murphy, M.A.; Linda Dunbar, R.N., Ph.D.
Presented by: Alyson Schuster, M.B.A., M.P.H.,
Research Associate, Research & Clinical
Outcomes, Johns Hopkins HealthCare LLC, 6704
Curtis Court, Glen Burnie, MD 21060, Phone: (410)
424-4958; Email: aschuster@jhhc.com
Research Objective: The study examined the effects of a combined pay-for-performance (P4P) and on-site disease management (DM) intervention
(P4P+DMinPC) upon both incentivized quality measures and non-incentivized quality measures.
The non-incentivized measures ranged from ones related to the incentivized index conditions, e.g. asthma, CAD, CHF, diabetes, to generic measures of total population health, e.g. ED utilization. Lastly, the study sought to examine the effects of the intervention on total healthcare costs.
Study Design: Employing administrative claims data from a Medicare Advantage (MA) plan members over age 65 for the period from January
2004 through March 31, 2007, we compared the quality of care and total medical costs for MA patients in nine Intervention practices with the
P4P+DMinPC Intervention to 945 Comparison practices without the Intervention. Among the
Comparison practices, 753 were composed of only one or two providers. We conducted a longitudinal data analysis using logistic regression with a binary outcome and an auto-regressive correlation structure. We examined the effects with both an
‘intention-to-treat” model that assigned all patients at each practice to the Intervention or Comparison group, and a per-protocol model that included patient specific information on DM enrollment status.
Population Studied: Individuals who were enrolled in MA and over the age 65 from January, 2004 through March, 2007 were the population studied
(N = 148,012; Intervention = 11,416 and
Comparison = 136,596). The Intervention and
Comparison primary care practices were located in
Tennessee, Alabama and Texas. The participants averaged 19.1 months of MA enrollment; were
58.9% female and had an average age of 72.9 years. Sixty nine percent of the MA participants had at least one index chronic condition; 32.6% had multiple index chronic conditions. Approximately
40% were in a DM program; two thirds of these for nine months or longer. Racial data were not available.
Principal Findings: For 4 of the 7 incentivized measures there was a general absolute improvement in quality for the Intervention group but the improvements were not significantly different from the Comparison group. For the 5 non-incentivized measures, there was no significant effect from receiving care at an Intervention practice or a Comparison practice. In terms of total
healthcare costs there were no differences in the cost trends between the two groups.
Conclusion: P4P+ DMinPC did not show significant positive effect relative to Comparison practices in this study. Improvements were observed in the quality of health care in both the
Intervention and Comparison groups, possibly reflecting secular trends in quality improvement.
Other factors such as MA patients comprising on average less than 10% of Intervention practices’ panels may have reduced the saliency of financial incentives. Concerning the non-P4P measures examined, there were neither a positive halo effects, nor negative unintended consequences.
Implications for Policy, Delivery or Practice: 1)
Financial incentives in quality improvement initiatives must be large enough to effect behavioral changes in providers. 2) “Unintended consequences” including possible racial disparities of P4P programs should continue to be examined.
3) Effects on quality and costs of disease management programs are examined more heuristically with comparison group designs.
Funding Source(s): Health Industry Forum
● Defining New Categories of Pregnancy
Intention in African-American Women
Alan Schwartz, Ph.D.; Nadine Peacock, Ph.D.;
Kenya McRae, M.P.H.; Rachel Seymour, Ph.D.;
Melissa Gilliam, M.D.; Linda Factor, M.P.H.
Presented by: Alan Schwartz, Ph.D., Associate
Professor, Medical Education, University of Illinois at Chicago, 808 South Wood Street (M/C 591),
Chicago, IL 60612, Phone: (312) 996-2070; Email: alansz@uic.edu
Research Objective: Pregnancies are categorized as intended, mistimed or unwanted, depending on the woman’s preferences just prior to conception.
Unintended (mistimed and unwanted) pregnancies are associated with a variety of adverse outcomes.
Our objective was to refine the definition and measurement of pregnancy intentions of young
African-American women in a way that reflects the complex interplay of motivations, preferences, desires and intentions that underlie contraceptive and reproductive behaviors.
Study Design: This interdisciplinary, multi-method study employed surveys, in-depth qualitative interviews, and Q-methodology -- a technique for eliciting systematic information on subjective viewpoints. For the Q-sort task, we compiled a set of statements (a concourse), drawn from prior qualitative studies, that reflect attitudes and preferences regarding pregnancy (e.g. “I want to get pregnant soon” or “if I got pregnant, my partner would be upset”.) Participants sorted the statements into a fixed distribution on a continuum from “least true for me” to “most true for me”.
Women participated in three sessions over a twoyear period. By-person factor analysis was conducted on the Q-sort data to derive new categories of pregnancy intention.
Population Studied: African American women age
15-25 were recruited at three clinics serving lowincome populations in Chicago; two clinics were located in communities on the city’s west and south sides, and the third was housed in a north side high school. Eligibility criteria differed slightly for adults and teens. The adult sample included women who were sexually active and neither surgically sterilized nor on long-acting methods of contraception (IUD or
Norplant.) 15-17 year-olds were not required to be sexually active at the time of recruitment.
Principal Findings: The factor analysis yielded six factors, which represent points of view about future pregnancy. One is a “pregnancy seeking” factor, one reflects ambivalence and lack of perceived control over reproduction, and four reflect a variety of “pregnancy avoidance” points of view, reflecting different constellations of perspectives on partner and family support, value of planning, perceived control over reproduction, and desired timing of future pregnancies. The distribution of the factors differs by age group, as well as for women who did and did not become pregnant during the course of the study.
Conclusion: This study yielded an inductivelyderived set of intendedness categories that, in comparison to more traditional “intention” questions from national surveys, captures more of the nuanced and evolving points of view of participants.
Implications for Policy, Delivery or Practice: A new measure of pregnancy intention that more accurately reflects the complexity of women’s lives should prove useful both in a clinical setting and for surveillance purposes. For the clinician, it can be used as a screening tool to help tailor family planning services to the unique needs of each woman. If used in surveys like the National Survey on Family Growth, the new measure may prove better at predicting adverse social and health outcomes, and thus inform more efficacious interventions.
Funding Source(s): CDC, Association of Schools of Public Health
Theme: Disparities
● Ethnicity, Acculturation & Clinical Values
Alan Schwartz, Ph.D.; Memoona Hasnain, M.D.,
M.H.P.E., Ph.D.; Jorge Girotti, Ph.D.; Angela Bixby,
Bianca Castellano; Michelle Garrido, Joe Gomez,
B.A.
Presented by: Alan Schwartz, Ph.D., Associate
Professor, Department of Medical Education,
University of Illinois at Chicago, 808 South Wood
Street, 986 CME, Chicago, IL 60612, Phone: (312)
996-2070; Email: alansz@uic.edu
Research Objective: To describe differences in clinical values held by Caucasian, Hispanic, and
African-American patients seeing a common pool of
family physicians, and to relate clinical values to health status.
Study Design: Ongoing prospective multisite patient survey including previously published measures of six clinical values (behaviors patients prefer in their physician; Schwartz et al 2006), SF-
12v2 health status items (Ware et al. 2007), and, for
Hispanic patients, the Bidimensional Acculturation
Language Proficiency Subscale (BAS/LP; Marin &
Gamba, 1996).
Population Studied: Caucasian (n=62 sampled),
Hispanic (n=115), and African-American (n=54) patients at two ambulatory clinic sites in Chicago staffed by physicians from the UIC Department of
Family Medicine. On the basis of the BAS/LP,
Hispanic patients were further subdivided into patients with Hispanic-only acculturation (n=45) and those with both Hispanic and non-Hispanic acculturation ("bicultural") (n=67).
Principal Findings: Clinical values differed significantly between ethnic and cultural groups.
African-American patients were significantly more likely than other patient groups to want their physician to consider their religious beliefs.
Hispanic patients were significantly less likely than other groups to want their physician to be familiar with herbal medicine. Among Hispanic patients, acculturation was a key predictor of both clinical values and health status. Bicultural Hispanic patients did not differ significantly from Caucasian or African-American patients in overall physical or mental health, or most clinical values. On the other hand, Hispanic-only acculturated patients were significantly less likely than any other group to want their physician to consider their religious beliefs and to expect their doctor to address their expectations for care. These findings persisted when controlling for gender and for clustering within physician.
Regression models predicting health scores from ethnicity/culture, gender, and clinical values found that African-American and Hispanic-only acculturated Hispanic patients had lower physical health scores than Caucasian or bicultural Hispanic patients; Hispanic-only acculturated Hispanic patients had lower mental health scores than other groups. Controlling for ethnicity/culture, patients who more strongly valued their physician addressing their expectations for care reported better physical and mental health.
Conclusion: Ethnic differences in clinical values and health status were found in patients of this group of physicians. Notably, among Hispanic patients, those with bicultural acculturation reported a different pattern of responses than those with
Hispanic-only acculturation.
Implications for Policy, Delivery or Practice:
Research into health disparities, particularly involving Hispanic patients in the United States, must go beyond consideration of ethnic categories alone and consider more nuanced concepts of culture and acculturation.
Funding Source(s): RWJF
Theme: Disparities
● A New Resource for Public Health Workforce
Researchers & Policymakers
F. Douglas Scutchfield, M.D.; Cynthia Lamberth,
M.P.H., C.P.H.; Susan Webb, M.S.W., M.P.H.;
Carol Gotway Crawford, Ph.D.
Presented by: F. Douglas Scutchfield, M.D.,
Professor of Health Services Research & Policy,
Health Services Research, University of Kentucky
College of Public Health, 121 Washington Avenue,
Suite 105, Lexington, KY 40536-0003, Phone:
(859) 218-2024; Email: scutch@uky.edu
Research Objective: To initiate and coordinate strategic efforts to improve the public health workforce through research.
Study Design: Creation of the Center of
Excellence in Public Health Workforce Research and Policy to provide a resource and forum for interdisciplinary researchers examining relationships between the workforce and public health incentives, costs, quality, accessibility, delivery, and outcomes of services.
Population Studied: U.S. Public Health Workforce
Principal Findings: To achieve ESPHS #8 (Assure a competent public and personal health care workforce), new directions in workforce analysis are required. Coordination of existing and new data, a venue for sharing research findings, and provision of technical assistance will enhance organized public health workforce research efforts.
Conclusion: There is need for strategic coordination of interdisciplinary research efforts regarding the public health workforce, including issues of monitoring, diversity, effectiveness and health impact, analysis of labor market forces, and education, training, and credentialing.
Implications for Policy, Delivery or Practice:
Research outcomes fostered by the Center can provide a basis for relevant enhancements and changes in public health policy related to the nation's public health workforce and the delivery of the 10 essential public health services.
Funding Source(s): CDC
Theme: Health Care Workforce
● Racial Disparities & Out-of-Pocket Payment in
Drug Utilization in Patients with Diabetes in the
United States
Arpamas Seetasith, B.Sc.; James Xuejie Zhang,
Ph.D.
Presented by: Arpamas Seetasith, B.Sc., Graduate
Student, Pharmacy, Virginia Commonwealth
University, 1112 East Clay Street, McGuire Hall
Annex, Room 218, Richmond, VA 23298-0533,
Phone: (757) 581-1767; Email: seetasitha@vcu.edu
Research Objective: Using a nationally representative sample of patients with diabetes in the United States, we sought to assess racial disparities in drug utilization in patients with diabetes measured through statin usage, adjusting for patients’ insurance status and other covaraites.
Study Design: We analyzed patients with diabetes in the 2003 Medical Expenditure Panel survey
(MEPS), a nationally representative survey in the
United States which was linked to drug utilization files. In order to control potential confounders due to insurance status, we categorized patients with diabetes into six mutually exclusive groups by public, private insurance categories, and prescription drug insurance coverage: Medicare,
Medicaid, Medicare and Medicaid (dual eligibility), private insurance, other public, and no insurance.
We performed bivariate chi-square tests to assess the association between race and statin use during the year. We further performed multivariate logistic regression analysis to assess the effect of race/ethnicity on statins usage for patients of different races/ethnicity, controlling for a quadratic form of age, gender, ethnicity (Hispanic), marital status, educational attainment, family size, family income relative to federal poverty line, prescription drug insurance coverage, completion of questionnaire by a proxy, and comorbid conditions including asthma, hypertension, heart disease, stroke, emphysema, joint pain, and arthritis. We evaluated the average out-of-pocket payments of statins per prescription in order to assess the price differences faced by patients by insurance status.
Population Studied: We studied 1,708 civilian, noninstitutionalized patients with diabetes aged between 20 and 85.
Principal Findings: Among 1,708 adult patients with diabetes, 369 (21.6%) were African American,
66 (3.9%) Asian, 396 (23.2%) Hispanic; and a minority, 47 (2.2%) were classified as other races.
The insurance groups included 585 (34.3%) patients with Medicare, 176 (10.3%) Medicaid, 239
(14.0%) dual eligible, 524 (30.7%), private insurance, 154 (9.0%) with no insurance, and 30
(1.8%) other public insurance. The mean out-ofpocket payment per prescription of statins was
$61.4 (SD $62.5) for Medicare patients, $ 24.0 (SD
$43.7) for Medicaid patients, $ 25.4 (SD $49.0) for patients with dual eligibility, $35.2 (SD $33.5) for those with private insurance, $76.7 (SD $47.8) for those without insurance, and $83.9 (SD $90.5) for others. In bivariate analysis, statin usage was found to be significantly different across races (p = 0.020) using chi-squared tests. In multivariate regression analysis, compared to White patients, African
American patients were less likely to use statins
(adjusted OR 0.57, 95% CI 0.43-0.76, p < 0.0001).
Asian and Hispanic patients were marginally less likely to use statins (adjusted OR 0.60, 95% CI
0.34-1.06, p = 0.078; and adjusted OR 0.75, 95%
CI 0.56-1.00, p = 0.055, respectively).
Conclusion: Racial/ethnic disparities are observed in drug utilization of patients with diabetes after adjusting for insurance status. African American patients were least likely to use statins compared to other races/ethnicity.
Implications for Policy, Delivery or Practice:
Policymakers need to consider individualized intervention to maximize benefits statins use to improve access to drugs and the overall welfare for patients with diabetes, and to eliminate racial/ethnic disparities.
Funding Source(s): Pharmaceuticsl Economics and Policy (PEP) Program: Graduate Fellowship
Theme: Disparities
● Barriers to Care Prospectively Predict Primary
Care Characteristics for Vulnerable Children with Asthma
Michael Seid, Ph.D.; Lisa Opipari-Arrigan, Ph.D.;
Adam Carle, Ph.D.
Presented by: Michael Seid, Ph.D., Director,
Health Outcomes & Quality of Care Research &
Professor of Pediatrics, Pulmonary Medicine &
Center for Health Care Quality, Cincinnati
Children's Hospital Medical Center, 3333 Burnet
Avenue, MLC 7014, Cincinnati, OH 45229, Phone:
(513) 803-0083; Email: Michael.Seid@cchmc.org
Research Objective: We sought to test the following two hypotheses in a sample of vulnerable children with asthma. 1) Barriers to care would prospectively predict primary care experiences after accounting for demographic characteristics, disease severity and financial, potential and realized access. 2) Barriers related to the process of care would account for more variance in primary care experiences than barriers related to accessing care.
Study Design: Bilingual, bicultural interviewers administered surveys in participants’ homes at baseline, 3-month and 9-month follow-up. Validated instruments were used to measure barriers to care
(Barriers to Care Questionnaire) and primary care experiences (Parent’s Perceptions of Primary Care measure).
Population Studied: Children (n = 252; ages 2-14 years) with persistent asthma and their parents
(83% Hispanic, 56.6% monolingual Spanish; 72.6% mother had less than a high school diploma) were recruited primarily from urban, Federally Qualified
Health Centers. Asthma severity was 27% mild persistent, 40.9% moderate persistent, and 32.1% severe persistent. At baseline, most children
(68.7%) had health insurance (financial access), an identified primary care provider (91.7%) (potential access) and had not experienced forgone care
(79.8%) (realized access).
Principal Findings: Consistent with our first hypothesis, generalized estimating equations accounting for the repeated measures over time showed that fewer barriers to care (Barriers to Care
Questionnaire) prospectively predicted better
primary care experiences (Parents Perceptions of
Primary Care), after controlling for demographics, disease severity, and financial, potential and realized access (p < .001). While having insurance
(standardized beta = -.12) and an identified care provider (standardized beta = -.18) did significantly predict better primary care experiences; barriers to care (standardized beta = .47) accounted for the most variance in outcome. In terms of unstandardized regression coefficients, a 1-point change on the Barriers to Care Questionnaire predicted a .69-point change in Parent’s
Perceptions of Primary Care. Consistent with our second hypothesis, fewer process of care barriers
(p < .001), but not access barriers (p = NS), predicted better primary care experiences after controlling for, demographics, disease severity and financial, potential and realized access. A 1-point change in process of care barriers predicted a .60point change in Parent’s Perceptions of Primary
Care.
Conclusion: For vulnerable children with asthma, barriers to care prospectively predict primary care experiences after accounting for demographic factors, disease severity and financial, potential and realized access. Furthermore, barriers related to the process of care as opposed to those related to accessing care, are central in predicting primary care experiences overtime.
Implications for Policy, Delivery or Practice:
Primary care is important, especially for children with special health care needs. These findings offer insight into how and why disparities in primary care experiences develop. As barriers are clearly modifiable, these findings can be used to inform practice and policy interventions designed to reduce health care disparities among vulnerable children with asthma. While access to care is necessary, it is not sufficient. Barriers in the actual delivery of care are important to address.
Funding Source(s): HRSA
Theme: Child Health
● Sharing the Cost of Reform in Massachusetts
Robert Seifert, M.P.A.; Paul Swoboda, M.S.
Presented by: Robert Seifert, M.P.A., Senior
Associate, Center for Health Law & Economics,
University of Massachusetts Medical School, 529
Main Street, 3rd Floor, Charlestown, MA 02129,
Phone: (617) 886-8065; Email:
Robert.Seifert@umassmed.edu
Research Objective: The concept of "shared responsibility" is a crucial underpinning of the health insurance expansions under the 2006
Massachusetts health care reform law. This research evaluates how the cost of coverage is shared among employers, individuals and government, pre- and post-reform. It also assesses the distribution of spending for uncovered services, as they are part of the spending system and should theoretically offset changes in spending on coverage.
Study Design: Compared aggregate spending by employers, individuals and government on insurance premiums, other expenses related to covered services, and expenses related to uncovered services in CY 2005 (pre-reform) and
2007 (first full year post-reform). Analysis used empirical premium and expense data, and interpolated estimates of values for which data were not available.
Population Studied: Massachusetts residents under age 65, and associated spending for health insurance and uncovered services.
Principal Findings: Total spending for coverage in
Massachusetts for the under-65 population was
$25.5 billion in 2007, up from $20.8 billion in 2005.
About half of the increase was attributable to premium increases independent of coverage expansion, 30% to enrollment in existing programs, and 8% to the introduction of the new publicly subsidized insurance program. Employers pay about half the costs of coverage: 49% in 2005 and
48% in 2007. The other half is divided roughly evenly between individuals and government. There was little change in the overall distribution between
2005 and 2007, with government's share growing slightly faster than the other two groups, but the basic relationship was unchanged. Payments to providers for uninsured people and other uncovered services declined by 43% from 2005 to 2007, with the government share by far the largest part of the decline.
Conclusion: The principle of shared responsibility has been sustained through the early phase of health care reform in Massachusetts. It is early in the reform era, however, and programs are still evolving, so potential future shifts bear monitoring.
Implications for Policy, Delivery or Practice: The
Massachusetts health care reform law struck a sensitive balance among constituencies with varied interests, and the future of the law depends in large part to maintaining that balance. A critical part of that is the concept of shared responsibility, that no one group should bear a disproportionate share of the expense. This analysis provides a baseline for policy makers to use in monitoring the finances of the reform law as it continues to unfold.
Funding Source(s): Blue Cross Blue Shield of
Massachusetts Foundation
Theme: Medicaid, SCHIP and State Health Reform
● Study of Trends in Radiation Exposure Due to
Diagnostic Imaging among Medicare
Beneficiaries from 1997 to 2007
Namrata Sen, M.H.A.; Joan Elizabeth DaVanzo,
Ph.D.; Susan Arday, B.S.P.H., M.H.S., C.H.E.S.;
Michael Pentecost, M.D.; Robert Lagalia; Mark
Anthony Zezza, Ph.D.
Presented by: Namrata Sen, M.H.A., Senior
Manager, The Lewin Group, 3130 Fairview Park
Drive, Suite 800, Falls Church, VA 22042, Phone:
(703) 269-5747; Email: Namrata.Sen@lewin.com
Research Objective: The increased use of diagnostic imaging requiring the use of “ionizing radiation” has raised concerns about both the long term impact of low level radiation exposure as well as cumulative “medical” radiation exposure. Since
2000, imaging services have been among the fastest growing services paid for by Medicare, with spending for advanced imaging services including computed tomography (CT) scans and nuclear medicine experiencing faster growth compared to other imaging services. The objective of this study is to determine trends in “annual radiation exposure” due to the various types of imaging exams among Medicare beneficiaries from 1997 through 2007.
Study Design: The study design is a retrospective series of cross-sectional analyses. Medicare claims for each year in the time period 1997 through 2007 will be used to examine annual trends in radiation exposure. Medicare claims for inpatient and outpatient hospital care, including emergency department care are used in this study. Using a defined list of procedure codes for diagnostic imaging services associated with exposure to ionizing radiation, we will identify all claims associated with the selected procedures for each year. Next, each beneficiary’s radiation exposure
“dose” will be computed by multiplying each of the selected procedures by the estimated milliseverts
(mSv) per exposure. We will determine the typical values for the procedures using the literature, since individual radiation doses differ because of differences in machines, and the individual’s metabolism. The summation of the radiation “dose” across all “exposed” beneficiaries represents the estimated cumulative exposure for a given year.
Beneficiaries identified as having a cancer diagnosis on radiology claims are treated separately.
Population Studied: Medicare beneficiaries having fee-for-service claims for a selected set of radiationbased imaging services from 1997 through 2007.
Principal Findings: For each year, we calculate descriptive statistics on the variables of interest, including: Number of beneficiaries exposed to each type of radiation-based imaging technology. We examine these groups by age, gender, geographic area, and clinical conditions identified through ICD-
9 diagnoses on the claims; The annual cumulative radiation exposure across all beneficiaries for each of the years. Similar to the first analysis, we will examine and compare the annual cumulative radiation exposure for each imaging service by age, gender, geographic area and clinical condition; and
Frequencies of the individual imaging services that expose beneficiaries to “ionizing radiation” for a sample of individual beneficiaries. Anecdotal evidence suggests that the time between receipt of multiple and/or concurrent imaging services is often too short for patient safety.
Implications for Policy, Delivery or Practice:
Diagnostic imaging services that are radiationbased have improved the care of a number of conditions prevalent among Medicare beneficiaries, such as cancer, brain disorders, and cardiovascular disease. Although the radiation exposure for a single individual from an isolated test is generally in a low range, Medicare beneficiaries can receive multiple imaging services within a short period of time. This study is important, as little is known about cumulative radiation exposure among
Medicare beneficiaries.
Funding Source(s): CMS
Theme: Quality and Efficiency: Measurement
● The Impact of Hematology/Oncology Bed
Management Strategies on Throughput: A
Simulation Model Approach
Shital Shah, Ph.D.; Alyssa, Howell, M.S.; Melinda
Dunham Noonan, M.S., R.N., C.N.A.A.; Henry
Fung, M.D.; Norma Melgoza, M.P.H.; J. Robert
Clapp, Jr., M.H.S.A., F.A.C.H.E.
Presented by: Shital Shah, Ph.D., Assistant
Professor, Health Systems Management, Rush
University, Rush University Medical Center, 1700
West Van Buren Street, 126B, Chicago, IL 60612,
Phone: (312) 942-7926; Email:
Shital_C_Shah@rush.edu
Research Objective: Growing health care demand, decreasing inpatient bed capacity, and inefficient processes can interfere with the ability to provide timely, appropriate, and effective patient care. To resolve these, effective bed management and patient throughput techniques are required. This study investigates the use of these techniques on two hematology/oncology specialized patient care units, which represent hospital wide strained bed capacity issues at an academic medical center. The two major objectives for this study were: 1) To describe the patient population and the management of supply and demand for beds; 2) To use simulation modeling to evaluate the impact of proposed policies and interventions through generation of projected post-intervention unit efficiency data, which included occupancy rate, wait time, length of stay, and bed availability upon request.
Study Design: This study is a pre-post study design.
Population Studied: The sample for this study consisted of patients for whom a bed was requested and who were subsequently admitted to either of the two hematology/oncology patient care units at an AMC during FY07. Data on this patient sample was utilized to generate a simulation model to mimic processes and operations on these units.
Simulation models based on the base (pre) and proposed interventions (post) provided data for
statistical analysis. Through the analysis of unit efficiency indicators using ANOVA or Chi-Square tests, the effectiveness of each proposed intervention was analyzed in comparison to other interventions and in comparison to the base, preintervention simulation model.
Principal Findings: The optimal strategy of reducing discharge time outliers (greater than 8 hours) and limiting admissions to a specific group of diagnoses would result in more than 10% additional bed capacity.
Conclusion: The application of simulation modeling to bed management and patient throughput strategies offers an effective, scientific, and methodological approach to identifying strategies for similar bed capacity and throughput issues, not only across the units within the hospital, but also across organizations.
Implications for Policy, Delivery or Practice:
There is the need to identify the steps that would be required to operationalize the selected intervention.
These would include a multi-disciplinary task force to develop and operationalize micro-level strategies
(e.g., 1 day prior discharge notification where possible, and reconciliation of discharge medication). On a broader level, the success of this study in identifying a feasible and effective intervention provides opportunity to apply this methodology to various hospital issues related to quality and efficiency.
Theme: Quality and Efficiency: Organized
Processes
● Factors Associated with Recent Budget
Reductions for Local Public Health Agencies.
Evidence from Two Nationwide Surveys
Gulzar Shah, Ph.D., M.Stat., M.S.; Carolyn Leep,
M.S., M.P.H.; Barbara Laymon, M.P.H.
Presented by: Gulzar Shah, Ph.D., M.Stat., M.S.,
Senior Analyst II, Research & Evaluation, Research
& Analysis, The National Association of County &
City Health Officials, 1100 17th Street NW,
Washington, DC 20036, Phone: (202) 507-4264;
Email: gshah@naccho.org
Research Objective: The general objective of this study is to examine the manner in which current economic downturn facing the country has impacted the budgets of local health departments
(LHDs) around the country, and to identify factors instrumental in predicting the vulnerability of a LHD to experiencing budget cuts. Specific research objectives include: 1. To determine the proportion of
LHDs that have had their budgets reduced compared to previous fiscal year; 2. To examine the extent to which LHD budgets have decreased over the past year; 3. To identify organizational characteristics that are associated with vulnerability to budget cut; and 4. To examine factors associated with level of reduction or increase in the LHD budgets.
Study Design: This study involves analysis of linked data from two recent cross-sectional national surveys of LHDs conducted by NACCHO in late
2008 – the National Survey of LHD Budget
Changes; and the 2008 National Profile of LHDs.
Potential explanatory variables for the multivariate models include population size and degree of urbanization of jurisdiction, type of governance, funding mix (sources of LHD revenue), per capita
LHD expenditures, type of services provided, and characteristics of the LHD’s top executive. We will use logistic regression to model the dichotomous variable representing whether or not LHD experienced a budget cut, as well as polytomous variable representing the extent of budget cut.
Population Studied: The target population for the study is all local health departments in the country, defined as “an administrative or service unit of local or state government concerned with health, and carrying some responsibility for the health of a jurisdiction, smaller than the state.” The Profile survey was a complete census of all LHDs (2795 at the time of the survey), whereas the Budget-Cut
Survey used predominantly census approach with exception of five states, where sampling was performed for only those groups of LHDs, for which the same contact person was listed for the entire group, and not separately for each LHD in the group.
Principal Findings: Our preliminary analysis shows that 27% of LHDs experienced a decrease in budget in the current year compared with the prior year and this percentage shows considerable variation across states and categories of jurisdiction population. We expect to find a significant impact of several organizational characteristics on vulnerability for budget cuts.
Conclusion: The proportion of LHDs experiencing budget reductions, and the level of reductions are substantial enough that it should draw attention of all stakeholders to the factors associated with vulnerability of LHDs, to be identified by our multivariate analyses.
Implications for Policy, Delivery or Practice: The findings of this study are very timely, given the current economic crisis. Identifying characteristics of local health departments that make them particularly vulnerable to budget cuts during an economic downturn will provide insights to leaders at all levels of the public health system. Local health department leaders and local government officials will be able to gauge the expected relative impact to their agency in a general economic downturn and consider whether changes to their structure, funding, or service mix might result in more stable funding levels in the future. A better understanding of the types of local health departments most severely impacted by an economic downturn may allow state and federal public health leaders to make better decisions about resource allocation, particularly at times when budgets are declining.
Funding Source(s): CDC, Robert Wood Johnson
Foundation
Theme: Public Health
● The Impact of New Medical Technologies on
Health Care Expenditures: Evidence from
External Beam Radiation Therapy to Treat
Prostate Cancer
Krishna Sharma, M.A.; Allen Goodman, Ph.D.;
Kendra Schwartz, M.D., M.S.P.H.
Presented by: Krishna Sharma, M.A., Doctoral
Student, Economics, Wayne State University, 656
West Kirby, #2074 FAB, Detroit, MI 48202, Phone:
(313) 577-3345; Email: ksharma@wayne.edu
Research Objective: To evaluate the short term effects on health care expenditures of recent innovations in external beam radiation therapy— three dimensional conformal radiation therapy (3D-
CRT) and intensity modulated radiation therapy
(IMRT)—to treat prostate cancer among Medicare patients in the United States.
Study Design: The study is designed as retrospective cohort and cross-sectional study using a large nationwide observational data. All patients who were receiving external beam radiation therapy and whose claim records were available for at least one year from the start of treatment were selected
(N=87783) and classified on the basis of treatment
IMRT, 3D-CRT or none of the two (control group).
One and two year total health care expenditures were calculated using all the claims and propensity score with local linear matching method was used to calculate the average expenditure for each of the treatment group versus the control group. Separate estimations were made for the whole study periods and for each cohort of the patients defined by the year of receiving treatment. The study period for
3D-CRT was from 1992-2004 and that for IMRT was 1998-2004.
Population Studied: All Medicare enrollees who were diagnosed with prostate cancer from 1991 to
2002, who received external beam radiation therapy from 1991 to 2004, and who were recorded by
Surveillance, Epidemiology, and End Results
(SEER) Program which covers 26% of the US population. The population includes only males predominantly aged 65 or higher. The SEER
Medicare-linked database was used.
Principal Findings: For the whole study period the inflation adjusted expenditures were at least 17% higher among those who received 3D-CRT compared to the control group who received standard radiation therapy. Average expenditure for those who received IMRT between 1998 and 2004 was at least 80% higher. In dollar terms, the increased expenditures varied from nearly $2,000 to $11,000. The results varied depending upon the duration of expenditure. A year by year analysis shows that both technologies have a significant impact on total expenditures after the first couple of years of adoption. The trend analysis for the study periods shows that the growth of expenditure is accelerated with the adoption of IMRT but not with
3D-CRT which accounted higher expenditures for a period of 10 years before they subsided.
Expenditure gap for IMRT remained significantly high.
Conclusion: Shown by both 3D-CRT and IMRT the recent innovations in external beam radiation therapy increased health care expenditures.
However, growth of expenditures is not necessarily accelerated with every innovation. The cost differentials taper off as the rate of adoption increases in the case of 3D-CRT. In the case of
IMRT, it is too early to say if the same rule applies.
Implications for Policy, Delivery or Practice:
Even if incremental cost per patient from a new technology seems small, the growth of overall expenditures from it can be substantial. It is, however, interesting to note that the spending differential also depends upon the rate of utilization.
One likely explanation for this can be the increased competition among the providers. The ability of providers to charge higher is also determined by their competitive edge arising from the early adoption of new technologies and increasing rate of acceptance of those technologies among users.
Funding Source(s): SEER Program (data support only)
Theme: Medicare
● Are Standing Order Programs Associated with
Increased Influenza Vaccination Coverage?
National Nursing Home Survey, 2004
Abigail Shefer, M.D.; Peng-jun Lu, Ph.D.; Barbara
Bardenheier, M.P.H.; Robin Remsburg, Ph.D.; Jill
Marsteller, Ph.D.
Presented by: Abigail Shefer, M.D., Medical
Epidemiologist, National Center for Immunization &
Respiratory Diseases, Centers for Disease Control
& Prevention, 1600 Clifton Road, MS E-52, Atlanta,
GA 30333, Phone: (404) 639-8233; Email: ams7@cdc.gov
Research Objective: To assess if standing order programs in long term care facilities are associated with increased influenza vaccination coverage among residents using a national survey. Standing order programs allow vaccines to be administered according to protocol without a physician’s verbal or written order prior to vaccine administration.
Study Design: Secondary data analysis using multinomial logistic regression of data from the
National Center for Health Statistic’s National
Nursing Home Survey 2004, a nationally representative sample survey. Main outcome measure was receipt of influenza vaccination within the last 12 months (yes, no, unknown). Facility’s vaccination program included standing orders versus other (pre-printed admission order, advance physician order, and personal physician order).
Multinomial logistic regression was used to examine the relationship between type of influenza vaccination program and receipt of vaccination, adjusted for resident and facility characteristics
Population Studied: A total of 1,152 U.S. long term care facilities systematically sampled with probability proportional to number of beds. A total of 11,940 randomly sampled people aged 65 years or older residing in long term care facilities between
July and December 2004 (up to 12 per facility).
Principal Findings: The proportion of residents 65 and older that received influenza vaccination was
64%; 41% of facilities had standing order programs.
Influenza coverage among residents residing in facilities with standing orders was 68% (range 59% to 63% for residents in facilities with other program types). Multinomial logistic regression showed that standing order programs compared to all other program types were independently associated with influenza coverage (adjusted OR 1.23); facility characteristics (ownership and census region) and resident characteristics (race and proportion of time residing in facility during influenza season) were confounders for the relationship between vaccination programs and vaccination status.
Conclusion: This study indicates that residents in long term care facilities having standing order programs for influenza had higher odds of being vaccinated than residents with other types of vaccination programs. This difference may be due to the coordination difficulties associated with having a physician specifically order vaccination for each resident. More research needs to be done to understand how to facilitate adoption of these programs.
Implications for Policy, Delivery or Practice:
Nursing facilities that implement standing order programs can expect somewhat higher vaccination coverage of residents as compared to other approaches.
Funding Source(s): CDC
Theme: Long Term Care
● Examination of Relationships Between Market
Factors & Community Services Activities of
General Acute Hospitals from 1994 to 2006
Jay Shen, Ph.D.; Greg Ginn, Ph.D.; Charles
Moseley, Ph.D.
Presented by: Jay Shen, Ph.D., Associate
Professor, Health Care Administration & Policy,
University of Nevada Las Vegas, 4505 Maryland
Parkway, Box 453023, Las Vegas, NV 89154-3023,
Phone: (702) 895-5830; Email: jay.shen@unlv.edu
Research Objective: : In today’s cost containment and competitive health care environment, market factors are among the strong forces that drive hospitals’ strategies and behavior. With more states implementing community benefit laws (CB
Laws) in recent years, it is important to understand how market forces affect hospital adaptation of community benefit activities The purpose of this study was to examine the relationship between market forces and community benefit activities of general acute hospitals.
Study Design: The study was a panel design that used 1994 and 2006 American Hospital Association survey data.Two response variables were an aggregated index of hospital community health orientation activities and an aggregated index of fifteen hospital-based health promotion services
The main independent variables included whether the hospital had revenue from a capitation payment system (indicating managed care influence), the
Herfindahl-Hirschman index (indicating the level of local market competition among hospitals), and the percentage of revenue from Medicaid patients
(indicating a resource constraint due to the lower level of Medicaid reimbursement ). A repeated measure fixed-effect model was applied in the multi-variable analysis. The multi-variable analysis controlled for hospital bedsize and ownership
(investor-owned versus not-for-profit). To control for the effect of the CB law, hospitals in states with the law and states without the law were analyzed separately
Population Studied: A total of 3080 private, acute care hospitals were included.
Principal Findings: For the hospital’s community orientation activities, in the states with CB Laws, as compared with hospitals without the capitation payment arrangement, hospitals with the capitation payment tended to offer more community orientation activities (on average, 0.63 more activity in 1994 and 2.07 more in 2006). In contrast, the percentage of revenues from Medicaid did not affect the community orientation activities in 1994 but showed strong effect (1% increase in the percentage of revenues from Medicaid resulted in reduction in community orientation activities by
3.33) in 2006. In addition, the market competition was not statistically significantly related to the hospital’s community orientation activities.
As for the hospital’s health promotion services, in the sates with CB Laws, as compared with hospitals without the capitation payment arrangement, hospitals with the capitation payment, on average, provided more health promotion services (on average, 0.77 more service in 1994 and 1.89 more in 2006). In contrast, the percentage of revenues from Medicaid was negatively related to the health promotion services (1% increase in the percentage of revenues from Medicaid resulted in reduction in health promotion services by 1.76 in 1994 and 3.96 in 2006). Finally, the market competition was not statistically significantly related to the hospital’s community orientation activities. Furthermore, in the states without the CB Laws, similar findings were observed as those in the states with the CB Laws.
Conclusion: Hospitals increased their community health orientation activities and health promotion services from 1994 to 2006. Financial incentives and resource capacity were strongly associated
with hospital community benefit activities after controlling for the influence of state CB laws on these activities.
Implications for Policy, Delivery or Practice:
Market forces appear to have as much influence on the provision of hospital community benefit activities as state CB laws. Reimbursement constraints influence whether hospitals will provide community benefit activities after controlling for the influence of state CB laws on these activities. States may need to do more than pass CB laws if they want their hospitals to increase these activities. They may also need to consider ways to enhance the reimbursement of these.
Funding Source(s): The UNLV Building Research
Infrastruture Award
Theme: Health Care Markets and Competition
● Prevalent & Incident Depression & Mortality among Veterans with Diabetes
Chan Shen, Ph.D.; Ranjana Banerjea, Ph.D.;
Patricia Findley, Dr.P.H., M.S.W., L.C.S.W.;
Leonard Pogach, M.D., M.B.A.; Usha
Sambamoorthi, Ph.D.
Presented by: Chan Shen, Ph.D., Health Science
Specialist, Veterans Health Administration,
Veterans' Affairs, 385 Tremont Avenue, Mail Stop
129, East Orange, NJ 07018, Phone: (973) 676-
1000 ext 1545; Email: chan.shen@va.gov
Research Objective: To examine the association between prevalent and incident depression and survival among individuals with diabetes mellitus after controlling for demographic, socioeconomic characteristics, physical and mental health status.
Study Design: Retrospective analysis of administrative data on veterans with diabetes using data from 2000 through 2004. Proportional hazards and logistic regressions at one year, two years and
5 year mortality were used to analyze the association between survival and incident, prevalent and no depression. Other independent variables consisted of demographic, socioeconomic and other chronic physical and mental illnesses.
Population Studied: Veterans with diabetes in
FY2000 (N = 526,823) were derived from the
Diabetes Epidemiologic Cohort. Depression was identified using ICD-9-CM codes of 296.2, 296.3,
309.1, 300.4, and 311. We distinguished between incident and prior episodes of depression by using
120-day (4-month) period without a depression or antidepressant prescriptions before the first observed depression diagnosis date in FY2000.
Principal Findings: Among VHA users with diabetes, 8.6% (N = 45,280) had an incident depression and 4.6% (N = 24,212) had a prevalent depression in FY2000. Over 5 years, 23% died. In two years, 7% died. After controlling for other factors, the hazard ratios were 1.14 for incident depression and 0.91 for no depression compared to prevalent depression. Compared to no depression group, the hazard ratios were 1.26 for incident and
1.10 for prevalent depression.
Conclusion: Among veterans with diabetes, both incident and prevalent depression had significant negative effect on mortality even after controlling for patient characteristics. However, incident depression had even greater effect.
Implications for Policy, Delivery or Practice:
While preventing clinical depression may not always be possible, it is important to routinely screen for depressive symptoms and occurrence of new episodes of depression so that depression can be recognized early on to minimize the negative consequences. Further research is needed to examine whether treatment for depression may reduce negative effect of new episodes of depression on survival.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● The Trials & Tribulations of Mental Health
Parity Legislation on the State & Federal Levels
Gordon Shen, M.Sc.; Lonnie Snowden, Ph.D.
Presented by: Gordon Shen, M.Sc., Student,
Health Services & Policy Analysis Ph.D. Program,
University of California, Berkeley, 3510 Deer Hill
Road, Lafayette, CA 94549, Phone: (415) 250-
2518; Email: gordon_shen@berkeley.edu
Research Objective: In the 110th Congress, the
Senate voted to pass the Mental Health Parity Act of 2007 (S. 558). Shortly after, in March 2008, the
House of Representatives also passed its version of the Mental Health Parity Act (H.R. 1424). In these initiatives, parity is broadly defined as insurance coverage for mental health benefits equivalent to those for medical and surgical services. These two landmark pieces of mental health and fiscal legislation is fraught with a history of bitter debate between employers and insurers on one side of the issue and medical practitioners and mental health advocates on the other. The aim of this legislative review is to trace the progress in the mental health parity movement, as it is mired by arguments related to the definition of mental illness and financial sustainability of the health system.
Study Design: In this study, the federal and state legislative history on mental health parity in the
United States is viewed from Charles Lindblom’s lens of incrementalism (Lindblom, 1959). Through a historical analysis of the 1996 Mental Health
Parity Act and state mental health parity laws, I predict the success of the implementation of a 2008 mental health parity law, based on the sociopolitical acceptance on the S. 558 and H.R. 1424 bills proposed in the 110th Congress. Particular attention is paid to the addition of innovative language for treatment of substance abuse and comprehensive definition of parity with regards to three components: financial limit, treatment limit, and cost-sharing requirements.
Population Studied: An extensive content analysis of all proposed and passed legislation found in the
Library of Congress, National Conference of State
Legislatures, and LexisNexis databases.
Principal Findings: The development of mental health law follows different patterns on the state and federal levels of analysis across time. The state legislatures made fundamental decisions that set the context for subsequent incremental decisions on both state and federal levels. On the federal level, the content of the mental health parity bills introduced in the 110th Congress build on the bills introduced in the 104th Congress. On the state level, however, some states seem to exhibit increasingly governance on insurance provision of coverage for mental disorders, while others gradually cut back coverage. In sum, some states move in a self-perpetuating direction, while others move in the self-defeating direction.
Conclusion: Future studies that seek to follow the incremental gains and losses of mental health parity legislation would benefit from applying Charles
Lindblom’s notion of incrementalism.
Implications for Policy, Delivery or Practice: The mental health parity movement is still far from attaining the policy goal of equity. A set of recommendations will be given for potential improvements to the language of future mental health parity proposals.
Theme: Behavioral Health
● Changes in Emergency Department Access
Between 2001-2005: Are Communities Affected
Equally?
Yu-Chu Shen, Ph.D.; Renee Hsia, M.D., M.S.
Presented by: Yu-Chu Shen, Ph.D., Assisstant
Professor of Economics, Graduate School of
Business & Public Policy, Naval Postgraduate
School, 555 Dyer Road, Code GB, Monterey,
93943, Mexico, Phone: (831) 656-2951; Email: yshen@nps.edu
Research Objective: We analyzed how access to
ED changed between 2001 and 2005, and examined whether access to ED for underserved population groups (such as minority, foreign-born, elderly, and economically disadvantaged) has deteriorated between 2001 and 2005. We also identified other health care market level factors that might contribute to decline access over time.
Study Design: We used retrospective zip-code level data from the 2000 Census Bureau and linked this with the American Hospital Association Annual
Surveys from 2001 and 2005 to identify nearest ED.
We first analyzed the trend in access change between 2001 and 2005 and compare area characteristics between communities that experience decreased distance or no change and those that experience increased distance to the nearest ED. For the multivariate analysis portion, we estimated logistic regression on the odds that distance to the nearest ED increased between 2001 and 2005.
Population Studied: All hospitals with ED service in the continental US between 2001 and 2005.
Analysis is done separately for urban and rural communities.
Principal Findings: Between 2001 and 2005, 10% of urban population (equivalent of 21.2 million population) and 8% of rural population (3.1 million) faced with increased distance to the nearest ED.
In urban areas, communities that experience increased distance to ED tend to be larger, less wealthy, have higher share of minority groups and economically disadvantaged groups. The difference in population characteristics is not as pronounced in rural areas. The multivariate analysis shows urban communities with high share of elderly is 1.34 times more likely to face increased distance
(p<0.05) than communities with low share; and communities with high share of unemployed population is 1.46 times more likely (p<0.05) to face increased distance to the nearest ED than communities with low share of unemployed population. The two groups that have deteriorating access in rural areas are: communities with high share of Hispanic population is 2.72 times (p<0.01) more likely to have decreased access to ED, and communities with medium and high share of families below poverty are 1.8 times more likely to face decreased access to ED (p<0.01). Lastly, in both urban and rural areas, communities with higher presence of FP hospitals are associated with higher odds of facing increased distance to the nearest ED (OR=1.66 for urban, p<0.01; OR=1.47 for rural, p<0.10).
Conclusion: A nontrivial number of US population faced increased distance to their nearest ED, and the affected rural populations having worse decline in access than the affected urban population.
Access deterioration is not even across communities and is more likely to happen in communities with higher share of underserved population.
Implications for Policy, Delivery or Practice: Our findings highlight disparities in ED access for some minority and economically disadvantaged groups, and for rural communities. Further research should focus on developing interventions to address systematic disparities that exist in ED access for vulnerable populations.
Funding Source(s): RWJF
Theme: Disparities
● How Can We Efficiently Cover the Uninsured?
A Micro-Simulation Model of Health Insurance
Coverage
Asako Shimazaki, M.A.
Presented by: Asako Shimazaki, M.A., Ph.D.
Candidate, School of Public Policy & Management,
Heinz College, Carnegie Mellon University, 239
Hamburg Hall, 4800 Forbes Avenue, Pittsburgh, PA
15217, Phone: (412) 537-9800; Email: ashimaza@andrew.cmu.edu
Research Objective: Extending health insurance coverage to the uninsured is one of the most important and urgent health-care issues in the
United States. In 2007, about 46 million Americans lacked health insurance. A large fraction of
Americans today face financial instability and have limited access to health care, placing large costs on the economy. Our objective is to develop a model of health insurance coverage based on economic principles and use that model to analyze the effects of various policies to extend health insurance coverage.
Study Design: This paper employs a simulationbased approach. The underlying model consists of three components: a health care consumption model; a health insurance purchase model; and an insurance premium setting model. Consumers choose health insurance options to yield the highest expected utility. Insurance companies set premiums with constant loading in a competitive market. The parameters of the utility function and other parameters of interest are calibrated by using a simulated method of moments (SMM). The model is calibrated to data from the Medical Expenditure
Panel Survey (MEPS) in 2003, which is a representative sample of the U.S. population. The model is then used to predict the change in the number of the uninsured and to evaluate the change in social welfare caused by the following policies to cover the uninsured: introduction of health insurance vouchers, CDHP (Consumer
Driven Health Plan)-type high deductible plans, and regulation of charity care.
Population Studied: This research focuses on the
U.S. non-elderly adults whose ages are between 25 and 64. It also limits its focus to the U.S. households of only one person for now to avoid further complications to the simulation model due to incorporating the decision problem of households with more than one person.
Principal Findings: Introduction of a health insurance voucher in the individual health insurance markets is found to reduce the uninsured rate from
20.4% to 16.4%, whereas introducing the voucher in all health insurance markets reduces the rate to
9.2%. The change in social welfare is almost negligible under the former policy, whereas social welfare decreases by 5.9% under the latter.
Mandating CDHP-type high-deductible plans decreases the rate of the uninsured to 13.2%, and social welfare increases by 21.4%.
Conclusion: This study is a first step towards policy analysis based on a comprehensive and coherent framework. Further development of the model, such as incorporating employer’s health insurance offer decisions, will enable us to make reliable predictions on the effects of the policies.
Implications for Policy, Delivery or Practice:
This paper shows that the model has the potential to provide reliable predictions of the effects of policies to extend health insurance coverage. It is an important first step towards providing an economic framework to evaluate various policies.
● Understanding the Role of Private Health
Insurance in the Universal Coverage System:
Jaeun Shin, Ph.D.; Sangho Moon, Ph.D.
Presented by: Jaeun Shin, Ph.D., Associate
Professor, Economics, KDI School of Public Policy
& Management, 87 Hoegi-ro Dongdaemun-gu,
Seoul, 130868, KR, Phone: +822-3299-1037;
Email: jshin@kdischool.ac.kr
Research Objective: In this study, we test hypotheses on the role of private insurance in a universal coverage system in the context of the
Korean National Health Insurance (NHI). We explore the macro and micro-evidence that whether the expansion of private insurance may be helpful to deal with financing and cost containment of universal health care system.
Study Design: Using data from the OECD Health
Data 2007, we assess the current status of health care financing in the Korean NHI in comparison of selective developed countries. We examine whether high out-of-pocket payment and low government financing in the Korean National Health
Insurance (NHI) system may be a legitimate basis to propose the expansion of private insurance. Also we seek for evidence that private insurance can be a solution to the financial deficit in the NHI account.
Further, we conduct a micro-level analysis to examine whether private insurance plans may selectively enroll the low-risk population or the wealthy population, and may crowd out other precautionary assets. The micro-evidence is conducted based on data from the Korean
Longitudinal Survey of Ageing (KLoSA) 2006. We apply the probit model to identify the presence of risk-selection in the private plan enrollment status.
Population Studied: As the KLoSA is a survey of individuals of aged 45 or more, we compare the control group of these older adults who have no private coverage with the counterpart group of holding private coverage.
Principal Findings: We find that high OOP payment associated with uncovered services in the
Korean NHI system are noted an important to warrant attention of policy makers, but find little evidence that the enhanced role of private insurance markets may ensure better financial protection for the population as well as relieve the financing burden for the NHI payer. Findings from the micro-analysis suggest that the younger and healthier population is more likely to be enrolled in private coverage. The affordability such as higher income and housing ownership is found important predictors for private insurance purchase. We find no evidence that private insurance participation may crowd out savings. The potential of informal
support from children, in particular, the eldest son seem to be a substitute to private coverage for aging parents. Life insurance/annuities plan purchase appears to be positively correlated with private coverage purchase, implying that both plans are considered as means for risk-spreading in later years of life for aging population. The accumulated value of life insurance/annuities plans is, however, negatively associated with private coverage purchase.
Conclusion: The expanded role of private insurance plans does not necessarily stabilize the health care financing in the national plan. There is no evidence to believe that the management scheme of private plans and market competition among these plans would bring the cost saving in health care. The private plans execute selective enrollment of low-risk population, although they may not crowd out other precautionary savings.
Implications for Policy, Delivery or Practice:
Since the early 1990s, the Korean NHI Corporation, the single payer for the entire NHI members, as well the Korean government and other stakeholders, have considered to promote more involvement of private entities in the provision of health insurance.
The controversy on the role of supplementary private plans was inflamed by serious financial deficit of the NHI fund in the late 1990s and by the persistently high OOP burden on patients. This study suggests that the role of private insurance is uncertain to benefit the NHI plan as it is expected.
Risk-selection and an increase in health care expenditure are potential consequences of the active role of private plans. Both the public and government need to pay a careful attention to whether they are willing to bear these outcomes which would undermine equity in health. Although further studies are warranted to provide conclusive policy suggestion regarding the adequate balance of private and universal coverage in the provision of health insurance, we should acknowledge ambiguity of the effectiveness of private initiative in health care delivery in dealing with difficulties of universal coverage system.
Theme: Coverage and Access
● Smoking, Time Preference & Educational
Outcomes
Jaeun Shin, Ph.D.
Presented by: Jaeun Shin, Ph.D., Associate
Professor, Economics, KDI School of Public Policy
& Management, 87 Hoegi-ro Dongdaemun-gu,
Seoul, 130868, KR, Phone: +822-3299-1037;
Email: jshin@kdischool.ac.kr
Research Objective: In this study, we examine the cost of smoking in student performance. We set up a model to describe the educational outcome and smoking decision of an individual as a function of time preference. Then, we conduct an empirical analysis to measure the relationship between smoking and academic perforamance among high school graduate cohort.
Study Design: We derive the estimation models of academic performance and smoking status based on the theroretal framework, assuming that time preference simultaneously determins these two decisions of an individual. Then, we estimate the effect of smoking during the senior year of high school on academic achievement in terms of the
Korean Scholastic Aptitude Test (KSAT) score and college attendance. We perform the Two Stage
Least Squares (2SLS) estimation to incorporate the possible endogeneity of smoking status due to unobserved individual heterogeneity. Further, we conduct three identification tests to confirm the validity of our choice of instrumental variables for smoking variable.
Population Studied: We use high school graduate cohort data from the Korean Education and
Employment Panel (KEEP) survey 2004-2005.
Principal Findings: Findings uniformly show that heavy smoking during the school year is negatively associated with test scores and college attendance.
The usual OLS estimates for the cost of youth smoking in educational outcomes are found underestimated.
Conclusion: As the model of time preference predicts, smoking is negatively associated with academic achievement of high school students.
This result is not attibuted to unobserved heterogeneity in smoking decisions.
Implications for Policy, Delivery or Practice:
Unhealthy behavior like smoking is a costly activity of an indiviual, in terms of educatinal outcomes, as this study suggest, as well as health risk, as prevalently noted. Since the schooling outcome of an individual is the key predictor of job performance in the labor market, the prevention of smoking and other substance use among the young population is stronly recommended for the improvement of public health and economic well-being of those under influence.
Theme: Behavioral Health
● Health Promotion for Kansans with
Disabilities: Can Targeted Case Managers
Effectively Use Health Information to Improve
Health Care Quality?
Theresa Shireman, Ph.D.; Aimee James, Ph.D.;
Allen Greiner, M.D., M.P.H.; James Backes,
Pharm.D.; Rebecca Ross, M.S.
Presented by: Theresa Shireman, Ph.D.,
Associate Professor, Preventive Medicine & Public
Health, University of Kansas School of Medicine,
3901 Rainbow Boulevard, MSN 1008, Kansas City,
KS 66160, Phone: (913) 588-2382; Email: tshireman@kumc.edu
Research Objective: Persons with physical (PD) and developmental disabilities (DD) are generally less likely to receive appropriate preventive health
care services, despite having higher rates of chronic medical comorbidities. Disabled persons also fare poorly when it comes to management of their chronic conditions. We hypothesized that case managers, with the appropriate resources, could intervene to improve the receipt of quality health care, given their close relationships with beneficiaries. Determine if case managers who provided targeted case management services for persons with PD and DD were able to improve the quality of health care for the individuals they serve.
Study Design: In February 2006, the Kansas
Health Policy Authority received a Medicaid
Transformation Grant from the Centers for
Medicare & Medicaid Services (CMS) to use electronic health-related claims data to promote health in persons with disabilities. The Health
Promotion for Kansans with Disabilities Project was a 13 month intervention (November 2007-
December 2008) involving four Community
Developmental Disability Organizations and three
Independent Living Centers from primarily rural locations across Kansas. During the intervention, case managers accessed an electronic, on-line tool called ImpactPro (Ingenix Corporation) that contained a 12-month Medicaid claims history for their consumers. Case managers were able to review inpatient visits, outpatient visits, and prescription medication claims to see what types of services their consumers had accessed. In addition, Impact Pro screened the claims to determine if there were any opportunities to improve the quality of care for the consumers based on current HEDIS indicators. At the beginning of the intervention, case managers and counselors attended a one-day workshop with hands-on training on the ImpactPro tool and small group sessions learning about the basics of chronic disease and risk factor management taught by clinicians from the University of Kansas Medical
Center. The clinical team also provided monthly newsletters covering a broad array of health topics designed for non-clinical readers.
Population Studied: Persons enrolled in Kansas
Medicaid’s PD and DD waivers.
Principal Findings: There were 1,654 Medicaid beneficiaries who met a minimum of six months of program enrollment. Using 12-month claims preceding the intervention period, we noted substantial deficiencies in the quality of care for enrollees with diabetes as well as standard age- and gender-appropriate cancer screening rates.
For individuals with diabetes, annual rates for standard HEDIS measures follow: HbA1c, 53.4%
DD and 45.7% PD; lipid testing, 27.4% DD and
30.6% PD; eye exam, 31.5% DD and 26.6% PD; and physician visit, 61.6% DD and 59.5% PD. With respect to cancer screening, annual rates were: cervical cancer, 21.5% DD and 11.3% PD; breast cancer (mammography), 30.2% DD and 21.0% PD; and colorectal cancer, 16.8% DD and 19.0% PD.
Since the intervention period has just concluded, we are actively calculating follow-up rates for each of these measures as part of the program evaluation. Those rates will be available in advance of the meeting
Conclusion: Our baseline analysis demonstrated substantial deficiencies in care for disabled enrollees. A program evaluation is in process to determine if case managers were able to correct or improve these deficiencies.
Implications for Policy, Delivery or Practice: The
CMS Transformation Grants have largely supported infrastructure development for information exchanges at multiple levels. The Kansas Medicaid
Transformation Grant will demonstrate the effectiveness of information dissemination through case managers who work directly with physically and developmentally disabled persons.
Funding Source(s): CMS
Theme: Medicaid, SCHIP and State Health Reform
● Adoption of Health Information Technology by
Case Managers to Improve Quality of Care for
Persons with Disabilities
Theresa Shireman, Ph.D.; Robert Clark, M.A.;
Sharon Johnson, B.S.W.; Debbie Huske, M.A.;
Presented by: Theresa Shireman, Ph.D.,
Associate Professor, Preventive Medicine & Public
Health, University of Kansas School of Medicine,
3901 Rainbow Boulevard, MSN 1008, Kansas City,
KS 66160, Phone: (913) 588-2382; Email: tshireman@kumc.edu
Research Objective: Case managers who provide targeted case management services for persons with physical (PD) and developmental disabilities
(DD) have unique relationships with the persons they serve. While social services dominate their interactions, they are also expected to address health care needs through Kansas Medicaid targeted case management. As part of a Medicaid
Transformation Grant from the Centers for
Medicare & Medicaid Services (CMS), we deployed an electronic, on-line tool based on Medicaid claims histories to case managers in primarily rural locations. We hypothesized that access to this health information technology would reduce disparities in the quality of care provided to persons with disabilities. We tracked the amount of time case managers were logged into the system as a measure of tool adoption.
Study Design: The Health Promotion for Kansans with Disabilities Project was a 13 month intervention
(November 2007-December 2008) involving
Community Developmental Disability Organizations and Independent Living Centers from primarily rural locations. Case managers were given access to an electronic, on-line tool called Impact Pro (Ingenix) that contained a 12-month Medicaid claims history
(inpatient, outpatient, and prescription claims) for their consumers. Impact Pro highlighted particular opportunities to improve consumers’ quality of care
through preventive or chronic health interventions.
Case managers attended a one-day workshop with hands-on tool training. Time logs at the case manager level during the intervention were tracked on a monthly basis. We conducted focus groups to determine strengths and limitations of the tool at the end of the intervention.
Population Studied: Case managers and independent living counselors at seven agencies serving rural Kansas locations.
Principal Findings: There were 78 case managers who completed the initial Impact Pro training: they provided a minimum of six months of targeted case management to 1,654 Medicaid PD and DD enrollees. The average time spent using the tool came to 22 minutes per beneficiary. Time per month ranged from a low of 978 minutes (16 hours) in the first month of the program (December 2007) to highs in excess of 4,900 minutes (82 hours) in
March and August 2008. The increase in time logged during March 2008 was associated with implementation of a gift card program for the three highest monthly users. Usage also increased in early fall we talked with agency directors and around the time we scheduled focus groups for program evaluation. Usage was lowest in June and
July due to staff vacations and changes in state rules for other service activities. As the project neared completion, usage rates dropped substantially.
Conclusion: While case managers had relatively ready access to detailed claims histories and care opportunity screens for the individuals they served, their actual adoption of the tool was quite limited.
Barriers to use included the need for static IP addresses, login/password recall, concerns over privacy, and inability to discern important medical information relevant to their consumers.
Implications for Policy, Delivery or Practice:
Simply providing HIT to service providers who have the capability to improve health care will not be sufficient to ensure its effective adoption.
Funding Source(s): CMS
● An Exploratory Study of Medication Therapy
Management (MTM): The Current State of
Evidence & the Implications for Medicare Part D
Sarah Shoemaker, Ph.D., Pharm.D.; Andrea
Hassol, M.S.
Presented by: Sarah Shoemaker, Ph.D.,
Pharm.D., Research Associate, Health Policy, Abt
Associates, 55 Wheeler Street, Cambridge, MA
02138, Phone: (617) 349-2472; Email: sarah_shoemaker@abtassoc.com
Research Objective: To understand MTM program
(MTMP) practice models and payment approaches.
To identify effective MTMP features from private and public sectors, the criteria used to evaluate effectiveness, strength of the evidence, and relevance for Medicare. To determine how MTMPs differ from disease/care management programs.
Study Design: The study included a literature review, 60 key informant interviews, and 4 in-depth case studies of MTMPs. A PubMed search was conducted for peer-reviewed MTM articles published in English between January 1, 2000 and
July 15, 2007, which yielded 26 relevant articles.
An Internet search was conducted of relevant grey literature, which yielded 33 relevant documents.
Sixty telephone interviews were conducted with representatives from 46 organizations including: pharmacy associations, Medicare PDPs/MA-
PDs/SNPs, employers, Medicaid, VA, pharmacy chains, LTC facilities/pharmacies, PBMs, MTM vendors, an AHRQ DEcIDE research center, and a
QIO. In-depth case studies were conducted of four
MTMPs, including: a Medicare Advantage plan, a
Medicare prescription drug plan, a Medicaid program, and an MTM vendor.
Principal Findings: There does not yet appear to be a set of clearly defined MTMP practice models that are reasonably stable and mutually exclusive.
There is wide variability in Medicare MTMP eligibility criteria, and criteria differ between
Medicare and non-Medicare MTMPs. MTMPs that use an opt-in approach have had less success than those using an opt-out approach. Most MTMPs provide a comprehensive medication review, at least annually. Some MTMPs use face-to-face interventions while others rely largely on telephone or mail interventions; no research has rigorously tested these different intervention modes. The majority of MTMP services are provided by pharmacists, either salaried or reimbursed on a feefor-service schedule. Medicare beneficiaries served by MTMPs may also participate in separate disease/care management programs; in non-
Medicare sectors separate MTMPs are uncommon.
Most Medicare MTMPs do not integrate MTM with nursing home care, or collaborate with consultant pharmacists. There is minimal evidence about the effectiveness of Medicare MTMPs. In the non-
Medicare MTM literature, costs were commonly measured but findings are inconsistent and input costs (e.g., administrative) were usually not measured. The strongest evidence regarding the impact of MTM was for a few intermediate clinical outcomes (i.e., HbA1c, LDL); fewer studies demonstrated effects of MTM on medical service use (e.g., ER visits).
Conclusion: Medicare MTMPs are still evolving and there is limited evidence to determine which beneficiaries would benefit most from MTM, which services or delivery modes achieve the desired outcomes, and which outcomes should be measured to compare MTMPs’ performance.
Implications for Policy, Delivery or Practice:
Some Medicare MTMPs favor strict eligibility criteria and an opt-in enrollment approach that may limit beneficiary access to MTM. Some Medicare
MTMPs are measuring the process of care, rather
than clinical or cost outcomes. Many of the significant findings in the literature involved populations not directly relevant for Medicare.
Some Medicare MTMPs are conducting research with strong methodological designs, and others are rapidly accumulating large databases, all of which could be of benefit to CMS in refining the Medicare
MTM program.
Funding Source(s): CMS
Theme: Medicare
● The Role of Parent Health Literacy among
Urban Children with Persistent Asthma
Laura Shone, Dr.P.H., M.S.W.; Kelly Conn, M.P.H.;
Lee Sanders, M.D., M.P.H.; Jill Halterman, M.D.,
M.P.H.
Presented by: Laura Shone, Dr.P.H., M.S.W.,
Assistant Professor, Pediatrics, University of
Rochester School of Medicine & Dentistry, 601
Elmwood Avenue, Box 777, Rochester, NY 14642,
Phone: (585) 273-4084; Email: laura_shone@urmc.rochester.edu
Research Objective: Health literacy (HL) affects adult asthma management, yet less is known about how parent HL affects child asthma care. Our objective was to examine associations between parent HL and measures related to child asthma.
Study Design: We performed secondary analyses of baseline data from the School Based Asthma
Therapy randomized trial. Data were collected via school health forms and structured in-home interviews. Measures included the Rapid Estimate of Adult Literacy in Medicine for parent HL; NHLBI criteria for asthma severity, and validated measures of asthma knowledge, beliefs, and experiences.
We conducted bivariate and multivariate analyses of associations between parent HL and measures related to child asthma.
Population Studied: Parents of 499 pre-school and school-age urban children with persistent asthma in Rochester, New York.
Principal Findings: Our response rate was 72% and mean child age was 7.0 years (range 3-10 years). Thirty-two percent had a Hispanic parent;
88% had public insurance. Thirty-three percent had parents with limited HL. Low parent HL was independently associated with greater parent worry parent perception of greater asthma burden, and lower parent-reported quality of life. Measures of health care use (e.g., emergency care, preventive medicines) were not associated with parent HL.
Conclusion: Parents with limited HL worried more and perceived greater overall burden from the child’s asthma, even though reported health care use did not vary.
Implications for Policy, Delivery or Practice:
Improved parent understanding and provider-parent communication about child asthma could reduce parent-perceived asthma burden, alleviate parent worry, and improve parent quality of life.
Funding Source(s): National Heart, Lung, and
Blood Institute; Centers for Disease Control and
Prevention, Office of the Director
Theme: Child Health
● How Do Self-Reported Health and
Absenteeism Measures Compare to
Administrative Data?
Meghan Short, M.P.H.; Ron Goetzel, Ph.D.;
Maryam Tabrizi, M.S., C.H.E.S.; Xiaofei Pei, Ph.D.;
Ronald Ozminkowski, Ph.D.; Teresa Gibson, Ph.D.
Presented by: Meghan Short, M.P.H., Senior
Analyst, Healthcare, Thomson Reuters, 4301
Connecticut Avenue, N.W., Washington, DC 20008,
Phone: (202) 719-7872; Email: meghan.short@thomsonreuters.com
Research Objective: Assessment of population health can be daunting, so researchers often use a self-reported measure, the Health Risk Assessment
(HRA) to evaluate health risk, healthcare utilization, and productivity. Although self-reported measures can be unreliable, they may be useful when access to administrative data (medical claims and absenteeism) is unavailable or expensive to analyze. An NHLBI-funded study assessed the accuracy of self-reported healthcare utilization and absenteeism, comparing self-reported responses to administrative records to determine the reliability of
HRAs in capturing these metrics.
Study Design: Monthly and yearly doctor visits, emergency room visits, hospital admissions, and annual absent days were compared for the same individuals using HRA and administrative data in
2005 and 2006, respectively. The analysis was a cross-sectional design, pooling both years of data to create a large sample. Percent agreement, overreporting, under-reporting, and Pearson's correlations were used to measure the validity and strength of the relationship between self-reported
HRA responses and claims data. A multivariate logistic regression model determined the likelihood of agreement between administrative and selfreported data. These models included the covariates of age, gender, race/ethnicity, job type, education, body mass index (BMI), the Charlson
Co-morbidity Index (CCI), and self-reported health status to predict the likelihood of agreement between self-report and administrative data.
Population Studied: Approximately 4,000-6,000 employees at The Dow Chemical Company were included in the analysis of these outcomes. To be included in the study, participants had to have selfreported HRA healthcare utilization and absenteeism data and corresponding administrative data. The majority of the sample consisted of white males (about 75%) and the average age was around 45.
Principal Findings: The majority of participants' self-reported estimates of prior year's healthcare utilization and absenteeism matched the data from
administrative claims. Rare events, such as emergency room visits and inpatient hospital admissions, had the highest percent agreement
(91.6% for annual ER visits, and 93.2% for inpatient hospital admissions). Annual doctor visits and absent days had the lowest percent agreement
(30.2% and 37.6%, respectively). Absenteeism showed the greatest discrepancy between selfreported and administrative measures, with respondents over-reporting by about one and 1/3 days. Analysis of the multivariate logistic regression model showed a detailed view of factors influencing the likelihood of agreement. Education, age, gender, and BMI were significant predictors of agreement in most of the models. Individuals with a bachelor’s degree or higher were 20% to 89% more likely to accurately recall their healthcare utilization and absenteeism compared to those who had less education.
Conclusion: Overall, this analysis is informative for researchers who rely upon self-reported HRA responses as a gauge of employee healthcare utilization and absenteeism rates when administrative claims are unavailable.
Implications for Policy, Delivery or Practice: The findings of this study should aid researchers that use self-reported measures to collect healthcare utilization and absenteeism data both in collection and analysis. HRA questions asking participants about these measures should be as specific as possible to accurately capture the utilization of their sample. Additionally, researchers can rely on selfreported measures of emergency room visits and hospital admissions but should be wary of selfreported annual rates for doctor visits and absenteeism.
Funding Source(s): National Heart Lung and
Blood Institute (NIH)
● Preliminary Findings: Evaluation of the
Advancing System Improvements to Support
Targets for Healthy People 2010 (ASIST2010)
Program
Emily Shortridge, Ph.D., M.P.H., M.P.P.; Barbara
James, M.P.H.; Caitlin Oppenheimer, M.P.H.;
Rachel Singer, Ph.D.
Presented by: Emily Shortridge, Ph.D., M.P.H.,
M.P.P., Research Scientist, Health Research,
National Opinion Research Center at the University of Chicago, 4350 East West Highway, Suite 800,
Bethesda, MD 20814, Phone: (301) 634-9343;
Email: shortridge-emily@norc.org
Research Objective: The Office on Women’s
Health-funded ASIST2010 program provides an ideal case for evaluating whether utilizing a sex- or gender-based public health systems or other collaborative partnership approach can have a positive impact on Healthy People 2010 targets.
ASIST2010 is a unique program in at least two respects: it places emphasis on sex- and genderbased care, and is designed to improve health outcomes within the selected Healthy People 2010 focus area – cancer, diabetes, heart disease and stroke, access to quality health services, educational and community-based programs, nutrition and overweight and physical activity and fitness. To accomplish this work, grantees have been asked to build on existing collaborative partnerships; use evidence-based strategies to achieve desired outcomes; track progress towards selected Healthy People 2010 objectives using gender-specific data; improve surveillance systems to allow such tracking of progress; and develop a plan to sustain the effort after the end of funding.
For this component of the study, we propose to identify the kinds of implementation-related and operational organizational changes that necessarily occur in implementing a sex- or gender-based approach towards improving community health status.
Study Design: This component of the study utilizes comparative logical analysis methods. We developed a data abstraction protocol, and conducted a comprehensive review of all project materials, including grant applications, progress reports, and publications and presentations. We then segmented and coded the data according to a priori codes. We reviewed the codes to develop a hierarchical system to develop relationships among categories.
Population Studied: The populations studied vary according to the specific program, although all programs utilize a gender approach to reduce health disparities and all grantees defined their target populations. Six out of the thirteen grantees, for example, explicitly target underserved populations. Six out of thirteen target rural communities.
Principal Findings: To date we have identified themes in the structural and process domains of the various organizations that map to implementation strategies for organizations that are creating collaborations between organizations with a history of sex- or gender-based work and organizations that focused in other areas. Some key areas of importance include making a case to the community for focusing exclusively on men or women; identifying evidence-based programs utilizing gender theory or a sex- or gender-based approach that could be modified to fit with community needs; and decomposing available data sources to measure target population improvement. Further, there are operational themes that we observe that point to greater success: the strength of the collaboration before adopting the sex- or genderbased approach is correlated with success of sex- or gender-based initiative implementation; the strength of organizational commitment to a sex- or gender-based approach has implications for the progress of the program; and creativity around problems solving, including bringing in more
members of the team to discuss problems, seem to link to program success.
Conclusion: This analysis provides us with information on the structures and processes that grantees have established to implement their initiatives (e.g., the proposed roles of the various partners, the framework the grantee established in order to provide sex or gender-based care, the
Healthy People 2010 objectives that the grantees propose to meet, and so on). The challenges that these grantees face are ones that will increasingly be facing community-based organizations that compete for health funding: how to identify likely community partners; how to develop a reasonable data collection and analysis plan to prove that the approach; how does a theoretical sex- or genderbased approach translate in implementation and operation; and how can an organization measure the success of utilizing a sex- or gender-based framework?
Implications for Policy, Delivery or Practice:
Community-based organizations are being asked to prove the efficacy of their interventions, and the
OWH’s ASIST2010 grantees are among the first to test the efficacy of a sex- or gender-based approach on health outcomes. Our findings will provide a blueprint for community organizations that are looking to develop partnerships around sex or gender approaches.
Funding Source(s): Office on Women's Health
Theme: Gender Health
● Improving Universal Inpatient Domestic
Violence Screening at an Academic Medical
Center
John Showalter, M.D. ; Stacy Andersen, M.D.;
Darla Shehy, B.S.N.; Noel Ballentine, M.D.; Jennifer
McCall-Hosenfeld, M.D.
Presented by: John Showalter, M.D., Medical
Informatics Fellow, Internal Medicine, Penn State
University, 500 University Drive, Hershey, PA
17033, Phone: (717) 531-8521; Email: jshowalter@hmc.psu.edu
Research Objective: Hershey Medical Center, a
500-bed academic medical center, is mandated by state law to screen all inpatients for domestic violence. In June 2005, universal screening was initiated using the electronic health record. The screen was a voluntary portion of the hospital’s electronic admission process, performed ad lib by admitting nurses as part of routine admissions assessment and was three multiple choice questions evaluating present and historic abuse.
The effectiveness of the electronic screening had not been evaluated. We began a quality initiative with three main objectives: 1) determine the frequency of domestic violence screening in the inpatient setting; 2) establish the validity of the screening instrument, and 3) determine the proportion of eligible patients receiving intervention by our in-house domestic violence advocate.
Study Design: The initiative was twofold. First, to establish baseline screening rates we used electronic data to assess screening frequency for seven consecutive days in August 2008. We then convened a committee of clinicians and clinical informaticists to evaluate screening frequency, the screening instrument, and intervention rates for positive screens.
Population Studied: All adult patients admitted for seven consecutive days.
Principal Findings: With a voluntary inpatient screening assessment, only 35% of adult inpatients were screened for domestic violence. An extensive literature review of published screening instruments for our population revealed that our screening tool was not adapted from existing instruments. Review of the screening tool development process revealed the instrument was developed based on consensus alone, and had never been validated. Comparison of actual interventions by our domestic violence advocate to the estimated frequency of positive screens for all adult admissions indicated that our local domestic violence medical advocate was notified for 2.47% of eligible patients.
Conclusion: Universal inpatient screening for domestic violence implemented by voluntary completion of an electronic screening yielded low
(35%) screening rates and lack of an automatic notification to the domestic violence advocate resulted in diminished access to services. To address these findings, the committee determined that domestic violence screening must be a mandatory part of the electronic admission form. A new single question screening instrument with multiple sections was developed based on published screens and pre-tested for ease of administration. To ensure high sensitivity, this instrument addresses three primary domains of violence: emotional, physical and sexual.
Additionally, the screen was limited to current/ongoing abuse. Plans for formal validation are in place. In addition, notification of the medical advocate will occur within the electronic health record via an automatic communication for all positive screens. As these modifications are being made, compliance with completion of the screening tool and frequency of notification of eligible patients will be followed. The electronic medical system is currently undergoing modifications to support clinical database queries that will track compliance with screening and the number of interventions by the medical advocate.
Implications for Policy, Delivery or Practice:
Leveraging the electronic medical record to track compliance with domestic violence screening will improve notification and the ability of our patients to access needed services. Future work will need to focus on validation of the screening instrument and on tracking compliance with screening and advocate notification.
● Race & Sex Differences in the Receipt of
Timely & Appropriate Lung Cancer Treatment
Lisa Shugarman, Ph.D.; Katherine Mack, M.P.P.;
Melony Sorbery, Ph.D.; Haijun Tian, Ph.D.; Arvind
Jain, M.A.; Steven Asch, M.D.,. M.P.H.
Presented by: Lisa Shugarman, Ph.D., Health
Policy Researcher, RAND Corporation, 1776 Main
Street, P.O. Box 2138, Santa Monica, CA 90407-
2138, Phone: (310) 393-0411 x 7701; Email: lisas@rand.org
Research Objective: Previous research suggests that disparities in non-small cell lung cancer
(NSCLC) survival can be explained in party by disparities in the receipt of cancer treatment. This study considers race and sex disparities in the timing and appropriateness of treatment across stages of diagnosis.
Study Design: Multivariate logistic regression models were estimated to evaluate the relationship of sex and race with the receipt of timely and clinically appropriate NSCLC treatment for each stage of diagnosis.
Population Studied: Included were Medicare beneficiaries included in the SEER cancer registries linked to Medicare claims data and diagnosed with
NSCLC between 1995 and 1999 (N=22,145).
Principal Findings: Overall adjusted rates of timely and appropriate treatment were 37.2%, 58.1%, and
29.2% for Medicare beneficiaries diagnosed with
Stage I/II, Stage III, and IV NSCLC, respectively.
Female and black Medicare beneficiaries diagnosed with Stage I/II NSCLC were less likely to receive appropriate and timely treatment. Black men were least likely to receive resection (22.2% compared to 43.7% for white men). Blacks were
34% less likely to receive timely surgery, chemotherapy, or radiation for Stage III disease and were 51% less likely to receive chemotherapy in a timely fashion for Stage IV disease relative to whites.
Conclusion: Significant variations in appropriate, timely treatment were found within and across stages of diagnosis, confirming that sex and race differences in NSCLC treatment exist.
Implications for Policy, Delivery or Practice: Like many conditions, NSCLC treatment falls short of the optimal. That we find substantial differences in the receipt of NSCLC treatment by race and sex at any stage of disease demands that health care professionals and policymakers direct more resources toward reducing disparities in treatment, which may reduce disparities in cancer outcomes.
Funding Source(s): HRSA
Theme: Disparities
● Evaluating the Equity of Primary Care
Capitation Payment Models in Ontario, Canada.
Lyn Sibley, Ph.D.; Richard Glazier, M.D. M.P.H.;
Julie Klein-Geltink; Alexander Kopp
Presented by: Lyn Sibley, Ph.D., Postdoctoral
Fellow, Honors Program in Medical Education,
University of Toronto, Health System Performance
Research Network, 155 College Street - 425,
Toronto, M5T 3M6, CA, Phone: (416) 978-5017;
Email: lyn.sibley@utoronto.ca
Research Objective: Several innovative models for primary care delivery have recently been introduced in Ontario, Canada. These models each have different formulae of blended fee-for-service, age and sex based capitation rates, and performance based incentives. There is concern that age and sex adjusted capitation rates alone do not take into account variations in morbidity burden and health care needs that are associated with socioeconomic status and rural residence. This study evaluates the morbidity burden of patients enrolled to primary care physicians in the two most common models --
Family Health Networks (FHN) and Family Health
Groups (FHG) -- in Ontario Canada.
Study Design: This cohort study uses administrative data collected by the Ontario Ministry of Health and Long-Term Care. Two measures of morbidity burden were calculated using The Johns
Hopkins Adjusted Clinical Groups (ACG) Case-mix
System. The number of Aggregated Diagnosis
Groups (ADGs) indicates the number of types of conditions for which individuals were diagnosed.
ACG weights are an indicator of the expected level of health care utilization relative to the population given the duration, severity, and aetiology of diagnosed conditions. The average number of
ADGs and ACG weights were compared across five-year age groups and among males and females. These comparisons were then stratified by socioeconomic status and rural/urban residence.
Population Studied: Eligible FHGs and FHNs were those in existence continuously from
September 1, 2005 to August 31, 2006 that had at least three physicians during the study time period.
The study sample was identified as those patients who were continuously enrolled to a FHG or FHN throughout the study period.
Principal Findings: The number of ADGs (types of diagnosed conditions) and ACG weights (relative expected resource use) was higher among women, and increased incrementally with increasing age.
Compared to FHNs, FHGs a had higher mean number of ADGs (4.62 vs. 4.01; P<0.001) and ACG weights (2.48 vs. 2.07; P<0.001). This finding was consistent across all age, sex, socioeconomic, and rural/urban categories. Lower socioeconomic status was associated with a higher average number of
ADGs and ACG weights -- overall and by all age and sex strata. People who resided in rural areas had fewer ADGs than those living in urban areas, however they had higher ACG weights. This same pattern held among women enrolled to FHGs in all age groups, but the pattern was not evident among men or FHN enrolees.
Conclusion: Age and sex adjusted capitation rates do take into account some of the variation in morbidity burden and expected health care resource utilization. The physician reimbursement system in FHGs and FHNs do not take into account all of the variation in morbidity burden that is associated with socioeconomic status and rural residence.
Implications for Policy, Delivery or Practice:
There is a risk that adjusting capitation rates for age and sex alone introduces an incentive to preferentially enrol patients with higher socioeconomic status. This policy may also incentivise physicians to practice in urban areas where residents have higher socioeconomic status.
Funding Source(s): CIHR Postdoctoral Fellowship
Theme: Quality and Efficiency: Policies and
Incentives
● Racial & Ethnic Differences in Antipsychotic
Use in Nursing Homes
Michele Siegel, Ph.D.; Judith Lucas, Ed.D., A.P.N.;
Ece Kalay, B.A.; Stephen Crystal, Ph.D.
Presented by: Michele Siegel, Ph.D., Assistant
Research Professor, Institute for Health, Rutgers
University, 30 College Avenue, New Brunswick, NJ
08901, Phone: (732) 932-6943; Email: msiegel@ifh.rutgers.edu
Research Objective: Use of antipsychotic drugs in nursing homes (NH) has become widespread, particularly for treating neuropsychiatric and behavioral symptoms of dementia. However, there is increasing concern that these drugs are used as
‘chemical restraints,’ and increase the risk of death among frail elderly patients. The aim of this study is to examine the effect of resident race/ethnicity and facility racial/ethnic composition on antipsychotic use.
Study Design: This was a cross-sectional, secondary data analysis of the effect of resident race/ethnicity (white, black, Hispanic) and facility racial/ethnic composition on off-label antipsychotic use among NH residents with and without dementia. Individual-level data from the Minimum
Data Set (MDS v.2.0) was merged with facility-level data from OSCAR. Logistic regression models controlled for: resident socio-demographics, physical and cognitive limitations, physical and psychiatric comorbidities, and symptom severity; and for facility ownership structure, size, staffing, quality indicators, and aggregate resident characteristics.
Population Studied: Nursing home residents in 8 states (CA, FL, GA, IL, NJ, NY, OH, TX), with an
MDS assessment in 2006. This included 278,850
(78%) whites, 52,443 (15%) blacks, and 25,867
(7%) Hispanics, for a total sample of 357,160 NH residents, including 203,095 (57%) with dementia and 154,065 (43%) without dementia. We excluded the 43,470 residents with schizophrenia or bipolar disorder.
Principal Findings: Bivariate results indicate that
31.4% of residents with dementia and 15% without dementia use antipsychotics. Among residents with dementia, 31.2% of whites, 29.9% of blacks, and
36.8% of Hispanics used antipsychotics (p<.0001).
Among residents without dementia, 14.6% of whites, 15.0% of blacks, and 18.8% of Hispanics used antipsychotics (p<.0001). Regression results indicate that, for the dementia subgroup, black race was associated with lower odds of antipsychotic use (OR=.884, CI .852-.918); Hispanic ethnicity was associated with greater odds of use (OR=1.126, CI
1.075-1.180). Residing in a facility with a higher proportion of black (OR=1.166, CI 1.039-1.308) or
Hispanic (OR=1.690, CI 1.469-1.945) residents was associated with higher odds of antipsychotic use. A higher proportion of residents on Medicaid was also associated with higher odds of use (OR=1.243, CI
1.120-1.379). More RN staff hours per-resident-day were associated with lower odds of use (OR=.884,
CI .813-.962). Regression results were similar for residents without dementia.
Conclusion: Elderly nursing home patients are often treated with antipsychotic drugs, despite the absence of a U.S. Food and Drug Administration
(FDA) approved indication (i.e., schizophrenia/bipolar disorder). While black residents have significantly lower odds of antipsychotic use than white residents, Hispanics have significantly greater odds of antipsychotic use.
Residents of facilities with a higher proportion of black or Hispanic residents, or a higher proportion of residents on Medicaid, have greater odds of antipsychotic use. These facilities may have fewer resources.
Implications for Policy, Delivery or Practice:
Nursing homes, and the government agencies that regulate them, should be mindful of the high rate of antipsychotic use among Hispanic residents when assessing treatment strategies. Regulators should be particularly cognizant of the greater use of antipsychotics in facilities with more black and
Hispanic residents, and more residents on
Medicaid.
Funding Source(s): AHRQ, Retirement Research
Foundation
Theme: Long Term Care
● Adoption of HIT by Primary Care Physicians &
Its Impact on Decison-Making
Carol Simon, Ph.D., M.S.; William White, Ph.D.;
Lauren Smith, M.D.
Presented by: Carol Simon, Ph.D., M.S., Vice
President, The Lewin Group, 3130 Fairview Park
Drive, Suite 800, Falls Church, VA 22042, Phone:
(703) 624-9615; Email: carol.simon@lewin.com
Research Objective: examine the role of practice and physician characteristics in adopting health
information technology (HIT) in office based primary care (PC) and pediatric medical practices and to explore the relationships between the presence of
HIT and clinical decision making
Study Design: This study uses data from a 2006-7 multi-mode survey of physicians in PC and pediatric practices linking patterns of HIT adoption and physician behavior to characteristics of the physician, his/her practice and the managed care environment. Data domains include physician demographic characteristics, income, practice revenues and structure (e.g., resources, payer type, and use of HIT, including electronic medical records
(EMRs) and decision support systems), and administrative controls. Physicians were also asked to respond to clinical vignettes (asthma and depression). Multivariate weighted regression and logistic regression are used to analyze HIT adoption and analyze physician treatment recommendations.
Data are weighted to account for sampling design and known sources of non-response.
Population Studied: A random sample of 1967 primary care (PC) and pediatric physicians in 5 states (California, Illinois, Georgia, Pennsylvania and Texas). The sample was derived from the
American Medical Association Physician Masterfile.
Pediatric and minority physicians were over sampled. Response rate was 69%
Principal Findings: HIT adoption in general was relatively low in the study sample. (e.g. 56% of study PC physicians and 64% of pediatricians made no use of EMRs for their patients). Our preliminary results indicate that HIT adoption is greater for physicians in larger practices and those with higher operating revenues per physician, while adoption was less likely for older and minority physicians, and those in pediatric practices and whose practices had high Medicaid loads. Preliminary analyses also suggest that along with facility practice size, Medicaid load, and physician demographics, HIT use was related to PC physicians’ propensity to follow recommended treatment guidelines for depression and asthma. IV techniques are used to control for the engogeneity of HIT adoption
Conclusion: there are significant differences between various physician practice organizations in use of HIT and that based on vignettes, HIT is associated with higher levels of compliance to practice guidelines.
Implications for Policy, Delivery or Practice: overcoming barriers to HIT adoption in physicain practices in small, rural, and underserved neighborhoods may contribute to significant improvements in quality of care for patients
Funding Source(s): CWF
Theme: Health Information Technology
● Comparing Safety Climate Between Two
Populations of Hospitals in the U.S.
Sara Singer, M.B.A., Ph.D.; Christine Hartmann,
Ph.D.; Amresh Hanchate, Ph.D.; Mark Meterko,
Ph.D.; David Gaba, M.D.; Amy Rosen, Ph.D.
Presented by: Sara Singer, M.B.A., Ph.D.,
Assistant Professor, Health Policy & Management,
Harvard School of Public Health, 677 Huntington
Avenue, Boston, MA 02115, Phone: (617) 432-
7139; Email: ssinger@hsph.harvard.edu
Research Objective: Benchmarking safety-climate survey results across healthcare systems is difficult due to coordination challenges. In this study, we used the same survey instrument, sampling, and administration procedures to benchmark safety climate in US and VA hospitals. This provided a unique opportunity to compare safety climate across two diverse populations and to identify any hospital features systematically related to safer care. Given potential advantages in promoting strong safety climate in a nationally integrated network, we hypothesized that safety climate among VA hospitals would be stronger than among
US hospitals. This paper also explored the factors influencing climate in US and VA hospitals. We hypothesized that variance in measured sample characteristics would explain more of the difference in safety climate between US and VA hospitals than would differential effects of sample characteristics on the two groups.
Study Design: Individual-level, cross-sectional comparisons linked safety-climate measures to characteristics of individuals and their work facilities. Safety-climate measures were derived from the Patient Safety Climate in Healthcare
Organizations survey. Forty-one items reflected 12 safety-climate dimensions. Psychometric analysis of response patterns supported construct validity.
We compared sample characteristics and safetyclimate perceptions of respondents in US and VA hospitals, and then used a regression-based decomposition approach (Oaxaca-Blinder decomposition) to examine differences in safety climate between US and VA hospitals. Oaxaca-
Blinder decomposition divides the difference between the two samples into (1) a samplecharacteristics component, which measures the difference in safety climate due to variance in sample characteristics and (2) the unexplained component, which measures the extent to which the effects of characteristics differed between US and
VA hospitals.
Population Studied: In 67 US and 29 VA hospitals—both representative of safety performance of their respective populations—we sampled 100% of senior managers, 100% of physicians, and a random 10% sample of all other workers, with selected over-sampling, in 2005-
2007. We received 13,841 completed surveys
(41% response) in US hospitals and 4,581 (50% response) in VA hospitals.
Principal Findings: Although there was substantial overlap between the two groups, the range in safety climate among individual US hospitals was larger than among VA hospitals. Contrary to our first hypothesis, however, safety climate was better on average in US hospitals than in VA hospitals on 10 of 12 safety climate dimensions. While decomposition results predicted 0.77 percentage points better safety climate in VA hospitals based on sample characteristics alone, it measured average safety climate as 2.2 percentage points better in US than VA hospitals. Thus, contrary to our second hypothesis, differential effects of sample characteristics explained more of the difference in safety climate between US and VA hospitals.
Conclusion: Results suggest that neither US nor
VA hospitals has achieved superior safety climate as a group. Rather, safety climate appears associated more with initiatives of individual hospitals than with participation in a nationally integrated system or to those characteristics of workers and facilities measured here.
Implications for Policy, Delivery or Practice:
Substantial unexplained variance among US and
VA hospitals suggests significant opportunities for improvement through individual hospitals’ creativity, innovation, and leadership.
Funding Source(s): AHRQ, VA HSR&D
Theme: Quality and Efficiency: Measurement
● The Need for a Better Definition of “Safety Net
Hospital” for Use in Allocating Medicare DSH
Funds & Other Public Resources
Sara Singer, M.B.A., Ph.D.; Jonathan Clark, M.S.;
John Petito; Melissa Valentine, M.P.A.; Gene Lee;
Nancy Kane, D.B.A.
Presented by: Sara Singer, M.B.A., Ph.D.,
Assistant Professor, Health Policy & Management,
Harvard School of Public Health, 677 Huntington
Avenue, Boston, MA 02458, Phone: (617) 432-
7139; Email: ssinger@hsph.harvard.edu
Research Objective: The burden of caring for this nation’s under- and uninsured, low-income population falls disproportionately on safety net hospitals. While there is consensus that such hospitals deserve public support for the unreimbursed costs of such care, there is disagreement regarding how to identify the hospitals most deserving of safety net status and related financial subsidies through Medicare and other programs. Prior research and resource allocation decisions have been based on various proxies that identify different hospitals as part of the safety net and that are typically cross-sectional, despite instability over time. The objective of this paper is to test the ability of various safety net measures to proxy for actual free care, a fairly direct measure of care delivered to under- and uninsured patients, but one that is not reliably available nationwide.
Study Design: In a two-phase approach, we first assigned “safety net” status according to four commonly-used proxies: (1) National Association of
Public Hospitals and Health System (NAPH) membership, (2) percent Medicaid inpatient days,
(3) percent minority-patient discharges, and (4)
Medicare disproportionate share (DSH) percentage.
For Medicaid share, minority share, and DSH percentage, safety net status was determined in comparison with local (percent Medicaid, minority,
DSH percentage one SD above the local market mean) and national (excluded if hospital’s measure was below national average; included if measure was in the top 10% nationally) benchmarks. Safety net status was deemed only if hospitals met the criterion for at least two of three consecutive years, including the most recent year, 2006. In phase-two, we are collecting free care data from audited financial statements on the sample of hospitals identified above through proxies. In combination, these data allow us to assess the extent to which measures of “safety net” status correlate with each other and are consistent over time. . Moreover, they will allow us to test the extent to which hospitals identified as “safety net” facilities using proxy measures also achieve such status based on reliable free care data. Finally, these data will allow us to statistically test the extent to which the proxy measures, in combination, could substitute for information about free care.
Population Studied: 3,599 US, non-rural, general acute care hospitals.
Principal Findings: A preliminary set of 803 safety net hospitals was initially defined based on NAPH membership, percent Medicaid days, percent minority discharges, and DSH status. None of the
“safety net” status categories based on proxies correlated with any of the others at more than 0.33.
Of those hospitals meeting safety net criteria in any one year, 32% did not meet the criteria in multiple years.
Conclusion: Procedures for identifying safety net hospitals that use different measures of services provided to vulnerable populations yield substantially different sets of hospitals.
Implications for Policy, Delivery or Practice:
Policymakers should use multiple and longitudinal proxies of “safety net” status when free care data is not available to better target resources to those hospitals most in need due to their charitable mission. Some proxies perform better than others in identifying hospitals that provide a significant amount of free care to their communities.
Funding Source(s): CWF
Theme: Medicare
● Access to Male Perpetrators of Domestic
Violence: New Methodology in Health Care
Vijay Singh, M.D., M.P.H.
Presented by: Vijay Singh, M.D., M.P.H., Robert
Wood Johnson Clinical Scholar, University of
Michigan, 6312 Medical Science Building I, 1150
West Medical Center Drive, Ann Arbor, MI 48109,
Phone: (734) 647-4844; Email: vijaysin@umich.edu
Research Objective: (1) to describe the reliability, validity and characteristics of domestic violence perpetration assessment tools used in health care settings, and (2) to assess the prevalence of domestic violence perpetration by male patients.
Study Design: A systematic review of English language articles was conducted in five databases:
MEDLINE, CINAHL, PsycInfo, Web of Science, and
Social Sciences Citation Index. Due to the large volume of literature and the variability of subject indexing among the databases, discrete searches were selectively performed and focused on: domestic violence, abuse, perpetration, male patient, and health care. The cited literature referenced in relevant studies was examined for possible studies, and experts in the subject area were contacted for additional articles.
Population Studied: This review examined male patients in family medicine, general medicine, emergency medicine, and psychiatry settings. Male partners of obstetrical patients were also included.
Principal Findings: 18 studies were found using a domestic violence perpetration assessment tool in a health care setting. The most frequently used instrument is the Conflict Tactics Scale, a 19 item questionnaire to assess emotional reasoning, verbal aggression and physical violence perpetrated over the past 12 months. Reliability of the scales is expressed in alpha coefficients, which range from 0.83 to 0.96. Validity of the Conflict
Tactics Scale has not been determined, though it is the standard by which other instruments are measured. The tool shows signs of face and content validity. Seven other instruments were identified in the remaining studies; however, many articles did not include information on the assessment tool’s reliability or validity. The instrument used in these 18 studies give a prevalence of 6-41% of male patients disclosing domestic violence perpetration.
Conclusion: Male patients can be assessed for domestic violence perpetration through a variety of instruments, of which the Conflict Tactics Scales is the most studied. The instrument shows good reliability, though its validity has yet to be determined. Using these instruments, male patients are willing to self-disclose perpetration of domestic violence.
Implications for Policy, Delivery or Practice:
Several instruments are available for clinicians and researchers to assess domestic violence perpetration among men in health care settings.
These instruments allow for abusive behavior to be identified in health care settings, from which treatment and community referrals can be initiated.
Male patients are willing to be identified as perpetrators of domestic violence, and this serves as a violence prevention opportunity.
Funding Source(s): RWJF
● E-prescribing: Will Medicare Incentives be
Enough to Encourage Adoption by Physicians?
Mark Siracuse, Pharm.D., Ph.D.; Kimberly Galt,
Pharm.D., Ph.D.; James Bramble, Ph.D.; Andjela
Drincic, M.D.; Karen Paschal, D.P.T., M.S.; Amy
Abbott, Ph.D.
Presented by: Mark Siracuse, Pharm.D., Ph.D.,
Assistant Professor, Pharmacy Sciences, Creighton
University, 2500 California Plaza, Boyne 145,
Omaha, NE 68178, Phone: (402) 280-3722; Email: msiracuse@creighton.edu
Research Objective: The objectives were to: 1) describe the current status of e-prescribing in
Nebraska and South Dakota, including type of health information technology used, and 2) determine physician attitudes and opinions about eprescribing.
Study Design: Using a modified Dillman technique, a cross-sectional study was conducted between
July and November 2007. The survey inquired whether physicians currently had an office-based practice. Physicians who indicated they did have such a practice were asked to complete and return the survey via postage paid envelope.
Population Studied: Physicians with an officebased practice in Nebraska and South Dakota.
Principal Findings: The survey was completed by
955 physicians. More than 90% of all respondents reported using handwritten prescription pads.
Nearly 25% reported using a desktop/laptop computer to generate prescriptions, and a negligible number used tablet PCs and Personal Digital
Assistants. Slightly more than 6% of respondents indicated they e-prescribe daily. Traditional methods of prescription delivery or transmission to pharmacies predominated, with over 90% indicating they telephone prescriptions to the pharmacy and/or have their patients take handwritten prescriptions to the pharmacy, while almost 84% use a fax machine to transmit them to pharmacies.
Of the respondents, 22% indicated their patients take computer generated prescriptions printed on a printer in the physician’s office to the pharmacy.
Approximately 62% believe using a desktop/laptop computer improves the accuracy and completeness of prescription information (p < 0.001). When asked about prescribing efficiency, the dominant response for both desktop/laptop computers and tablet PCs was uncertainty regarding whether e-prescribing takes longer than handwriting prescriptions (p =
0.803), with 44.2% and 56.7%, respectively.
Conclusion: Traditional methods of prescription delivery to the pharmacy still predominate, with the majority generated using handwritten prescriptions.
Although adoption of e-prescribing by physicians has been slow, physicians have a positive attitude
regarding the accuracy and completeness of eprescribing.
Implications for Policy, Delivery or Practice:
Potential benefits resulting from e-prescribing include enhanced patient safety through a reduction in medical errors, improved workflow efficiency in the physician’s office and in the pharmacy, more cost-effective prescribing as a result of improved formulary compliance, and more attention paid to lowering drug costs. Some major barriers to eprescribing include lack of infrastructure and financial support, difficulty integrating into the physician work flow, the fact that regulations require prescriptions for controlled substances to be handwritten, and that approximately 30% of pharmacies currently do not accept e-prescriptions.
Most health care providers and policy makers believe the benefits to e-prescribing far outweigh the drawbacks. Starting in 2009, physicians who eprescribe for their Medicare patients can receive a
2% bonus in their fees, and beginning in 2012 physicians who do not e-prescribe will face reduction in Medicare fees initially at 1% and rising to 2% by 2014. Will government incentives alone provide enough to gain widespread adoption for something that faces so many barriers? As shown in this study, actual use of e-prescribing technology has lagged behind expectations, indicating difficulty incorporating this technology into practice. Further study of physician practices is needed to determine if Medicare incentives are working and reasons why physicians are not adopting e-prescribing.
Funding Source(s): AHRQ, Nebraska Medical
Association Foundation
Theme: Health Information Technology
● Participatory Evaluation of a Community
Youth Violence Prevention Initiative
Rachel Skeete, M.D.; Barbara Tinney, M.S.W.;
Georgina Lucas, M.S.W.; Leslie Curry, Ph.D.,
M.P.H.; Maya Greene, M.P.H.; Marjorie Rosenthal,
M.D., M.P.H.
Presented by: Rachel Skeete, M.D., Postdoctoral
Fellow, Internal Medicine, Yale Robert Wood
Johnson Clinical Scholars Program, IE-61 SHM;
P.O. Box 208088, New Haven, CT 06520, Phone:
(203) 785-5469; Email: Rachel.Skeete@yale.edu
Research Objective: Youth violence in the United
States disproportionately affects urban minority racial and ethnic communities. Homicide is the leading cause of death among African Americans between the ages of 10 and 24, and the second leading cause for Hispanics/Latinos. The Street
Outreach Worker Program, a tertiary violence prevention initiative operated by a communitybased organization in New Haven, CT, is designed to reduce gun violence by connecting at-risk youth with adults previously involved in violence but now committed to mentoring youth. Evaluation of the process of such a complex undertaking early on is paramount, as it provides accountability and can aid in program development and improvement. A community-academic partnership was formed to conduct a participatory evaluation of this program.
Study Design: We conducted a qualitative study using in-depth interviews with a sample of 13-24 year old youth program participants, the eight Street
Outreach Workers and the three program administrators. Our interview guide was designed to determine, from the perspective of the three groups sampled: 1) program elements that keep youth engaged; 2) program elements that prevent gun violence and; 3) barriers and facilitators to youth program participation. With both community members and academics on the research team, interviews were audiotaped, transcribed and synthesized into common themes using the constant comparative method of qualitative analysis.
Population Studied: We conducted in-depth interviews with a sample of 13-24 year old youth program participants, the eight Street Outreach
Workers and the three program administrators.
Principal Findings: A unifying theme in keeping youth engaged, preventing gun violence, and overcoming barriers to participation was the relationship between youth participants and Street
Outreach Workers. Interviews from all three parts of the sample described this relationship as a proxy family. A second theme was how the program changed perceptions of the youth about themselves and others. A third theme was that while the youth described program elements that occupy their time, provide exposure to pro-social alternatives to gun violence and bring together youth from different neighborhoods, many of them did not believe these activities would prevent violence.
Conclusion: Eighteen months after the start of the program, many youth who enjoyed and participated in program activities, did not believe the activities could reduce violence; they did, however, believe mentoring relationships were important aspects of violence reduction. Further study should assess the extent to which youth who participate in different aspects of the program are able to avoid violence.
Implications for Policy, Delivery or Practice: The study highlights the perceived importance of human capital and relationships for preventing youth violence; if such programs are found successful, the importance of sustaining the relationships in the proxy family between mentors and youth should not be underestimated.
Funding Source(s): RWJF
● The RN, ARNP & LPN Workforce in
Washington State: Employment, Education &
Demographic Characteristics
Susan Skillman, M.S.; C. Holly Andrilla, M.S.; Linda
Tieman, R.N., M.N.; Mark Doescher, M.D.,
M.S.P.H.
Presented by: Susan Skillman, M.S., Deputy
Director, Center for Health Workforce Studies/Rural
Health Research Center, University of Washington,
4311 11th Avenue Northeast, Suite 210, Seattle,
WA 98105, Phone: (206) 543-3557; Email: skillman@u.washington.edu
Research Objective: This study examines the employment, education and demographic characteristics of registered nurses (RNs), advanced registered nurse practitioners (ARNPs), and licensed practical nurses (LPNs) in Washington state to provide information for assessing current and future nurse supply.
Study Design: Data sources were professional licenses (age, address and gender) and stateconducted surveys of the professions in 2007 and
2008. Response bias was estimated for each profession and survey responses were weighted to reflect each profession’s actual age and geographic distribution. The resulting data were used for descriptive analyses of the three nursing professions and the resulting estimates were compared with national statistics, where available.
Population Studied: Licensed RNs, ARNPs and
LPNs in Washington State.
Principal Findings: Among licensed RNs, ARNPs and LPNs in Washington, 64%, 92% and 72%, respectively, were estimated to practice in-state.
The largest percentage of RNs work in hospitals
(49%), ARNPs in ambulatory/outpatient settings
(67%), and LPNs in nursing home/extended care facilities (37%). On average, practicing RNs in
Washington work 34 total hours per week with 22 hours in direct patient care, ARNPs work 37 hours per week with 30 hours in direct patient care and
LPNs work 37 hours per week with 25 hours in direct patient care. Overall, 57% of Washington’s
RNs obtained at least some of their nursing education in-state, 53% of ARNPs received graduate nursing education in-state, and 73% of
LPNs received their nursing education in-state. The average age of Washington’s RNs, ARNPs and
LPNs is 49, 50 and 47 years, respectively: somewhat older than national averages. Statewide, the ratio of licensed RNs per 100,000 population is
962, dropping to 767 if only practicing RNs are included. The ratio of licensed ARNPs to 100,000 population is 55, or 50 practicing ARNPs per
100,000, and the ratio for licensed LPNs is 209, or
157 for practicing LPNs. These ratios are somewhat lower than national numbers.
Conclusion: Large portions of Washington state’s
RN, ARNP and LPN workforces are nearing retirement age. Many nursing professionals in the state hold licenses, but are not actively employed in nursing. Among those practicing, many work parttime. ARNPs spend more time in direct patient care than do RNs or LPNs. More of the LPN workforce obtained their professional education in-state than did RNs or ARNPs. Washington has fewer nurses per population than national averages.
Implications for Policy, Delivery or Practice:
These findings demonstrate that having estimates of the proportion of professionals who are actively practicing, whether they work full or part time, and in what areas they work, adds important information about the current and future workforce. The survey findings show that the available nursing workforce in Washington is much smaller than a simple examination of license numbers would imply.
Knowledge of the education background of a workforce is important for planning in-state education programs as well as for recruitment and retention programs geared toward professionals from other states. This information is only available through surveys of the workforce: a resource not currently available in many states.
Funding Source(s): Washington Center for
Nursing
● Unexpected Differences in the Dietary Intakes of Obese, Overweight & Healthy Weight
Adolescents: Implications for Obesity
Interventions
Asheley Skinner, Ph.D.Eliana Perrin, M.D., M.P.H.;
Suzanne Havala Hobbs, Dr.P.H., M.S., R.D.
Presented by: Asheley Skinner, Ph.D.,
Postdoctoral Fellow, Health Policy & Management,
University of North Carolina at Chapel Hill, CB
7411, Chapel Hill, NC 27599, Phone: (919) 966-
6394; Email: asheley@unc.edu
Research Objective: Attempts to reduce weight in obese children through dietary interventions have yielded little to no sustained success. There is scant previous research on the dietary consumption differences between obese and healthy weight children, though a few small studies report obese children consume no more or only slightly more. In order to develop and promote public health and clinical recommendations, we need data demonstrating how diets of obese and healthy weight children differ.
Study Design: We analyzed data from six years
(2001-2006) of repeated cross-sections from the
National Health and Nutrition Examination Survey
(NHANES). Measured height and weight were used to calculate BMI percentile and categorize weight status according to accepted standards:
>=95th percentile were obese; >=85th & <95th percentile were overweight; and <85th and >5% percentile were healthy weight. Food intake was collected using the Automated Multiple Pass
Method (AMPM), which involves a computerassisted interview with multiple food-specific questions, and a collection of measuring devices to help participants estimate portion sizes. This method has been repeatedly validated and prior research has indicated weight-related reporting bias is unlikely in this sample. Dietary information about the foods consumed was then determined based on the Food and Nutrient Database for Dietary
Surveys. “Adequate intake” of calories and micronutrients was based on current Dietary
Reference Intakes.
We use adjusted Wald tests to compare nutrition intake differences by weight status. We then used multiple logistic or ordinary least squares regression as appropriate to control for age, sex, income, race, ethnicity, and insurance status.
Population Studied: Adolescents aged 12-17 years in NHANES (2001-2006).
Principal Findings: Obese adolescents reported consuming fewer total calories than either overweight or healthy weight adolescents (2028 vs.
2175 vs. 2373 kcal, p<0.001). Additionally, obese adolescents are more likely than overweight or healthy weight adolescents to have inadequate caloric intake (27% vs. 18% vs. 16%, p<0.001), as well as inadequate intake of fiber, most B vitamins, calcium, phosphorous, iron, zinc, and sodium (all p<0.01). These differences persisted in multivariate analyses.
Conclusion: Our findings challenge the prevailing wisdom that overweight and obese adolescents eat significantly more than healthy weight adolescents, and indicate they actually likely consume less food than their healthy weight counterparts. Of additional concern is that obese children may be at greater risk for nutrient deficiencies than healthy weight children. Our results suggest weight differences may be more related to variations in activity levels or metabolism than persistent dietary differences.
Implications for Policy, Delivery or Practice:
Although improving dietary habits in all children is a worthy goal, obesity interventions focusing on dietary changes may be asking children who are already eating less than peers to restrict consumption further. In doing so, we may be placing them at risk of dietary deficiencies and risking significant unintended consequences, including negative effects on self-esteem, mental health, and physical health. Our findings suggest that dietary advice needs to be carefully planned and individualized. It is possible that weight loss may be more appropriately targeted by physical activity than by calorie restriction.
Funding Source(s): NINR & NICHD
Theme: Obesity Prevention and Treatment
● BMI versus BMI Percentile to Estimate
Cardiovascular Risk Factors in Children
Asheley Skinner, Ph.D.; Michael Steiner, M.D.;
Frederick Henderson, M.D.; Eliana Perrin, M.D.,
M.P.H.
Presented by: Asheley Skinner, Ph.D.,
Postdoctoral Fellow, Health Policy & Management,
University of North Carolina at Chapel Hill, CB
7411, Chapel Hill, NC 27599, Phone: (919) 966-
6394; Email: asheley@unc.edu
Research Objective: Interventions to improve weight and health in children use BMI percentile cut-offs to identify which children are overweight or obese and thus at risk for future weight-related health problems. However, the absolute BMI value corresponding to a given BMI percentile changes dramatically through childhood. Because the BMI percentile cutoffs are based on historical distributions, not the current population, children of different ages classified as obese using BMI percentile vary in their actual adiposity and cardiovascular (CV) risk factors. We examined the relative strength of associations between CV risk factors and both BMI percentile or absolute BMI in order to determine which best identifies children at risk for future weight-related health problems.
Study Design: We performed a cross-sectional analysis of eight years of the National Health and
Nutrition Examination Survey, 1999-2006.
Measured height and weight were used to determine absolute BMI and age- and sex-specific
BMI percentiles according to CDC growth charts.
CV risk factors included were total cholesterol,
HDL, LDL, triglycerides, blood pressure and a marker of inflammation, C-reactive protein (CRP).
Ordinary least squares regression was used to compare the relationships between log-adjusted, standardized values for risk factors, BMI, and BMI percentiles, controlling for race/ethnicity and income. We then used adjusted Wald tests for differences in the magnitude of the coefficients relating CV risk factors and either 95% BMI or absolute BMI cutoffs across different ages.
Population Studied: Children ages 2-17 years in
NHANES.
Principal Findings: The mean BMI in the sample was 19.8 and mean BMI percentile was 63.9
(N=13,328). Absolute BMI was more strongly related than BMI percentile to HDL (b=-0.33 vs. b=-
0.21, p<0.001), triglycerides (b=0.25 vs. b=0.14, p<0.001), and CRP (b=0.43 vs. b=0.28, p<0.001), but there were no significant differences for total cholesterol, LDL, or blood pressure. The risk for abnormal CV risk factors using BMI percentile increased with age in every measure. However, an absolute BMI value was associated with more consistent risk of abnormal results regardless of patient age.
Conclusion: Absolute BMI is a stronger and more consistent predictor of abnormal CV risk factors across different ages than is BMI percentile. This could be due to clustering of an increasingly heterogeneous group of obese children above the
95% BMI, while absolute BMI values allow for a continuous distribution. Alternatively, CV risk factors may be dependent on absolute fat mass, which is more consistently reflected by actual BMI instead of
BMI percentile.
Implications for Policy, Delivery or Practice:
While BMI percentile predicts future weight, absolute BMI is more strongly associated with the current presence of abnormal cardiovascular (CV)
risk factors. Clinical practice and public health interventions aimed at identifying and treating the negative health effects of obesity will require consideration of actual BMI in addition to percentilebased obesity definitions. Further research is needed to clarify the best use of the two measures to identify children most likely to develop clinicallysignificant CV disease.
Funding Source(s): NINR & NICHD
Theme: Obesity Prevention and Treatment
● Assessing Assertive Community Treatment:
Program Effects on Mental Health Costs
Eric Slade, Ph.D.; John McCarthy; Marcia
Valenstein; Lisa Dixon; Rose Ignacio; Stephanie
Visnic
Presented by: Eric Slade, Ph.D., Research
Scientist, VISN5 Mental Illness Research, Eduation
& Clinical Center, U.S. Department of Veterans
Affairs, 809 Chumleigh Road, Baltimore, MD 21212,
Phone: (410) 377-2734; Email: eslade@psych.umaryland.edu
Research Objective: Assertive Community
Treatment (ACT) is a team-based case management model for persons with serious mental illness (SMI). A central goal of ACT teams is to reduce clients' dependency on psychiatric inpatient care. In ACT, teams of mental health clinicians provide clinical and case management services to
ACT clients for periods of a year or longer. ACT teams are mobile and available 24/7. Teams have frequent contact with clients, and client-to-staff ratios in ACT programs are kept low (usually <12-
1). ACT services may be cost-neutral when access to ACT services is limited to SMI patients who are at high risk for re-hospitalization. However, given the striking decline in U.S. psychiatric inpatient utilization during the past 25 years and the rapid expansion of ACT programs in state public mental health systems and in the Department of Veterans
Affairs (VA), it is important to reassess the impacts of ACT on mental health costs. In this study, we assessed the effects of ACT initiation on mental health costs and services utilization during the 12 months following enrollment.
Study Design: Using administrative data from the
VA, we identified all VA patients who enrolled in
ACT between fiscal years 2001 to 2004 and a comparison sample of 28204 patients who were
ACT-eligible in this period but did not enroll (ACTeligible non-enrollees). ACT eligibility criteria required >30 days or >=3 episodes of psychiatric inpatient utilization, diagnosis of schizophrenia or bipolar disorder, and residence within 60 miles of a
VA hospital. ACT enrollees (N=2102) were propensity score matched to 2102 ACT-eligible non-enrollees. Matched non-enrollees closely resembled (abs(t)<1.10) ACT clients on all study covariates, which included inpatient and outpatient mental health services use in the prior 12 months, demographics, homelessness, VA disability rating, and substance use diagnosis. We estimated twopart generalized linear model (GLM) models of matched data to assess the effects of ACT initiation. Instrumental variables (IV) GLM estimates were compared with standard GLM estimates. Instruments were distance to the nearest
ACT team and whether or not an ACT team was located at the hospital where the patient was most recently discharged.
Population Studied: 4204 ACT-eligible seriously mentally ill VA patients
Principal Findings: In standard GLM models, ACT initiation was associated with a 9% reduction in psychiatric inpatient days (33.0 vs. 36.3 for nonenrollees) and a 141% increase in partial hospitalization days (7.3 vs. 3.0 for non-enrollees).
In instrumental variables analyses, ACT initiation was associated with a 59% reduction in inpatient days (14.9 vs. 36.3 for non-enrollees) and a 595% increase in partial hospitalization days (21.1 vs.
3.0). IV-GLM estimates indicated that ACT program costs ($9680 per client) were fully offset by savings for other services ($28190 vs. $40120 for non-enrollees; net savings=11930).
Conclusion: ACT programs still reduce use of inpatient psychiatric care, and do not increase clients' mental health costs.
Implications for Policy, Delivery or Practice: secular declines in psychiatric inpatient utilization have diminished ACT programs' potential to reduce mental health costs, and savings from reduction of inpatient utilization are attenuated by additional costs for partial hospitalization services.
Funding Source(s): VA
Theme: Behavioral Health
● Operating Room Clinicians’ Individual & Team
Workload, Quality of Care & Non-Routine
Events
Jason Slagle, Ph.D.; Matthew Weinger, M.S., M.D.;
Robert Greevy, Ph.D.; Samuel Nwosu, M.S.; Ken
Wallston, M.A., Ph.D.
Presented by: Jason Slagle, Ph.D., Assistant
Professor, Center for Perioperative Research in
Quality, Vanderbilt University Medical Center, 1211
21st Avenue, MAB 732, Nashville, TN 37212,
Phone: (615) 936-6598; Email: jason.slagle@vanderbilt.edu
Research Objective: Increased clinical workload can cause adverse events occur and decrease quality of care. Current methods of measuring surgical workload are crude (e.g., staffing ratios & patient volume), retrospective (e.g., volume of work units performed), and not tailored to the unique attributes of individual clinical encounters. We developed an instrument, the Quality and Workload
Assessment Tool (QWAT), to measure the perceived clinical workload, and quality of care delivered by, of operating room (OR) clinicians.
Study Design: Prospective observational pilot study to assess the association between clinical workload and quality of care ratings among three professional disciplines on the OR team.
Anticipated individual and team workload ratings were collected independently from each OR clinician (surgeons, anesthesia providers (APs), and OR nurses) prior to incision (pre-case; scale from 1 (very easy) to 10 (very difficult)) at two academic medical centers (VUMC 56 cases and VA
54 cases). At the end of the case, perceived individual and team workload (post-case) and perceived quality of care (1-poor to 10-excellent) were again obtained. Participants were also asked if any Non-Routine Events (NRE; a measure of quality of care) occurred during the case.
Multivariate linear regressions were used to test the associations between QWAT measures and NRE occurrence adjusting for clinician type and site. Ttests were used to compare between clinician types and sites.
Population Studied: 695 Surgeons, Nurses, and
Anesthesia Providers.
Principal Findings: Overall, surgeons’ individual
(mean±SD; pre: 4.5±2.0; post: 4.4±2.3) and team
(pre: 4.0±1.8; 3.9±2.1) case difficulty ratings were consistently higher than other clinicians’ individual
(APs: pre: 3.4±1.7, post: 3.0±1.7; RNs: pre:
3.2±2.6, post: 2.7±2.2) and team (APs: pre:
3.8±1.8, post: 3.2±1.7; RNs: pre: 3.3±2.4, post:
3.0±2.2) ratings (p<0.001). Surgeons’ care quality ratings (9.0±1.1) were lower than were other clinicians’ (APs: 9.4±0.9, RNs: 9.5±0.8; p<0.001).
While surgeons and APs workload ratings were not significantly different between the two sites, VUMC nurses’ individual case difficulty ratings (pre:
3.9±2.8, post: 3.3±2.3) were significantly higher than VA nurses’ ratings (pre: 2.8±2.4, p=0.002; post: 2.4±2.1, p=0.001). Thirty-four cases (31%; VA
19, VUMC 15) were reported to contain at least one
NRE. NRE cases had higher individual (pre:
4.1±2.2, p=0.003; post: 3.8±2.4, p=0.002) and team case difficulty ratings (pre: 3.9±1.9, p=0.013; post:
3.8±2.3, p<0.001) than non-NRE cases (Individual: pre: 3.5±2.2, post: 3.2±2.1; Team: pre: 3.6±2.1, post: 3.2±2.0) ratings (p<0.001).
Conclusion: Surgeons rated their own workload higher and quality of care lower than other OR clinicians. NREs occurred in almost 1/3 of cases and correlated with higher anticipated (pre) and perceived (post) workload and lower quality of care ratings. Organiza¬tional metrics of surgical workload (“OR productivity”) are typically limited to the number, type, and duration of cases performed.
Such aggregated output measures provide limited guidance for addressing problems with quality.
Perioperative workload and quality may be influenced by many factors not readily discerned by traditional perioperative productivity metrics.
Implications for Policy, Delivery or Practice:
New approaches using more informative point-ofcare process measures are needed to improve care quality. Our results suggest that real-time perceptions of OR team members provides insight into the relationship between case difficulty, clinical events, and quality of care.
Funding Source(s): VA
Theme: Quality and Efficiency: Measurement
● Barriers to Health Care Impact Utilization,
Quality of Life & Asthma Symptoms/Severity in
Patients Receiving Telephonic Disease
Management
Brad Smith, Ph.D.; Emma Forkner, M.S.N.; Claudia
McCoy, M.S.N; Pamela R. Wood, M.D.; Jay Peters,
M.D.; Autumn Dawn Galbreath, M.D.
Presented by: Brad Smith, Ph.D., Senior Analyst,
Medical Care Systems - Health Quality Research,
Altarum Institute, 3737 Broadway, Suite 205, San
Antonio, TX 78209, Phone: (210) 832-3000; Email: brad.smith@altarum.org
Research Objective: How do barriers to health care affect asthma severity, quality of life and health care utilization in patients receiving disease management services?
Study Design: Multi-method analysis of data from
551 subjects (263 adults; 288 children) drawn from the intervention groups of a 12 month clinical trial of telephonic asthma disease management.
Qualitative coding techniques were applied by trained coders to progress notes written by nurse disease managers to identify barriers to proper asthma care. Barriers which emerged included environmental allergens, lack of compliance, communication difficulties, and lack of access to necessary medicines. We examined the relationship between the total number of barriers identified and asthma symptoms/severity (Lara score and National Asthma Education and
Prevention Program (NAEPP) clinician-assessed severity), quality of life (QOL, measured by AQLQ or PAQLQ), and health care utilization (rates of verified asthma-related admissions, ED visits and acute asthma-related office visits). Statistical analyses were conducted using least-squares, ordinal logistic and negative binomial regression techniques. QOL and severity analyses were adjusted for sex and race/ethnicity while utilization analyses were adjusted for sex, race/ethnicity and asthma severity.
Population Studied: The study population consisted largely of traditionally underserved groups, including racial and ethnic minorities, children with Medicaid, and underinsured and uninsured adult patients in South Texas.
Principal Findings: At baseline, the total number of barriers (mean: 8.5, range: 1-33) was positively associated with Lara symptom score (p<0.01) and negatively associated with AQLQ overall score
(p=0.04) in adults. Adults with the highest barrier counts had AQLQ scores that were 1 point lower than those with the lowest barrier counts. Adults
with higher barrier counts also tended to show increases in NAEPP severity over the course of the trial compared to adults with fewer barriers, controlling for initial severity, ethnicity/race and sex
(p<0.001). However, there was no relationship between number of barriers and asthma symptoms/severity or QOL in children. Analyses of utilization patterns showed that the number of barriers was not associated with the rate of inpatient admissions in either adults or children.
Barriers were positively related to the rate of ED visits in children (p=0.02); children with the most barriers had 0.6 more ED visits/yr than children with the fewest barriers. The number of barriers was also inversely related to the rate of office visits in adults (p=0.02). Adults with a large number of barriers had 0.5 fewer office visits per year than adults with few barriers.
Conclusion: Barriers to asthma care significantly impact QOL and symptom scores for adults with asthma; however, similar findings were not seen in children with asthma. These barriers significantly increased the rate of ED visits in children and decreased the number of acute office visits in adults.
Implications for Policy, Delivery or Practice:
Health care systems and health care providers should assess barriers to care in order to improve outpatient health care utilization and optimize quality of life in individuals with asthma.
Funding Source(s): Health and Human Services
Theme: Prevention and Treatment of Chronic
Illness
● The Impact of Home Health Agency Market
Competition on Risk for Hospital Readmission among Post-Acute Care Medicare Home Health
Beneficiaries
Laura Smith, Ph.D., M.A.; Vincent Mor, Ph.D.; Kate
Lapane, Ph.D.; Mary Fennell, Ph.D.; Edward Alan
Miller, Ph.D., M.P.A.
Presented by: Laura Smith, Ph.D., M.A., Health
Policy Analyst, Aging, Disability & Long-Term Care,
RTI International, 1440 Main Street, Suite 301,
Waltham, MA 02451, Phone: (781) 434-1733;
Email: lsmith@rti.org
Research Objective: To quantify the impact of market competition on the quality of care provided by freestanding home health agencies (HHAs) to post-acute care Medicare home health beneficiaries and to determine if market competition differentially effects the quality of care being provided by forprofit and not-for-profit agencies.
Study Design: We conducted a retrospective cohort study using multi-level analysis to account for clustering of patients in home health agencies, and agencies within markets. We measured competition using the Herfindahl index, a measure of market concentration. We entered the Herfindahl index into our regression as a categorical according to the US Department of Justice thresholds as follows: below 0.1 (competitive), 0.1-0.18
(moderately competitive), above 0.18
(concentrated). We accounted for competition from other HHA, but also considered the presence of other types of post-acute care providers including skilled nursing facilities, long-term care hospitals and inpatient rehabilitation facilities and also if HHA in the market were owned by other types of facilities such as hospitals.
Population Studied: Non-institutionalized
Medicare post-acute home health beneficiaries 65 years and older, admitted to freestanding HHAs from October 2003 to September 2004. Individuallevel data were obtained from the Outcome
Assessment Information Set (OASIS) for patients of
HHAs identified as for-profit or non-for-profit in the
Medicare Healthcare Cost Report Information
System. We linked this data to market level
Medicare data from the Dartmouth Atlas and
Medicare Provider of Service Files. Markets were defined using the Dartmouth Atlas Hospital Referral
Region.
Principal Findings: The median agency-level proportion of patients rehospitalized during the study period was 17.0% with a range from 0 to
50%. Multilevel logistic regression results, adjusted for market, agency and patient characteristics suggest market competition does not affect risk for rehospitalization for patients receiving care from free-standing HHA. Findings persisted with the
Herfindahl index entered as a continuous measure.
We found that patients in markets with more
Medicare hospital discharges per 1,000 Medicare enrollees were more likely to be rehospitalized. This variable accounted for a portion of the market-level variation in rehospitalization by competition. Patient clinical, demographic, functional characteristics were similar across levels of market competition.
Patients receiving care from agencies in high competition markets had slightly higher disability scores, and more severe dyspnea. Patients in moderately competitive markets had better rehabilitation prognoses than those in highly competitive or in concentrated markets.
Conclusion: Findings from the current analysis suggest competition does not influence the quality of care provided to post-acute care patients by freestanding HHA, and that agency profit status does not impact rehospitalizaiton risk regardless of market competition. Findings suggest variation in rehospitalization may be driven more by regional practice patterns and propensity to utilize inpatient care than by competition.
Implications for Policy, Delivery or Practice:
Many have commented on the increasingly fuzzy distinction between for-profit and not-for-profit health care organizations, and observed general shifts in health care management culture to emphasize competition and strategy. Regulation of not-for-profit HHA compliance with non-distribution constraints may be lacking. Characteristics of post-
acute care and home health may make competition irrelevant to rehospitalization. Efforts to improve quality through increased competition may be less effective among free-standing HHA.
Funding Source(s): AHRQ
● Criterion Validity of the Newcastle Diabetes
Symptoms Questionnaire (NDSQ)
Kevin Smith, M.A.; Joseph Burton, M.S.; Douglas
Kamerow, M.D.; Smith Family Foundation Diabetes
Initiative Group
Presented by: Kevin Smith, M.A., Senior Health
Research Analyst, Health Care Quality &
Outcomes, RTI International, 1440 Main Street,
Suite 310, Waltham, MA 02451, Phone: (781) 434-
1748; Email: kevinsmith@rti.org
Research Objective: Diabetes-related symptoms are a means of monitoring the health status of diabetes patients. The purpose of this analysis was to test the criterion validity of the Newcastle
Diabetes Symptoms Questionnaire (NDSQ).
Study Design: The NDSQ measures the frequency of 9 diabetes-related related symptoms in the past month. Total scores may range from 0-100 with higher scores indicating greater frequency and symptoms. In theory, symptom frequency should affect both physical functioning and emotional wellbeing. The criterion validity of the NDSQ was tested by associations with survey measures of physical functioning (the RAND HSI Physical Health
Composite (PHC)), utility (EQ-5D), depression
(PHQ-9), and treatment regimen (insulin vs. oral medications and diet).
Population Studied: The target population consisted of racially and ethnically diverse adult patients with Type II diabetes sampled from the diabetes registries of four Boston inner-city health centers.
Principal Findings: A total of 375 patients completed a self-report survey. Eighty-seven percent of the respondents were members of racial and ethnic minorities. The most frequent symptoms were passing a lot of water (72% of respondents), feeling unusually hungry (61%), and feeling very sleepy during the day (60%). The mean NDSQ score was 22.3 (SD=16.6). NDSQ scores were significantly correlated with depression (r=0.58), the
PHC (r=-0.39), and the EQ-5D (r=-0.39). NDSQ scores were also significantly higher for patients receiving insulin injections (26.1) than for those treated with oral medications or diet (20.2; p=.002).
Conclusion: The statistically significant crosssectional correlations between NDSQ scores and multiple indicators of functional status suggest that the NDSQ is a valid measure of diabetes-related symptomotology.
Implications for Policy, Delivery or Practice: The
NDSQ may be used as a proximate measure of symptom problems in Type II diabetes.
Funding Source(s): The Richard and Susan Smith
Family Foundation; the Paul and Phyllis Fireman
Charitiable Foundation
● Creating an Essential Benefit Package as Part of a State Coverage Initiative
Ariel Smits, M.D., M.P.H.; Jeanene Smith, M.D.,
M.P.H.; Darren Coffman, M.S.
Presented by: Ariel Smits, M.D., M.P.H., Medical
Director, Health Services Commission, Oregon
Office of Health Policy & Research, 1225 Ferry
Street Southeast, 1st Floor, Salem, OR 97301,
Phone: (503) 373-1647; Email:
Ariel.Smits@state.or.us
Research Objective: To design an essential benefit package as part of a state initiative to expand coverage for all state residents.
Study Design: Health policy initiative. The Oregon
Health Fund Board was created by the Oregon
Legislature in 2007 to create a plan for providing insurance coverage for all Oregonians. The
Benefits Committee was charged by the Board to design, in an open and public manner, a benefits package to act as the minimum coverage allowed under the health reform initiative. The essential benefit package was required to be evidencebased, promote primary care and the medical home, emphasize improvement of the health of the public, and be fiscally sustainable for both patients and the state.
Population Studied: The essential benefit package was designed for all Oregonians by a volunteer committee consisting of representatives from the physician, dental, complimentary and alternative medicine, nursing, hospital, managed care organization, brokerage, public health, and patient advocate communities and state government.
Principal Findings: The Essential Benefit Package
(EBP) recommended by the Benefits Committee is based on the Prioritized List of Health Services, which has been used by the Oregon Health Plan since 1994 as an evidence-based method of rationing health services. The EBP builds upon the
Prioritized List by dividing the List into four tiers, with increasing cost sharing for services with lower priorities on the List. To encourage the development and utilization of the medical home, lower cost sharing was incorporated for services provided in the medical home setting; services provided in the specialist office or hospital settings have higher cost sharing. “First-dollar” coverage
(no deductible applied) with little or no copayment required is provided for “value-based services,” shown to reduce complications of disease and the overall cost of care, as well as basic diagnostic services and comfort care services. A high deductible is required on other services to maintain fiscal soundness, with an out-of-pocket maximum to protect patients from catastrophic medical costs.
Other diagnostic services are provided, subject to
prior authorization and/or evidence-based guidelines and other cost controls. Services which do little to improve overall health, such as eyeglasses, are subject to limitations as
“discretionary services.” The EBP is designed to act as a minimum level of coverage; health plans can offer more comprehensive plans within the reformed Oregon heath care system proposed by the Oregon Health Fund Board.
Conclusion: An essential benefit package can be created for assisting state coverage initiatives.
Such a package requires careful consideration of cost sharing for both the type of service and place of service, and should promote “value-based services” which improve health and lower overall health care costs. An essential benefit package should also protect patients from catastrophic loss while maintaining fiscal soundness.
Implications for Policy, Delivery or Practice:
State policy makers attempting to improve coverage and/or create an essential benefit package for state coverage will find valuable insights in Oregon’s experience. The innovative design of the EBP should be considered by other states in their coverage expansion initiatives.
Funding Source(s): State of Oregon
Theme: Coverage and Access
● What Works in Chronic Care Management?
Julie Sochalski, Ph.D., R.N.; Harlan Krumholz,
M.D.; Ann Laramee, M.S., R.N.; Mary Naylor,
Ph.D., R.N.; Michael Rich, M.D.; Barbara Riegel,
D.N.Sc., R.N.
Presented by: Julie Sochalski, Ph.D., R.N.,
Associate Professor, School of Nursing, University of Pennsylvania, Claire M. Fagin Hall, 418 Curie
Boulevard, Philadelphia, PA 19104-4217, Phone:
(215) 898-3147; Email: julieas@nursing.upenn.edu
Research Objective: Private health plans and public payers including Medicare have implemented a diverse array of chronic care management programs to better manage the costly care of patients with chronic illnesses. These programs have produced an equally diverse range of encouraging and disappointing results, yielding an evidence base with more holes in than evidence of what works in chronic care management. The purpose of this study was to distill the elements of post-discharge chronic care management programs that produce the greatest positive outcomes among heart failure patients.
Study Design: We pooled and reanalyzed individual-level data from 10 randomized clinical trials conducted over a 14-year period in four countries that tested the effectiveness of postdischarge chronic care management program among heart failure patients in order to identify program elements that significantly influenced program outcomes. Program outcomes included allcause hospital readmissions and readmission days per month. We constructed a program elements assessment instrument from published taxonomies of chronic care management programs to elucidate and measure the key program elements in each clinical trial. The 10 trials employed chronic care programs that were similar on five of seven program elements we assessed and differed on two—delivery personnel and method of communication—which were the focus of our analysis.
Population Studied: 2,028 heart failure patients—
961 program patients and 1,067 routine care patients—enrolled in 10 randomized clinical trials conducted between 1990 and 2004 in four countries: Australia (2), Netherlands (1), United
Kingdom (1), and United States (6).
Principal Findings: Program patients had 25% fewer unadjusted readmissions and 30% fewer unadjusted readmission days than patients receiving routine care. After adjusting for key patient socio-demographic and clinical characteristics, patients in programs carried out by a single heart failure expert relying on telephone follow-up did not see a significant reduction in hospital readmissions (0.4%, p= n.s.) and readmission days (1.5%, p= n.s.) per month over patients in routine care. In-person communication greatly improved the outcomes of programs using a single heart failure expert, reducing hospital readmissions per month by 1.8% (p = 0.05) and readmission days per month by 4.3% (p=0.06), on average, over routine care. Patients in programs using a multidisciplinary team and in-person communication had significantly fewer hospital readmissions (2.9%, p < 0.001) and readmission days (6.4%, p < 0.001) per month than did routinecare patients. Based on national estimates of hospitalization rates among Medicare heart failure patients, these reductions could produce from
14,700 and 29,140 fewer hospital stays annually among Medicare beneficiaries.
Conclusion: Chronic care management programs delivering state-of-the-art content to post-discharge heart failure patients via multidisciplinary teams and in-person communication achieve significant reductions in all-cause hospital readmissions and readmission days per month over routine care.
Implications for Policy, Delivery or Practice:
This study offers important guideposts for developing an actionable evidence base upon which to build future initiatives that reward plans and practices that adopt effective chronic care management programs.
Funding Source(s): CWF
Theme: Prevention and Treatment of Chronic
Illness
● State-Level Wireless Substitution Estimates from the NHIS
Karen Soderberg, M.S.; Stephen Blumberg, Ph.D.;
Julian Luke; Gestur Davidson, Ph.D.; Michael
Davern, Ph.D.; Tzy-Chyi Yu, Ph.D.
Presented by: Karen Soderberg, M.S., Research
Coordinator, State Health Access Data Assistance
Center, 2221 University Avenue Southeast, Suite
345, Minneapolis, MN 55405, Phone: (612) 625-
5062; Email: soder145@umn.edu
Research Objective: To present state-level estimates of the percentage of households that do not have a landline telephone but do have at least one wireless telephone. These estimates will assist survey researchers to develop more accurate estimates of health insurance coverage and access to care.
Study Design: A two-sample modeling strategy was used to estimate the prevalence of wirelessonly households and adults by state. This modeling was based on data from the 2007 National Health
Interview Survey (NHIS) and the 2008 Current
Population Survey's Annual and Social Economic
Supplement (CPS).
Population Studied: The NHIS and CPS target population includes the civilian non-institutionalized
U.S. population.
Principal Findings: The results show that the prevalence of wireless-only households and adults in 2007 varied substantially across states. Statelevel estimates ranged from 5.1% (Vermont) to
26.1% (Utah) of households and from 4.1%
(Delaware) to 24.6% (Oklahoma) of adults.
Conclusion: The increasing prevalence of the wireless-only household population poses several significant problems for researchers. Wireless-only households are typically excluded from randomdigit dial telephone surveys of health insurance coverage and access to care. Furthermore, results from the NHIS show that these households have different characteristics from those with landline phone service. Our results show great state-level variation in the prevalence of wireless-only households, suggesting that national estimates provided will not provide adequate information to inform state analysts. Telephone surveys are expected to continue to be a common form of data collection in the foreseeable future and states will need to rely on these state-level estimates to conduct survey and data adjustments to improve the accuracy of their estimates.
Implications for Policy, Delivery or Practice:
Because most of the state surveys do not purposively sample cell phones they can all suffer bias as a result of not including the wireless-only in their sample. Researchers have advocated using post-stratification adjustments based on known differences between wireless-only and landline populations. State-level estimates of the wirelessonly population and their characteristics will greatly improve the post-stratification adjustment technique for state-level health surveys. Furthermore, the state-level wireless-only estimates can be used to help state survey analysts decide whether or not to consider including a cell phone sample in their study. These estimates will help improve the accuracy and credibility of state health survey results.
Funding Source(s): RWJF, NCHS
● State Regulation of Practice & the Utilization of Nurse-Midwives for Medicaid Funded
Prenatal Care
Andrea Sonenberg, N.P., C.N.M., D.N.Sc.
Presented by: Andrea Sonenberg, N.P., C.N.M.,
D.N.Sc., Assistant Professor, Lienhard School of
Nursing, Pace University, 861 Bedford Road,
LH311, Pleasantville, NY 10570, Phone: (914) 773-
3435; Email: asonenberg@pace.edu
Research Objective: The aims of this descriptivecorrelational design health policy study utilizing multiple secondary data sets were to: 1) describe any association between a state’s CNM Scope
Index score and the proportion of Medicaid funded services delivered by CNMs in each of the fifty states of the United States, 2) describe any association between a state’s CNM Scope Index score and the proportion of Medicaid-funded prenatal care delivered by CNMs in exemplar states representing the extremes on the Scope of Practice
Index; 3) describe the category of regulation of practice (legal status, reimbursement policies, or prescriptive privileges) within the index that has the greatest association with the proportion of
Medicaid-funded services delivered by CNMs; and
4) describe which of the factors within the most influential category had the greatest association with the proportion of Medicaid-funded services delivered by CNMs.
Study Design: This was a descriptive-correlational health policy study. The association between State regulation of nurse-midwifery practice and total number of nurse-midwives, total births, total CNM births, total Medicaid claims, total CNM Medicaid claims, total MA births, percent 1st trimester prenatal care, percent low-birth weight, percent prematurity, proportion of minorities, and percent below poverty level in 2000-2003 was determined.
Correlations were then performed between the categories of Scope Score (legal status, reimbursement, and prescriptive authority) with each of the outcome variables instead. Finally, in an attempt to determine if extreme differences in
Scope Scores influenced the outcome variables more than the overall difference of all 50 states, the states with Scope Scores on either end of the spectrum were analyzed.
Population Studied: Nurse-midwifery practice and maternal/neonatal health outcomes of 50 states of
United States of America.
Principal Findings: The four key findings were: 1) there is a paucity of data related to individual and specific Medicaid funded services provided by nurse-midwives; 2) states with the fewest barriers to practice have the greatest proportion of CNM births;
3) states with the most restrictive practice regulation also have the highest proportion of vulnerable populations (minority and populations in poverty) and higher percents of adverse perinatal outcomes
(LBW, prematurity, and neonatal mortality); and 4) states with the greatest autonomy in three areas: professionalism (legal status), business practices
(reimbursement), and clinical practice (prescriptive authority) may have the greatest access to care for vulnerable populations.
Conclusion: Despite the evidence of the clinical effectiveness and cost savings of the nursemidwifery model of care and the variety of governmental and organizational recommendations for the increased utilization of certified nursemidwife (CNM) services, there continues to be an impact of state regulation of practice on access to nurse-midwifery care, as well as continued disparities in health outcomes for mothers and newborns, for certain vulnerable populations. Due to a paucity of data related to medicaid funded nurse-midwifery services in the United States, final conclusions could not be drawn from this study.
The finding of this methodological limitation was significant in and of itself. Future study of health policy related to practice of advanced practice nurses, as members of the health workforce, requires improved data collection of their services provided under various payers.
Implications for Policy, Delivery or Practice:
Future studies should focus on improving research methodologies in the areas of state regulation and outcomes of nurse-midwifery practice, specifically those related to data collection. Health policy implications fall into three categories: organizational policy related to research and nurse-midwifery practice, regulation of nurse-midwifery practice, and access to nurse-midwifery care for vulnerable populations of women.
Theme: Health Care Workforce
● Hospital Characteristics Associated with
Racial & Ethnic Disparities in Time to Treatment at U.S. Emergency Departments
Nancy Sonnenfeld, Ph.D.; Stephen Pitts, M.D.;
Susan Schappert, M.A.
Presented by: Nancy Sonnenfeld, Ph.D., Public
Health Analyst, Division of Health Care Statistics,
National Center for Health Statistics, 3311 Toledo
Road, Hyattsville, MD 20782, Phone: (301) 458-
4156; Email: nls2@cdc.gov
Research Objective: Racial and ethnic differences in time to treatment (TTT) at U.S. emergency departments (EDs) have previously been reported.
We sought to understand the extent to which these differences were attributable to differences within and among hospitals, and to identify hospital attributes that might explain “among hospital” disparities.
Study Design: We constructed cross-sectional multi-level models comparing TTT among non-
Hispanic (NH) white, NH black, and Hispanic patients at U.S. EDs using the National Center for
Health Statistics’ 2005 and 2006 National Hospital
Ambulatory Medical Care Surveys (NHAMCS).
NHAMCS collects data from hospitals through induction interviews with hospital administrators, and from representative visits by abstracting ED visit records. TTT was defined as the time from arrival until the patient saw a physician. TTT data were log-transformed because they were skewed; consequently, geometric means are reported.
Patient and visit characteristics examined were triage status (immediate, emergent, urgent, and not urgent), hour of visit, patient age, sex, and expected payment source. NHAMCS hospital characteristics examined were location within a metropolitan statistical area, ED volume, ownership, teaching hospital status, trauma level, and percentages of visits with uninsured, Medicaid, or S-CHIP as expected payment source. Additional hospital characteristics of interest and obtained from other sources were median household income, percent in poverty, and population density for each ED zip code (from 2000 U.S. Census), and 12-month average hospital inpatient bed occupancy and annual ED volume (from Verispan LLC data products).
Population Studied: We analyzed data from a nationally representative sample of 56,026 patient visits at 456 U.S. hospital EDs.
Principal Findings: In unadjusted models, geometric mean TTTs were 25, 34, and 36 minutes for NH white, NH black, and Hispanic patient visits in the U.S (p < .001 for differences among NH white and each of the other population groups). Adjusting for patient and visit characteristics accounted for less than 20% of these differences. Within-hospital differences were examined by controlling for each
ED's geometric mean TTT; predicted geometric mean TTTs were 27, 29, and 32 minutes for NH white, NH black, and Hispanic patients respectively
(p = .021 NH white versus NH black; p < .001: NH white versus Hispanic). In preliminary models of among-hospital differences, geometric mean TTTs for patients with immediate triage status were 89% longer in high volume EDs (p < .001), and 21% longer in high poverty zip codes (p <. .001). The percent differences in TTT between urgent and immediate visits and between non-urgent and immediate visits were significantly greater in teaching hospitals than in non-teaching hospitals.
Hospital bed occupancy was not associated with
TTT.
Conclusion: Although small racial and ethnic disparities among patients within the same hospital remain, most racial and ethnic disparities in TTT may be explained by where patients receive care.
High volume EDs everywhere, but especially in poor zip codes, appear to have the longest TTT.
The relative TTT for urgent and non-urgent visits are longest in teaching hospitals.
Implications for Policy, Delivery or Practice: To reduce racial and ethnic disparities in TTT, identifying general strategies to reduce TTT in high volume EDs may matter more than focusing on
“within-ED” disparities.
Funding Source(s): CDC
Theme: Disparities
● Diagnostic Breast MRI & Surgical Treatment for Breast Cancer
Melony Sorbero, Ph.D., M.S., M.P.H.; Andrew Dick,
Ph.D.; Ellen Burke Beckjord, Ph.D., M.P.H.;
Gretchen Ahrendt, M.D.
Presented by: Melony Sorbero, Ph.D., M.S.,
M.P.H., Health Policy Researcher, RAND, 4570
Fifth Avenue, Suite 600, Pittsburgh, PA 15213,
Phone: (412) 683-2300; Email: msorbero@rand.org
Research Objective: The use of magnetic resonance imaging (MRI) in the pre-operative assessment of women with newly diagnosed breast cancer to identify occult ipsilateral or contralateral disease is increasingly common. Preoperative use of MRI in women with breast cancer may increase rates of mastectomy. This study investigated relationships between breast MRI and therapeutic mastectomy and contralateral prophylactic mastectomy (CPM) in women with breast cancer.
Study Design: Retrospective analysis of diagnostic and surgical treatment data for 2 cohorts of women diagnosed with breast cancer in two time periods:
1998 through 2000 (early period) when use of preoperative MRI was uncommon and 2003 through
2005 (late period) when its use was becoming more common. Patients were identified and patient demographic and clinical characteristics were obtained from our institution's tumor registry. MRI utilization in the diagnostic evaluation was obtained from a prospective radiology database. Rates of therapeutic mastectomy, CPM and associations with breast MRI were compared between the two time periods using multiple logistic regressions controlling for disease stage, age, family history, and calendar year of diagnosis.
Population Studied: The study population consisted of all women diagnosed with primary unilateral DCIS, stage I, II, or III breast cancer treated at the Magee-Womens Hospital of the
University of Pittsburgh Medical Center during the early (n=1,743) or late (n= 1,863) time periods.
Women with bilateral disease were excluded from the study.
Principal Findings: 14.2% of women underwent
MRI, 29.0% had mastectomy, and 5.3% had CPM.
Use of breast MRI increased substantially between the two time periods (4.1% to 23.7%, p<0.001). The rate of mastectomy grew slightly from 28% to 30%, but the change was not statistically significant
(p=0.184). The rate of CPM increased by more than
50% from the early to late period (4.1% to 6.4%, p<0.002). Women who underwent MRI were substantially more likely to undergo mastectomy
(38.5% versus 27.5%, p<0.001) and nearly twice as likely to have CPM (9.2 vs. 4.7%, p<0.001).
Multivariate models found MRI was associated with increased rates of mastectomy for women with stage I disease (OR = 1.430 p=0.005) and CPM for women with stage I or II disease (OR 2.036 p=
0.001). Limiting our sample to those women receiving a therapeutic mastectomy, we found a statistically significant association between MRI and
CPM for women who had stage I disease (OR =
1.754 p = 0.031).
Conclusion: Our results suggest that the preoperative use of MRI as part of the diagnostic workup for breast cancer changes the surgical treatment of breast cancer among subsets of women diagnosed with breast cancer.
Implications for Policy, Delivery or Practice:
Pre-operative breast MRI has become an increasingly common part of the diagnostic evaluation of an identified breast cancer without a thorough evaluation of its effects on women’s wellbeing, including long-term outcomes and quality of life. The associations between breast MRI and mastectomies, both therapeutic and CPM, suggest there are hidden monetary and nonmonetary costs from its use.
Theme: Gender and Health
● Patients Cared for by Physicians with a
Tendency Toward Shorter Lengths-of-stay May
Have Increased Mortality
William Southern, M.D., M.S.; Eran Bellin, M.D.;
Julia Arnsten, M.D., M.P.H.
Presented by: William Southern, M.D., M.S.,
Hospitalist, Division of General Internal Medicine,
Montefiore Medical Center/Albert Einstein College of Medicine, 111 East 210th Street, Bronx, NY
10467, Phone: (718) 920-4774; Email: wsouther@montefiore.org
Research Objective: Since Medicare introduced the prospective payment system in 1982, hospitals have been financially incentivized to reduce inpatient length-of-stay (LOS). Though average
LOS for hospitalizations has since become shorter, few studies have examined the impact of shorter
LOS on outcomes of care. Such studies are particularly difficult because of the need to eliminate confounding. We used a unique study design to examine associations between LOS and outcomes of care. Our analysis was based on physician LOS tendencies, which were determined for each physician and then used to group hospital admissions and assess associations with readmission and mortality.
Study Design: On the teaching service at our institution, at the time of admission, patients are
assigned to an attending physician without bias.
For this analysis, we included only attending physicians to whom admissions had been assigned in both years of the study, and whose mean LOS was not significantly different between study years
1 and 2, and then we divided the attending physicians into quartiles based on their mean LOS during the study period. Mean LOS for physicians ranged from 3.9 to 4.8 days in the first quartile, 4.9 to 5.2 days in the second quartile, 5.3 to 5.9 days in the third quartile, and 6.1 to 8.2 days in the highest quartile. Next, we created four admission groups, which we defined according to the physician to whom the admission had been assigned.
Admissions were thus assigned to an attending physician group that shared a LOS tendency. We then compared admissions assigned to physicians in the highest LOS quartile group (the longest LOS physicians) to admissions assigned to physicians in the lower three quartiles (the shorter LOS physicians), with respect to baseline characteristics
(age, sex, race/ethnicity, insurance, Charlson comorbidity score, number of prior admissions, and admission albumin), and 30-day readmission and mortality rates, using t-tests, chi-squared, and
Wilcoxon rank-sum tests. Finally, we constructed mixed-effects logistic regression models using physician random intercepts to assess independent associations between physician LOS tendency and readmission and mortality rates.
Population Studied: All admissions to the medical teaching service of a single large urban medical center from 7/1/02 through 6/30/04.
Principal Findings: There were 3221 admissions and 23 physicians during the study period. 2763 admissions were assigned to physicians in the lower three quartiles of LOS (shorter LOS physicians), and 458 admissions were assigned to physicians in the highest quartile of LOS (longest
LOS physicians). There were no significant differences between admission groups with respect to age, sex, race/ethnicity, insurance, Charlson comorbidity score, number of prior admissions, or admission albumin. There was also no difference in
30-day readmission rate between admissions assigned to longest LOS physicians and admissions assigned to shorter LOS physicians
(16.4% vs. 16.5%, p = 0.96). In univariate analysis, admissions assigned to longest LOS physicians had a lower 30-day mortality rate than admissions assigned to shorter LOS physicians, but this difference was not significant (7.4% vs. 9.6%, p =
0.13). After adjustment for covariates (age,
Charlson score, admission albumin), admissions assigned to the longest LOS physicians had lower
30-day mortality than patients assigned to shorter
LOS physicians (OR 0.74, 95% CI 0.47-1.12), but this difference was still not significant.
Conclusion: We used a unique study design to examine associations between physician LOS tendency and mortality, but had limited power to detect differences between patient groups.
Compared to admissions assigned to physicians with shorter length of stay tendencies, admissions assigned to physicians with the longest LOS tendency had lower 30-day mortality, but the difference was not significant. Future studies should apply this study design to a larger sample size to better assess the association between physician length of stay tendency and patient outcomes.
Implications for Policy, Delivery or Practice:
Systems that incentivize shorter LOS may lead to worse patient outcomes.
Funding Source(s): National Center for Research
Resources
Theme: Quality and Efficiency: Policies and
Incentives
● Exploring the Impact on Patient Care & the
Workforce of a New ‘Higher-Level’ Nurse
Support Worker Introduced into Acute Hospital
Ward-Based Nursing Teams in the UK
Karen Spilsbury, B.A., Ph.D.
Presented by: Karen Spilsbury, B.A., Ph.D., Senior
Research Fellow, Department of Health Sciences,
The University of York, Area 2 Seebohm Rowntree
Building, York, YO10 5DD, UK, Phone: +00 44
1904 321331; Email: ks25@york.ac.uk
Research Objective: Changing societal expectations and beliefs, new ways of perceiving health and illness, the introduction of new technologies and nurse staffing shortages have created global changes in nursing occupational structures and roles. Changes in occupational boundaries occur through (i) identification of new areas of work and (ii) the transfer of work between occupational groups. These changes have important implications for patient care and outcomes and the entire nursing workforce, both registered and non-registered. The last decade has witnessed significant growth in nursing assistant workforce numbers and their scope of practice. An important assistant worker development in the
United Kingdom (UK) is the Assistant Practitioner
(AP) role. APs are ‘higher level’ assistant workers, introduced in the UK to complement the work of registered professionals and work across professional groups in both hospital and community settings. APs have a remit to deliver protocolbased clinical care, under the direction and supervision of a state registered practitioner, and cover activities previously associated with the work of registered practitioners. We are conducting the first national evaluation of this role to understand how its introduction into ward-based nursing teams in acute hospital settings is impacting on the organization, management and quality of patient care.
Study Design: In-depth case studies of three UK acute hospitals (total 13 wards) have been completed (April 2007-December 2008) to establish
the impact of ward-based AP roles at varying stages of implementation. Mixed quantitative and qualitative methods have been used including documentary analysis, questionnaires, focus groups, interviews, direct observation of activities and quality of care delivery.
Population Studied: The population studied are
Assistant Practitioners working in ward-based nursing teams in acute hospital settings in the UK.
However, these workers are not considered in isolation but located within their occupational structures. Therefore, registered nurses, other nursing assistant staff, physiotherapists, occupational therapists and nurse managers have been studied.
Principal Findings: This paper explores and describes the extent of AP role development in acute hospitals in the UK and the impact of the role on the existing workforce, service delivery and patient care. It highlights relationships between formal policy expectations of the role, expectations of colleagues in practice and factors that act as facilitators or barriers to development, implementation and integration of the AP role. The relevance of these findings for international health care contexts is described by considering this ‘new’ assistant worker role development within occupational roles and structures, the contexts of care and service delivery to patients.
Conclusion: These findings will help policy makers, organisations and practitioners understand factors that enable and or inhibit the integration of
‘new’ assistant roles within existing occupational structures to develop and innovate service delivery and enhance patient care. The findings provide evidence of the wider impacts of introducing new roles on occupational structures, workload and practices and their potential impact on patient care.
These factors are important when considering how care will be delivered to maximise the skills of the entire nursing workforce.
Implications for Policy, Delivery or Practice:
Workforce planning is a key for the successful organisation and management of health care delivery internationally. This study makes an important contribution to the evidence-base about introduction and development of ‘new’ assistant workers and adds to broader knowledge base on optimal utilisation of the health care workforce.
Funding Source(s): NIHR Service Delivery and
Organisation (UK)
Theme: Health Care Workforce
● Mapping the Evidence-Base for a Relationship
Between Nurse Staffing & Quality of Nursing
Home Care for Resident: Informing Future
Research Agendas
Karen Spilsbury, B.A., Ph.D.
Presented by: Karen Spilsbury, B.A., Ph.D., Senior
Research Fellow, Department of Health Sciences,
The University of York, Area 2 Seebohm Rowntree
Building, York, YO10 5DD, UK, Phone: +00 44
1904 321331; Email: ks25@york.ac.uk
Research Objective: Care homes (nursing) - also called long-term, aged or skilled care facilities – have an important role in the provision of care for dependent older people. Ensuring quality of care for residents in these settings is the subject of ongoing international debates. Poor quality care has been associated with inadequate nurse staffing and poor skills mix. It is therefore important to understand more about the relationship between care home nurse staffing (proportion of RNs and support workers) and how this affects quality of care for nursing home residents. This review maps the evidence-base and explores methodological lessons for studying this topic within an international context.
Study Design: A systematic mapping review. A total of 13,411 references were identified from 7
OVID databases (using a search strategy). These references were screened to meet inclusion criteria.
A total of 80 papers were subjected to full scrutiny and checked for additional references (n=3). Of the
83 papers, 50 were included. Papers included and excluded were checked by a second reviewer.
Data were extracted by one reviewer (KS) and checked by another.
Population Studied: Studies were selected if they were carried out post-1980 and evaluated the relationship between nurse staffing and the quality of care delivered to nursing home residents or residents’ outcomes. Studies of residential care homes were excluded. The concept of quality was not limited so as to incorporate all potentially relevant studies.
Principal Findings: A growing body of literature is examining the relationships between nurse staffing levels in nursing homes and quality of care provided to residents. The research evidence to date has focused predominantly on US nursing facilities. It is difficult to offer conclusions and recommendations based on the existing research evidence: research has produced inconsistent and contradictory results. Main reasons for this are that the literature presents (i) a wide range and varied mass of findings that (ii) use disparate methods for defining and measuring quality (42 measures of quality identified) and nurse staffing (52 ways of measuring staffing identified). A focus on numbers of nurses fails to address the influence of other staffing factors (such as turnover, staffing levels, worker stability and agency staff use), the training or experience of individuals, or the ways in which care is organised and managed. ‘Quality’ is a difficult concept to capture directly and the measures are often used as a proxy and focus mainly on ‘clinical’ outcomes for residents. Only limited studies take a broader view of resident outcomes (such as quality of life measures and social indicators).
Conclusion: This systematic mapping review highlights important lessons for future international
studies. This includes meaningful measurements of quality for different key stakeholders with a mix of quality of care and quality of life indicators. In addition, there should be consideration of which measures of nurse staffing could routinely be collected across studies for the purposes of comparisons as the evidence-base grows.
Implications for Policy, Delivery or Practice:
This study makes an important contribution to the evidence-base of the relationship between the nursing workforce and quality of care and resident outcomes in nursing home settings and outlines methodological challenges for future studies of this topic internationally.
Funding Source(s): NIHR Post Doctoral Personal
Award (UK)
Theme: Long Term Care
● Fostering Quality Improvement at Primary
Care Practices Using the Chronic Care Model
Joseph Stankaitis, M.D., M.P.H.; Deborah Peartree,
M.S., B.S.N.; Howard Brill, Ph.D.; Laura Weisbein,
Ph.D.
Presented by: Joseph Stankaitis, M.D., M.P.H.,
Chief Medical Officer, Monroe Plan for Medical
Care, 2700 Elmwood Avenue, Rochester, NY
14618, Phone: (585) 256-8425; Email: jstankaitis@monroeplan.com
Research Objective: It is often suggested that primary care practices should adopt the Chronic
Care Model (CCM) to provide their patients with high-quality coordinated care. But many practices, particularly safety net practices, do not have resources to implement the changes necessary to work toward this goal. This study evaluated a framework for providing support through a nontraditional partnership in which Monroe Plan for
Medical Care assisted practices with clinical transformation and practice improvement activities to help them move toward the CCM.
Study Design: A needs assessment was conducted for each practice. For two years,
Monroe Plan helped practices implement the CCM, engage in open access scheduling, develop a team environment for process improvements, and implement PDSA (Plan, Do, Study, Act) cycles.
Monroe Plan also provided infrastructure support for clinical services at some practices. It was hypothesized that improvement at the practices would increase patient satisfaction and ultimately lead to better health outcomes. Change at the practices was assessed through structured interviews at the beginning and end of the initiative using a modified version of the Assessment of
Chronic Illness Care (ACIC). Patient outcomes were measured by surveys (based on the SF-36v.2) as well as by utilization, cost, and quality data.
Population Studied: The study included nine practices that serve mainly the low-income population in urban Rochester, NY. Most serve a significant proportion of African-American, Latino, or refugee patients.
Principal Findings: Qualitative data showed that practice characteristics leading to successful change included strong leadership, champions for change, stability, team cohesion, clear aims and strategies to meet them, acceptance of the scientific method and measurement, realization that change takes time, and commitment to service (i.e., not focusing on the academic side of medicine).
Between effects regression analysis found that practice change as measured by increase in ACIC scores was linked with increased patient satisfaction in a small but statistically significant way (at the p<.05 level in the general sample and the p<.01 level in the chronically ill sample).
Utilization, costs, and health outcomes at these practices did not differ significantly from seven comparison practices (which had nearly identical chronicity of their patient populations). While health status did not improve during the initiative period, quality did not suffer and costs did not increase during the transition period while changes were being implemented.
Conclusion: Practices were able to improve in a way that increased patient satisfaction. Clinical outcomes may take more time to become evident.
Change at the practice level should be conceptualized as a process: improvements need time to evolve and take root. Furthermore, practices embrace change in their own way and at their own pace.
Implications for Policy, Delivery or Practice:
Increased patient satisfaction is connected to practices’ financial well-being, which may serve as an incentive for change. The longer-term impact on clinical outcomes likely to accompany the CCM should be assessed in future research. Each practice’s environment needs to be understood by those involved in change planning; a practice’s priorities, resources, structure, and personnel may greatly affect its ability to change.
Funding Source(s): Center for Health Care
Strategies
Theme: Quality and Efficiency: Organized
Processes
● Low Burden, High Impact: Integration of
Technology Improves Community Health
Practice
Elizabeth Stanton; Deborah Main, Ph.D.; Wilson
Pace, M.D.; Elizabeth Whitley, Ph.D., R.N.
Presented by: Elizabeth Stanton, Department of
Family Medicine, University of Colorado Denver,
12631 East 17th Avenue, Mail Stop F496, Aurora,
CO 80045, Phone: (303) 724-9717; Email: elizabeth.staton@ucdenver.edu
Research Objective: We designed and implemented a unique electronic data collection and management system for a statewide cardiovascular
disease (CVD) risk reduction program in 14
Colorado communities. The program's goal was to reduce CVD risk by conducting risk assessment for clients, providing risk-reduction counseling, followup with clients to facilitate behavior change, and directing clients to local health and wellness resources. The software system was designed to limit the data collection burden on the community health workers (CHWs), facilitate follow up with clients, and collect the data necessary for a robust evaluation of the program.
Study Design: CHWs use pen-tablet computers to enter all screening and other client data. The software system helps the CHW manage their work by guiding them through the screening process, calculates a Framingham risk score, tracks both medical and community referral resources and tracks referrals made at the individual level. The pen-tablet tracking system creates follow up reminders for CHWs and summary reports of all screenings for program managers to monitor CHW activities. The pen-tablet interface can support multiple languages and is currently available in
English and Spanish.
Population Studied: The program currently includes data from over 12,000 individuals.
Principal Findings: The software system consists of a centralized Microsoft SQL server database that supports both a secure administrative web and a
CHW-specific, distributed database. The distributed database contains a full dataset of all screened individuals for each CHW, and is housed on a
Toshiba M700/T8100XP pen-tablet computer used by each CHW. The distributed database operates in a stand-alone fashion to guide the screening and tracking processes of the CHWs. The distributed databases are synchronized weekly with the central database through a secure Internet connection.
The system is fully HIPAA compliant. The centralized database is located behind a double firewall-protected server with highly restricted access. All outside connectivity is handled through a separate web server that also sits behind the firewall. The pen-tablet computers are password protected, and the local database is encrypted.
The CHWs report great satisfaction with the pentablet computers due to the portability, ease of data entry, and convenience of the BMI and Framingham risk calculations and the prompts built into the program. However, Internet connectivity, required for synchronizing the databases, proved to be challenging for some of our rural and frontier community partners. Ultimately, all were able to sync, even if it required going to an internet café.
Conclusion: Data are used for program management, evaluation and for program improvement. The program management and evaluators observed that the use of pen-tablet computers improved data quality and quantity over other CHW projects using alternative data collection and management strategies. Additionally, the pentablet approach made data transfer, quality improvement and program evaluation almost seamless. The cost per pen-tablet computer was approximately $1,600. Programming and support costs were almost $36,000 over three years with the majority of the expense during the start up phase of the program. We continue to refine the program for use with other research and service projects.
Implications for Policy, Delivery or Practice:
Pen-tablet computers are convenient for data collection in multiple community settings.
Funding Source(s): Cancer, Cardiovascular
Dease and Pulmonary Disease Competitive Grants
Program, Colorado Department of Public Health and Environment
Theme: Health Information Technology
● Barriers & Motivators to Using Patient
Experience Survey Data for Improving Quality:
Experiences of Physician Group Leaders in
Massachusetts
Gillian SteelFisher, Ph.D., M.Sc.; Mark Friedberg,
M.D., M.P.P.; Melinda Karp, M.B.A.; Eric Schneider,
M.D., M.S.
Presented by: Gillian SteelFisher, Ph.D., M.Sc.,
Research Scientist, Health Policy & Management,
Harvard School of Public Health, 677 Huntington
Avenue, Boston, MA 02115, Phone: (617) 432-
2859; Email: gsteel@hsph.harvard.edu
Research Objective: Surveys of patient experiences (or satisfaction) are frequently used to guide leaders of health organizations in making improvements in the quality of care. However, little is known about what motivates leaders to use this information for quality improvement (QI), and what barriers prevent them from doing so. Since 2005,
Massachusetts Health Quality Partners (MHQP) has conducted and publicly reported results from its biennial statewide survey of patients’ experiences with primary care physicians. We sought to measure the extent to which physician groups are using these PES results for QI; to gather data on group leaders’ perceptions of motivators and barriers to PES-based QI; and to determine whether motivators and barriers differed between groups that are using PES results for QI and those that are not.
Study Design: From June to November 2008, we surveyed leaders of all 122 Massachusetts physician groups containing at least 3 physicians and providing primary care to adults. In 30-minute semi-structured interviews, respondents discussed whether their groups were using PES reports to guide QI and perceived motivators and barriers to doing so. Groups were divided into those using
PES to guide QI and those who were not.
Differences between these groups in the frequencies with which respondents mentioned particular motivators and barriers were assessed using Chi squared tests.
Population Studied: Leaders of physician groups in Massachusetts
Principal Findings: We obtained responses from
72 physician group leaders (62% response rate).
Two thirds of them (67%) were using PES reports for QI and the remaining third (33%) were not.
Across all respondents, the five most frequently mentioned motivators for using PES data for QI included professional pride (37%), organizational culture (37%), financial incentives (30%), the desire to recruit or retain patients (27%) and pressure from public release (19%). Groups using PES reports to implement QI were more likely than their counterparts to note that organizational culture was an important motivator (45% vs. 21%; p=.05) and less likely to mention financial incentives as motivators (22% vs. 46%; p=.04). Across all respondents, the five most commonly reported barriers to using PES data for QI included physician resistance (40%), too little time or money (37%), lack of financial incentives (16%), problems with the survey reports (15%), and lack of competition from other physician groups (14%). Groups using PES for QI were less likely than their counterparts to mention a lack of financial incentives as a barrier
(8% vs. 33%; p=.01).
Conclusion: Physician group leaders appear to be motivated to use PES reports for QI by a number of factors relating to group culture, physician attitudes, financial factors, competition, and aspects of the
PES reports. Groups conducting QI are more likely to focus on motivators and barriers related to culture and physician attitudes, while other groups focus on financial factors.
Implications for Policy, Delivery or Practice:
Financial incentives may be needed to motivate groups that have not taken advantage of PES data for QI. Organizations measuring patient experience need to more clearly demonstrate the relevance of the results to physician group leaders in order to support them in efforts to overcome physician resistance.
Funding Source(s): CWF
Theme: Quality and Efficiency: Policies and
Incentives
● Substance Abuse Detoxification & Residential
Treatment: Rates & Duration of Subsequent
Treatment
Bradley Stein, M.D., Ph. D.; Jane Kogan, Ph.D.;
Mark Sorbero, M.S.
Presented by: Bradley Stein, M.D., Ph. D.,
Associate Professor, Psychiatry, University of
Pittsburgh School of Medicine, 3811 O'Hara Street,
Pittsburgh, PA 15213, Phone: (412) 454-8633;
Email: steinbd@upmc.edu
Research Objective: Individuals who receive intensive substance use treatment may not receive the necessary care subsequent to their discharge from that service [Fulco et al., 1995; Stein et al.,2000]. Although treatment for substance use can be as effective as treatment for other chronic disorders, fragmented or insufficient care is likely to result in less than optimal outcomes [McLellan et al., 2000]. The aim of this study is to provide empirical information to clinicians and policymakers by examining rates and predictors of treatment engagement and retention subsequent to detoxification and residential substance abuse treatment to help improve the quality of care and outcomes for Medicaid-enrolled adults being discharged from these treatment services.
Study Design: We conducted univariate and bivariate analysis, examining the relationship between receiving follow-up care within the 30 days following discharge from detoxification and residential treatment and race, Medicaid eligibility category, age, gender, and comorbid mental health disorders. Cox proportional hazard models were used to examine sociodemographic predictors of duration in treatment for those individuals receiving follow-up care within 30 days of detoxification or residential treatment.
Population Studied: Using administrative data from the largest Medicaid managed behavioral health organization (MBHO) in a large mid-Atlantic state, we identified 8005 Medicaid-enrolled adults
(age 18-64) with a new episode of detoxification
(inpatient or outpatient) or residential substance abuse treatment during 2004-2006.
Principal Findings: Slightly less than half (47%) of the sample received follow-up care within 30 days of discharge. Rates of follow-up were significantly higher in individuals with a comorbid psychiatric disorder, and significantly lower in African-American individuals, males, individuals with disabilities, and those who received detoxification without residential treatment. Duration of treatment was longer among individuals with comorbid schizophrenia, females, and Caucasians. Even after controlling for individuals’ socio-demographic and clinical characteristics, there was substantial variation in time to follow-up among discharging providers.
Conclusion: The relatively low rates of follow-up care and brief duration of treatment for many of those who received follow-up care is concerning in a population receiving intense substance abuse treatment. The markedly lower rates among those receiving detoxification alone without subsequent residential treatment and among those without a comorbid mental health disorder suggest that efforts targeting those individuals may be of particular benefit. Though it is unrealistic to expect every individual undergoing detoxification to participate in subsequent treatment, our findings highlight the importance of targeting treatment retention efforts to individuals being discharged to the community from detoxification.
Implications for Policy, Delivery or Practice: The ability to use existing information to identify the populations and providers having the most difficulty in successfully transitioning patients from
detoxification and residential treatment settings to subsequent care will allow states and other oversight entities to work with clinicians in developing and implementing interventions designed to increase the rate and duration of followup care to those individuals most in need of such programs. Successful efforts will require an integrated effort with multiple strategies and different approaches to implement effective practices across a range of individuals and organizations, all of whom can make an important contribution to improving the care of individuals with serious substance use disorders.
Funding Source(s): Community Care Behavioral
Health Organization
Theme: Behavioral Health
● Continuity of Care Following Hospitalization for Bipolar Disorder in Private Payer Claims
Michael Stensland, Ph.D.; Xiaomei Peng, M.D.
Presented by: Michael Stensland, Ph.D., Sr.
Research Scientist, Global Health Outcomes, Eli
Lilly & Company, Lilly Corporate Center, D.C. 4133,
Indianapolis, IN 46285, Phone: (317) 433-7119;
Email: stenslandmd@lilly.com
Research Objective: Poor continuity of care occurs when patients do not follow up after being discharged from the hospital. In bipolar disorder, problems with continuity of care and medication adherence may lead to expensive relapses. The objectives of this research were threefold: (1) examine the frequency of poor continuity of care for individuals with bipolar disorder, (2) identify the predictors of poor continuity of care, and (3) evaluate the potential increased psychiatric hospitalization costs.
Study Design: We identified 319 individuals who had a primary discharge diagnosis of bipolar disorder in the PCD and who maintained continuous eligibility for the 60 days prior and the
180 days following the hospitalization. Good continuity of care was operationally defined as the presence of a psychiatric visit in the 60 days following discharge. Propensity scores were used to correct for background differences between patients with good continuity of care and patients with poor continuity of care.
Population Studied: Premier linked their
PerspectiveTM Comparative Database (PCD), the largest, most detailed, U.S. hospital database, with the administrative claims from a large managedcare organization.
Principal Findings: 34.5% of individuals had poor continuity of care. Prior to discharge, individuals who would later have poor continuity of care could be identified. Some variables that independently predicted later poor continuity of care included no psychotherapy visits, no psychiatric visits, substance use diagnoses, and psychiatric hospitalizations. After correcting for background differences, patients with poor continuity of care had reduced mood stabilizer use (94 vs. 116 days, p=.008) and twice the psychiatric hospitalization charges ($10,027 vs. $4,892, p=.03).
Conclusion: One in three individuals with bipolar disorder did not have a psychiatric follow-up visit after discharge from the hospital. This poor continuity of care appears to lead to decreased medication adherence and increased psychiatric hospitalizations.
Implications for Policy, Delivery or Practice: An effective, targeted intervention could potentially prevent relapses and reduce health care costs.
Funding Source(s): Eli Lilly and Company
● Identifying Potentially Cost Effective Chronic
Care Programs for People with Heart Failure,
Diabetes or COPD
Lotte Steuten, Ph.D., M.Sc.; Bert Vrijhoef, Ph.D.,
M.S.
Presented by: Lotte Steuten, Ph.D., M.Sc., Senior
Research Fellow, MacColl Institute / Department of
Health Organization, Policy & Economics, Group
Health (WA)/Maastricht University Medical Centre
(The Netherlands), 5789 Wimsey Lane, Bainbridge
Island, WA 98110, Phone: (206) 842-7791; Email: lotte.steuten@beoz.unimaas.nl
Research Objective: Objective: to review published evidence regarding the cost effectiveness of multi-component chronic care programs and to illustrate how potentially cost effective programs can be identified.
Study Design: Methods: systematic literature search.
Population Studied: Medline and Cochrane databases were electronically searched for evaluations of multi-component disease management or chronic care programs for adults with heart failure, diabetes or COPD, describing process, intermediate and end results of care. Data were independently extracted by two reviewers and descriptively summarised.
Principal Findings: Results: 132 original articles and 35 reviews were included. There is evidence for significant improvements in process and intermediate outcomes for heart failure and diabetes, but less so for COPD. Overall, programs generate end results equivalent to usual care, but programs containing more than 2 components show significantly lower relative risks for hospitalisation than less comprehensive programs and usual care.
There is limited scope for programs to break-even or save money.
Conclusion: Conclusion: There is limited published evidence that multi-component chronic care programs currently save money, but some programs may be considered cost effective depending on the willingness-to-pay threshold for improved quality care and quality of life. Identifying
potentially cost-effective programs remains a challenge due to scarce methodologically sound studies that demonstrate significant improvements on process, intermediate and end results of care.
Implications for Policy, Delivery or Practice:
Estimations of potential cost effectiveness of specific programs, illustrated in this presentation, may in the absence of ‘perfect data’ support timely decision-making regarding these programs.
Nevertheless, well-designed health economic studies are needed to decrease the current decision uncertainty.
Theme: Quality and Efficiency: Organized
Processes
● The Health, Education, Advocacy & Law
(HEAL) Project: Medical-Legal Collaboration to
Alleviate Health Disparities
Brenda Stevenson Marshall, Ph.D., M.P.H., M.A.E.;
David Neumeyer, J.D.; Koren Goodman, M.S.Ed.,
Ph.D.
Presented by: Brenda Stevenson Marshall, Ph.D.,
M.P.H., M.A.E., Associate Dean, Community &
Environmental Health, Old Dominion University,
2114 HLTH Sciences Building, Norfolk, VA 23529,
Phone: (757) 683-6482; Email: bmarshal@odu.edu
Research Objective: This study seeks to
(1)evaluate the impact and the effectiveness of the
HEAL Project, an innovative medical-legal collaboration designed to address non-health problems that interfere with health treatments and disease prevention; (2) develop profiles for clients seeking HEAL services; (3)identify the most prevalent health interfering problems; (4)train and orient all health providers in the service area to
HEAL benefits.
Study Design: We used the social-ecological model (Dahlberg&Krig 2002) to develop the HEAL
Project, a two year retrospective cohort study. All clients are interviewed by especially trained attorneys in the Virginia Aid Legal Society offices.
Providers are also trained by these attorney advocates. The pre and post assessments for this study include a referral form and the client file closing form/checklist. Provider training materials also featured pre and post questionnaires and assessments. We used evidence based practice, as data extracted revealed favorable outcomes, which are operationalized here as financial recovery/avoidance and number of people assisted.
Population Studied: All individuals and families referred by physicians, social workers, free clinic staff and other health providers trained to identify candidates for the HEAL Project. Eligibility is met through residence in the Western Tidewater area of
Southern Virginia, the existence of a health problem, an interfering non-health problem, and household income below $200 of the Federal
Poverty Level. Examples of interfering non-health problems are benefit issues with Medicare or
Medicaid, broken heating, cooling and plumbing systems, wage claims, a child's special educational needs, child support, and medical powers of attorney.
Principal Findings: Over 100 individuals and/or families referred have received assistance from the
HEAL Project. Of these 100, all required legal intervention to correct a non-health problem that either caused or exacerbated a health ailment or posed a threat to the prevention of disease.
Participants self-reported problems with
Medicaid/FAMIS, Medicare, Unemployment benefits, and Social Security or SSI/Disability payments. HEAL has recovered $11,050.00 in lump-sum payments, over $48,000.00 in monthly payments and protected clients against over
$16,000 in unjust claims. Over 20 training sessions were conducted with multiple providers present, 25 regular meetings with local healthcare providers were held and 250 informational "rack" cards alerting prospective participants to the existence of
HEAL and how to access its resources were distributed.
Conclusion: The HEAL Project has made a difference in the treatment and prevention of health problems in economically disadvantaged families and individuals in the Southern VA Western
Tidewater area. The Project has also increased the purchasing power of Westerm Tidewater residents through its financial recovery efforts. With an increased demand for HEAL Services, the second year of funding will focus on training more providers and expanding the project to the Virginia Legal Aid
Society service area in Northern VA.
Implications for Policy, Delivery or Practice: The
HEAL Project demonstrated that assisting economically disadvantaged individuals and families to satisfy basic needs will bring about improved health outcomes. Over 40 of the 100 cases were resolved indicating the long term success of such a project could be substantial.
Expanding the project to include two or more service areas will permit a more rigorous assessment of the outcomes, clients served, case profiles,and strategies for case resolution. Funding has been sought for this expansion. More health providers need to receive training familiarizing them with the impact of non-health events on health problems. Long term goals include local, state, and federal legislation that acknowledges a public health responsibility to resolve those non-health problems that pose a significant deterrent to the treatment or prevention of a health problem.
Funding Source(s): OBICI Healthcare Foundation
Theme: Disparities
● Antecedents & Consequences of
Organizational Loyalty & its Role in Decreasing
Nursing Turnover
Jocelyn Steward, M.S.M.
Presented by: Jocelyn Steward, M.S.M., Graduate
Research Assistant, Administration - Health
Services, University of Alabama at Birmingham,
2909 Highland Avenue South, #803, Birmingham,
AL 35205, Phone: (706) 394-0284; Email: jsteward@uab.edu
Research Objective: Nursing turnover has resulted in a mass shortage of qualified nurses in healthcare organizations which resulted in a loss in positive organizational culture, financial consequences, and negative outcomes in patient care. Nursing turnover refers to nurses that leave (voluntarily or involuntarily) an organization in which another nurse must be hired to replace them. I viable option to reduce nursing turnover is the increase of organizational loyalty in nurses. Organizational loyalty refers to the employee’s commitment to the organization and the employee’s acceptance, involvement, and dedication towards achieving the organization’s goals. Organizational loyalty is measured by intentions to leave the organization and the actual decision to stay or leave the organization. In this project, the current literature is examined to determine what constructs are related to organizational loyalty in the nursing field, how these constructs are related, and the role of organizational loyalty in nursing turnover. The focus of this project is to develop a parsimonious model looking at what constructs thoroughly identify what organizational loyalty is in the nursing profession.
Due the high-touch and high-need environment of the nursing field, constructs in use related to organizational loyalty may not be sufficient or even applicable in nursing. This model will serve as a springboard for future research. A theoretical model that can be quantifiably researched will be presented. Constructs found in the literature related to organizational loyalty in nurses include: organizational commitment, job dissatisfaction, happiness, job embeddness, demographics, and internal and external environmental factors.
Conclusion: The research in this project is used to develop a theoretical framework to determine how each construct relates together in efforts to understand the nature of organizational loyalty in nurses. The model, although constructed with the intentions of reducing nursing turnover, is broad enough to be universally used in any organization dealing with high turnover rates.
Implications for Policy, Delivery or Practice: The model will be used as an avenue for future research and as an option for managerial practices to increase nursing organizational loyalty and decrease turnover. This research will allow healthcare organizations the opportunity to decide on the best methods to increase organizational loyalty and decrease turnover.
● Quality of Care in Homeless Patients and their
Providers: Do Priorities Differ in Veteran &
Gender Status?
Jocelyn Steward, M.S.M.; David Pollio, Ph.D.; N.
Kay Johnson-Roe, M.P.H.; Stefan Kertesz, M.D.
Presented by: Jocelyn Steward, M.S.M., Graduate
Research Assistant, Research Enhancement Award
Program, Birmingham Veterans Health
Administration, 2909 Highland Avenue South, #803,
Birmingham, AL 35205, Phone: (706) 394-0284;
Email: jsteward@uab.edu
Research Objective: Homeless person’s primary health care may not be well assessed by traditional quality of care measures. The first step in developing quality quantitative metrics for homeless primary care (PC) is to establish priorities valued by homeless patients and their health care providers.
The Veterans Health Administration (VA), a highly organized integrated healthcare system, and independent federally funded nonprofit health centers have provided homeless health care for over 20 years. Therefore, the objective of the research is to determine what constructs related to quality are most valued in primary care related to this unique group of individuals.
Study Design: Using the Institute of Medicine (IoM) reports, we identified 16 primary care quality constructs and mailed a card-sort to currently or previously homeless patients and to providers of healthcare for homeless patients. We asked participants to rank the 16 constructs from most important to least important, and to indicate gender and whether they were a veteran. We measured agreement in rankings between providers and patients, veterans and nonveterans, and males and females.
Population Studied: Card-sort packets were distributed to (1) VA patients; (2) Non-VA patients using non-VA nonprofit health centers; (3) VA providers; and (4) Non-VA providers from 16 states and the District of Columbia. Similarity between the group mean ranks was examined.
Principal Findings: Overall, we received a response rate of 71%. We found that VA patients’ top constructs and Non-VA patients’ priorities were moderately correlated with those of Non-VA providers (r=0.58; p= 0.02), but there was no correlation between VA patients and VA providers (r
= 0.0; p = 0.99). VA patients were distinct in prioritizing transparency and safety, compared to
Non-VA patients. Patient control of care was prioritized highly by all groups other than VA providers. Priority rankings were strongly correlated between men and women (r = 0.86588; p
= <0.0001); and between veterans and nonveterans
(r = 0.78081; p = 0.0004); there was high correlation by gender among providers (r =
0.82609; p =<.0001); and moderate correlation by gender among patients (r = 0.55882; p = 0.0244).
Both veterans and non-veterans valued evidencebased decision making very highly and both males and females highly prioritized accessibility of care.
Conclusion: Primary care quality priorities are not strongly correlated between homeless patients and providers in this sample, and correlation was even lower in the VA setting. Conversely, in samples combining providers and patients, neither veteran status nor gender predicted disagreement. Samples in this qualitative analysis were small and convenience-based, but geographically diverse.
Implications for Policy, Delivery or Practice:
Many constructs relevant to quality have been elaborated by the Institute of Medicine.
Operationalizing these constructs to assess the care of special populations may require special attention to differences in the priorities held by patients and their providers, and to differences the setting in which are is obtained.
Funding Source(s): VA
● The Impact of Patient-Physician Racial
Concordance on Press Ganey Satisfaction
Scores
Jessica Stewart, B.A.; Diane Howard, Ph.D.;
Kimberly Skarupski, Ph.D.; Francis Fullam, M.A.
Presented by: Jessica Stewart, B.A., Intern, Health
Systems Management, Rush University Medical
Center, Rush University Medical Center, Chicago,
IL 60612, Phone: (773) 946-9162; Email: jessica_stewart@rush.edu
Research Objective: This study aimed to determine whether or not racial/ethnic concordance between patients and physicians results in greater patient satisfaction with physicians than discordant patient-physician dyads. Additionally, the study assessed whether or not patient-physician racial/ethnic concordance impacts the patient’s satisfaction with their health care experience overall. The study will use results from Press Ganey satisfaction surveys to measure patient satisfaction and quality of care.
Study Design: We employed a secondary data analysis using a cross-sectional study design of data from the Rush University Internist outpatient clinic during fiscal years 2004-2008. A simple t-test was conducted to test the null hypothesis that mean patient satisfaction scores do not differ between racially concordant and racially discordant physician- patient dyads. The independent variable was patient- physician racial/ ethnic concordance to which there are two levels; the patient-physician dyad can be either racially/ethnically concordant or racially/ethnically discordant.The dependent variables for this study were overall patient satisfaction and the patient’s satisfaction with their provider. A linear regression model was run to control for age and sex.
Population Studied: Rush University Internist physicians and patients between fiscal years 2004-
2009 (July 1, 2003 - June, 30, 2008).
Principal Findings: The principle findings are forthcoming and will be available by the end of
February 2009. The findings will show statistically to what extent racial/ethnic concordance influences the patient’s perception of quality care which is measured through Press Ganey satisfaction surveys. Based on my review of the available literature I hypothesize that patient-physician racial/ethnic concordance has a statistically significant impact on Press Ganey satisfaction scores.
Conclusion: Although some preliminary results are available I am not prepared at this time to state any final conclusions. To maintain the integrity of the research I will wait until all analyses of the data have been completed.
Implications for Policy, Delivery or Practice:
This study highlights the importance of training and recruiting minority physicians. However, given the shortage of minority physicians in America, many healthcare organizations are serving diverse populations with limited opportunities to increase the diversity of their medical staff. Thus, in understanding the impact of patient-physician racial concordance, managers might be inclined to focus on developing ways to increase the cultural competency of their practicing physicians.
● Geographic Access to OB/GYN Care: Does
Medicaid Plan Type Make a Difference?
John Stewart, M.S., M.P.H.; Ana, Lopez De Fede,
Ph.D.; Kathy Mayfield-Smith, M.A., M.B.A.; Qiduan
Liu, Ph.D.
Presented by: John Stewart, M.S., M.P.H., Senior
GIS Research Associate, Institute for Families in
Society, University of South Carolina, 1600
Hampton Street, Columbia, SC 29208, Phone:
(803) 777-5516; Email: jstewart@mailbox.sc.edu
Research Objective: Geographic proximity to health care providers is an important component of access to health care services. State managed care programs set standards within their managed care plans to ensure enrollees have adequate access to care within provider networks. Although geographic access to health care is critical for all women, proximity to care may be especially important for low income, pregnant women. This study explores geographic access to health care for
Medicaid women needing OB/GYN services. In particular, the study examines road distance to
OB/GYN care for Medicaid enrollees with live births in South Carolina in fiscal year 2008. Separate analyses compare distance to OB/GYN care for
Medicaid women receiving fee-for-service versus managed care plan health services.
Study Design: Medicaid OB/GYN providers and women in Medicaid with a delivery claim in fiscal year 2008 were geocoded (spatially located) using
MapInfo MapMarker Plus. The ESRI ArcView extension, Shortest Network Path, was used to calculate the shortest distance on the South
Carolina road network (comprised of all local
streets, roads, state, U.S. and Interstate highways) between each individual study population residence and the nearest OB/GYN provider. Notably, these computationally intensive spatial analyses generated point-to-point travel distances representing Medicaid recipients’ optimal path of travel to OB-GYN providers, rather than generalized straight line or zone-based distance-to-care estimates.
Population Studied: South Carolina Medicaid women with live births in fiscal year 2008 enrolled in managed care. For comparison with the managed care population, a stratified sample of twenty percent of all South Carolina Medicaid fee-forservice women with live births was examined.
Principal Findings: Ninety-eight percent of
Medicaid recipients with live births in fiscal year
2008 had access to an OB/GYN provider within 25 miles of their place of residence. As measured across South Carolina’s four health regions, geographic access to all Medicaid OB/GYN providers for Medicaid managed care recipients was equal to or better than access for women enrolled in fee-for-service. When restricting travel to providers enrolled with their managed care plan, ninety-one percent of Medicaid managed care recipients had access to a network OB/GYN provider within 25 miles of their home.
Conclusion: Using the ESRI ArcView extension,
Shortest Network Path, to calculate optimal distance-to care, we found that over ninety percent of all Medicaid recipients lived within 25 miles of an
OB/GYN provider. This finding held true regardless of the choice of Medicaid recipients to be enrolled in a managed care plan or to remain in fee-for-service care.
Implications for Policy, Delivery or Practice:
Information about distance to OB/GYN care may be useful to state Medicaid agencies, local perinatal health planners, and perinatal health providers, particularly when approving network providers within managed care plans, locating new perinatal care facilities, recruiting OB/GYN care providers to existing facilities, or planning mobile OB/GYN transportation routes.
Funding Source(s): SC DHHS
Theme: Coverage and Access
● AHRQ-Funded Investigator-Initiated Research on Health Care Costs, Productivity,
Organization & Market Forces: Goals,
Dissemination Strategies & Impacts
Kate Stewart, Ph.D.; Tim Lake, Ph.D.; Marsha
Gold, Sc.D.
Presented by: Kate Stewart, Ph.D., Researcher,
Mathematica Policy Research, 600 Maryland
Avenue, SW, Suite 550, Washington, DC 20024,
Phone: (202) 484-4529; Email: kstewart@mathematica-mpr.com
Research Objective: Health services research
(HSR) is a well-established and growing discipline that provides important information on the healthcare system. While HSR may be used to inform policy debates and decisions, the extent to which investigators target dissemination to various end-users and the extent to which this research is used by policymakers is not well understood. AHRQ has historically been a key funder of investigatorinitiated HSR. The goals of this project were to describe recent AHRQ-funded investigator-initiated research on healthcare costs, productivity, organization and market forces, and to understand how this research was disseminated and its perceived impact.
Study Design: We conducted a web-based survey with AHRQ principal investigators (PIs) between
July and August 2008. The survey included both open- and closed-ended questions about research goals, key findings, dissemination strategies, assistance with dissemination and perceived impacts of the research. We tabulated responses to close-ended responses and analyzed open-ended responses for common themes.
Population Studied: 149 PIs who received funding from AHRQ between 1998 and 2006. Ninety-seven
PIs responded for a 70% response rate.
Principal Findings: Studies were generally designed to address policy-relevant questions
(60%) or basic knowledge or methods development
(39%). Research questions often related to issues of quality of care, costs and expenditures, healthcare access and utilization, and disparities.
Findings described how economic factors, organizational characteristics, systems and markets, public policies and patient characteristics and preferences were associated with relevant outcomes. PIs focused dissemination strategies on publication in peer-review journals (97%) and conferences (64%). They received little assistance from their institutions for dissemination activities.
Among 85 PIs who rated the impact of their study to date, 87% reported at least some impact and 44% reported large impacts. Large impacts were characterized as 1) providing important new information; 2) widely cited and used by other researchers; 3) informing policymakers and endusers through testimony and presentations; 4) actual use by policymakers; and 5) widespread media attention. While PIs may be motivated to overstate the effects, the open-ended descriptions they provided support claims of large impacts. In addition, PIs report that they spend an average of
10% of their time interacting with policymakers, but desire to spend 15% of their time in this manner.
Conclusion: AHRQ has funded a broad array of
HSR that has made its way into the policy arena.
While most investigators focus dissemination on peer-reviewed publications and conference presentations, much of this research has been picked up and used by policymakers. PIs desire to
spend more time interacting with policymakers compared to current levels.
Implications for Policy, Delivery or Practice:
AHRQ-funded investigator-initiated research has played an important role in informing research and policy. This research has made its way into the policy arena primarily through publications and presentations. Funders such as AHRQ may be able to improve the impact of research through targeted dissemination activities to relevant audiences, leveraging the desire of researchers to spend more time interacting with policymakers. Relatively few organizations fund research on broad-based healthcare systems-related issues, and AHRQ has played an important role in supporting research that can inform policymaking.
Funding Source(s): AHRQ
Theme: Health Care Markets and Competition
● Ring Maps: Multivariate Geographic Data
Presentation for Behavioral Health Planners
John Stewart, M.S., M.P.H.; Ana Lopez De Fede,
Ph.D.; Kathy Mayfield-Smith, M.A., M.B.A.; Dawn
Sudduth, M.Ed.
Presented by: John Stewart, M.S., M.P.H., Senior
GIS Research Associate, Institute for Families in
Society, University of South Carolina, 1600
Hampton Street, Columbia, SC 29208, Phone:
(803) 777-5516; Email: jstewart@mailbox.sc.edu
Research Objective: Thematic maps allow researchers to represent geographic variation in behavioral health conditions, socioeconomic status, demographic characteristics, and other data relating to the behavioral well-being of local populations.
The use of multiple thematic maps in behavioral health planning, however, can be cumbersome, both in terms of map production and interpretation.
This study uses an innovative ring map design to present on a single map the prevalence rates of seven behavioral diagnoses among Medicaid recipients across 46 South Carolina counties. The same ring map shows geographic statistical variance across diagnosis categories.
Study Design: A geographic information system,
ArcGIS 9.3, was used with Adobe Illustrator CS3 and Adobe Photoshop CS3 to create and populate with behavioral health data a county level ring map for the state of South Carolina. The ring map shows county prevalence rates of ADHD, anxiety, delirium, schizophrenia, substance abuse, developmental disorders, and mood disorders among South Carolina Medicaid recipients in fiscal year 2008.
Population Studied: South Carolina Medicaid recipients with selected behavioral health diagnoses in fiscal year 2008 (56,092 children;
28,226 adults).
Principal Findings: High prevalence rates of
ADHD, anxiety, delirium, schizophrenia, substance abuse, developmental disorders, and mood disorders tended to geographically coincide, particularly in counties in the South Carolina
Upstate. This region of the state is predominantly white and relatively affluent, with a greater number of mental health facilities and service providers.
Low prevalence rates across behavioral diagnoses also tended to geographically coincide, most notably in South Carolina Low Country counties.
This region of the state has greater economic and demographic diversity, with fewer mental health care providers
Conclusion: Relatively low prevalence rates of behavioral disorders in the South Carolina Low
Country may reflect a paucity of behavioral health resources and diagnostic services in the area rather than higher rates of behavioral well-being. The ring map is an innovative and effective data presentation tool, allowing behavioral health planners to present on a single map multiple sets of place-specific behavioral health data for both general and targeted populations (e.g., children, minorities, Medicaid managed care plan enrollees).
Implications for Policy, Delivery or Practice: Like standard thematic maps, ring maps can contribute critical information about geographic variation in behavioral health, behavioral disorders, and the behavioral health resource environment. Because they depict multiple sets of data on a single map, ring maps may be especially useful to state and local health planners, particularly when combined with other complementary quantitative and qualitative data methods.
Funding Source(s): SC DHHS
Theme: Behavioral Health
● Cost-Effectiveness of a Community Based
Patient Assistance Program
Anne Stey, M.Sc., M.D.; Kezhen Fei, M.S.; Jessica
Weidmann; Ali Mendelson; Nina Bickell, M.D.,
M.P.H.
Presented by: Anne Stey, M.Sc., M.D., Health
Policy, Mount Sinai Hospital, 1 Gustave Levy Plaza,
New York, NY 10029, Phone: (646) 895-1956;
Email: as013j@gmail.com
Research Objective: Poor, less educated and minority women with breast cancer have poorer quality of breast cancer care and worse survival.
Patient needs & barriers to breast cancer care can impede their ability to obtain and complete treatment. High quality community-based patient assistance programs to help women with breast cancer abound. Few women are aware of and access these programs. It is unknown if patient assistance programs improve the quality of life of women with breast cancer. A RCT is underway to assess whether patient assistance programs improve the quality of breast cancer care and women's health status. We performed an analysis to determine the cost effectiveness of patient assistance programs.
Study Design: We surveyed 89 women with new, primary early stage breast cancer patients in NYC and assessed their informational, psychosocial, and practical-access needs and their quality of life and health status (SF12). We derived Quality Adjusted
Life Years (QALY) from SF12 data adjusted for baseline QALY using multiple regressions and then performed bivariate comparisons of women who did and did not attend patient assistance programs. We derived costs from the annual cost per visit of New
York's Breast Health Resource Center multiplied by the hours utilized by each level of socioeconomic need using third party payer perspective. Tree-age cost-effectiveness software was used and willingness to pay was set at the USA's conventional $50,000 threshold.
Population Studied: We prospectively surveyed women status post definitive surgical procedure of a new primary breast cancer in stage I or II from eight
New York City hospitals servicing marginalized populations including Columbia, Mount Sinai,
Einstein, Montefiore, Queens Cancer, Elmhurst,
Metropolitan, Harlem Hospital. These surgical procedures included in order of hierarchy: mastectomy, axillary evaluation and Lumpectomy.
Principal Findings: Compared to women who did not attend community based patient assistance programs, women who sought patient assistance for 1 need (N=22), 2 needs (N=14), and 3 needs
(N=26) had improved quality of life (QALY=0.68 vs
0.65 p<0.000, QALY=0.69 vs 0.66 p=.302,
QALY=0.68 vs 0.632 p<0.000). Patients with no needs who attended community based patient assistance programs with no needs (N=22) actually had a loss in health status (QALY= 0.66 vs 0.65 p<0.000). On average, patients with one need used
5 hours, two needs used 16 and three needs used
28 hours of community based patient assistance.
The average cost of each hour of community based patient assistance was $73. Accordingly, treating patients with one need was $365, two needs $1148 and three needs cost $2049. The intervention, when targeted to patients with needs only, was shown to be cost-effective with an incremental cost effectiveness ration of $33,374 per quality adjusted life year.
Conclusion: Community based patient assistance program are a cost-effective way to improve quality of life in women with breast cancer who express ongoing needs even after accessing care. Patient assistance programs should be aimed towards breast cancer patients with needs.
Implications for Policy, Delivery or Practice:
Community based patient assistance programs may be a cost-effective way of addressing disparities in care and quality of life in women with substantial practical, psychosocial or informational needs.
Funding Source(s): NCI
Theme: Disparities
● Understanding Variation in Quality Indicator
Adherence & Patient Satisfaction in the
Department of Veterans Affairs
Kelly Stolzmann, M.S.; Mark Meterko, Ph.D.; David
Mohr, Ph.D.; Michael Shwartz, Ph.D.; Bert White,
M.B.A.
Presented by: Kelly Stolzmann, M.S., Research
Analyst, Center for Organization, Leadership, &
Management Research, VA Boston Healthcare
System, 150 South Huntington Avenue (Mail Stop
152M), Boston, MA 02130, Phone: (857) 364-5355;
Email: kelly.stolzmann@va.gov
Research Objective: The delivery of health care depends not only on the skills of individual providers but also on the coordination of the teams in which they practice and the structure of the work setting. We sought to determine the amount of variation in primary care adherence to quality indicators and patient satisfaction accounted for at the provider, primary care team, and medical center levels.
Study Design: Data from the Department of
Veterans Affairs (VA) External Peer Review
Program (EPRP) was used to calculate a single measure of provider adherence using twelve key performance indicators of high quality care associated with diabetes, cancer, acute myocardial infarction, hypertension and immunizations. Data from the Survey of Healthcare Experiences of
Patients (SHEP) was used to calculate a composite measure of outpatient satisfaction. Both the SHEP and EPRP programs are administered by the VA
Office of Quality and Performance. On average data was provided by 7.9 and 17.4 patients per provider for EPRP and SHEP, respectively.
Hierarchical linear models including random intercepts only with no explanatory variables were used to partition the variability in each compliance measure by level.
Population Studied: The EPRP measures were obtained from 37,304 patients of 4,715 primary care providers within 1,274 teams at 128 medical centers. The SHEP measures were obtained from
83,454 patients of 4,784 providers within 1,318 teams at 129 medical centers. VA outpatients who were included in the EPRP or SHEP programs in
FY2007 were subsequently matched to a primary care team. Providers were assigned to teams based on their assignment in the Primary Care
Management Module (PCMM).
Principal Findings: Mean provider adherence across all patients was 83.0% (SD 26.9%; interquartile range: 75.0%-100%). Mean patient satisfaction was 84.0% (SD: 18.7%; inter-quartile range: 76.6%-98.1%). For provider adherence the percent of variance accounted for at the provider, team and medical center level was 2.5%, 0.8% and
0.7%, respectively. For patient satisfaction the percent variance accounted for at the provider,
team, and medical center level was 4.0%, 2.0%, and 2.4%, respectively.
Conclusion: Overall, both provider adherence and patient satisfaction were quite high. In both of these domains, differences between patients were the predominant source of variance. The percent of variance in either domain accounted for by differences between providers would generally be regarded as noteworthy but small. Team and medical level factors each accounted for less than one percent of variance in provider adherence.
Regarding patient satisfaction, the percentages of variance accounted for at the provider, team and medical center levels were roughly double those observed for adherence, and medical center factors appeared relatively more important than team level factors.
Implications for Policy, Delivery or Practice:
Provider, team and medical center level characteristics vary in their respective influence on both the care and satisfaction of patients. These findings can help managers improve provider adherence and patient satisfaction by targeting interventions at specific organizational levels in the
VA.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● Prevalence of Illnesses Associated with
Changing Body Mass Index in Diverse
Populations
Manfred Stommel, Ph.D.
Presented by: Manfred Stommel, Ph.D., Senior
Service Fellow at National Center for Health
Statistics/Associate Professor, District Health
Information System/Nursing, National Center for
Health Statistics/Michigan State University, 3311
Toledo Road., Room # 2216, Hyattsville, MD
20782, Phone: (301) 458-4357; Email:
MStommel@cdc.gov
Research Objective: It is well established that obesity is a risk factor for many chronic diseases and conditions, including type 2 diabetes, hypertension, coronary heart disease (CHD), asthma and arthritis. Although average obesity rates are known to vary across racial and ethnic groups, it is less clear whether the health risks associated with variation in the Body Mass Index
(BMI) are uniform across these major racial and ethnic groups. Thus, at similar levels of overweight or obesity, different racial and ethnic groups may experience different risks of various chronic diseases and conditions.
Study Design: The analysis presented here relies on information combining 11 years (1997-2007) of
Sample Adult data from the National Health
Interview Survey (NHIS). The NHIS is a household, multi-stage probability sample survey conducted annually by the CDC’s National Center for Health
Statistics. The average annual conditional response rate for the 1997-2007 NHIS Sample Adult component was 83%. The BMI was computed from self-reported height and weight, but adjusted for reporting biases using estimates from the National
Health and Nutrition Examination Survey.
Information on the prevalence of chronic illnesses was obtained through survey responses. Potential confounders included in all analyses are sociodemographics (age, gender, education, marital status, income, home ownership, health insurance, urban or rural resident, born abroad or in the U.S) as well as health-related behaviors (smoking, physical activity, alcohol use).
Population Studied: The data include 219,521
Non-Hispanic whites, 47,468 Non-Hispanic African
Americans, 56,964 Hispanics and 5,553 East
Asians (of Chinese, Vietnamese, Japanese or
Korean origin).
Principal Findings: Overall prevalence estimates for the U.S. civilian resident population between
1997 and 2007 were 6.5% for type 2 diabetes,
24.7% for hypertension, 3.9% for coronary heart disease, 10.0% for asthma, and 10.3% for arthritis, with African Americans having the highest unadjusted prevalence rates for diabetes, hypertension, asthma and Non-Hispanic whites having the highest prevalence rates for CHD and arthritis. Asian Americans consistently exhibit the lowest unadjusted prevalence rates. However, the adjusted prevalence odds show changes in the rankings: Hispanics, Asians and African Americans all have higher risks of diabetes than Non-Hispanic whites (respective ORs=1.54, 1.51, 1.49, p<0.01), while Non-Hispanic whites exhibit greater risks than all minorities with respect to CHD, asthma and arthritis. The adjusted risk of hypertension is greatest among African Americans, followed by
East Asians, whites and Hispanics. Finally, an examination of the relative odds of illness among racial and ethnic groups shows a few consistent trends as the BMI increases: The prevalence odds of diabetes among Asian Americans are increasing with higher BMI relative to all other racial and ethnic groups, while Hispanics experience lower odds of
CHD with increasing BMI relative to the other racial and ethnic groups. For all other conditions examined here, the onset and magnitude of increases in risk associated with rising BMIs appear to be similar.
Conclusion: For some chronic conditions, the association between variations in BMI and health risks varies according to ethnic or racial group.
Implications for Policy, Delivery or Practice:
Clinical practice recommendations concerning health risks associated with the BMI may need to be varied according to different ethnic or racial groups.
Funding Source(s): CDC
Theme: Disparities
● Does A Targeted Bundled Approach Improve
Quality in ICUs?: A Nationwide Study
Patricia Stone, Ph.D., M.P.H., R.N., F.A.A.N.; E.
Yoko Furuya, M.D., M.S.; Eli Perencevich, M.D.,
M.S.; Monika Pogorzelska, M.P.H.; Andrew Dick,
Ph.D.
Presented by: Patricia Stone, Ph.D., M.P.H., R.N.,
F.A.A.N., Associate Professor of Nursing, School of
Nursing, Columbia University, 630 West 168th
Street, Mail Code 6, New York, NY 10032, Phone:
(212) 305-1738; Email: ps2024@columbia.edu
Research Objective: Efforts to implement bundles of interventions to improve the quality of patient care and reduce hospital infections have become popular in intensive care units (ICU). Each bundle targets a specific patient outcome. However, the importance of any single component versus the total bundle is not known, nor is it known if the focus on reducing one type of infection has benefits in reducing other infections. The objectives of this study were: a) to examine the extent of adoption of two separate infection-related intervention bundles in ICU; b) test the effectiveness of the components of the bundles on reducing infections; and c) test for potential cross-over effects of the bundles.
Study Design: As part of a larger study, a survey was sent to select hospitals participating in the
National Healthcare Safety Network. Respondents were infection preventionists trained in surveillance and monitoring of infection outcomes and processes. The following data were collected: ICU specific policies regarding adoption of a central line associated bloodstream infection (CLBSI) bundle (4 infection components) and a ventilator bundle (2 infection components and 2 general care components); how often bundle components were correctly implemented; and CLBSI and ventilator associated pneumonia (VAP) rates. Descriptive statistics and ordinary least square regressions were conducted controlling for setting characteristics.
Population Studied: 250 hospitals across the nation were enrolled (response rate 57%). The respondents provided data on 415 ICUs of which
223 (54%) were medical/surgical units.
Principal Findings: Most ICUs (n = 284, 68%) had adopted the full ventilator bundle and 204 (49%) adopted the full CLBSI bundle. The mean infection rates were 2.6 (sd = 4.1) for VAP and 2.1 (sd = 2.8) for CLBSI per 1000 device days. In ICUs where all components of the bundles were always implemented correctly (n = 73 for ventilator and n =
35 for CLBSI) there were decreased device associated infection rates (VAP, beta = -0.6, p =
0.005; CLBSI, beta = -0.2, p = 0.06). No single component was significantly related to the infection rate and implementing the 2 infection components of the ventilator bundle increased the magnitude of
VAP reduction (beta = -2.3. p = 0.001). Moreover, always implementing the ventilator bundle correctly was not significantly associated with decreased
CLBSI and vice versa.
Conclusion: Bundles of interventions are fairly widely adopted and these interventions are associated with reductions in the specific targeted infections. However, they need to be implemented correctly for an impact to be seen. Furthermore, we found no evidence that targeting one problem, such as VAP, will improve another problem, such as
CLBSI.
Implications for Policy, Delivery or Practice:
Clinicians must understand that focusing on just one infection has little crossover in preventing other infections. Policymakers must consider that different settings may have different problems when developing recommendations and regulations.
Funding Source(s): National Institute of Nursing
Research
Theme: Quality and Efficiency: Policies and
Incentives
● Public Health Systems Research in
Emergency Preparedness: Systematic
Literature Review 1997-2008
Michael Stoto, Ph.D.; Elena Savoia, M.D., M.P.H.;
Sarah Massin-Short, M.P.H.; Angiemae Rodday,
M.S.; Lisa Sahagian, M.S.C; Melissa Higdon
Presented by: Michael Stoto, Ph.D., Professor,
Health Systems Administration, Georgetown
University, 3700 Reservoir Road, NW, Washington,
DC 20007, Phone: (202) 687-3292; Email: stotom@georgetown.edu
Research Objective: To identify and characterize the U.S. public health systems research (PHSR) literature related to public health emergency preparedness (PHEP).
Study Design: Systematic literature review.
Population Studied: PHSR papers and reports published by U.S institutions regarding the preparedness of U.S. public health systems. Gray literature and publications indexed in Medline and
Embase, were included.
Principal Findings: From 1997 through 2008, 547 articles or reports meeting the inclusion criteria were published, representing a growth rate of about
33% per year. 314 (57%) articles fell into at least one of the four PHEP research priority areas identified by the IOM: 61 (11%) addressed the usefulness of training, 39 (7%) addressed communications in preparedness and response,
193 (35%) addressed sustainable preparedness and response systems, and 39 (7%) addressed criteria and metrics to measure PHEP.
The majority of the articles (82%) were based on qualitative analysis. This included commentaries or reviews (62%), and case studies (10%). Other qualitative studies (9%) used formal approaches such as focus groups, Delphi techniques, or content analysis. The most frequently used study design was cross-sectional (15%). Few publications used a quasi-experimental design (1%) or longitudinal design (1%), and no study could be classified as
having an experimental design. Mathematical models were the focus 1% of studies.
Conclusion: Coincident with the vastly increased interest in and funding for public health preparedness since September 11, 2001, the
PHSR literature on PHEP has grown exponentially at about 33% per year, with 547 articles or reports having been identified between 1997 and 2008. The range of substantive topics addressed by this literature was broad. However, as might be expected for a new field, more than half (62%) of the publications are commentaries or reviews rather than primary research. Case studies and other qualitative research comprise an additional 21% of all studies, so studies using quantitative methods comprised less than 20% of studies published during this period. While case studies and other qualitative research approaches can be rigorous, substantially less than half of the PHSR literature presents new evidence about public health preparedness. The relatively small fraction of studies using quasi-experimental and longitudinal designs (there were no true experiments) or mathematical models calls the validity of the literature as a whole into question. Moreover, a large portion of the primary research in the current literature is made up of case studies and crosssectional needs assessments, raising questions about the generalizability of the results. Thus, the promise of PHSR to improve the preparedness of public health systems has yet to be fulfilled.
Implications for Policy, Delivery or Practice:
Since September 11, 2001, the Federal government has invested over $7 billion in PHEP at the state and local level, but very little in the research needed to ensure that state and local public health systems use these resources effectively and that benchmarks and standards are correctly developed.
Continued support of PHEP research, especially with a PHSR focus and efforts to ensure its validity and generalizeability, is needed.
Funding Source(s): CDC
Theme: Public Health
● Public Health Emergency Preparedness at the
Local Level: Results of a National Survey
Michael Stoto, Ph.D.; Elena, Savoia, M.D., M.P.H.;
Angiemae Rodday, M.S.
Presented by: Michael Stoto, Ph.D., Professor,
School of Nursing & Health Studies, Georgetown
University, 3700 Reservoir Road, NW, Washington,
DC 20007, Phone: (202) 687-3292; Email: stotom@georgetown.edu
Research Objective: To study the relationship between aspects of public health organization and financing and local public health emergency preparedness (PHEP).
Study Design: Cross-sectional analysis of the
NACCHO 2005 National Profile of Local Health
Departments.
Population Studied: US Local Health Departments
(LHDs).
Principal Findings: The NACCHO profile has a sample size of 2,864 LHD and high response rate
(>80%), providing a national representative sample of LHD’s activities. Although developed to address essential public health services of all types,
NACCHO’s Profile of Local Health Departments also provides useful information on PHEP activities and capacities. Elements of public health infrastructure related to LHDs levels of PHEP are assessed in the survey: population size, presence of a Board of Health (BOH), participation in coalitions and received preparedness funding.
Across the board, LHDs in the United States that serve larger populations, and LHDs in states where more federal funds were available at the local level, are more likely to have undertaken emergency preparedness activities and have preparedness staff and capacities in place. Adjusting for population size, LHDs served by a BOH are more likely to have undertaken preparedness activities such as written an emergency response plan, reviewed legal authorities, participated in drills and exercises, assessed the emergency preparedness competencies of their staff and provided them with training. LHDs that participated in emergency preparedness coalitions were more likely to have met preparedness goals, but participation may just reflect a greater interest in preparedness.
Conclusion: The most consistent result in this analysis is that LHDs’ emergency preparedness activities and outcomes are strongly and consistently associated with the size of the population that the department serves. LHDs serving larger populations, and states with more federal funds available at the local level, are more likely to have staff and capacities in place, as well as have undertaken emergency preparedness activities. Population size, of course, is not what causes these LHDs to be better prepared, rather it likely serves as a proxy for the resources and general public health capacities available in the departments and the communities they serve.
Implications for Policy, Delivery or Practice:
Despite substantial investments in public health emergency preparedness since September 11,
2001, there is little systematic data on the impact of that investment on local health departments. To the extent that these results suggest that greater resources for public health systems in general, and investments in PHEP capacities in particular, translate into greater levels of preparedness, they provide some evidence about the effectiveness of those investments. Increasing the size of the population served is obviously not a policy lever, except if health departments merge or otherwise work together in regional structures. While not providing any direct evidence of the effectiveness of regionalization, the results in this paper are consistent with other studies showing that effectiveness is related to the size of the population
served, and at least raise the question of regionalization.
Funding Source(s): CDC
Theme: Public Health
● Predialysis Care Across Healthcare Systems among Patients with Chronic Kidney Disease
Kevin Stroupe, Ph.D.; Michael Fischer, M.D.,
M.S.P.H.; Denise Hynes, R.N., M.P.H., Ph.D.; Ann
O'Hare, M.D.; Min-Woong Sohn, Ph.D.; James
Kaufman, M.D.
Presented by: Kevin Stroupe, Ph.D., Research
Scientist, Center for Management of Complex
Chronic Care, Department of Veterans Affairs, 5000
South 5th Avenue, Hines, IL 60141, Phone: (708)
202-3557; Email: kevin.stroupe@va.gov
Research Objective: Patients with end stage renal disease (ESRD) require chronic hemodialysis to replace lost kidney function. For patients with severe chronic kidney disease, timely nephrology care before chronic dialysis initiation is associated with better health outcomes after dialysis initiation.
Moreover, clinical practice guidelines recommend that permanent vascular access (PVA) placement occur before dialysis initiation to avoid temporary catheters, which are associated with excess morbidity, mortality, hospitalization, and cost.
There is limited information about the role of the healthcare financing system (Department of
Veterans Affairs [VA], Medicare) in determining receipt of appropriate predialysis care. We examined the impact of the healthcare financing system on the likelihood of timely predialysis nephrology care and predialysis PVA placement.
Study Design: We conducted retrospective crosssectional analyses of veterans and non-veterans.
For this study, veterans included patients who received care at a VA facility or received benefits from VA. Data sources included national VA administrative data, Medicare claims data, and data from the US Renal Data System, a national registry of patients with ESRD. We used multiple regression analyses to examine associations of healthcare system use with the probability of any predialysis nephrology care, early predialysis nephrology care
(>=9 months before dialysis initiation), and predialysis permanent vascular access placement.
Population Studied: We compared veterans
(n=8,032) and non-veterans (n=54,927) who initiated hemodialysis in 2000-2001. Based on overall outpatient care during the 12-month predialysis period, we further classifying veterans as VA-only outpatient users (n=1,394), Medicareonly outpatient users (n=3,093), or dual
(VA+Medicare) outpatient users (n=3,545). To ensure that we captured nearly all healthcare events for study patients, we included only patients who were eligible for Medicare at least one year before dialysis initiation and excluded patients who used Medicare managed care or had Medicare as a secondary payer.
Principal Findings: Predialysis nephrology care occurred in 63% of veterans versus 52% of nonveterans (P<0.001). Among veterans, 66% of VAonly outpatient users, 66% of dual users, and 58% of Medicare-only outpatient users had predialysis nephrology care (P<0.001). Non-veterans were
~20% less likely than VA-only veterans to have any predialysis nephrology care (non-veterans: risk ratio
[RR]:0.81, 95% CI:0.78-0.62; Medicare-only veterans: RR:0.91, CI:0.86-0.95; dual veterans:
RR:1.02, CI:0.97-1.07). Early predialysis nephrology care occurred in 20% of VA-only users,
17% of dual users, 10% of Medicare-only users, and 10% of non-veterans (P<0.001). Non-veterans were ~40% less likely than VA-only veterans to have early nephrology care (non-veterans: RR:0.56,
CI:0.50-0.63; Medicare-only veterans: RR:0.52,
CI:0.45-0.61; dual veterans: RR:0.85, CI:0.75-0.97).
Predialysis PVA occurred in 26% of VA-only veterans, 33% of dual use veterans, 29% of
Medicare-only veterans, and 25% of non-veterans
(P<0.001). Dual use veterans were the most likely to receive a permanent vascular access (nonveterans: RR:1.07, CI:0.97-1.18; Medicare-only veterans: RR:1.14, CI:1.02-1.28; dual veterans:
RR:1.33, CI:1.20-1.49).
Conclusion: VA-only veterans were more likely to have timely predialysis nephrology care than nonveterans. However, VA-only veterans had similar low rates of predialysis PVA as non-veterans.
Implications for Policy, Delivery or Practice:
Although veterans using VA-only outpatient care had better access to timely predialysis nephrology care than non-veterans, additional efforts to improve PVA placement are warranted to reduce excess morbidity, mortality, hospitalization, and cost.
Funding Source(s): VA
Theme: Military and Veterans Health Care
● The Impact of Canadian & U.S. Cancer
Screening Guidelines on Cancer Detection
Erin Strumpf, Ph.D.; Srikanth Kadiyala, Ph.D.
Presented by: Erin Strumpf, Ph.D., Assistant
Professor, Economics Dept & Epidemiology &
Biostatistics Dept, McGill University, 855
Sherbrooke Street West, Montreal, AL H2W 2G6,
CA, Phone: (514) 398-2880; Email: erin.strumpf@mcgill.ca
Research Objective: Cancer screening guidelines provide information on screening effectiveness in detecting cancer and the expected mortality impacts. Guidelines generally specify an age at which screening should begin, implicitly recommending that screening not occur for younger asymptomatic individuals. We examine compliance with screening guidelines in the U.S. and Canada
and estimate the effect of screening on cancer detection.
Study Design: We examine breast, colorectal, and prostate cancer screening guidelines from US and
Canadian federal agencies and cancer societies.
We calculate age-specific screening rates, and measure guideline compliance with respect to the recommended initiation age by estimating the difference in screening rates on either side of that age cutoff. We analyze national health survey data for both countries using multivariate logistic regression and controlling for age trends and sociodemographic characteristics. We then use a regression discontinuity design to estimate the effect of screening on age-specific incidence rates and stage at detection, taking advantage of the discrete increases in screening rates at recommended initiation ages.
Population Studied: Adults in the US and Canada who are the appropriate gender and within 10 years of age of the recommended initiation age for different kinds of cancer screening
Principal Findings: Age-specific screening rates for breast, colorectal and prostate cancer rise by
33%, 50% and 25% at the guideline recommend ages in the US, controlling for trends in age and other covariates. Canadian mammography rates increase by 23% but there is no discrete change in colorectal cancer screening rates. Preliminary results from the regression discontinuity analysis indicate substantial effects of US breast and colorectal cancer screening on cancer detection, but no identifiable impact of prostate cancer screening on cancer detection. We expect to find a positive effect of screening on breast cancer detection in Canada as well. Subsequent analysis of the effect of screening on the stage of cancer, conditional on diagnosis, will allow us to examine the role of screening in detecting cancer at early, treatable stages.
Conclusion: We provide evidence on how effective clinical guidelines are in affecting screening rates.
Estimates of the causal effect of cancer screening on detection will inform cost-benefit analyses.
Variation in breast cancer screening guidelines across countries will enable a comparison of the benefits and costs of breast cancer screening at different age thresholds.
Implications for Policy, Delivery or Practice:
Evidence from this observational study will complement evidence from clinical studies. It will inform clinicians and guideline makers regarding at what ages screening is most effective and the associated costs and benefits of population screening at different ages.
Theme: Prevention and Treatment of Chronic
Illness
● Primary Health Care Reform in Quebec,
Canada: Who Participates?
Erin Strumpf, Ph.D.; Pierre Tousignant, M.D.
Presented by: Erin Strumpf, Ph.D., Assistant
Professor, Economics Dept & Epidemiology &
Biostatistics Dept, McGill University, 855
Sherbrooke Street, West, Montreal, AL H2W 2G6,
CA, Phone: (514) 398-2880; Email: erin.strumpf@mcgill.ca
Research Objective: To identify the characteristics of patients and physicians who voluntarily participate in integrated forms of primary care delivery, and to what extent they differ from nonparticipants.
Study Design: In 2002, Québec established family medicine groups (GMFs) to increase access to primary care providers and to improve the quality of primary care. We examine early adopters, participants and non-participants in this form of integrated primary care using a rich data set based compiled by the Montreal Agency for Health and
Social Services specifically for the purpose of evaluating the impacts of GMFs. These data combine administrative and survey data on the population of vulnerable patients in Quebec from
2000-10 and their providers. We use multivariate regression analysis to compare GMF and non-GMF patients and physicians over time as the policy is implemented.
Population Studied: Quebec residents who are characterized as vulnerable due to having one of eleven chronic conditions or being age 70 and over.
The physicians from whom these patients receive health care services are also studied.
Principal Findings: Initial analysis reveals that participant and non-participant patients look very similar in terms of demographic and health status measures. Physicians, on the other hand, exhibit greater differences in terms of demographics, practice type, and main revenue source. Further analysis will examine GMF participation across geographic regions in Quebec.
Conclusion: Important differences across participant and non-participant physicians in GMFs in Quebec suggest that an important part of the impact of this policy might be due to the types of providers that are attracted to this style of practice.
Analysis that wishes to estimate the impact of integrated primary care delivery separately from this mechanism will have to pay important attention to these physician selection issues.
Implications for Policy, Delivery or Practice:
Understanding the characteristics of patients and physicians that voluntarily participate in integrated primary care delivery will inform our understanding of changes that may need to be made, or incentives provided, to attract current non-participants. It will also inform an analysis of the impacts of this policy, improving our understanding of whether we can expect similar results if the policy is implemented in other populations.
Theme: Public Health
● Short-Term Health Impacts of Medicare Part D
Enrollment for Beneficiaries with No Prior
Prescription Drug Coverage
Bruce Stuart, Ph.D.; Amy Davidoff, Ph.D.; Thomas
Shaffer, M.S.; J. Samantha Shoemaker, B.S.;
Christopher Zacker, Ph.D.
Presented by: Bruce Stuart, Ph.D., Professor,
Peter Lamy Center, University of Maryland
Baltimore, 220 Arch Street, Baltimore, MD 21201,
Phone: (410) 706-5389; Email: bstuart@rx.umaryland.edu
Research Objective: The objective of the Medicare
Modernization Act of 2003 was to improve access to prescription drug coverage for the Medicare population within the larger programmatic goal of improving beneficiary health status. The beneficiaries with the most to gain from the new benefit were those who previously had no prescription benefits. The aim of this paper is to evaluate short-term health impacts of Part D enrollment for these individuals.
Study Design: We used unadjusted and adjusted difference-in-difference (DID) models to compare selected outcomes in accessibility (financial hardship in paying for drugs and cost-related nonadherence) and health (changes in self-reported health from a year ago, count of ADL limitations, and newly incident chronic conditions) for those without prescription benefits in 2005 who did versus did not enroll in Part D in 2006. We conducted 2 sensitivity analyses: (1) comparing the newly enrolled to a sample of beneficiaries with continuous employment-sponsored drug benefits in
2005 and 2006, and (2) comparing outcomes for the full sample of previously uncovered beneficiaries regardless of Part D participation to the employment-sponsored coverage sample. The latter test is equivalent to an intention to treat analysis.
Population Studied: Community-dwelling
Medicare beneficiaries participating in both the
2005 and 2006 Medicare Beneficiary Survey
(MCBS) Access to Care modules who had no prescription coverage in 2005 (N=3,038), and a secondary sample who maintained continuous coverage with an employer-sponsored prescription drug plan in 2005 and 2006 (N=2,493).
Principal Findings: 63% of beneficiaries with no drug coverage in 2005 enrolled in Part D in 2006.
Both Part D enrollees and non-enrollees experienced less financial hardship in pharmaceutical access between 2005 and 2006, but the improvement for the Part D group (8 percentage points) was double that of the nonenrolled group (4 percentage points), p<.001. In adjusted comparisons, non-enrollees were 3 percent less likely to report improved health in 2006 compared to 2005, relative to enrollees p<.01. Both groups reported similar counts of ADL limitations in
2005 (0.62 versus 0.61), but the rate declined by
0.05 limitations for Part D enrollees and rose by the same amount for non-enrollees, p<.01. This difference was also significant in the adjusted comparisons. There was no significant difference between the two groups in the incidence of chronic conditions in either the unadjusted or adjusted comparisons. The sensitivity test results supported the findings from the primary analysis.
Conclusion: The Medicare Part D program resulted in small but statistically significant improvements in pharmaceutical access and health status for previously uninsured beneficiaries in
2006.
Implications for Policy, Delivery or Practice: Our findings suggest that in its inaugural year, the program was modestly successful in achieving its stated aims. In that first year, just shy of two-thirds of beneficiaries without drug coverage signed up, and they experienced outcomes consistent with policy goals. Future research is necessary to determine whether the short-term gains we identified were maintained or improved in future years.
Funding Source(s): Novartis Pharmaceutical Corp.
Theme: Medicare
● Persistency in Health Behaviors among
Medicare Beneficiaries
Bruce Stuart, Ph.D.; Amy Davidoff, Ph.D.;
Francoise Pradel, Ph.D.; Ruth Lopert, M.D.; Lirong
Zhao, M.S.; Franklin Hendrick, M.A.
Presented by: Bruce Stuart, Ph.D., Professor,
Peter Lamy Center, University of Maryland
Baltimore, 220 Arch Street, Baltimore, MD 21201,
Phone: (410) 706-5389; Email: bstuart@rx.umaryland.edu
Research Objective: Persistency in practicing healthy behaviors is a key factor in reducing future disability and chronic disease burden for children and adults. However, studies of health behaviors among the elderly are sparse and none has examined the issue of persistency over time. We designed this study to fill an important gap in the literature by examining persistency in 8 common preventive health practices (influenza vaccination, eye examination, cholesterol testing, routine exercise, abstinence from smoking, routine mammography, and cervical and prostate cancer screening) for a nationally representative sample of
Medicare beneficiaries over 4-year observation periods.
Study Design: We tracked self-reported practice of each behavior annually using 1997 to 2005 using data from the Medicare Current Beneficiary Survey
(MCBS). Behavioral persistency was defined as the proportion of the observation period beneficiaries participated in each activity. We estimated behavioral persistency as a function of baseline demographic, socioeconomic, and health
characteristics using partial proportional odds and
OLS regression models.
Population Studied: Six cohorts of communitydwelling Medicare beneficiaries inducted into the
MCBS between 1997 and 2002. Study sample limited to survivors completing four annual surveys
(N=13,913).
Principal Findings: Beneficiaries were most persistent in non-smoking (81% reported not smoking at all, 11% always smoked, and 8% smoked at some point), followed by cholesterol testing (68% reported it every year), and least persistent with routine exercise (47% reporting none). 52% reported influenza vaccinations every year as did 30% for eye exams and prostate cancer screening (among men). Except for eye exams, at least half the sample reported practicing each behavior in 2 of 4 years observed. Regression results showed a consistent pattern of greater persistency among married, higher educated, and higher income beneficiaries. Increasing age was significantly associated with greater persistency in influenza vaccination, eye exams, and nonsmoking, but lower persistency in exercise and cancer screening. Poor health and ADLs generally predicted less persistent behavior. We found that
African Americans had slightly lower rates of influenza vaccination and routine exercise compared to whites, but were significantly more likely to get annual cholesterol tests and cancer screening.
Conclusion: Medicare beneficiaries are not consistently persistent with common preventive health practices. Persistence is high in smoking abstinence and cholesterol screening, moderate in influenza vaccination and prostate screening, and low in eye exams and routine exercise. Low socioeconomic status and poor health are common factors in non-persistent behavior. Lack of persistency in routine mammography, cervical, and prostate cancer screening may nevertheless be appropriate in light of uncertainty about the value of these tests in elderly populations. Research is needed to study persistency in other beneficiary health behaviors and evaluate its impact on health outcomes and Medicare spending.
Implications for Policy, Delivery or Practice: The
Medicare program has gradually expanded preventive health services to include an array of screening tests, other vaccinations, care management programs, and smoking cessation aids. The success of these and other preventive activities will depend largely on how persistent beneficiaries are in using them. Our results suggest that particular effort will need to be devoted to encouraging healthy behaviors among beneficiaries with low socioeconomic status and poor health.
Funding Source(s): RWJF
Theme: Prevention and Treatment of Chronic
Illness
● Insurance Status & Outcomes of Ectopic
Pregnancy
Debra Stulberg, M.D., M.A.; Christina Jarosch, B.A.;
Stacy Tessler Lindau, M.D., M.A.P.P.
Presented by: Debra Stulberg, M.D., M.A.,
Instructor, Family Medicine, The University of
Chicago, 5841 South Maryland Avenue, MC 7110
Room M-156, Chicago, IL 60637, Phone: (773)
834-1356; Email: stulberg@uchicago.edu
Research Objective: Ectopic pregnancy is an acute life-threatening emergency and a leading cause of maternal morbidity and mortality. This study aimed to assess the effect of insurance status on ectopic pregnancy utilization and outcomes at an institution serving a high-risk urban patient population.
Study Design: Data were abstracted from medical chart review. Chi-sqare and Student’s t-test were used to compare outcomes between groups.
Population Studied: Adults with ectopic pregnancy cared for at the University of Chicago, January 1,
2004 through June 30, 2006.
Principal Findings: Of 191 analyzable cases of possible or definite ectopic pregnancy, 114 patients required hospitalization with a mean length-of-stay of 1.7 days (range 1-12 days). Among patients with an admitting diagnosis of ectopic pregnancy, those without insurance had longer hospitalizations (mean
= 2.3 days) compared to those with any insurance
(mean = 1.6 days, p<0.05). Uninsured patients received care, on average, at a later gestational age than those with insurance (mean difference =
4.0 days), although this trend was not statistically significant (p=0.17). Rates of hospital admission and surgical treatment did not differ across insurance groups.
Conclusion: Lack of insurance is associated with costlier, more burdensome care for ectopic pregnancy.
Implications for Policy, Delivery or Practice:
Insurance coverage that allows women to receive pre-conception and prenatal care in a timely fashion may prevent longer hospitalizations for pregnancyrelated complications.
Theme: Disparities
● Measuring Disparities: Bias in Self-reported
Health among Spanish-speaking Patients
Joseph Sudano, Ph.D.; Adam Perzynski, Ph.D.;
Thomas Love, Ph.D.; Steven Lewis, M.S.; Bernice
Ruo, M.D.; David Baker, M.D.
Presented by: Joseph Sudano, Ph.D., Assistant
Professor, Medicine, Case Western Reserve
University, Center for Health Care Research &
Policy, Rammelkamp 236a, 2500 MetroHealth
Drive, Cleveland, OH 44109, Phone: (216) 778-
1399; Email: jsudano@metrohealth.org
Research Objective: Many national surveys have found substantial differences in self-reported health
(SRH) between Spanish-speaking Hispanics and other racial/ethnic groups. However, because cultural and language differences may create measurement bias, it is unclear whether observed differences in SRH reflect true differences in health.
This study investigates psychometric properties of the SF-36v2 for subjects across four racial, ethnic and language groups.
Study Design: The SF-36v2 was administered to non-Hispanic Whites and Blacks, and English and
Spanish-speaking Hispanics by trained interviewers. Multi-group latent variable modeling was used to test increasingly stringent criteria for measurement equivalence. We first tested weak
(metric) factorial invariance, which examines whether SF-36v2 items have the same relationship with their latent variables (factors) across all four groups. We then tested strong (scalar) factorial invariance, which examines whether the conditional expectation of SF-36v2 item responses (intercepts) are invariant across the four groups. Strong factorial invariance is necessary for the direct comparison of group means.
Population Studied: Our sample (N = 1281) included 383 non-Hispanic whites, 368 non-
Hispanic blacks, 206 Hispanics interviewed in
English and 324 Hispanics interviewed in Spanish recruited from outpatient medical clinics in two large urban areas.
Principal Findings: Weak factorial invariance across the four racial, ethnic and language groups was confirmed; equality constraints on all factor loadings did not significantly degrade model fit (CFI change < .01, RMSEA change < .001). However, strong factorial invariance did not hold, as the overall fit of the model was substantially worse
(change in CFI > .02, RMSEA change > .004) after requiring equal intercepts across the four groups.
Further comparisons established that constraints on the intercepts for Spanish-speaking Hispanics were responsible for the poor model fit. Equality constraints on factor loadings and intercepts did not result in a significant degradation of model fit when we included only non-Hispanic Whites, Blacks, and
English speaking Hispanics (CFI change < .01,
RMSEA change < .001). These results suggest that observed differences between SF-36v2 scores for Spanish speaking Hispanics are systematically biased, relative to the three other groups.
Conclusion: Measurement equivalence testing for this study supports correlational, multivariate, or latent variable analyses of SF-36v2 responses across all four subgroups, since these analyses require only weak factorial invariance. However, the lack of strong (scalar) invariance suggests the need for caution when comparing SF-36v2 mean scores of Spanish-speaking Hispanics with those of other groups.
Implications for Policy, Delivery or Practice:
There is variation in SF-36v2 mean scores for
Spanish-speaking Hispanics that may be traced to linguistic and/or cultural differences in the way that
Spanish speakers respond to particular SF-36v2 items. Future studies are needed to establish procedures for accurately comparing the selfreported health of Spanish speakers with that of
English speakers. Policy makers and health providers should be aware that SF-36v2 mean scores for Spanish speakers cannot be compared to those of English speakers with confidence using current methods of data collection, analysis and reporting.
Funding Source(s): NIA
Theme: Disparities
● Improving Hand Hygiene Compliance: The
Influence of Organizational Factors on Clinical
Innovation
Jennifer Sullivan, Ph.D.; Carol Van Deusen Lukas,
Ed.D.; Ryann Engle, M.P.H.; Marjorie Nealon
Seibert, M.B.A.
Presented by: Jennifer Sullivan, Ph.D., Research
Health Scientist, Center for Organization,
Leadership, & Management Research, VA Boston
Healthcare System, 150 South Huntington Avenue
(152M), Boston, MA 02130, Email: jennifer.sullivan@va.gov
Research Objective: As part of a larger Veterans
Health Administration study to increase facilities’ use and adoption of evidence-based hand hygiene
(HH) practices, we identified organizational factors associated with perceived HH compliance. Our objective is to examine change over time in both
HH compliance and organizational factors.
Study Design: In 2006 and 2008, staff participating in a HH intervention were surveyed to describe their work environment, perceptions of quality care and compliance with HH procedures. The overall response rates were 41% in 2006 and 40% in 2008.
Organizational factors include 1) facility priority to quality care, 2) work team learning, 3) work team skills in improvement, 4) accountability, and 5) management support. Alpha values for 5 scales were above .80. Change scores from time 1 to time
2 were created for each factor. The dependent variable, change in overall mean compliance from time 1 to time 2, was the average of respondents’ estimates of the perceived compliance levels for 6 types of staff within the facility. Respondents had to answer 50 percent of the 6 compliance items to be included in this analysis. Linear regression analysis was conducted at the facility-level because respondents were unpaired.
Population Studied: VA facility employees on the test units and redesign teams involved in the HH intervention in 17 participating VA medical centers in three VA regional networks. There were 550 respondents in 2006 and 650 respondents in 2008.
Change scores were aggregated to the facility-level.
Principal Findings: Overall model fit was statistically significant with R-squared = .76. Two organizational factors, management support (help with resource allocation, and coordination across services) and accountability (celebrate successes, rewards, improvement structure in place) were positively and significantly associated with perceived HH compliance. One organizational factor, team skills (work team skills in quality improvement), contradicting expectations, was negatively and significantly associated with change in perceived HH compliance. To explore this surprising finding, we entered team skills into a regression model alone and then added it into a second model with management support. We found a positive relationship for team skills with the
HH outcome for both models. The team skills coefficient turns negative when accountability is added to the model.
Conclusion: Our findings suggest providing management support and accountability are promising strategies to improving HH compliance.
Our team skills finding suggests that an effective team that is aligned with the organization can improve HH compliance. Yet, when accountability and larger organizational support are not present, much less improvement occurs.
Implications for Policy, Delivery or Practice:
These findings are consistent with the literature on implementation of clinical innovations, suggesting broader application to organizational approaches to integrating evidence-based practices into clinical operations. These findings are also useful for managers looking to improve HH compliance rates.
Funding Source(s): VA
● Impact of a Simple Intervention that Improves
Colon Cancer Lymph Node Yield &
Assessement
Shahnaz Sultan, M.D.; Lukasz Kowalczyk, M.D.;
Phalgoon Shah, M.B.B.S.; Thomas George, M.D.;
Li Lu, M.D.; George Sarosi, M.D.; Rebecca Beyth,
M.D.
Presented by: Shahnaz Sultan, M.D., Assistant
Professor, Dept of Medicine, University of Florida,
Box 100214, Gainesville, FL 32610, Phone: (352)
226-3815; Email: shahnaz.sultan@medicine.ufl.edu
Research Objective: The National Quality Forum has endorsed the 12 lymph node (LN) benchmark as a quality metric. Currently, less than 40% of institutions meet this requirement. The purpose of this study was to determine whether implementation of a simple pathology template with dedicated fields for LN reporting led to an increase in the number of colon cancer resections where more than or equal to 12 LNs were reported
Study Design: A simple pathology template, derived from the College of American Pathology, using standardized terminology and dedicated fields for LN reporting was implemented in August 2007.
Using a pre and post-test design, all consecutive pathology cases were retrospectively reviewed.The primary outcome was the percentage of cases in which >12 LNs were assessed between the pre and post-template group. Age, gender, anatomic location, and stage were also collected. Statistical comparisons were made using chi-square and
Fisher's exact t-test.
Population Studied: Inclusion criteria consisted of all stage 0-IV colon cancer patients who underwent surgical resection at a single Veterans Affairs
Medical Center.
Principal Findings: 111 pre-template and 71 posttemplate cases were analyzed. The majority of patients were Caucasian (74%) males (97%). There were no significant differences between the two groups (see Table 1), however there was a trend towards more right-sided colon cancers in the pretemplate group. 51% of all pre-template pathology reports evaluated >12 LNs compared to 68% of post-template reports (33% improvement in LN yield; p=0.03).
Conclusion: Examination of >12 LNs has important therapeutic and prognostic implications in colon cancer patients. Use of a standardized pathology template with dedicated fields for LN reporting is a simple intervention that can increase yield of LN reporting.
Implications for Policy, Delivery or Practice:
This can have a significant impact for institutions striving to reach the 12 LN quality metric.
Theme: Quality and Efficiency: Measurement
● Who is Better Served in Medicaid Managed
Care: Racial Differences in Health Care Access
Chengxiu Sun, M.A.
Presented by: Chengxiu Sun, M.A., Ph.D. Student in Public Policy, Ph.D. Program in Public Policy,
University of North Carolina at Charlotte, 9201
University City Boulevard, Charlotte, NC 28223,
Phone: (980) 225-2663; Email: csun1@uncc.edu
Research Objective: The issue of securing access to needed health care without regard to race has been a part of the mission in most public health programs, and unhindered access has been regarded as one of the most important instruments for public health program to achieve effectiveness and efficiency in delivering services to their beneficiaries. The objective of this paper was to investigate whether there were significant racial differences in accessing health care, both in adults and children, by using the self-reported information from enrollees in the Medicaid program in North
Carolina. If there were racial differences, additional targeted efforts might be needed to ensure that those racial groups with the greatest need can access and secure the needed care.
Study Design: A stratified random sample of adults was derived from the 2006 sampling frame of
Carolina Access-II eligible population provided by the North Carolina Department of Health and
Human Services. Similarly, a stratified random sample of children was obtained from the 2007 sampling frame of Carolina Access-II eligible population. The adult survey of 2815 interviews was conducted using computer-assisted telephone interviews between October 16, 2006 and May 8,
2007, and the child survey of 2827 interviews of parents, guardians, or other knowledgeable adults was conducted between May 21, 2007 and July 10,
2007. The response rates for the adult survey and the child survey were 27.1% and 30.0% respectively.
To answer the research question, we first ran Chisquare tests to detect whether there were statistically significant differences in health access among racial groups, which include blacks, whites, and other races that contained most of the
Hispanics. A logistic regression analysis further investigated the contribution of such variables age, gender, region, urbanicity, and chronicity to the racial differences in health care access.
Population Studied: The studied populations consist of 100,014 adults and 374, 667 children who had been continuously enrolled for at least six months in the network programs of Community
Care of North Carolina (CCNC), one of the state’s
Medicaid primary care case management arrangements. In terms of race, 51.1% of the entire adult population was black, with whites accounting for 40.3% and “other” races comprising slightly less than 9%. Among the child population, 40.3% was black, with white 38.2% and “other” races 21.5%.
Principal Findings: Adult Survey: In the adult survey, the race variable generated 11 statistically significant differences. Except for two statistically significant distributions, a recurring pattern in the health access saw blacks and whites at the extremes and “other” races, the category that contained most of the Hispanics, sitting between them. Child Survey: Overall, there were large variations in accessing health care for children, with
14 out of the 21 access questions being statistically significant across racial groups.
Conclusion: Overall, the Medicaid program represented by this sample of beneficiaries is doing very well in assuring access to care for its enrollees. However, there were some statistically significant racial differences in accessing health care, both in adults and children. In addition, it is important to emphasize that those statistically significant racial differences mainly related to the perception of needing care and seeking it, rather than getting the care.
Implications for Policy, Delivery or Practice: The statistically significant racial differences in accessing health care indicate that there was room for N.C. Medicaid to improve in some of these areas.
● Who is Healthier Under Medicaid Managed
Care: Racial Differences in Health Status
Chengxiu Sun, M.A.; Lutchmie Narine, Ph.D.;
William Brandon, Ph.D., M.P.H.; Nancy Schoeps,
Ph.D.
Presented by: Chengxiu Sun, M.A., Ph.D. Student in Public Policy, Ph.D. Program in Public Policy,
University of North Carolina at Charlotte, 9201
University City Boulevard, Charlotte, NC 28223,
Phone: (980) 225-2663; Email: csun1@uncc.edu
Research Objective: One goal of public health programs in the U.S. is to provide effective and efficient health care to their beneficiaries. Selfreported health status can be expected be an important determinant of health behavior and the effort to seek care. The objective of this study was to investigate whether there were significant racial differences in health status, both in adults and children, by using the self-reported information from enrollees in the Medicaid program in North
Carolina. If there were racial differences, additional targeted efforts might be needed to ensure that those racial groups with the greatest need get the needed care.
Study Design: A stratified random sample of adults was derived from the 2006 sampling frame of
Carolina Access-II eligible population provided by the North Carolina Department of Health and
Human Services. Similarly, a stratified random sample of children was obtained from the 2007 sampling frame of Carolina Access-II eligible population. The adult survey of 2815 interviews was conducted using computer-assisted telephone interviews between October 16, 2006 and May 8,
2007, and the child survey of 2827 interviews of parents, guardians, or other knowledgeable adults was conducted between May 21, 2007 and July 10,
2007. The response rates for the adult survey and the child survey were 27.1% and 30.0% respectively.
To answer the research question, we first ran Chisquare tests to detect whether there were statistically significant health differences among racial groups. A logistic regression analysis further investigated the contribution of such variables as age, gender, chronicity, urbanicity, and region to the racial differences in health status.
Population Studied: The studied populations consist of 100,014 adults and 374, 667 children who had been continuously enrolled for at least six months in the network programs of Community
Care of North Carolina (CCNC), one of the state’s
Medicaid primary care case management arrangements. In terms of race, 51.1% of the entire adult population was black, with whites accounting for 40.3% and “other” races comprising slightly less than 9%. Among the child population, 40.3% was black, with white 38.2% and “other” races 21.5%.
Principal Findings: Adult Survey: Overall, very few
(24%) adults reported having an excellent or very good health condition; Regarding race, black adults were more likely than white adults or adults from
“other races”, the category that contained most of the Hispanics, to report having a better health status.
Child Survey: Overall, the majority of children
(71.9%) were reported having an excellent or very good health condition; Regarding race, white children were reported to be in better health than black children or children of “other” races, the category that contained most of the Hispanics.
Conclusion: The general white populations in the
U.S. typically have better health than their black counterparts. However, our study shows that this may not be the case with Medicaid populations. In terms of adults, black adults had the best health, while white children had the best health.
Implications for Policy, Delivery or Practice:
Future public health programs need to take such racial differences into consideration in order to deliver services to their beneficiaries more effectively and efficiently.
Theme: Medicaid, SCHIP and State Health Reform
● Will Americans Support the Individual
Mandate to Achieve Universal Coverage?
Tara Sussman, M.P.P.
Presented by: Tara Sussman, M.P.P., Doctoral
Candidate, Harvard University, 677 Huntington
Avenue, Room 417, Boston, MA 02115, Phone:
(617) 970-3996; Email: tsussman@fas.harvard.edu
Research Objective: The individual mandate, a requirement that every individual be insured, is an important feature of many recent plans to achieve universal coverage within the current U.S. health care financing system. One reason why health care reform has failed in the past is because of a profound partisan divide. This study examines whether there are significant differences between the two major parties in their views on the individual mandate or whether these differences disappear once sociodemographic factors are taken into account.
Study Design: I analyzed data from a national telephone-based, random-digit-dial survey conducted in English and Spanish in February,
2008. The survey was designed by the Kaiser
Family Foundation, National Public Radio, and the
Harvard School of Public Health. The data were weighted to be representative of the nation’s adult population. Support for individual-mandate based reform is defined as occurring when respondents say they support reform once they are asked about two different mandate-based plans. A random half of the sample was asked about their support for a plan, the “stand-alone mandate,” in which everyone would have to get health insurance or be subject to a fine. The government would subsidize coverage for low-income individuals. The other half of the sample was asked about “shared responsibility,” a plan with the same mandate on individuals and assistance for low-income people, coupled with additional requirements on employers, government, and insurance companies to enable coverage expansion.
Population Studied: The sample included 1,704 randomly selected adults across the country.
Principal Findings: Fifty-three percent of the public supports mandate-based reform overall.
Democrats and Independents are significantly more likely to support reform than Republicans, controlling for sociodemographic factors. Though partisan differences persist, the shared responsibility plan is significantly more popular than the stand-alone mandate for all groups. We see the largest differences in support for shared responsibility versus the stand-alone mandate among people earning between $80,000 and
$100,000 and African Americans, as compared to lower income and white respondents, respectively.
A majority of all groups, including Republicans, supports the proposal that parents should be required to provide insurance for their children.
Conclusion: Democrats are significantly more supportive of mandate-based reform than
Republicans and these differences cannot be explained by sociodemographic factors. A policy that includes an individual mandate as part of a
“shared responsibility” plan does not bridge the partisan gap but does attract higher overall support and is more appealing to some important groups, as compared to a stand-alone individual mandate.
Implications for Policy, Delivery or Practice: A universal coverage policy that hinges on the individual mandate does not transcend the partisan divide that characterizes our political system. Some legislators may still decide to pursue a mandate in hopes of achieving the substantive goal of universal coverage through our current health care financing system. In this case, policymakers may be able to expand the base of supporters by incorporating the mandate into a “shared responsibility” plan that also includes requirements for employers, government and insurers.
Funding Source(s): Harvard Merit Fellowship
● Pharmacists’ Prescriptions for Medicare Part
D.
Tara Sussman, M.P.P.
Presented by: Tara Sussman, M.P.P., Doctoral
Candidate, Harvard University, 677 Huntington
Avenue, Room 417, Boston, MA 02115, Phone:
(617) 970-3996; Email: tsussman@fas.harvard.edu
Research Objective: Retail pharmacists are on the front lines of Medicare Part D implementation. We know from previous health care initiatives that the way programs like this one actually function can depend on how they are perceived by the
professionals implementing them. The overarching aim of this study is to examine pharmacists’ perceptions of Medicare Part D.
Study Design: The primary data for my analyses are from a national random sample survey of pharmacists conducted between April 21 and June
27, 2006 and designed by the Kaiser Family
Foundation. Using factor analysis, I created scales to measure pharmacists’ perceived administrative burden and financial burden. I developed a third scale to measure pharmacists’ perceptions of how effectively Medicare Part D was helping beneficiaries. In addition, I interviewed
Massachusetts pharmacists, recruited through nonrandom snowball sampling, to illustrate some of the results that emerged in the quantitative analysis.
Population Studied: The national survey included
224 independent pharmacists and 545 chain pharmacists. Twenty-six pharmacists in
Massachusetts were also interviewed.
Principal Findings: Forty-six percent of pharmacists surveyed were favorable toward
Medicare Part D. Controlling for pharmacy characteristics and pharmacist demographics, pharmacists who, under the Medicare Part D program, perceived more administrative burden, more financial burden, or less programmatic success at helping beneficiaries, were associated with a lower likelihood of favorability toward Part D than those who professed less burden or viewed the program as more helpful for beneficiaries, respectively. Also, independent pharmacists were significantly less favorable toward Part D than chain pharmacists. Pharmacists interviewed mentioned several specific implementation challenges. For example, some pharmacists noted the burden caused by the lack of plan standardization and said that this can be especially frustrating when patients switch plans. Several pharmacists said that they advise beneficiaries not to switch their prescription drug plans even when other plans may better suit their needs.
Independent pharmacists, and pharmacists whose customer base included a high percentage of
Medicare beneficiaries, perceived significantly more administrative and financial burdens, than did pharmacists who worked in chain pharmacies or who worked with fewer Medicare customers, respectively. Also, rural pharmacists perceived significantly more administrative burden than urban pharmacists.
Conclusion: Our study has found that many of the same factors, namely administrative and financial burden, and concern for patients, that have been significant to the way that physicians and other health care professionals have perceived previous health care initiatives, are also relevant here.
Independent pharmacists are also significantly less favorable toward Medicare Part D than those working in chains, even beyond the financial burden controlled for in our models. This may be because independent pharmacies do not have the economies of scale to handle additional unmeasured challenges associated with Part D.
Implications for Policy, Delivery or Practice:
Pharmacists have the potential to significantly affect the implementation of Medicare Part D. Their perceptions of the program may influence the way they execute it, so addressing pharmacists’ perceptions of burden may in turn also help the program to achieve its goals for beneficiaries. To that end, we recommend: standardizing plan procedures; providing more information to pharmacists about plans and beneficiary spending; and, increasing pharmacy reimbursements.
Funding Source(s): Charles H. Farnsworth Trust
● Using Automated Laboratory Data to Identify
Patient at High Risk: Analysis of Creatinine
Change at Early Hospitalization & Associated
Mortality among 852,734 Admissions Between
2003 & 2006
Ying Tabak, Ph.D.; Xiaowu Sun, Ph.D.; R.S.
Johannes, M.D., M.S.
Presented by: Ying Tabak, Ph.D., Director,
Biostatistics, Clinical Research, Cardinal Health,
400 Nickerson Road, Marlborough, MA 01752,
Phone: (508) 571-5120; Email: ying.tabak@cardinalhealth.com
Research Objective: Automation of clinical data has been under rapid development in the USA.
Using automated clinical data to identify patients with increased risk for mortality at early hospitalization could be a cost efficient way to aid decision making and quality improvement. Renal function is a key indicator of a patient’s clinical status. As an end organ, kidney closely manifests the functions of cardiovascular, liver, and other vital organ systems. We hypothesized that deterioration of renal function, as indicated by elevated serum creatinine level, would be associated with increased mortality.
Study Design: We selected 74 acute care hospitals that captured and imported laboratory data electronically between 2003 and 2006. We included patients with length of stay >1 day and had at least 2 measures of serum creatinine within 48 hours from the baseline. We classified the creatinine change within 48 hours from the baseline into 4 levels: <0.30 mg/dL (stable or improved),
0.30-0.50 mg/dL (mild deterioration), 0.51-1.00 mg/dL (moderate deterioration), and >1.00 mg/dL
(severe deterioration). We fit a mixed model to estimate the mortality risk associated with each level of worsening renal function within 48 hours of hospitalization, controlling for the baseline creatinine, demographics, comorbidity, and disease severity.
Population Studied: A total of 852,734 adult admissions met the criteria were included for this study.
Principal Findings: The median age was 71 years old (IQR: 57, 81) and 52% were women. The overall in-hospital mortality was 3.5%. The top 5 diseases in volume were heart failure, pneumonia, acute myocardial infarction, chest pain, and arrhythmia, which accounted for 26% of total study cohort. The prevalence of creatinine change within
48 hours was 92.6% (stable or improved), 3.0%
(mild deterioration), 2.5% (moderate deterioration), and 1.9% (severe deterioration). The corresponding crude mortality rate was 2.9%, 7.9%, 12.4%, and
14.9% respectively (p<0.0001). Compared to patients with stable or improved creatinine in the first 48 hours, the multivariable adjusted mortality odds ratio and 95% confidence intervals were 1.89
(1.80, 1.99), 2.73 (2.60, 2.86), and 3.73 (3.53, 3.93) for mild, moderate, or severe creatinine deterioration groups respectively (all p<0.0001).
Conclusion: Worsening creatinine within 48 hours was associated with significant increase of mortality for hospitalized patients. There was a doseresponse relationship between the degree of creatinine increase and the risk of mortality.
Implications for Policy, Delivery or Practice:
Creatinine is an objective and routine laboratory test captured electronically in most USA hospitals.
Future study should examine whether automated surveillance system using creatinine and other automated clinical data could provide early alert to clinicians and improve favorable management of patient outcomes.
Theme: Quality and Efficiency: Measurement
● The Impact of Health Plan Physician-Tiering on Access to Care:
Sean Tackett, B.S.; John Adams, Ph.D.; Elizabeth
McGlynn, Ph.D.; Ateev Mehrotra, M.D.
Presented by: Sean Tackett, B.S., Medical
Student, Medicine, University of Pittsburgh School of Medicine, 128 North Craig Street, Apartment
814, Pittsburgh, PA 15213, Phone: (412) 370-8610;
Email: tackett.sean@medstudent.pitt.edu
Research Objective: In an attempt to improve quality and control costs, health plans have created tiered physician networks. In such a network, enrollees have a financial incentive via lower copayments (e.g. $10 vs. $30) to seek care from
“high-value” physicians. A high-value physician is one, based on health plan metrics, who is both high quality and low cost. One concern with physician tiering that has yet to be addressed empirically is whether enrollees have reasonable access to the limited pool of high-value physicians. The concern is that tiered networks financially penalize enrollees without providing a real opportunity to switch physicians. In this research we compared access, as measured by travel time, to primary care physicians (PCP) and orthopedic surgeons in one state under two scenarios: (1) enrollees can see all physicians or (2) enrollees can see only high-value physicians.
Study Design: Using techniques similar to those used by health plans, we created cost and quality scores for Massachusetts practicing physicians using an aggregated commercial claims data set from four health plans for 2004-2005. A physician cost score was determined by identifying healthcare episodes, assigning each episode to the physician who had the highest proportion of professional costs, and then creating a ratio of actual to expected costs for all episodes assigned to that physician. A physician quality score was determined using 131 clinical process indicators for common chronic and acute medical conditions. A physician was labeled high-value if he or she was statistically different than the mean physician on both cost and quality metrics. Using ArcGIS we determined the travel time from each of more than eighty-thousand Massachusetts census blocks to the closest physician and then calculated the population-weighted average across all census blocks. Sensitivity analyses included using less stringent definitions of high-value.
Population Studied: We studied Massachusetts practicing physicians with self-identified specialties of family practice (FP), general internal medicine
(IM), or orthopedic surgery who submitted at least one claim in the two years.
Principal Findings: The analysis focused on 3,677
PCPs (2708 IM, 969 FP) and 529 orthopedic surgeons of whom only 98 (2.7%) and 6 (1.3%) were identified as high-value respectively. The average travel time to the nearest PCP was 2min
50sec and to the nearest high-value PCP was
11min 9sec. The average travel time to the nearest orthopedic surgeon was 4min 3sec and the nearest high-value orthopedic surgeon was 37min 31sec.
Conclusion: Under a physician-tiering network, enrollees would have much greater travel times to see the closest high-value PCP or specialist, which may adversely impact access to care. However, even with a small number of high-value physicians, travel times to a high-value physician may not be prohibitive (<15 minutes for PCP and <40 minutes for specialist).
Implications for Policy, Delivery or Practice:
These results help inform the concerns that physician-tiering products might adversely impact access to care. Future work will have to investigate what fraction of patients switch care to high-value physicians and whether high-value physicians are accepting new patients.
Theme: Quality and Efficiency: Policies and
Incentives
● Use of Complementary or Alternative
Medicine (CAM) to Prevent or Treat Cognitive
Problems & Communication about CAM Use with Health Care Providers
Elizabeth Tait, M.H.S.; James Laditka, D.A., Ph.D.;
Sarah Laditka, Ph.D.; Manana Tsulukidze, M.D.,
M.P.A., M.P.H.
Presented by: Elizabeth Tait, M.H.S., Ph.D.
Student & Research Assistant, Public Health
Sciences, University of North Carolina at Charlotte,
9201 University City Boulevard, Charlotte, NC
28223, Phone: (828) 775-9095; Email: emtait@uncc.edu
Research Objective: About half of U.S. adults ages 50 and older report using complementary or alternative medicine (CAM). Adults are concerned about maintaining cognitive health. Yet, research suggests health providers rarely discuss maintaining cognitive health with patients, despite growing evidence that health behaviors play an important role in maintaining it. Although some
CAMs are being studied for cognitive health effects, little conclusive evidence exists. Relatively little research has examined individuals’ use of CAM to promote cognitive health. We compared CAM use among people with and without reported cognitive problems, including memory problems, Alzheimer’s disease, and related disorders, and also the likelihood that they would discuss CAM use with their physicians, using data from a recent nationally representative survey. We examined both CAM use in general, and also specifically to treat, cure, or prevent cognitive problems; reported results are for the latter.
Study Design: Data represented in-person surveys from the 2007 National Health Interview Survey
(NHIS), which included a detailed CAM supplement.
The NHIS, conducted annually in the U.S, collects information about health status, use of medical services, and other measures. Data from the CAM supplement were linked to the adult and person
NHIS files. Analyses included chi-square and multivariate logistic regression, accounting for the complex survey design and weighted for national representativeness. Controls included age, sex, race and ethnicity, education, marital status, health insurance, chronic conditions, self-reported health, recent changes in health status, health behaviors
(smoking, body mass index), and region. We examined CAM defined both including and excluding use of vitamins; reported results, which include vitamins, are for ages 50 and over.
Population Studied: 10,104 adults ages 50 or older in the 2007 NHIS and CAM supplement.
Principal Findings: Nearly 11% reported a cognitive problem. Of those without cognitive problems, 8.4% (95% Confidence Interval, CI 7.7%
9.1%) used CAM; of those with cognitive problems,
12.1% used CAM (CI 10.1%-14.1%) (p<.01). In adjusted results, people reporting cognitive problems were notably more likely to use CAM than others (Odds Ratio, OR 1.56, CI 1.21-2.00). Of those using CAM, 27% reported that they do not tell their health providers. In adjusted results, those with cognitive problems did not differ from others in reporting CAM use to providers. There were analogous results for CAM defined without vitamins.
Conclusion: A substantial number of older
Americans use CAM specifically to prevent cognitive problems. Many individuals with cognitive problems use CAM as treatment. People with cognitive problems are notably more likely to use
CAM than those without them, but no more likely to tell health providers about CAM use.
Implications for Policy, Delivery or Practice:
Individuals with cognitive problems should be encouraged to discuss CAM use with health providers, as some CAMs, particularly herbal supplements, may interact with conventional therapies. Health providers should inquire about
CAM use for cognitive health, as it may be a marker for motivation to adopt healthy behaviors. Patients using CAM to maintain cognitive health should be encouraged to engage in behaviors for which there is substantial scientific evidence of association with cognitive health, including physical activity and heart-healthy diets.
Theme: Behavioral Health
● An Inconvenient Truth: Purchaser-Provider
Split Ends in Inter-Organizational Hostility:
Lessons from Implementing Family Medicine in
Iran
Amirhossein Takian, Ph.D.; Mohammad Javad
Kabir, M.P.H.
Presented by: Amirhossein Takian, Ph.D., Ph.D.
Candidate, Public Health & Policy, London School of Hygiene & Tropical Medicine, Room 105, 51
Bedford Square, London, WC1B 3DP, UK, Phone:
+44(0)2072994639; Email: amirhossein.takian@lshtm.ac.uk
Research Objective: To investigate the process of health sector reform in the Iranian health system
(with focus on implementing family medicine: a model of primary care, within which a family doctor
(FD) is the first level of public contact with the health system. Gate-keeping which rations referral to specialist care is at the core of FM)in the period of last five years, in order to identify facilitators of and barriers to implementation of change.
Study Design: A multi-level study of the process of implementing a series of reforms into the Iranian health system, focusing on comprehensive changes in primary care (promoting family physician and referral system), and universal coverage (providing a comprehensive insurance package for all residents in rural areas counting for 25 million population), using qualitative (interviews, focus groups, narrative documentary analysis and observation), as well as quantitative methods
(survey). Analysis: A mixed technique of inductive/deductive (framework approach)for the qualitative part, using QSR Maxqda 2007 to keep and retrieve the data, based on a selective
framework of theories of public policy implementation.
Using SPSS 14.00 to keep the data from quantitative leg (survey), employing multiregression tests for analysing and interpreting the results.
Population Studied: 1- Qualitative leg: Three levels study at the national (19 stakeholders), provincial (28 from three purposefully selected provinces)and local (95 participants including service providers, administrative staff, representatives of the public, and ordinary people as user of health services in 12 rural health centres, where primary care services are delivered to the public). 2- Quantitative leg (survey):
183 participants at the local level including service providers, administrative staff, and people who utilized health services.
Principal Findings: The study revealed the severe hostility between the two implementers of the reform: The Ministry of Health (MOH) as provider, and the Medical Insurance Organization (MIO) as
Purchaser of services, as the main barrier to their satisfactory cooperation in the common task of implementing reform. The hostility was not only due to their diverse goals. For two reasons, whatever policy was putting into practice by these two bodies, might have experienced more or less a similar destination. First, their organizational perspectives were fundamentally different that was even recognizable for several staff at the local level. The
MOH was branded a preventive focused organization whose attempt was promoting health and improving primary care. Whereas, many described the MIO as a curative focused body who by default was reluctant to pay for primary care.
Such an accusation was ruled out by the MIO after they started implementing FM with the MOH, although it was proved with several evidence in practice. Second, their organizational past bubbled up their tension because of their effort to prove their supremacy. The MOH was unhappy at losing control over the MIO. On the other hand, the MIO’s focus was proving its independency to the MOH.
They looked to each other as rivals not partners.
The MOH accused the MIO of not releasing money so paralyzing the MOH to fulfil its goals. Whereas, the MIO restricted payments to the MOH, accusing it to not meet the relevant criteria of implementation.
This was why their cooperation in the purchaserprovider split failed. The concept of purchaserprovider separation and the benefits it might bring into the health system was attractive to many interviewees. However, the separation resembled a political game for organizational benefits, which reduced their cooperation considerably, diminished their trust in each other and collapsed their volatile partnership. Time and resources of the two were wasted fighting and accusing each other, rather than attempting to improve implementation.
Conclusion: Purchaser-provider split that was applied for the first time at this huge scale in Iran, did not succeed in changing the status quo of primary care. Similar experiences have been reported in decentralisation of secondary care services in Iran (Jafarisirizi et al,2008). Instead, the separation became a reason for fighting, misunderstanding, lack of cooperation and failure of the fragile partnership.
Ironically, the majority of interviewees praised the concept of separation and branded it as a necessary action to ensure good services. The majority of respondents eulogized about the separation for three reasons. First, they were unaware of either the consequences of separation or the historical barriers before its implementation.
Second, they knew the barriers, but supported the idea to attain their own benefits. Third, they knew the consequences or barriers, but followed the common rule of ceremonial talk in Iran and supported something in which they did not believe.
The separation that aimed to improve the relationship between the principal and agents, created in a bitter relationship between the two implementers themselves, mainly because of their institutional past. Devil shift (Sabatier et al,1987) encompassed their relationship, particularly on the
MIO side. Most actors from each group assumed that they are right-thinking, virtuous, and fair in their judgments. Thus anyone who disagreed with them must be mistaken about the facts, operating from the wrong value premises, or acting from evil motives (Harrison,1976). Such a relationship became a strong barrier to effective cooperation.
On the one hand the MIO, that used to be under the
MOH, looked at implementation as an opportunity to prove its highness, independence and superiority to its rival (MOH). On the other hand, it was hard for the previously superior institution (MOH) to recognize the MIO, which it viewed as a child, to control the funds. Although both organizations occasionally showed some degree of common language and agreed upon certain objectives, they never became close enough to mutually cooperate.
The separation harmed implementation and wasted resources in fighting, when unity and cooperation were essential.
Implications for Policy, Delivery or Practice:
Given the adverse effects of the current hostility between implementers on either the public or practitioners, as well as the high faith in the concept of such a reform as the only prescription for reengineering the whole health system, careful attention towards histroy of co-implementers seems to be essential.
Funding Source(s): Ave-Sina Charity Foundation
Theme: Organizational Performance and
Management
● The Dynamic Model of Health Policy Analysis:
An Alternative Framework to Study the Process of Implementing Change in Health Systems
Amirhossein Takian, Ph.D.; Arash Rashidian,
Ph.D.; Mohammad Javad Kabir, M.P.H.
Presented by: Amirhossein Takian, Ph.D., Ph.D.
Candidate, Public Health & Policy, London School of Hygiene & Tropical Medicine, Room 105, 51
Bedford Square, London, WC1B 3DP, UK, Phone:
+44(0)2072994639; Email: amirhossein.takian@lshtm.ac.uk
Research Objective: 1- To investigate the process of health sector reform in the Iranian health system in the period of last five years, in order to find facilitators of and barriers to implementation of change; and
2- To assess the appropriateness of current widelyused models of health policy analysis, particularly
Walt & Gilson (1994) triangle to address the reality of implementing change in the context of a developing country (Iran); and 3- To develop a tailored conceptual framework to analyze the process of implementing change into the health systems in the context of an autocratic developing country.
Study Design: A multi-level study of the process of implementing a series of reforms into the Iranian health system, focusing on comprehensive changes in primary care (promoting family physician and referral system), and universal coverage (providing a comprehensive insurance package for all residents in rural areas counting for 25 million population), using qualitative (interviews, focus groups, narrative documentary analysis and observation), as well as quantitative methods
(survey). Analysis: A mixed technique of inductive/deductive (framework approach)for the qualitative part, using QSR Maxqda 2007 to keep and retrieve the data, based on a selected framework of theories of public policy implementation.
Using SPSS 14.00 to keep the data from the quantitative leg (survey), employing multiregression tests for analysing and interpreting the results.
Population Studied: 1- Qualitative leg: Three levels study at the national (19 stakeholders), provincial (28 from three purposefully selected provinces)and local (95 participants including service providers, administrative staff, representatives of the public, and ordinary people as user of health services in 12 rural health centres, where primary care services are delivered to the public). 2- Quantitative leg (survey):
183 participants at the local level including service providers, administrative staff, and people who utilized health services.
Principal Findings: The study adopted four main themes to categorize facilitators of and obstacles to implementation of the reform, within four interrelated and dynamic groups, which contributed to the emerged dynamic model for health policy analysis. They include: aspects of the policy
(content); the existing environment (context); the experience of implementation (process); and impact of the policy on local staff and the public (actors).
Challenging to some extent Walt and Gilson’s
(1994) triangle for health policy analysis that focuses on content, context, process and actors, the dynamic model highlights the importance of the dynamic relationship between the four main components of health policy. It reveals a clear distinction between central and peripheral stakeholders, which reflects the consequences of imposing policies and the top-down approach to policy. The distinction challenges the capability of
Walt and Gilson’s triangle to explain the health policy chain in the autocratic environment of policy making in Iran and perhaps similar developing countries.
Conclusion: The dynamic model of health policy analysis is a simplified approach to a complex set of inter-relationships. It seeks to understand the reality of implementation by appreciating how actors influence the context within which they live and work. Therefore, while it helps to systematically investigate factors which influence the policy, it acts as a map that shows the main roads. Buse et al’s
(2005) description of Walt & Gilson’s triangle applies here that has yet to have contours, rivers, forests, paths and dwellings added to it.
Implications for Policy, Delivery or Practice: The model helps analysing and understanding the policy itself (analysis of policy) or it can be applied to plan for improving the policy (analysis for policy). This study was seeking to do both: a retrospective exploration of the determination of the policy and what it consisted of, and suggestions for the future of the policy to improve its implementation, by reflecting on practical implications.
Funding Source(s): Ave-Sina Charity Foundation
Theme: Quality and Efficiency: Measurement
● Scaling-up Primary Care & Universal
Coverage Based on PHC Infrastructure: The
Case of Iranain Health System Reform.
Amirhossein Takian, Ph.D.;Arash Rashidian, Ph.D.
Presented by: Amirhossein Takian, Ph.D., Ph.D.
Candidate, Public Health & Policy, London School of Hygiene & Tropical Medicine, Room 105, 51
Bedford Square, London, WC1B 3DP, UK, Phone:
+44(0)2072994639; Email: amirhossein.takian@lshtm.ac.uk
Research Objective: To investigate the process of health sector reform (with focus on implementing family medicine: a model of primary care, within which a family doctor (FD) is the first level of public contact with the health system. Gate-keeping which rations referral to specialist care is at the core of
FM) in the Iranian health system in the period of last five years, in order to identify facilitators of and barriers to implementation of change.
Study Design: A multi-level study of the process of implementing a series of reforms into the Iranian health system, focusing on comprehensive changes
in primary care (promoting family physician and referral system), and universal coverage (providing a comprehensive insurance package for all residents in rural areas counting for 25 million population), using qualitative (interviews, focus groups, narrative documentary analysis and observation), as well as quantitative methods
(survey and analysing data from the annual vital census) . Analysis: A mixed technique of inductive/deductive (framework approach)for the qualitative part, using QSR Maxqda 2007 to keep and retrieve the data, based on a selective framework of theories of public policy implementation. Using SPSS 14.00 to keep the data from quantitative leg (survey and the vital census), employing multi-regression tests for analysing and interpreting the results.
Population Studied: 1- Qualitative leg: Three levels study at the national (19 stakeholders), provincial (28 from three purposefully selected provinces)and local (95 participants including service providers, administrative staff, representatives of the public, and ordinary people as user of health services in 12 rural health centres, where primary care services are delivered to the public). 2- Quantitative leg: A.(survey):
183 participants at the local level including service providers, administrative staff, and people who utilized health services. B. The verified annual vital census from all 30 provinces in Iran.
Principal Findings: There was a consensus that the pre-existing extensive PHC network which covers more that 90% of Iran’s rural population, facilitated the implementation of FM. Its position for the public, accessibility in terms of location and service provision, efficient hierarchy and organization with almost three decades of experience, preventive focus rather that curative, and particularly its reliance on community-based health workers, alongside a very weak presence of the private sector in rural areas, made the PHC the only available media for hosting FM in rural areas of
Iran. Within the past three years of implementation, more than 3000 doctors and 3200 midwives have been added to the existing healthcare practitioners, settling in almost 2300 rural health centres nationwide.As a result, health houses saw doctors more often and regularly, which sharply increased utilization of services. Out of the 23 million registered population in 2005, almost 6 million visited FDs. The number rose to 14 million in early
2007. Predictably, the number of visits to pharmacies also increased, but not in line with the increase in visits to FDs (from 2.8 million in 2005 to
8.3 million in early 2007), which might indicate improvements in prescription behaviour.
Further, FM led to improvements in the PHC’s physical infrastructure. Because of doctors’ night shifts in health centres to provide 24-hour emergency services, health authorities attempted to prepare proper living-in facilities for personnel.
Number of physical facilities attached to health centres for staff’s accommodation jumped from
1445 in 2005 to 2292 in 2006.
Conclusion: Being described as an “incredible masterpiece” (Tavassoli,2008), the Iranian primary health care (PHC) networks devised in 1981 by creating health houses in rural areas, which emphasized community participation and intersectoral cooperation, and focused on basic health care financed by governmental budget. As the first points of seeking services, health houses have a fundamental role in providing basic care in rural areas, where 35% of the population live.In line with the World Health Report 2008, the pre-existing
PHC provided a reasonable host for enjoying the universal coverage and led to equipping rural health centres. Despite enhancing the PHC in villages, shifting towards primary care and prevention, establishing a long-term relationship between doctors and people, and health promotion have not happened yet.
Implications for Policy, Delivery or Practice: The pre-existing PHC network was an important facilitator of the next steps to scale-up primary care in the format of family medicine. Providing universal coverage is also essential to increase utilization of services. However, sustainable change in the peoples’ well-being requires multi-dimensional reforms that take into account the diverse factors affecting health, while inspiring leadership armed with long-run vision and appropriate expertise to provide high-level political support.
Theme: Coverage and Access
● The Impact of “Present on Admission”
Designations for Failure to Rescue-Based
Complications
AkkeNeel Talsma, Ph.D., R.N.; Katherine Jones,
Ph.D., R.N., F.A.A.N.; Darrell Campbell, M.D.;
Margaret Freundl, M.S., R.N.; David Ronis, Ph.D.
Presented by: AkkeNeel Talsma, Ph.D., R.N.,
Assistant Professor / Senior Clinical Information
Analyst, School of Nursing, University of Michigan,
400 South Ingalls, Room 4154, Ann Arbor, MI
48109-5482, Phone: (734) 763-5199; Email: antalsma@umich.edu
Research Objective: Since the passage of CMS
IPPS ruling (CMS-1553-FC, October 2008), the acute care setting is focused on managing the impact of reduced payments for patients that acquired specific complications and conditions during their hospital stay. One measure considered for inclusion in future IPPS rulings is Failure to
Rescue (FTR). FTR captures patient mortality based on secondary diagnoses that describe the following complications: acute renal failure,
PE/DVT, pneumonia, sepsis, cardiac arrest/shock, or GI hemorrhage. Research has shown a link with lower nurse staffing levels and other health systems issues. The aims for this study were to validate the
FTR complications through medical record review,
to determine FTR complications present on admission (POA) and associated FTR complication mortality rates.
Study Design: A multi-year (2003-05) seven hospital study was conducted to evaluate the relationship between nurse staffing levels and failure to rescue outcomes. Subsequent efforts have focused on validation of the specific FTR complication and the determination of whether the patient entered the system with the FTRcomplication (POA). The FTR Validation Tool was developed to validate the specific FTR complications, determine whether the FTRcomplication was POA and examine the trajectory of care provided, using medical record reviews by trained bachelor’s prepared nurses.
Population Studied: Seven acute care hospitals were included in the study. AHRQ HCUP specifications were used to select the cases that met the inclusion criteria of the failure to rescue measure. Subsequent analyses identified the specific FTR complication each case experienced.
A random sample of 235 cases representing all six
FTR related complications was prepared.
Principal Findings: The FTR complications ranged from 15% (sepsis) to 19% (GI hemorrhage) of the sample. For 22.5% of the sample the complication was POA. PE/DVT showed the highest POA status
(32%) while GI hemorrhage was least likely identified as POA (11%). The mortality rate for cases that had the complication POA was 13.2% compared to 24.2% for cases without the FTRcomplication POA.
A logistic regression model was tested to determine predictors of overall mortality (model p<.001;
Nagelkerke R2=.545). Included in the model were the six FTR complications, gender, and POA status of the complication. Significant predictors were sepsis, cardiac arrest, and acute renal failure as the patient complication. Cases with pneumonia, subject gender and POA status approached statistical significance. Findings suggest that males and hospital-acquired complication had higher odds of dying (ns).
Conclusion: Better understanding the characteristics and distribution of the population that is included in the Failure to Rescue measure will help hospitals identify sub-populations at risk of developing the complication within the hospital and implementation of timely treatment plans. Specific admission protocols can be developed to assure
POA cases are appropriately identified and treated upon the point of entry into the health system.
Implications for Policy, Delivery or Practice:
Implementation of the CMS IPPS ruling has prompted acute care hospitals better understand its patient populations and pre-existing conditions. This study showed that patient outcomes vary pending the location of presenting the condition.
Funding Source(s): Blue Cross Blue Shield of
Michigan Foundation
Theme: Medicare
● Bridging the Gap: Impact of Part D Low
Income Subsidy Coverage or Branded Gap
Coverage on Medication Continuity Among
Severe Osteoporosis Patients Treated with
Teriparatide
Leonardo Tamariz, M.D., M.P.H.; Claudia Uribe,
M.D., Ph.D.; Jiacong Luo, Ph.D.; John Hanna,
M.B.A.; Daniel Ball, Dr.P.H.; Kelly Krohn, M.D.; Eric
Meadows, Ph.D.
Presented by: Leonardo Tamariz, M.D., M.P.H.,
Assistant Professor, Medicine, University of Miami,
4718 Northwest 103 Court, Miami, FL 33178,
Phone: (305) 626-5716; Email: ltamariz@med.miami.edu
Research Objective: Low income subsidy (LIS) is a benefit that reduces out-of-pocket medication spending for Medicare Part D (Part D) patients.
Premature discontinuation of osteoporosis medications can lead to significant long term complications. We determined the discontinuation rates of Forteo (teriparatide), an injectable and relatively expensive treatment for patients with severe osteoporosis, in Part D patients. The primary covariates of interest were the presence of
LIS benefits and/or the absence of a coverage gap
(“donut hole”) in the patient’s Part D benefits.
Study Design: We performed a retrospective analysis of multiple Part D plans from a single large insurance provider in 2006 to assess continuity of prescription refills for teriparatide. We defined and calculated discontinuation as stopping the medication after reaching the coverage gap or the equivalent threshold of total pharmaceutical costs for those patients without a coverage gap. Patients needed to fill a teriparatide prescription prior to the coverage gap and reach the coverage gap (or equivalent cost for patients without a coverage gap) to be at risk for discontinuation.
Population Studied: We identified patients using pharmacy claims for the National Drug Code for teriparatide (54868540600). Patients were stratified by LIS status and whether non-LIS patients had a gap in coverage due to the Part D plan design. LIS patients are not subject to a coverage gap but were categorized by presence or absence of a coverage gap in the selected plan to facilitate comparisons.
Principal Findings: We identified 9,068 teriparatide users, of which 31% received LIS benefits. Those with LIS were younger (73.0±12.3 vs. 76.6±8.3), had higher chronic disease score
(CDS) (5,626±3,435 vs. 4,280±2,819) and higher pill burden (18.2±10.0 vs. 12.3±7.4) when compared to teriparatide users without LIS (all p<0.01). Among the 6,221 non-LIS patients, 34% enrolled in a plan with a coverage gap compared to
92% of the 2,847 LIS patients. Of the at risk population (n=5,837), non-LIS patients with a coverage gap discontinued teriparatide 25% (298 of
1,187) of the time compared to 13% (191 of 1,449)
for those with LIS (p-value<0.01). Among patients in a plan without a coverage gap, 8% without LIS discontinued teriparatide (229 of 3,031) compared to 9% with LIS (16 of 170), p-value=0.37).
Conclusion: In one Part D provider, the majority of non-LIS teriparatide patients in 2006 enrolled in a plan with benefits continuing throughout the federally designated coverage gap. Non-LIS patients exposed to the coverage gap discontinued teriparatide at higher rates than either LIS cohort and non-LIS patients without a gap in coverage.
Notable differences in characteristics existed among the cohorts, which should be further investigated to better elucidate the underlying factors associated with discontinuation.
Implications for Policy, Delivery or Practice:
Nearly all currently offered Part D plans have the statutorily defined standard benefit that includes a coverage gap. Part D LIS patients are not subject to a gap in coverage. These data suggest that the absence of a gap in coverage (with or without LIS) is associated with lower teriparatide discontinuation in Part D. Lower osteoporosis medication adherence associated with coverage gaps could translate into future adverse health care outcomes
(e.g. fractures).
Funding Source(s): Eli Lilly and Company
Theme: Medicare
● Adding Socioeconomic Statis to Framingham
Risk Scoring to Reduce Disparities in Coronary
Risk Assessment
Daniel Tancredi, Ph.D.; Kevin Fiscella, M.D.,
M.P.H.; Peter Franks, M.D.; M.D., M.P.H.
Presented by: Daniel Tancredi, Ph.D., Assistant
Professor, Pediatrics, University of California,
Davis, 2103 Stockton Boulevard, Suite 224,
Sacramento, CA 95618, Phone: (916) 734-3293;
Email: djtancredi@ucdavis.edu
Research Objective: Epidemiological studies suggest that the increased coronary heart disease
(CHD) associated with low socioeconomic status
(SES) is not adequately captured by traditional
CHD risk factors, resulting in potential underestimation of CHD risk for low SES persons and thus under treatment, further exacerbating
SES disparities in CHD. We examined the impact of adding SES to Framingham Risk Scoring (FRS) to improve CHD prediction for persons with low
SES.
Study Design: Using an inception cohort, we developed a model for CHD risk assessment that adds a robust measure of SES (<12 years of education or <150% of federal poverty) to FRS and compared the discrimination and calibration of this model with one based on FRS alone. We validated the findings in a second, nationally representative cohort.
Population Studied: We developed the model using data from the Atherosclerosis Risk in
Communities (ARIC) Study, which was initiated in
1987 and included 4,000 persons aged 45-64 from a four communities: Forsyth County, North
Carolina; Jackson, Mississippi; the northwestern
Suburbs of Minneapolis, Minnesota; and
Washington County, Maryland. ARIC contains detailed clinical data relating to CHD events including mortality. The National Health and
Nutritional Examination Survey (NHANES III) linked to the National Death Index served as the validation cohort and allowed us to assess risk for CHD mortality.
Principal Findings: Based on FRS alone, persons of higher and lower SES had a predicted CHD risk of 3.7% and 3.9%, respectively, compared to observed risks of 3.2% and 5.6%. Adding SES to a model with FRS improved calibration, with predicted risk estimates of 3.1% and 5.2% for those with higher and lower SES, mitigating the discrepancy between predicted and observed CHD events for low SES persons. Model discrimination (Area Under the Receiver Operator Curve) was not significantly affected. Adding SES to FRS lowered treatment thresholds such that an additional 15% of low SES persons become eligible for statin treatment based on the Third Adult Treatment Panel (ATP III) guides for treatment of high cholesterol in adults.
Consistent findings were observed in the validation sample.
Conclusion: Standard FRS underestimates CHD risk for those at low SES; treatment decisions ignoring SES may exacerbate SES disparities.
Adding SES to CHD risk assessment reduces this bias.
Implications for Policy, Delivery or Practice:
Current ATP III guidelines for cholesterol treatment may result in undertreatment of persons at low
SES. Further study is needed, both to determine the optimal means for assessing SES and incorporating it into treatment guidelines, as well as to quantify the potential that revised guidelines offer for reducing SES disparities in CHD.
Funding Source(s): NHLBI
Theme: Disparities
● A Longitudinal Study of Missed Opportunities for Diagnosis of Overweight among Young
Adults in an Academic General Medicine Clinic
Joyce Tang, M.D.; Robert Kushner, M.D.; Jason
Thompson, B.A.; David Baker, M.D., M.P.H.
Presented by: Joyce Tang, M.D., Postdoctoral
Research Fellow, Institute for Healthcare Studies,
Northwestern University, 750 South Lake Shore
Drive, 10th Floor, Chicago, IL 60605, Phone: (312)
503-5577; Email: joyce-tang@northwestern.edu
Research Objective: Young adults (aged 18-35 years) who are overweight (Body Mass Index [BMI]
25-<30) are at increased risk for obesity. Although one cross-sectional study suggested low rates of physician diagnosis of overweight, no prior studies
have examined diagnosis rates longitudinally. We conducted a retrospective cohort study to determine the proportion of young adults that were overweight or became overweight during a 2-4 year period, the proportion of overweight patients who received a physician diagnosis of overweight, and the proportion of patients with missed opportunities for diagnosis at new or preventive visits.
Study Design: We calculated BMI at each visit that weight was available using the last available height.
Patients with a baseline BMI>=30 were excluded from these analyses. We classified patients as overweight if they had a BMI 25-<30 kg/m2 at any visit between 2004-6. We determined whether patients were diagnosed as overweight based on whether any physician recorded an ICD-9 code
(278.00-278.02) for diagnosis of overweight or obesity in the past medical history, problem list, or encounter diagnosis between 2004-8, allowing for a two year minimum follow-up time. We classified patients as becoming obese if they developed a
BMI>=30 during any follow-up visit. To examine opportunities for diagnosis, we identified the number and types of visits (new or preventive, acute or other) for each patient. New or preventive visits were defined as those including ICD-9 codes specific to preventive visits (V70.0, V70.3, V70.5,
V70.8, V72.31) or which were slotted to be 40 minutes in duration.
Population Studied: We used our electronic medical record (EMR) to identify a cohort of 6461 individuals who had a first visit to our academic general medicine clinic between 2004-6, were 18-
30 years old, and who had at least one subsequent visit between 2004-8.
Principal Findings: The sample included 6461 patients with a median of 4 clinic visits. 68.2% of patients had at least one new or preventive visit.
BMI could not be calculated for 38.4% of patients, with virtually all cases due to missing height data.
Weight was recorded at an average of 57% of visits. Of the 3931 individuals with measured height and weight, 29.6% were overweight between 2004-
2006. 21.2% (530/2500) of females and 44.3%
(634/1430) of males were overweight. Only 3.0% of overweight patients received a physician diagnosis of overweight over a follow-up period of 2-4 years.
12.8% of overweight patients became obese
(BMI>30) during subsequent follow-up. There were no differences in diagnosis rates by number or type of visit. Most overweight patients (73.7%) had a missed opportunity for diagnosis during a new or preventive visit at the time of or subsequent to first overweight BMI.
Conclusion: In this longitudinal analysis of young adults seen in an academic general medicine clinic, a third were overweight, but few (3.0%) of these individuals received a physician diagnosis of overweight. The vast majority of overweight patients were not diagnosed despite having had a new or preventive visit. Height is often not recorded, which poses a barrier to identifying and recognizing overweight patients.
Implications for Policy, Delivery or Practice:
Interventions are needed to decrease missed opportunities for diagnosis of overweight.
Funding Source(s): AHRQ
Theme: Obesity Prevention and Treatment
● Cancer Cost in the US: Has the Burden
Shifted over Time?
Florence Tangka, Ph.D.; Justin Trogdon, Ph.D.;
Lisa Richardson, M.D.; Sue Sabatino, M.D.; David
Howard, Ph.D.; Eric Finkelstein, Ph.D.
Presented by: Florence Tangka, Ph.D., Health
Economist, Division of Cancer Prevention &
Control, Centers for Disease Control & Prevention,
4770 Buford Highway Northeast, MS K-55, Atlanta,
GA 30341, Phone: (770) 488-1183; Email:
Ftangka@cdc.gov
Research Objective: The objective of this analysis is to use nationally representative data spanning two decades to 1) present updated estimates of the prevalence and direct medical cost of cancer for select payers and how these have changed from
1987 through 2005 and 2) for each payer, present the distribution of payments by point of service across the two time periods to assess whether there have been shifts in where cancer is treated.
Study Design: This study used data from the 2001 through 2005 Medical Expenditures Panel Survey
(MEPS) and its predecessor, the 1987 National
Medical Care Expenditure Survey (NMES). We used an econometric approach to estimate cancerattributable medical expenditures by payer
(Medicare, Medicaid, private insurers, other public payers, and self-pay) and point of service (inpatient, prescription, and other). The estimates from the econometric models were used to quantify medical expenditures for each payer/service attributable to cancer. All costs are reported in 2007 dollars.
Population Studied: U.S. civilian noninstitutionalized population
Principal Findings: The prevalence of cancer patients increased between 1987 and the 2001 to
2005 period among all payers. This, combined with an increase in the number of enrollees for each payer, led to substantial increases in the number of cancer cases, ranging from an 81.1% increase in the number of cases for those covered by other public payers to a 250.8% increase in the number of cases for those covered by Medicaid. In 1987, the direct medical cost of cancer was $24.5 billion.
Between 1987 and the 2001-2005 period, the direct medical cost of cancer increased by 98%, to $48.6 billion. Medicaid costs increased by 357% ($1.5 billion), costs to private insurers increased by 137%
($13.6 billion), and Medicare costs increased by
103% ($8.5 billion). Across payers, the share of cancer medical costs that resulted from inpatient admissions fell from 64.8% of total medical cancer
costs in 1987 to 27.2% in 2001-2005. Prescriptions accounted for 6.0% of all cancer-attributable medical expenditures in 2001-2005, up from 1.8% in 1987.
Conclusion: Although there has been a shift away from inpatient care, a growing population combined with higher treated prevalence has resulted in large increases in cancer related medical costs to all payers.
Implications for Policy, Delivery or Practice: As cancer medical cost continue to rise, it is important to understand the changes in the burden of cancer care by payer and the setting where care is provided. Such information is critical for allocating resources and planning cancer care.
Funding Source(s): CDC
Theme: Public Health
● Appropriate Processes of Care for Bipolar
Disorder
Carrie Farmer Teh, Ph.D.; Amy Kilbourne, Ph.D.;
Elaine Lasky, R.N.; Deborah Welsh, M.S.; Karen
Austin, M.P.H.
Presented by: Carrie Farmer Teh, Ph.D.,
Associate Policy Researcher, RAND Corporation,
4570 Fifth Avenue, Suite 600, Pittsburgh, PA
15213, Phone: (412) 683-2300; Email: tehc@upmc.edu
Research Objective: Studies regarding quality of care for mental health conditions generally focus on adherence to medication treatment guidelines.
However, quality measures based solely on medication algorithms are difficult to operationalize because of the need for timely and accurate psychiatric diagnosis data. Moreover, patientcentered measures that reflect good clinical practice for the whole person, such as assessment of medication side effects, suicidal ideation, noshows follow-ups, and comorbidities, have not been operationalized. We used chart review methodology to define measurable aspects of patient-centered processes of care for bipolar disorder and examined whether veterans with bipolar disorder receive care concordant with these practices.
Study Design: Data from the Continuous
Improvement for Veterans in Care – Mood
Disorders (CIVIC-MD) study were used to assess chart documentation of key processes of care for
435 veterans with bipolar disorder. Processes of care (good clinical practice) were defined as chart notation of bipolar symptoms, substance use and psychiatric comorbidity, medical assessment, and documentation of patient’s treatment experience.
We assessed the rate of receipt of these important processes of care as well as variables associated with receiving good care.
Population Studied: Patients who were receiving inpatient or outpatient treatment for bipolar disorder at a large urban VA facility in Western Pennsylvania from July 2004 to July 2006 (n=435).
Principal Findings: Overall, 51% had documented assessment of bipolar symptoms, 78% had documented assessment of substance abuse, and
6% had documented screening for PTSD. Suicidal ideation was assessed in only 34% of patients. Side effects of bipolar medications were noted 48% of the time and no-show visits were followed up 45% of the time. Medical assessment was high - 80% had documented weight, blood pressure, and pulse, though obesity was noted in only 46% of the charts of obese patients (BMI >=30). In multivariate analyses, we found that homelessness (OR=1.61;
95% CI= 1.05-2.46) and use of complementary therapies (OR=1.90; 95%CI= 1.06-3.41) were associated with increased documentation of side effects, no-show visit follow-up and patient reasons for discontinuing medications.
Conclusion: Only half of veterans diagnosed with bipolar disorder received care concordant with good clinical practice. Of particular concern is the fact that assessments of suicidal ideation and PTSD were infrequent.
Implications for Policy, Delivery or Practice:
High quality treatment of bipolar disorder includes not only adherence to treatment guidelines but also the routine conduct of appropriate care processes.
The development of interventions to routinize these care processes is essential in order to improve the quality of care for this group.
Funding Source(s): VA
Theme: Behavioral Heath
● The California Minimum Staffing Law, Nurse
Satisfaction, Hours Worked & Wages: Analysis of 1997, 2004, 2006 & 2008 Survey Data
Michelle Tellez, R.N., M.S., Ph.D.; Joanne Spetz,
Ph.D.; Mitchell Watnik, Ph.D.
Presented by: Michelle Tellez, R.N., M.S., Ph.D.,
Assistant Professor, Nursing & Health Sciences,
California State University, East Bay, 4700 Ygnacio
Valley Road, Concord, CA 94521, Phone: (510)
654-7104; Email: michelle.tellez@csueastbay.edu
Research Objective: The purpose of this study is to explore changes in the nursing workforce that occurred after the implementation of California’s minimum nurse-to-patient ratios law in 2004. Using survey data collected by the California Board of
Registered Nursing from 1997 (before the ratios implementation), 2004 (immediately after implementation), 2006 (mid-term) and 2008 (longerterm), this study has two objectives. First, we describe changes in the California nurse population, satisfaction rate, hours worked and hourly wages across the four survey years. Second, we examine the relationships between wages, average hospital nurse staffing per patient in a nurse’s region, and labor supply, and whether these relationships changed after the ratios were implemented.
Study Design: This is a secondary data analysis of cross-sectional survey data. We first describe the
nursing workforce, and changes in demographics and satisfaction, from 1997 through 2008. Then, we estimate two multivariate regression equations, which estimate labor supply as a function of wages, average nurse staffing per patient in the nurse’s county, and human capital variables. In the first equation, the dependent variable is hours worked, and the equation is estimated using a two-stageresidual-inclusion regression with instrumental variables to address endogeneity of hours and wages. The second equation’s dependent variable is an indicator of whether a nurse is working fulltime versus part-time. This equation is estimated with the logit method. The equations are estimated for each of the years separately, and for all four survey years together (1997, 2004, 2006, and
2008).
Population Studied: The population of interest consists of California nurses who are able to respond to the regulatory changes taking place in this health care work environment. The inclusion criteria are that nurses must be between the ages of 20 and 65, reside in the state, and be working in nursing at the time of the survey.
Principal Findings: Preliminary results for the equation regressing hours worked indicate that the wage coefficient was significant, although small and negative, while the coefficient for year of the survey was large and positive. Since nurses worked more hours in the latter years, these coefficients suggest that it was not in response to wages, but in response to other factors that have changed over time, which may include changes in nurse staffing due to the ratio law.
Conclusion: Findings from this study suggest that the California law is having a positive effect on the nurse shortage in California. Between now and the
AcademyHealth meeting, we will compute measures of average hospital staffing in the regions in which RNs work, and determine whether such changes had a significant effect on labor supply.
Implications for Policy, Delivery or Practice:
Results from this study inform policy-makers in states without ratios as to their effects on the labor force. This study reminds front-line managers and workforce researchers about the importance of understanding workload and scheduling effects on staff retention. Finally, this study highlights the value of periodic and consistent data collection by institutions, such as the California BRN, to the evaluation of regulatory practices and the health of the nursing workforce.
Theme: Health Care Workforce
● Predictors of Consumer Decision-Making
Support under Florida’s Medicaid Reform
Keva Thompson, M.P.H.; Gail Young, M.A.;
Allyson Hall, Ph.D; Kimberly Elliott, B.A.; Lilly Bell,
M.H.A.
Presented by: Keva Thompson, M.P.H., Ph.D.
Student, Department of Health Services Research,
Management & Policy, University of Florida,
P.O.Box 100195, Gainesville, FL 32610, Phone:
(325) 273-6068; Email: kethomps@phhp.ufl.edu
Research Objective: In the past decade policymakers have increasingly espoused the notion of patient decision-making and empowerment in healthcare. In the state of Florida,
Medicaid Reform was implemented in several demonstration counties (Broward & Duval) with a view to control cost and improve quality of care.
Reform efforts required active consumer decisionmaking from a selection of Health Maintenance
Organizations (HMO)and Provider Sponsored
Networks (PSN). Consequently, Choice
Counseling Program was implemented by the state to support enrollee selection. The purpose of this study is to evaluate whether enrollee attributes are related to the utilization of Florida’s Choice
Counseling Program.
Study Design: This is a secondary analysis of data used in the evaluation of the Florida Medicaid
Reform. Two logistic regression models and multinomial regression were performed.
Respondents were identified as a part of a weighted sample of second wave of Consumer Assessment of Health Care Provider and Systems (CAHPS) in the demonstration counties. A total of 7,206 interviews were completed. Primary dependent variables of interest were: if respondents had heard of choice counseling services; utilized the services or experienced any problems with the services.
Population Studied: All Medicaid enrollees residing in the demonstration counties who were consecutively enrolled for six months and participated between June and November 2007 were included in the analyses. Analysis was restricted to individuals in Broward and Duval counties that were at least 18 years old.
Principal Findings: Blacks had a lower odds (OR
=0.66, CI: 0.51-0.86) of having heard of choice counseling when compared to Whites. Similarly, males were less likely to report hearing about
(OR=0.68, CI: 0.52-0.89) these services. Enrollees reporting poorer health status were also less likely
(OR=0.36, CI: 0.21-0.60) to hear of the services relative to enrollees with excellent health status.
With respect to use of the services, males were less likely (OR=0.6, CI: 0.41-0.85) to have utilized choice counseling when compared to their female counterparts. Enrollees residing in Broward were more likely (OR=2.7, CI: 1.1-6.7) to report having a big problem with choice counseling.
Conclusion: Disparities existed for groups who heard about, utilized and encountered problems regarding the Choice Counseling Program. Race, sex and health status were all attributes that predicted use of the decision-making services offered by the state.
Implications for Policy, Delivery or Practice:
Providing enrollee decision- making support services in Florida was devise to empower
individuals to make choices in a complex environment. To reach maximum program effectiveness greater outreach must be employed to ensure certain groups are adequately reached.
These findings have important implications to efforts that seek to support consumer decisionmaking.
Funding Source(s): State of Florida
Theme: Consumer Choices in Health Care
● Rural-Urban Differences in Perceived Access to Health Services in Older Adults: The Role of
Workforce Supply.
Joshua Thorpe, Ph.D., M.P.H.; David Mott, Ph.D.
Presented by: Joshua Thorpe, Ph.D., M.P.H.,
Assistant Professor, Social & Administrative
Sciences, University of Wisconsin - Madison School of Pharmacy, 777 Highland Avenue, Madison, WI
53705, Phone: (608) 890-2091; Email: jthorpe@pharmacy.wisc.edu
Research Objective: This study aimed to (1) identify rural-urban disparities in perceived access to a range of health services in community-dwelling elderly; and (2) assess the relative contribution of healthcare workforce supply, above and beyond rural-urban differences in individual characteristics, in explaining identified disparities in perceived access.
Study Design: We used data from the 2004 wave of the Wisconsin Longitudinal Study (WLS), a longitudinal study of a random sample of 5,741 men and women who graduated from Wisconsin high schools in 1957, subset to those currently living in
Wisconsin (n=4,898). Indicators of perceived access barriers were derived from the Group Health
Association of America (GHAA) questionnaire of consumer satisfaction. The questionnaire assessed perceived access to mental health, specialty,
“routine”, and emergency care, as well as hospitals and services for filling prescription medications.
Respondents were asked to rate access to each of the services as poor, fair, good, very good, excellent; responses of poor or fair were categorized as a perceived barrier. Provider-topopulation ratios (primary care doctors, specialists, psychiatrists, hospitals, emergency rooms), HPSA primary and mental health shortage area designations, and county-level rurality (rural urban continuum codes), were appended from the Area
Resource File. We appended community pharmacy-to-population ratios from the Wisconsin
Department of Regulation and Licensing. Selection of individual-level control variables was guided by the Andersen behavioral model (i.e., predisposing, enabling, need). Unadjusted logistic regression models were first conducted to determine the total effect of rurality, then were adjusted to control for rural-urban differences in individual-level predisposing, enabling, and need factors.
Population Studied: Community-dwelling older adults (age 63-67 years) who graduated from
Wisconsin high schools in 1957.
Principal Findings: The percentage of respondents perceiving an access barrier ranged from 3.1% for access to hospitals to 17.5% for access to “routine” care. Unadjusted logistic regression results revealed that, compared to older adults living in metropolitan counties, those in rural counties were more likely to perceive barriers for all healthcare services except for routine care.
Unadjusted odds ratios comparing the likelihood of perceived barriers between rural and metropolitan counties were: emergency services (OR=2.99, p<.01) hospital care (OR=2.8, p<.01), specialty care
(OR=2.7, p<.01), mental health care (OR=2.6, p<.01), and prescription services (OR=1.8, p<.05).
These disparities persisted even after adjusting for individual-level predisposing, enabling, and medical need. For prescription services, the disparity was no longer significant after adjusting for workforce supply variables. However, workforce supply did not statistically explain the rural-urban disparities for emergency, hospital, mental health, and specialty care.
Conclusion: Compared to older adults in metropolitan counties, those residing in rural counties were less satisfied with access to five of six healthcare service types. These disparities were not explained by rural-urban differences in individual-level characteristics. Disparities also persisted for all service types except for prescription services, even after controlling for workforce supply variables (primary care, mental health, specialists, hospitals, emergency services, pharmacies).
Implications for Policy, Delivery or Practice:
These results suggest that older adults in rural areas -- a group with the potentially the greatest healthcare needs-- may face greater challenges in accessing a number of different health services.
Further, the rural-urban disparity persisted even adjusting for differences in workforce supply, suggesting that policies designed to increase workforce supply in rural areas, while necessary, may not be sufficient for ensuring access to care for older adults living in rural communities.
Theme: Health Care Workforce
● Diabetes Performance Measurement at the
Physician Group Practice Level: A Comparison of Methods for Assigning Medicare
Beneficiaries with Diabetes to an Outpatient
Provider Group
Carolyn Thorpe, Ph.D., M.P.H.; Grace Flood, M.D.,
M.P.H.; Sally Kraft, M.D., M.P.H.; Christine Everett,
P.A.; Maureen Smith, M.D., Ph.D., M.P.H.
Presented by: Carolyn Thorpe, Ph.D., M.P.H.,
Assistant Research Scientist, Department of
Population Health Sciences, University of
Wisconsin, E5/724 Clinical Sciences Center, 600
Highland Avenue, Madison, WI 53792, Phone:
(608) 262-4051; Email: cthorpe@wisc.edu
Research Objective: Performance measurement and pay-for-performance initiatives at the physician group practice (PGP) level are increasingly advocated; however, little research has evaluated how different methods of assigning patients to
PGPs affect performance measurement. This study compared three methods for assigning Medicare patients with diabetes to a PGP.
Study Design: We analyzed Medicare data for diabetes patients who were seen at least once in any setting during 2003-2004 by a physician within a large Midwestern academic PGP. Patients were assigned to the PGP for 2004 using three common methods: 1) the “Most Visited Provider” method includes patients who had the greatest number of their 2004 outpatient Evaluation & Management
(E&M) visits at the PGP (requires all claims); 2) the
“Provider Home” method includes patients with two diabetes-coded and two primary care (or 1 primary care/1 endocrinology) outpatient E&M visits at the
PGP during 2003-2004 (requires PGP claims only); and 3) the “Any Diabetes Care” method includes patients with any 2004 diabetes-coded outpatient
E&M visits at the PGP (requires PGP claims only).
We examined means, proportions, and overlap of
95% bias-corrected bootstrapped confidence intervals to compare the three non-mutually exclusive groups regarding descriptive characteristics and the PGP’s provision of hemoglobin A1c and LDL monitoring.
Population Studied: 21,619 diabetes patients who were continuously enrolled in Medicare fee-forservice and seen by a physician within the PGP at least once in 2003-2004.
Principal Findings: Only 28% (n=6,072) of patients were assigned to the PGP under at least one method; 15% (n=3,158) were assigned under all three methods. Fewer patients were assigned under Provider Home (16.5%; 16.0%-17.0%) than
Most Visited Provider (22.4%; 21.9%-23.1%) and
Any Diabetes Care (23.0%; 22.4%-23.6%) methods, due to insufficient visits coded for diabetes or to primary care physicians/endocrinologists. Diabetes complications were most common in the Any
Diabetes Care group and least common in the
Provider Home group. Visits to non-PGP physicians were more likely in the Any Diabetes
Care group than the other groups. Further, when compared to the Most Visited Provider group, patients in the Provider Home group were more likely to receive all of their primary care (89% vs.
83% of patients) and diabetes visits (82% vs. 79% of patients) at the PGP. Based on PGP claims only, diabetes performance was significantly higher under the Provider Home method, with 58% of patients having >1 A1c tests in 2004 compared to
46% and 46% in the other groups, and 62% having at least 1 LDL test in 2004, compared to 52% and
52% in the other groups.
Conclusion: Compared to other methods, the
Provider Home method assigned the PGP responsibility for fewer patients with less complicated diabetes whose primary and diabetes care was more concentrated within the PGP. The
PGP’s diabetes performance levels appeared more favorable under the Provider Home method.
Implications for Policy, Delivery or Practice: The method used to assign diabetes patients to PGPs may substantially affect estimates of diabetes performance, and should be carefully considered when designing pay-for-performance initiatives and comparing performance measures gathered from different sources.
Funding Source(s): AHRQ
Theme: Prevention and Treatment of Chronic
Illness
● Payment-efficiency Analysis of Joint
Replacement Rehabilitation in Skilled Nursing 7
Inpatient Rehabilitation Facilities
Wenqiang Tian, Ph.D.; Gerben DeJong, Ph.D.;
Randall Smout, M.S.; Koen Putman, Ph.D.; Joan
DeVanzo, Ph.D.
Presented by: Wenqiang Tian, Ph.D., Senior
Research Associate, National Rehabilitation
Hospital/Medstar Research Institute, 102 Irving
Street, NW, Washington, DC 20010, Phone: (202)
877-1038; Email: wenqiang.tian@medstar.net
Research Objective: Joint replacement is the largest category of patients receiving rehabilitation services in both skilled nursing facilities (SNF) and inpatient rehabilitation facilities (IRF). The absence of a uniform outcomes assessment tool creates a barrier for cost-effectiveness comparisons between the two settings. The objective of this study is to assess the payment-efficiency for joint replacement rehabilitation using the same measurement tool in both settings.
Study Design: This study is a prospective observational cohort study. Data were collected from 10 SNFs and 11 IRFs across the nation between Feb. 2006 and Feb. 2007. We determined payment for IRF patients based on case-mix group
(CMG) and tier, and calculated SNF payment by resource utilization groups (RUGs) and associated length of stay. All payments were calculated as the base payment without any adjustments for teaching status, urban location, etc. The outcome was measured by observing the change in the functional independence measure (FIM) motor score between admission and discharge. Payment-efficiency was measured by dividing changes in motor FIM score by log-transformed payment because costs increase exponentially with each additional point of motor FIM gain.
Population Studied: The sample consisted of approximately 1,600 participants including 948 knee
replacement and 618 hip replacement patients receiving rehabilitation care at either a SNF or IRF.
The selection criteria were broad: (1) 21 years or older, (2) admitted following hip or knee replacement of any type and (3) did not have a hip fracture prior to hip replacement.
Principal Findings: Preliminary results show that
IRF patients incur higher costs than SNF patients while having shorter lengths of stay. This is true for both knee replacement ($9610±$2394 in 9.0±3.7 days vs. $5549 ± $2870 in 14.2 ±7.2 days) and hip replacement ($10380±$2614 vs. $6164 ± $3489 in
10.1 ± 4.8 days vs. 15.9±8.6 days). IRF patients entered with lower motor FIM scores (for knee: 42.4 vs.50.7, p<.001; for hip: 39.1 vs. 47.8, p<.001) and achieved larger motor FIM gains than SNF patients
(for knee: 25.9 vs.21.2, p<.001; for hip: 26.5 vs.
22.7, p<.001). Payment-efficiency in IRFs is higher than in SNFs for both knee replacement (2.82±0.77 vs. 2.49±0.84, p<.001) and hip replacement
(2.87±0.82 vs. 2.65±0.81, p=.007). Controlling for patient characteristics and severity, multivariate analyses showed that IRFs were still more payment-efficient for knee replacement patients, but not for hip replacement patients.
Conclusion: Despite receiving a higher payment from Medicare, IRFs, as a group, achieved higher motor FIM gains for both types of joint replacement patients. IRFs are more payment-efficient in providing rehabilitation care to knee replacement patients but not hip replacement patients.
Implications for Policy, Delivery or Practice:
Knee replacement can obtain payment-efficient rehabilitation services in IRFs. Hip replacement can obtain payment-efficient rehabilitation services in either a SNF or an IRF.
Funding Source(s): Jointly funded by the
HealthSouth Corporation, ARA Research Institute of the American Rehabilitation Providers
Association, Brooks Health, National Rehabilitation
Hospital, American Hospital Association, the
Federation of American Hospitals, and others.
Theme: Quality and Efficiency: Organized
Processes
● Analysis of Five Health Insurance Expansion
Options for New York State
Nicholas Tilipman, B.A.; Sherry Glied, Ph.D.;
Nicholas Tilipman, B.A.; Olveen Carrasquillo, M.D.,
M.P.H.
Presented by: Nicholas Tilipman, B.A., Research
Assistant / Project Coordinator, Department of
Health Policy & Management, Columbia University
Mailman School of Public Health, 600 West 168th
Street, New York, NY 10032, Phone: (212) 305-
7975; Email: nt2232@columbia.edu
Research Objective: In this study, we present cost and coverage estimates of five different options for expanding health insurance in New York State: a comprehensive coverage single-payer proposal, three increasingly generous variations of building block or incremental reform proposals, with and without individual mandates, and a more marketoriented approach that would eliminate service mandates and community rating in the individual market and require employers to offer Section 125
Cafeteria plans.
Study Design: Our analysis employs a simple cellbased modeling approach to provide estimates of the likely costs and consequences of selected expansion proposals. For each proposal, we estimate the number of people in New York State eligible for the expansions by categories defined by age, poverty level, and insurance status from the
2004-2006 Current Population Survey (CPS). We apply a take-up rate for each program to the aggregate population within each cell. We obtained estimates of national per capita health expenditures stratified by age, gender and type of service using the 2004 Medical Expenditure Panel Survey
Household Component, adjusted to reflect aggregate NYS level expenditure data from the
2004 National Health Accounts. We used these data to calculate the effects of each proposal on coverage and expenditures for the state by population.
Population Studied: Coverage effects of the five proposals were measured for the 2004 NYS population by age, poverty level and insurance status using data from the CPS.
Principal Findings: We estimate that extending coverage to all NYS residents, including the 2.6 million uninsured, through a single payer plan would result in a net increase in overall health care spending of $5.8 billion. To provide this coverage,
NYS would have to generate about $66.8 billion through taxes on individuals and employers, a substantial portion of which could be offset by recapture of current employer contributions and individual out-of-pocket payments. The building block approaches would cover between 500,000 and 1,000,000 uninsured people without a mandate at costs to NYS between $1 billion and $3.6 billion.
Adding a mandate would increase coverage to between 2.1 million and 2.3 million uninsured, and would increase costs to between $4 billion and $5.2 billion. The market-based approach would cover between 100,000 and 130,000 uninsured persons at a cost to NYS of $130 million.
Conclusion: We find that no single approach clearly dominates the others. Any coverage expansion in NYS will involve tradeoffs between numbers covered, costs, generosity and comprehensiveness of coverage.
Implications for Policy, Delivery or Practice: By providing a reference of the broad implications and tradeoffs of the various reform options and philosophies, we hope our findings will stimulate debate and discussion on health insurance expansions among NYS policymakers and stakeholders.
Funding Source(s): New York State Health
Foundation
● Development of a Summary Measure of
Disparities in Access to Health Services &
Outcomes: A Coefficient of Health Equity
Brenda Tipper, M.H.Sc.; Adalsteinn Brown, Ph.D.;
Carey Levinton, M.S.
Presented by: Brenda Tipper, M.H.Sc., SPWIn
Consulting, 201 Von Neumann Drive, Princeton, NJ
08540, Phone: (609) 279-2803; Email: btipper@spwinconsulting.com
Research Objective: Disparities in health outcomes and access to services are commonly analyzed through stratifying results of measures and analyzing the differences across sub-groups.
This approach provides important information about access and outcome disparities across the groups of interest. However, the results may not indicate if disparities are changing over time, or if the extent of disparities varies across health systems. Our goal was to develop a summary measure of health equity, ranging in value between 0 and 1 (similar in concept to the Gini coefficient of income inequality), to be reported in the Ontario Health System
Scorecard. Such a coefficient would help us understand changes in health system equity over time and variations across health regions and enable us to more directly measure the impact of strategies designed to improve equity.
Study Design: A conceptual model of relationships between selected SES and demographic characteristics and “unmet health care needs” was developed. Classification and regression tree
(CART) analysis was used to determine which of the characteristics were most significant in segmenting the population group into relatively homogeneous sub-groups based on rate of unmet health care need. The results were used to plot a
Lorenz curve of cumulative percent of total population versus cumulative percent of reported unmet need. A Gini coefficient of equity for unmet health care need was calculated. Gini coefficients were determined for all of Ontario and for regional health systems.
Population Studied: Data from the 2003 and 2005
Canadian Community Health Surveys (CCHS) were used for the analysis. The CCHS is a representative survey of Ontario households that includes series of questions on health status, utilization of health care services, and SES and demographic characteristics.
Principal Findings: The percentage of respondents reporting unmet health care needs increased from 11% in 2003 to 12% in 2005. The equity coefficient for unmet health care needs increased from 0.167 to 0.185, indicating that the distribution of unmet health care need became less equitable. The percentage of respondents reporting unmet need varied across regions from 10.0% to
14.3%. The equity coefficients varied from 0.125
(most equitable) to 0.231 (least equitable). Results for both the equity coefficient and percent reporting unmet health care needs improved in three of the fourteen regions; results for both measures declined in six regions.
Conclusion: A coefficient of equity is a summary measure of disparities in access to health services or outcomes across population sub-groups. This measure can highlight differences in equity across health systems and trends in equity over time. The methodology was applied to the measure of unmet health care needs; it could be applied to other measures of access or outcome such as rates of angiography or hospitalization for ACSC.
Implications for Policy, Delivery or Practice: A summary measure of health system equity such as the equity coefficient can assist policy-makers in formulating and measuring targets for health system equity. The impact of strategies and policies can be tracked from the perspective of equity over time and we can assess whether improvement in equity coincides with improvement on health system measures of access or outcomes.
Funding Source(s): Canadian Institutes of Health
Research
Theme: Disparities
● The Effect of Live & Web-Based Training
Sessions on the Accuracy of Blood Loss
Estimation in Simulated Obstetric Delivery
Scenarios
Paloma Toledo, M.D.; William Grobman, M.D.,
M.B.A.; Robert McCarthy, Pharm D.; Cynthia Wong,
M.D.
Presented by: Paloma Toledo, M.D., Instructor in
Anesthesiology, National Research Service Award
Fellow, Anesthesiology, Northwestern Memorial
Hospital, 251 East Huron Street, F 5-704, Chicago,
IL 60611, Phone: (312) 479-4493; Email: ptoledo@md.northwestern.edu
Research Objective: Clinicians routinely estimate blood loss using visual assessments and have been shown to overestimate small volumes, and underestimate large, clinically significant amounts of blood loss. The objective of this study was to evaluate the impact of a didactic training program on the accuracy of blood loss estimates and to determine the effectiveness of this training when provided via a web-based session compared with the live training.
Study Design: Labor and delivery patient care providers volunteered to participate in this study as part of an educational program. The study was divided into two phases, live and web-based training and testing. Expired packed red blood cells were used to create five pretest stations.
Participants estimated the blood volume for each station and then received structured training.
Following the didactic session, a five station post-
test was administered. The stations were the same as the pretest; however, the volumes of blood used to create the stations were varied from the pretest.
Providers who did not participate in the live training program completed a web-based version of the training program. Participants completed a pretest in which they were asked to estimate blood loss in 5 photographs of the pre-test stations used in the live sessions. After completing the pre-test the participants advanced through the same didactic materials presented at the live sessions. The participants then estimated blood volumes from photographs taken of the five post-test scenarios used in one group during the live sessions.
Accuracy was determined as the percent error from the actual volume for each scenario.
Population Studied: All labor and delivery providers: obstetric and anesthesia residents and attendings and labor and delivery nurses.
Principal Findings: In the pretest scenarios, the only difference between the live and web-based testing groups was in the vaginal delivery drape scenario where the electronic group underestimated the blood loss more than the live group (-71.5 ±
17.0 versus -52.5 ± 23.4) (P<0.001). Following training, estimates of blood volumes improved for all scenarios regardless of method of training
(P<0.001). At the pretest, providers in both groups overestimated volumes in small volume scenarios, but better approximated the true value after training.
Despite a significant reduction, both groups continued to underestimate blood volumes in postpartum hemorrhage scenarios.
Conclusion: The important finding of this study was that didactic training on postpartum hemorrhage and blood loss estimation techniques significantly improved the accuracy of estimated blood loss in clinical scenarios. Our study supports the use of web-based training, although live training appears to be more effective when estimating large volumes.
Implications for Policy, Delivery or Practice:
Delay in the diagnosis and treatment of postpartum hemorrhage may place a parturient at increased risk for adverse outcomes. The use of these educational programs could help improve the accuracy of blood loss estimation and possibly improve early detection of blood loss, allowing for improved maternal outcomes.
Funding Source(s): Departmental funding
Theme: Organizational Performance and
Management
● "Lost" in the Emergency Room: A Profile of
Uninsured Young Adult Use of the Emergency
Department
Jonathan Tolentino, M.D.; Lisa Simpson, M.B.,
B.C.; Anthony Goudie, Ph.D.; Gerry Fairbrother,
Ph.D.
Presented by: Jonathan Tolentino, M.D., House
Staff, Cincinnati Children's Hospital Medical Center,
3333 Burnet Avenue, ML 5018, Cincinnati, OH
45229-3039, Phone: (513) 258-5431; Email:
Jonathan.Tolentino@cchmc.org
Research Objective: The young adult population has had the highest uninsured rates. This has been attributed to aging out of private and public coverage and lack of jobs offering employer-based insurance. Young adults have reported a high rate of Emergency Department (ED) use, second only to the most elderly population. The purpose of this study is to determine whether ED visits for young adults are more likely to be uninsured compared to other populations seeking care and to describe how socioeconomic factors influence payer status for emergency room visits.
Study Design: This study was a secondary analysis of de-identified data that used the
Healthcare Cost and Utilization Project (HCUP)
2006 State Emergency Department Datasets
(SEDD), a large administrative dataset of state emergency department visits for those patients who are treated and released and uniquely allows for evaluation of variation between states. We studied
Maryland, New Jersey, Arizona, California, and
Florida data. Univariate and bivariate analyses were performed using SAS version 9.1. Insurance status was defined using expected payer with uninsured individuals having "no expected payer."
Age groups were defined as “young adult” ages 19
– 26, “adolescent” ages 11 – 18, and “adult” ages
27 – 49. Regions were defined as large metropolitan, small metropolitan, micropolitan, and rural. Race was reported as white, black,
Hispanic/Latino, or other. Multivariable analysis will use log-linear regression to evaluate the interaction between payer status and socio-demographic variables.
Population Studied: Young adult population ages
19 - 26 using emergency department data from the
2006 HCUP SEDD Database. We examined five states - Arizona, California, Florida, Maryland, and
New Jersey
Principal Findings: Over 2.9 million ED visits were made by young adults in the five states studied.
35.2% of ED visits made by young adults had no payer source, compared to 15.1% of adolescent visits and 27.1% of adult visits. Compared to visits for young adult women, a higher percentage of visits by young adult men had no payer source
(44.1% vs. 29.47% respectively), and was higher compared to adolescent (15.4% vs.14.73% respectively), and adult (33.4% and 22.2% respectively) visits. The patterns of uninsured ED visits were similar among the defined regions.
However, the percentage of uninsured ED visits varied between states, and the pattern consistently showed young adult uninsured visit rates higher than adolescent and young adult visits. The largest proportion of uninsured visits was found in Florida
(41.1%) and New Jersey (40.2%) with the smallest proportion found in Arizona (26.4%) and California
(29.9%). The analysis also found that 39.4% of
Hispanic young adult visits had no payer source, closely followed by 39.2% for blacks, which compares to 33.9% of visits for whites.
Multivariable results will be reported.
Conclusion: This study suggests that young adults have a higher proportion of uninsured or non-payer
ED visits compared to adolescent or adult patients, with greater than a third ED visits essentially uninsured. ED visits by women have a higher proportion of insurance in the young adult population is likely the reflection of the greater availability of public healthcare programs to women.
While little variation was seen between rural and metropolitan areas, there was significant variation between states, which may be attributed to the differences in state public coverage programs.
Implications for Policy, Delivery or Practice:
There are several opportunities for federal and state governments to improve access to care to reduce the number of uninsured young adults and the frequency of non-payer visits to the emergency department.
● Measuring the Effect of Hospital Palliative
Care Programs on Discharge Destinations of
Patients with Life Limiting Illness
Rachel Toungate, M.I.S.
Presented by: Rachel Toungate, M.I.S., Clinical
Outcomes Research Team Manager, Clinical
Outcomes Research Team, Seton Family of
Hospitals, 1201 West 38th Street, Austin, TX
78705, Phone: (512) 324-3303; Email: rltoungate@seton.org
Research Objective: Measure the effect of hospital palliative care programs on discharge destinations of patients with life limiting Illness. The study hypothesis is that those hospitals with a palliative care program will have a statistically significant higher rate of referral to hospice than those that don’t.
Study Design: The study is a retrospective cohort study that looks at whether Texas hospital-based palliative care programs have an effect on the discharge destination of the cohort of patients with life limiting illness. Data from the National Palliative
Care Research Center was used to determine whether the hospital has a palliative care program.
The study looks for correlation between two variables. The first variable is the number of patients with life limiting illness at hospitals that have a palliative care program. This is calculated by querying all patient’s ICD-9 diagnosis codes for conditions that fall into the following categories:
AIDS, Alzheimer’s Disease, ALS (also known as
“Lou Gehrig’s Disease”), Cancer, Cardiomyopathy,
Congestive Heart Failure, Chronic Pulmonary
Obstructive Disease, Dementia, Emphysema, Liver disease, Stroke, Renal or respiratory/pulmonary disease (end stage). Each patient is counted one time per hospital visit. This first variable is defined as the cohort destination per palliative care program hospital. The second variable is the number of patients with life limiting illness at hospitals that do not have a palliative care program. This is calculated by querying all patient’s ICD-9 diagnosis codes for conditions that fall into the categories mentioned above. Each patient is counted one time per hospital visit. This second variable is defined as the cohort destination per non palliative care program hospital.
Once the two variables are calculated, they are compared to determine if they are statistically correlated.
Population Studied: The population studied is the cohort of patients with an ICD-9 diagnosis code suggesting a life-limiting illness during January 1,
2007 through December 1, 2007.
Principal Findings: Analysis shows that there is a strong correlation showing that the life-limiting patient cohort in a hospital with a palliative care program is more likely to go to a hospice discharge destination.
Conclusion: Based on the positive results of this analysis, hospitals should be encouraged to put palliative care programs in place.
Implications for Policy, Delivery or Practice:
Palliative care programs have grown in popularity in the past years, with more and more hospitals adding them to their complement of programs.
Whether patients with a life-limiting illness are discharged to hospice, rather than an alternative environment, is an important measurement of these program’s effectiveness, as well as the patient’s quality of life. According to the National Hospice and Palliative Care Organization, along with other palliative care groups, the earlier in the course of a life limiting illness that a patient is given the option for palliative care, the better their quality of life is. In fact, the Medicare Payment Advisory Commission
(“MedPAC”) released a report in June of 2008 addressing the continuing trend of patients using hospice for more than the traditional six month period that had been established, suggesting that lengths of stay in excess of 180 days be certified.
Recent studies also show that those patients with a life-limiting illness admitted into a hospice live longer than those who remain in the hospital, further reinforcing the quality of life for the patient in a hospice program.
● Analyzing the Rhetoric of Obstetrician
Oriented Literature on Vaginal Birth after
Previous Cesarean Delivery
Felicia Trembath, B.S.; Robin Jensen, Ph.D.
Presented by: Felicia Trembath, B.S., Graduate
Student, Health & Kinesiology, Purdue University,
800 West Stadium Avenue, West Lafayette, IN
47907, Phone: (765) 838-1369; Email: ftrembat@purdue.edu
Research Objective: Following the surgical delivery of a child through cesarean section, future pregnancies will be delivered either by repeat cesarean or vaginal birth after previous cesarean delivery (VBAC). Controversy over selecting a repeat cesarean or VBAC centers on cost, health outcomes and malpractice risk. A VBAC costs significantly less than a cesarean delivery and is believed to lead to better health outcomes. VBACs can lead to rare adverse outcomes including uterine rupture, which has led to the development of strict policies on how and when a VBAC may be offered.
The International Cesarean Awareness Network found that over 300 hospitals in the United States had instituted a ban on VBACs, and refused to offer this procedure. Since the 1980s physician oriented literature on the use of VBAC has largely originated from the American College of Obstetrics and
Gynecology (ACOG). This study examines how the rhetoric of these publications is reflected in physician practice and patient agency and considers the question of whether the guidelines reflect goals of professional autonomy and/or quality care.
Study Design: Physician oriented publications on
VBAC issued by ACOG from 1988 to 2004 served as the focus for this analysis. These consisted of 3
Committee Opinions, 1 Practice Patterns, and 3
Practice Bulletins. Texts were analyzed in chronological order for insertions, deletions, and format changes. Each edition was analyzed for the rhetorical strategies utilized and its impact on physician and patient agency. This is compared to changes in the VBAC rate in the US.
Population Studied: Births in the US from 1988 to
2004
Principal Findings: As the reports evolve from committee opinions to evidence based clinical guidelines a variety of rhetorical themes are utilized, including medicalization of the procedure, the creation of a cultural narrative, physician as a gatekeeper, and loss of patient agency. The nature and tone of these reports contribute to the cultural narrative of cesarean births as normal and VBAC’s as abnormal, resulting in the assumption of greater decision-making authority by the physician, accompanied by a loss of patient agency. ACOG’s reports have indirectly resulted in over 300 hospitals adopting policies discouraging woman from having or even attempting a VBAC. These trends are illustrated by the sharp decline in
VBAC’s since 1997 and the steady increase in cesarean sections since 1999. Despite the increase of resulting cesareans, maternal and fetal outcomes have not improved in the United States.
Conclusion: Rhetorical themes that have evolved in ACOG guidelines on the appropriateness of
VBAC births are consistent with increased medicalization of childbirth and the transfer of decision-making from the patient to the physician.
The development of this rhetoric is consistent with changes seen in VBAC rates, and appears to have no advantageous impact on maternal and fetal outcomes.
Implications for Policy, Delivery or Practice:
ACOG recommendations are consistent with Starr’s model of the medical profession seeking increasing professional autonomy. In some instances, clinical guidelines may represent a mechanism for the assertion of professional prerogatives rather than best practices for the delivery of high quality health care.
● Physician Specialty & Symptoms from
Medicare Claims Data Prior to Ovarian Cancer
Katrina Trivers, Ph.D.; Matthew Urato, M.A.; Joseph
Burton, M.S.; Lisa Richardson, M.D. M.P.H.; Sun
Hee Rim, M.P.H.; Don Blackman, Ph.D.
Presented by: Katrina Trivers, Ph.D.,
Epidemiologist, Division of Cancer Prevention &
Control, Centers for Disease Control & Prevention,
4770 Buford Highway, Northeast MS K55, Atlanta,
GA 30341, Phone: (770) 488-1086; Email: ktrivers@cdc.gov
Research Objective: Ovarian cancer is the leading cause of death among all gynecologic malignancies. Local stage disease has a 5 year survival rate of 92%, but most women (81%) are diagnosed with regional or late-stage disease with
5-year survival rates of 71% and 30%, respectively.
Recent research indicated that late-stage disease was associated with gastrointestinal symptoms
(abdominal pain/distention, constipation, diarrhea, nausea/vomiting, etc.) and early stage disease with gynecologic symptoms (abdominal/pelvic swelling, abnormal bleeding, and genital organs pain).
Women with gynecologic symptoms also received abdominal/pelvic imaging sooner than women with gastrointestinal symptoms. The reasons for these results are unknown. One possibility is that pathways to diagnosis may differ for women who present with gynecologic versus gastrointestinal symptoms. Physician specialty may explain some of this difference. The overall aim of this research is to examine how physician specialty and symptoms influence the number of visits and time to diagnosis for ovarian cancer after symptoms are reported.
Study Design: Linkage study of Surveillance,
Epidemiology, and End Results (SEER) cancer registry program data with Medicare claims data.
Population Studied: Invasive, incident ovarian cancer cases (n = 8,152), diagnosed between 1996 and 2002, age 66 years and older, were identified through SEER. Of these women, 7,878 had at least one Part B Medicare claim in the 12 months prior to diagnosis, and this represents the sample for analysis (linkage success rate = 97%). For each case, we defined an “index” claim as the first claim with one of 4 types of symptoms - gynecological, gastrointestinal, other symptoms related to ovarian cancer, and any claim). Other symptoms included conditions like fatigue, benign neoplasm of the
ovary, and non-inflammatory disorders of the ovary/fallopian tube. We recorded the physician specialty on that index claim for each woman and quantified the physician visits and the number of months from the index claim to diagnosis for the various combinations of symptom types and physician specialties.
Principal Findings: Women with a gynecologic index claim were less likely to first see a general/family practitioner or internal medicine physician (19%) and more likely to see a obstetrician/gynecologist or other specialist (81%) than were women with a gastrointestinal index claim (45% and 55%, respectively). Women with a gynecologic index claim had fewer visits between symptoms and diagnosis (average 7.2 visits [95%
CI: 6.9 – 7.6]) than those with gastrointestinal symptoms (average 11.4 visits [11.1 – 11.8]), regardless of physician specialty on the index claim.
However, this was not true for women who saw obstetrician/gynecologists (8.2 versus 8.7 visits, respectively). Also, women with a gynecologic index claim had a shorter time to diagnosis (average 1.8 months (1.7 – 1.9)) versus those with gastrointestinal symptoms (average 3.9 months (3.7
– 4.0)), regardless of physician specialty.
Conclusion: In the year before diagnosis, physician specialty on the index claim may have less of an impact on pathways to ovarian cancer diagnosis than the type of symptoms at presentation.
Implications for Policy, Delivery or Practice: For all physician specialties, knowledge of ovarian cancer associated symptoms (including nongynecologic symptoms) is crucial and prompt attention to unresolved symptoms may be warranted.
Funding Source(s): CDC
Theme: Health Care Workforce
● A Review of Federally Funded Studies for the
Medical Home for People with Disabilities
Rebecca Trocki, A.B.; Harvey Schwartz, Ph.D.,
M.B.A.; Padmini Jadgadish, B.A., M.P.P.; Kishena
Wadhwani, Ph.D., M.P.H.
Presented by: Rebecca Trocki, A.B., Program
Analyst, Department of Health & Human Services,
Agency for Healthcare Research & Quality, 540
Gaither Road, Suite 2040, Rockville, MD 20850,
Phone: (301) 427-1541; Email: rebecca.trocki@ahrq.hhs.gov
Research Objective: Our goal was to synthesize what we know about federally-funded research studies on the medical home for people with disabilities.
Study Design: We searched National Institutes of
Health Information, Management, Planning,
Analysis and Coordination Systems II (IMPACII) and the Query, View and Report (QVR) and
Computer Retrieval of Information on Scientific
Projects (CRISP) systems. Search terms included medical home, disabilities, children with special health care needs, care coordination, patient centered care, and family centered care. Studies that met defined criteria (e.g., people with disabilities, United States-based, quantitative) were selected. We extracted data, including design, population characteristics, sample size, intervention, and expected outcomes.
Population Studied: The majority of the studies focused on children with special healthcare needs with specific chronic or multiple diseases that require care coordination. Some of the studies focus on “high risk” adolescents in underserved areas (e.g., inner cities) with sexually transmitted diseases, and a few address adults with a specific
“disability” such as depression or PTSD.
Principal Findings: We selected 35 studies, many of which addressed children with asthma, autism, spina bifida, sickle cell, cancer, and other chronic conditions. A small portion of the studies address issues about these specific conditions across the lifespan. These studies emphasize assisting patients with “navigating” the healthcare system.
Only two of the studies were comparison group studies. None of the studies examined the medical home or populations with disabilities in its entirety.
Although research designs, medical home-related interventions, populations and associated disabilities varied across studies, the intent of many of the studies suggests a positive relationship between medical home and desired outcomes.
Conclusion: The synthesis suggests that for people with disabilities, the medical home takes on a greater importance in integrating health and human services for individuals receiving public or private insurance. In addition, the synthesis holds promise for the hypothesis that medical homes will provide enhanced health-related outcomes for people with disabilities. Additional synthesis of published research on the medical home for people with disabilities and comparisons of published and currently funded research should be performed in order to identify possible trends and gaps. This synthesis suggests that further research on health care transition and preparation for adult care, especially including a deliberate, coordinated provision of developmentally appropriate and culturally competent health assessments, counseling, and referrals needs to be undertaken.
Further research with comparison groups encompassing many of the attributes of the medical home and consistent definitions also needs to be undertaken.
Implications for Policy, Delivery or Practice:
How can the expectations and preferences of patients with disabilities be incorporated into a medical home model? Can a medical home model be expected to address disparities in care delivery between people with disabilities and those without disabilities? Will the payment system make it viable for practices to provide medical home services for
people with disabilities? Will practices be able to transform into medical homes that will make available to all patients, including people with disabilities whatever “helpful” resources the medical home has available?
● The Effect of Hospital Global Budgeting on
Outcome of Care for Acute Myocardial
Infarction Patients
Yu-Chi Tung, Ph.D.; Guann-Ming Chang, M.D.,
M.S.; Ming-Chin Yang, Ph.D.
Presented by: Yu-Chi Tung, Ph.D., Assistant
Professor, Healthcare Information & Management,
Ming-Chuan University, Room AA611, No.5, De-
Ming Road, Gui-Shan Township, Taoyuan County,
333, TW, Email: yctung@mail.mcu.edu.tw
Research Objective: As health care costs continue to rise, cuts in reimbursement have become central of health care reform efforts such as global budgeting or the Balanced Budget Act in the United
States. An important-but-largely unanswered question is whether such reductions are associated with changes in patient outcomes. Since July 2002 in Taiwan, the Bureau of National Health Insurance
(BNHI) has adopted the hospital global budgeting system designed to replace a fee-for-service payment system. Using nationwide populationbased data from Taiwan, we examined the effect of hospital global budgeting on patient outcomes, adjusting for important hospital, physician and patient characteristics.
Study Design: We used the National Health
Insurance Research Database (NHIRD), provided by the BNHI and managed by the National Health
Research Institutes. With all discharges from 1327 physicians and 282 hospitals, we identified patients discharged between years 1997 and 2007 with a diagnosis of acute myocardial infarction (AMI) based on the International Classification of
Diseases, Ninth Revision, Clinical Modification
(ICD-9 CM) codes. We used multilevel logistic regression to examine the impact of hospital global budgeting on 30-day mortality for AMI patients, after taking into account patient-level, physician-level and hospital-level covariates. The hospital-level covariates included ownership, accreditation level, teaching status, annual hospital volume, geographic location, and so on. The physician-level covariates included age, specialty, and annual physician volume. The patient-level covariates included gender, age, comorbid conditions, weekend admission, and so on. Both time (years 1997–2007) and global budgeting phase (Before 1997–
6/30/2002, After 7/1/2002–2007) were incorporated as fixed effects. We took the time variable to be related to changes in hospital management, and the global budgeting phase variable to represent system change.
Population Studied: Access to discharge data, amounting to 110,422 records, was secured for all 282 hospitals in Taiwan over the period 1997–2007.
Principal Findings: We found that 30-day mortality of AMI patients in Taiwan increased after the implementation of the hospital budgeting system, adjusting for patient gender and age, comorbidities, physician age, specialty, and volume, hospital ownership, accreditation level, teaching status, geographic location, and so on.
Conclusion: Global budgeting created a moderate financial strain on hospitals. Outcome of care for
AMI patients was affected by these revenue reductions. The results in this study apply only to
AMI patients; we do not know the degree to which these findings can generalize to other conditions.
Implications for Policy, Delivery or Practice:
With cuts in reimbursement, measuring and monitoring the processes and outcomes of care regarding AMI may be necessary.
Theme: Quality and Efficiency: Policies and
Incentives
● Comparison of Health Plan Diabetes Costs for
Commercial HMO & PPO Plans
Sally Turbyville, M.A., M.S.; L. Gregory Pawlson,
M.D., M.P.H.; Kiran Johal, M.P.H.
Presented by: Sally Turbyville, M.A., M.S.,
Research Scientist, Performance Measurement,
National Committee for Quality Assurance, 1100
13th Street, NW, Suite 1000, Washington, DC
20005, Phone: (202) 955-1756; Email: turbyville@ncqa.org
Research Objective: In 2007, the National
Committee for Quality Assurance (NCQA) introduced standardized measures of health plan costs of care. Previous studies have found that while HMOs and PPOs provide roughly comparable quality of care, HMOs may use hospital and other expensive resources less. This study examines whether health plan performance differs between
HMO and PPOs using the NCQA cost of care measure diabetic members.
Study Design: Health plans submit data on costs of care for diabetics in four service categories: inpatient facility, evaluation and management, procedure and surgery and ambulatory pharmacy services. Plans apply standardized prices to utilization to estimate cost. Data are risk adjusted to account for variation in health plan case and riskmix of members. NCQA calculates an observed-toexpected (O/E) ratio for each plan comparing the actual costs to expected costs (given the casemix/risk of their members). All plans applied the same standardized prices for units of services to estimate cost. Observed costs are the plan submitted cost; expected costs are the plan risk adjusted costs. NCQA established the range of O/E results and the mean risk adjusted costs for each service category for the eligible population (the
PMPM for those with diabetes) for both commercial
HMO and PPO plans.
Population Studied: 206 commercial HMO and
135 commercial PPO plans submitted summarized cost and utilization data based on claims data for adult diabetic members. Diabetic members were enrolled in a plan for at least 11 months during the measurement year (2007). Plans included in this study were from all regions of the U.S.
Principal Findings: Among commercial PPO plans the inpatient O/E ranged from 0.35 to 2.01
(minimum to maximum), for procedure and surgery results ranged from .02 to 1.82, evaluation and management from .59 to 2.05, and for pharmacy from .37 to 2.2. Among HMO commercial plans, the inpatient O/E ranged from 0.11 to 1.78, for procedure and surgery results ranged from .00 to
2.21, evaluation and management from .49 to 3.45, and for pharmacy from .34 to 3.90. The total medical (sum of PMPM costs fro all service categories except pharmacy) mean risk adjusted among PPO plans was greater, $270 compared to
$244 among HMO plans.
Conclusion: While variation in PPO cost was narrower, the risk adjusted mean costs for their diabetic patients as on average greater when compared to diabetic HMO members. For both
HMO and PPO plans, substantial variation in costs for diabetics exists across plans.
Implications for Policy, Delivery or Practice:
Understanding and containment of health care costs is a rapidly emerging policy issue in the
United States. Examining and understanding differences in utilization or cost patterns with varying plan type along with relevant quality results is critical to developing practical and effective policies related to the delivery and use of health services.
Theme: Quality and Efficiency: Measurement
● Catalyzing Innovation at the Front Lines of
Health Care
Leah, Tuzzio, M.P.H.; Judith Schaefer, M.P.H.;
David McCulloch, M.D.; Sarah Greene, M.P.H.;
Susan Bennett; Brian Austin
Presented by: Leah Tuzzio, M.P.H., Project
Director, Group Health Center for Health Studies,
1730 Minor Avenue, Suite 1600, Seattle, WA
98101, Phone: (206) 287-2109; Email: tuzzio.l@ghc.org
Research Objective: To find new ways for Group
Health Cooperative, its research center and its community Foundation to work together, we pilot tested the Partnership for Innovation, a grant program to encourage and seed organization-wide innovation. Innovation is defined as new ideas or approaches that have the potential to add value and change the way the organization works. The program gives staff funding to develop, implement and evaluate innovations within the healthcare system. Evaluations assess the effect of the innovations in terms of costs, patient satisfaction, organizational readiness, and other aspects. These results inform decisions about adopting the innovation organization-wide.
Study Design: We disseminated an initial call for proposals to medical center leaders, and developed a two-step selection process in which we first reviewed brief concept proposals for their innovativeness, likely effectiveness, impact, feasibility, congruence with strategic priorities, and overlap with other efforts. We next invited selected applicants to expand their proposals for review by the program’s steering committee. The Partnership program itself was refined and improved through
Plan-Do-Check-Act cycles after each of the two funding rounds in 2008.
Population Studied: We invited medical center leaders across Washington State to submit proposals, including physicians, pharmacists, nurses, and clinical program managers.
Principal Findings: We received 19 proposals in
2008 (12 in the first round). Funded proposal topics included nasal flu vaccine for children, an on-line chronic disease self-management program, infectious disease management for long-term care facilities, and enhancing medication safety for patients with complex medication regimens. The first round of four pilots are currently underway and the second round of two funded pilots will begin in
2009. Viable but unfunded proposals were rerouted to other departments to encourage discussion about whether the idea can be developed through other channels.
Conclusion: Offering a pilot grant program to frontline staff may enhance engagement, empowerment, and creativity. Evaluating the pilot programs is an essential component of this initiative, both to improve the pilots themselves, and the innovation program as a whole. The partners found the collaboration promising and worth continued funding in 2009.
Implications for Policy, Delivery or Practice: The ability to mount and test innovative ideas safely and more rapidly than most research studies is attractive to delivery system leaders for whom the timeline for change and practice improvement does not align with the typically more protracted timeline in academic research. The model for funding clinical staff’s innovative programs to improve the quality of care hopefully will contribute to other funding programs and processes to translate patient-centered care initiatives into practice safely, rapidly, and effectively.
Funding Source(s): Group Health Cooperative
● Understanding Attitudes Regarding Healthy
Lifestyles: A Qualitative Study
Beverly Tyler, B.A.; Amanda Phillips Martinez,
M.P.H.
Presented by: Beverly Tyler, B.A., Senior
Research Associate, Georgia Health Policy Center,
14 Marietta Street, Atlanta, GA 30303, Phone:
(404) 413-0288; Email: btyler@gsu.edu
Research Objective: Understand attitudes regarding healthy living, specifically barriers to making heathy livestyle choices
Study Design: Qualitative research consisting of
17 focus groups with 165 Georgians representing 8 of 18 demographic clusters. The Demographic
Profiles of Georgia developed by the Georgia
Division of Public Health was used as the methodology for identifying and recruiting participants for the focus groups. The Demographic
Profiles methodology creates Georgia-specific demographic clusters from census data variables including: age, income, family structure, housing value and housing type, and education and employment type. These variables are analyzed using a classification model that creates four major groups, which are further partitioned into a total of
18 distinct demographic clusters. Once the clusters were identified, zip codes within Georgia that contained the largest percentage of residents within each individual cluster were used to determine the locations for the 17 focus groups. A screener for each cluster was developed for the purposes of recruiting focus group participants. Questions contained within the screeners assured that those being recruited for the focus group would meet the desired profile for the specific cluster. A professional recruitment firm utilized a call list obtained from Experian to randomly recruit participants.
Population Studied: One hundred sixty-five
Georgia residents from 8 of the 18 Georgia
Demographic Profiles of Georgia. Participants represented 60.6 percent of the Georgia population.
Principal Findings: The majority held a similar concept of what it means to lead a healthy lifestyle; they mentioned eating right, exercising, getting sufficient sleep, managing stress, getting annual physicals and avoiding drugs, tobacco and excessive alcohol. The barriers to living a healthier lifestyle include: lack of time due to demands from work and family, cost, the school system and lack of motivation.
Lack of time: Participants reported working long hours and multiple participants work more than one job to support themselves and their families. Their work load leaves them feeling stressed and exhausted at the end of the day and unable to carve out time to exercise on a regular basis.
Coupled with long work days, many parents in the focus groups described the time and energy required to keep up with their children’s busy activity schedules (e.g. sports games, classes, and other after school activities). These demands on families’ time often lead participants to make meal choices based more on convenience than nutrition.
Many described stopping at a fast food restaurant between activities to pick up dinner for the family.
Cost: For a large number of focus group participants, the cost of healthy foods such as fresh fruits and vegetable and lean meats and fish is prohibitive. In every focus group, participants observed that a meal from a fast food restaurant is significantly cheaper than buying vegetables from the grocery store to make a salad. Cost is also a factor in people’s access to physical activity opportunities. Multiple participants noted that most families cannot afford to join a gym. School system:
Parents often cited schools as barriers to their children’s healthy behaviors. Throughout the state, they complained about the quality and nutritional value of the food served in school cafeteria and vending machines. Much of the food served, they observed, is fried and high in fat and calories. In addition, schools are cutting physical education classes and recess. Parents stated that schools should be playing a more active role in teaching children healthy behaviors.
Lack of motivation: In many cases, participants acknowledged that they could overcome external barriers to living healthier lifestyles, but lacked the discipline and the motivation to do so. When asked what might motivate them to take steps towards leading a healthier lifestyle, participants most often responded that “more time and more money” would be the greatest motivators. Multiple participants observed that they are not likely to make lifestyle and health-related behavior changes until they are faced with a health issue. Across the state, participants reported that having a heart attack, developing diabetes or high blood pressure, or having someone close to them fall ill would most likely motivate them to change their health-related habits. Education on strategies for disease prevention and healthy living was also mentioned as key to motivating healthier behaviors.
Participants stressed the importance of teaching good habits to children early and expressed their hope that schools might take a more active role in instilling those good habits. Participants across the state said that decreases in insurance premiums and employer incentives for healthy living would be strong motivators to change their lifestyles.
Environmental factors like more access to safe recreational spaces in their communities and community-wide health information campaigns and challenges were also cited as potential motivators.
Conclusion: Focusing on personal responsibility will not motivate Georgians to live healthier lifestyles to avoid chronic illnesses and obesity.
Incentives for living healthier lifestyles must come externally from employers, insurance companies combined with environmental changes that encourage more active lifestyles.
Implications for Policy, Delivery or Practice:
This study identified a number of implications for changes in policy, delivery and practice. Employers may have the most significant opportunity to impact
health by alloting time for and encouraging physical activity, sponsoring health challenges, providing healthy food choices in vending machines and employee cafeterias and encouraging their insurance carriers to build in incentives for healthy living. Governments should sponsor social marketing campaigns focused on the consequences of poor life style choices. Schools must improve the quality of food served, provide more opportunities for physical activity and put a stronger emphasis on health education. In order to improve the health of their residents, local governments will have to allocate more resources to create environments which encourage physical activity and healthy eating.
Funding Source(s): Georgia Department of
Community Health
Theme: Prevention and Treatment of Chronic
Illness
● Prevalence of Nursing Assistant Training and
Certification Programs within Nursing Homes
(1997-2006)
Denise Tyler, Ph.D.; Hye-Young Jung, B.A.;
Zhanlian Feng, Ph.D.; Vince Mor, Ph.D.;
Presented by: Denise Tyler, Ph.D., Investigator,
Center for Gerontology & Healthcare Research,
Brown University, 121 South Main Street, Box G-
S121-6, Providence, RI 02912, Phone: (401) 863-
3894; Email: denise_tyler@brown.edu
Research Objective: Certified Nursing Assistants
(CNAs) provide as much as 80% of the care received by nursing home (NH) residents, so the quality of care residents receive is largely dependent on their skills. The federal government mandates that all CNAs receive at least 75 hours of training and be certified before working. This rule went into effect in 1991 and information about whether or not NHs were providing this training and certification in-house began to be collected as part of the annual survey process in 1997. This study sought to describe how the prevalence of these programs has changed over the past decade and to describe the differences between NHs that dropped the program and those that kept the program.
Study Design: We used descriptive statistics to examine three groups of NHs over the period 1997 to 2006: NHs that never provided a CNA training and certification program, those that provided the program in at least one period and dropped it, and those that kept the program for the entire study period. We compared these groups on a number of characteristics, including location, ownership type, profit status, patient payer mix, and CNA staffing level, all available from the Online Survey,
Certification and Reporting (OSCAR) data.
Population Studied: All free-standing
Medicare/Medicaid certified NHs with 20 or more beds in the 48 contiguous states for the years 1997 to 2006.
Principal Findings: Prevalence of NH-based CNA training and certification programs dropped throughout the study period with 37.6% of facilities providing the program in 1997 and 27.8% providing the program in 2006. NHs that kept the program throughout the study period were larger and had higher occupancy rates on average. A greater percentage of urban, for-profit, and chain-affiliated
NHs dropped the program at some point and NHs in urban areas accounted for the highest percentage of those that never offered a program.
NHs that dropped the program at some time during the study period were found to have the lowest average CNA staffing. Interestingly, CNA staffing increased among all three groups over the study period with the sharpest increase in staffing observed in 2001, which coincides with a rapid decrease in the prevalence of NH-based CNA training and certification programs.
Conclusion: While CNA staffing increased among all three groups of facilities, prevalence of CNA training and certification programs in NHs declined.
This suggests that many CNAs may now be receiving their pre-employment training in other venues, such as community colleges. Increased availability of these alternate training venues in urban areas may explain why facilities in urban areas were both more likely to have never provided the program and to have stopped providing it.
Implications for Policy, Delivery or Practice:
With more CNAs likely receiving their preemployment training outside the NH setting, NHs may need to provide new CNAs with more training and orientation when they begin employment.
Future research will be needed to test this hypothesis.
Funding Source(s): NIA
● Cover Michigan: A Report Card on the State of
Health Care Coverage in Michigan
Marianne Udow-Phillips, M.H.S.A.; Jonathan
Linder, M.H.S.A.; Tomi Ogundimu, M.P.H.
Presented by: Marianne Udow-Phillips, M.H.S.A.,
Director, University of Michigan, Center for Health
Care Research & Transformation, 2929 Plymouth
Road, Suite 245, Ann Arbor, MI 48105, Phone:
(734) 998-7555; Email: mudow@umich.edu
Research Objective: To establish a single source picture of health coverage in Michigan, including essential information on the uninsured, publicly insured, privately insured and health care safety net.
Study Design: Data was pulled from publicly available sites (principally SHADAC and CPS data) as well as information privately available from Blue
Cross and Blue Shield of Michigan. Trend data and
50 state comparisons were provided where available.
Population Studied: The entire population of the state of Michigan
Principal Findings: Even given the economic challenges of the state, Michigan continues to have a higher rate of private coverage than the country as a whole because more individuals in the state continue to be employed by large employers that provide better than US average health benefits.
Nevertheless, private employers have been dropping coverage in Michigan and the numbers of uninsured have increased considerably over 2006, going from 10.5% of the state’s population to
11.6%. Particularly concerning is the increase in uninsured children and the disproportionate numbers of African Americans without health coverage in the state. Even those with private health coverage have experienced significant increases in cost sharing over the past several years. From 2000 to 2005, average employment related premiums increased 68%. In essentially the same time frame, family deductibles increased
25.3% and single deductibles, 52.3%. The impact on the safety net in the state is notable, with a 68% increase in uncompensated care in hospitals from
2004 to 2007. Despite some increase in Federally
Qualified Health Centers in the state, almost half of the state’s Medically Underserved Areas lack an
FQHC.
Conclusion: While the picture of health coverage in the state of Michigan is generally better than in the country as a whole, the trends in Michigan are in the wrong direction and the recent downturn in the economy with its particularly devastating effects on the auto industry means that the picture in 2008 compared to 2007 will be even worse. Those who are African American, low income and with less than a bachelor’s degree are disproportionately likely to be uninsured. There is also a significant increase in the numbers who could be considered under insured. That is, having health insurance coverage does not assure access to health care.
Those with Medicaid coverage often have difficulty finding a provider who will accept their coverage. In addition, affordability of care is a major issue for those with private coverage but high cost sharing requirements. The safety net is clearly affected by the increasing the number of the under insured and uninsured in the state.
Implications for Policy, Delivery or Practice:
Coverage is often the focus of discussion on health reform. But coverage alone, while important, is not enough. Reform efforts need to focus on access to care and the stability of the health care delivery system as well as on the availability of health insurance. In addition, special attention must be paid to key population groups who face a disproportionate burden with regard to access to health care.
Theme: Coverage and Access
● Regression Discontinuity Evidence of the
Effect of CHIP Disenrollment on Asthma-Related
Hospitalizations of Children
Mark Unruh, M.S.; Hye-Young Jung, B.A.; Vincent
Mor, Ph.D.
Presented by: Mark Unruh, M.S., Graduate
Student, Community Health - Health Services
Research Program, Brown University, 121 South
Main Street, Box G-S121, Providence, RI 02912,
Phone: (831) 420-7039; Email: mark_unruh@brown.edu
Research Objective: The primary objective of this study is to estimate the increase of asthma-related hospitalizations of children 18 years-old and younger following the disenrollment of thousands from Texas Children’s Health Insurance Program
(CHIP). Prior research has shown an association between insurance status and hospitalizations of children with asthma. However, existing estimates of the magnitude of this association are primarily based on research designs which are susceptible to bias arising from omitted variables and unobservable influences. We exploit the disruption in the individual selection process caused by the policy change in order to derive estimates less vulnerable to biases of this nature.
Study Design: The Texas Public Use Data File was used to assess hospital admissions of individuals 18 years-old and younger. Monthly enrollments for Texas CHIP were obtained from the
Texas Health and Human Services Commission.
The policy change, which led to large numbers of children being dropped form Texas CHIP, disturbs the individual selection process. Since the timing of the policy change is known with certainty, and is outside the individuals’ control, an assignment rule based on whether asthma-related admissions were before or after the policy intervention was utilized.
This rule was incorporated into a reduced form regression discontinuity model based on a smooth function representing changes in the Texas CHIP enrollment profile, such as a low-order polynomial, with a binary outcome indicating whether the admission was asthma-related or not. Additional adjustments, including a quadratic for age, and indicators for Medicaid coverage, gender, race/ethnicity, and weekend admission were included. Estimates of the causal effect of the state’s policy change on asthma-related hospital admissions of children were obtained with probit estimation of the model parameters.
Population Studied: The study population included individuals 18 years-old and younger admitted to hospitals licensed by the State of Texas between
January 1, 2001 and December 31, 2003.
Admissions to hospitals with less than 100 licensed beds, or in counties with fewer than 35,000 residents were excluded from the data.
Principal Findings: Using several specifications for our models, we obtain estimates of the effect of the state’s policy change on emergency department admissions associated with asthma diagnoses ranging from approximately 9% to 11%. Likewise,
we estimate that the policy change is responsible for an increase of approximately 9.9% to 10.5% for all asthma-related hospital admissions for individuals 18 years-old and younger.
Conclusion: We conclude that the state’s policy change, and the associated disenrollment of children from Texas CHIP, is responsible for a significant increase in asthma-related hospitalizations. Additionally, we find that alternate hypotheses, including changes in Medicaid enrollment, private coverage, and potential influxes of individuals from outside the state, are unlikely to explain our results.
Implications for Policy, Delivery or Practice:
States’ decisions to reduce public health insurance enrollments potentially lead to declines in health status among individuals who lose coverage, consequently increasing utilization of inpatient services. The budgetary impact resulting from this tradeoff may be much less beneficial than anticipated.
Funding Source(s): NIA
● Iatrogenic Conditions Accompanying
Hospitalized Patients to Post-Acute Nursing
Home Care in the U.S.
Mark Unruh, M.S.; Vincent Mor, Ph.D.
Presented by: Mark Unruh, M.S., Graduate
Student, Community Health - Health Services
Research Program, Brown University, 121 South
Main Street, Box G-S121, Providence, RI 02912,
Phone: (831) 420-7039; Email: mark_unruh@brown.edu
Research Objective: Since the landmark Institute of Medicine study "To Err is Human" researchers have examined the determinants of many different types of errors compromising patient safety in the hospital from wrong sided surgeries and anesthesia errors to the myriad of medication errors made, both those resulting in an adverse drug reaction and those associated with “near misses.” While researchers have examined inappropriate medication use among the frail elderly residing in long term care facilities for the past several decades, it has only recently become possible to identify individuals entering nursing homes for the first time directly from hospitals by linking the US
Minimum Data Set Resident Assessment
Instrument (RAI) clinical and functional information with hospital discharge records. As part of a large national study of the changing role of US nursing homes, we estimated the rate of adverse drug events (ADE) that either caused hospitalizations or occurred during hospitalizations that ended in patients’ admission to a nursing home.
Study Design: We conducted a cross-sectional analysis to identify factors associated with new nursing home admissions following hospitalizations with at least one diagnosis of an ADE. In particular, we examined the prevalence of these events associated with the five most frequent diagnoses, illness severity, 30 day mortality, anti-psychotic use, number of drugs administered, age, and nursing home characteristics, using data linking the RAI with hospital discharge records.
Population Studied: Our study includes over
700,000 first-time admissions to nursing homes in the year 2000. These nursing home admissions followed a hospital discharge associated with an
ADE within 30 days.
Principal Findings: We find that 1.0 percent of new nursing home admissions following hospitalization had at least one discharge diagnosis associated with an ADE. The rate dropped slightly with age and was more common among patients with hip fracture as their primary reason for hospitalization, and less for CVA’s or heart failure.
Additionally, the rate for African-American patients was much lower than the national rate at 0.7 percent. Admissions to not-for-profit nursing homes had a rate of 0.9 percent, while for-profit facilities were above the national rate at 1.2 percent.
Consistent with much other research about health care quality, we observed considerable inter-state variation from over 2 percent in several states to less than 0.7 percent in others.
Conclusion: These results imply a link between hospitalizations associated with adverse drug events and nursing home admission.
Implications for Policy, Delivery or Practice: To the extent that iatrogenic adverse drug events in hospitals complicate patients’ recovery and future function, the scope and consequences of the problem of medical errors particularly affects this highly vulnerable population.
Funding Source(s): NIA
● Student Response Systems: Keeping
Students Interested
Maribel Valle, Ph.D., M.P.H.; Carolinda Douglass,
Ph.D.
Presented by: Maribel Valle, Ph.D., M.P.H.,
Assistant Professor of Public Health, School of
Nursing & Health Studies, Northern Illinois
University, 250 Wirtz Hall, Dekalb, IL 60112,
Phone: (815) 757-0889; Email: mvalle@niu.edu
Research Objective: Large interdisciplinary courses that draw students with varying knowledge and interest are a special challenge to those preparing the healthcare workforce of the future.
The competition for the attention of instructors, the lack of student interaction, and the distancing which occurs in classes of over 200 students combine to make a complex and sometimes difficult teaching and learning experience. New technologies, such as electronic student response systems (clickers) have emerged as a way to keep students engaged and provide instructors with quick assessment information. This study addresses the question "Do
student response systems in large health related courses improve learning outcomes?"
Study Design: This quasi-experimental nonequivalent group design compared exam scores of two sections of an interdisciplinary course on the
American health care system with and without the use of a student response system.
Population Studied: 388 college students at a large midwestern regional university participated in this study, Subjects were divided into treatment and control groups. There were no statistically significant differences between groups on gender, race/ethnicity, G.P.A., major, or academic years.
Principal Findings: There was at least a four point increase in each of three exam scores, 2 of the increases were significant at the .01 level and the third at the .05 level. The one exam which was significant only at the .05 level occurred after the campus experienced a very traumatic event.
Conclusion: Given its limitations, this study provides evidence that clickers can be a useful tool in the classroom.
Implications for Policy, Delivery or Practice: As educators deliver course content in new and innovative ways, clickers could be an effective and inexpensive way to improve learning outcomes.
Funding Source(s): NIU, Committee to Improve
Undergraduate Education
● Evaluating Perceptions of Quality: A
Comparison of CEO & Board Chair Responses
Kevin Van Dyke, M.P.P.; Frances Margolin, M.A.
Presented by: Kevin Van Dyke, M.P.P., Research
Manager, Health Research & Educational Trust,
American Hospital Association, One North Franklin,
Suite 3000, Chicago, IL 60201, Phone: (312) 422-
2641; Email: kvandyke@aha.org
Research Objective: There have been few empirical studies that have compared and contrasted CEOs and Board Chairs’ perceptions of the importance of quality of care in hospital governance. The purpose of this study is to identify where such differences, if any, exist.
Study Design: Data from the 2005 American
Hospital Association governance survey will be used to match responses from CEOs and their
Board Chairs to identify differences in perception of the importance of quality of care in board recruitment; selection, monitoring, and evaluation of the organization and CEO performance; and selfassessment of the board.
Population Studied: The 2005 AHA governance survey includes data from 1,587 CEOs and 906
Board Chairs. The governance survey data was merged with descriptive information on each hospital from the 2005 American Hospital
Association Annual Survey of Hospitals. Data was then weighted to reflect the national hospital population on five variables: ownership, bed size,
Census region, metro/non-metro location, and whether or not the hospital is part of a multihospital system.
Principal Findings: Initial non-matched results showed a difference between CEO and Board Chair perceptions of quality in several areas. For example, Board Chairs were more likely to indicate that quality of care and outcome management received at least considerable weight when evaluating the CEO (86%) than were CEOs (76%).
Also, Board Chairs were more likely to indicate that safety and quality received at least heavy emphasis when selecting new board members (19% versus
11%). However, we now are looking at whether these differences will hold under a more systematic design using matching techniques. We anticipate that our findings will be available by early March,
2009.
Conclusion: There are several areas where CEOs and Board Chairs have different perceptions regarding the importance of quality in board processes. We speculate that these findings will hold up under a technique that matches CEO and
Board Chair responses.
Implications for Policy, Delivery or Practice:
Differences in perceptions of quality between CEOs and Board Chairs indicate the need for greater investment in time and resources for joint educational activities that promote greater alignment between CEOs and their Board Chairs in the area of quality of care.
Funding Source(s): American Hospital Association
● The Economic Burden of Providing Informal
Care to Lung & Colorectal Cancer Patients
Courtney Harold Van Houtven, Ph.D.; Scott
Ramsey, M.D., Ph.D.; Mark Hornbrook, Ph.D.;
Audie Autienza, Ph.D.; Michelle van Ryn, Ph.D.
Presented by: Courtney Harold Van Houtven,
Ph.D., Assistant Professor, VA Health Services
Research & Development & General Internal
Medicine, Durham VA & Duke University, 508
Fulton Street, Durham, NC 27514, Email: courtney.vanhoutven@duke.edu
Research Objective: The provision of informal care entails economic costs to caregivers. The paper aims to quantify the direct and indirect economic burden of providing informal care to lung and colorectal cancer patients.
Study Design: Cross-sectional survey of caregivers of lung and colorectal cancer patients participating in the Share Thoughts on Care survey conducted by the Cancer Care Outcomes Research and Surveillance (CanCORS) consortium. Direct and indirect economic burden of caregiving were assessed and calculated using multivariate statistical methods.The economic burden of caregiving is calculated for caregiver sub-samples using the estimated value of caregiver time and work and leisure hours lost, out-of-pocket expenditures for cancer-related care and related
non-medical expenditures, and expenditures for household assistance. Multivariate models of economic burden controlled for key characteristics of the caregiver and care recipient: cancer type
(lung or colorectal), stage of disease, phase of disease, recent treatment, comorbid conditions, demographics, and economic resources.
Population Studied: Caregiver-reported data from a 2005 Share Thoughts on Care mailed survey.
Caregivers were nominated by a consecutive subsample of recently diagnosed lung and colorectal cancer patients in the Cancer Care Outcomes
Research and Surveillance (CanCORS) consortium study. 1,629 primary caregivers of lung and colorectal cancer patients completed the survey.
662 caregivers were surveyed at the patient’s initial phase of disease, 822 in continuing phase, and 144 in terminal phase.
Principal Findings: The economic burden for cancer caregivers was estimated to be $8,595 for caregivers of initial phase patients, $24,209 for caregivers of continuing phase patients, and
$16,927 for caregivers of terminal patients.
Economic burden was higher for caregivers of lung cancer patients and for caregivers of patients diagnosed at stage 4. There was an additional increase in economic burden for caregivers of lung cancer patients and a diagnosis at stage 4.
Spouses faced significantly higher economic burden than other relatives or friends who provided care. Age of the patient had no influence on economic burden of the caregiver.
Conclusion: The total economic burden of informal care is equivalent to between 3 to 8 months of an average worker’s annual gross salary. Economic burden is high even if one values a caregiver’s time at its replacement cost—that of a home health aide wage or a minimum wage worker’s wage—rather than the opportunity cost of a caregiver’s time—his or her own wage rate.
Implications for Policy, Delivery or Practice: The economic burden of informal care should be included in estimates of the cost of cancer care.
Omitting these significant costs will prevent an accurate assessment of the net benefits and costs of interventions, cancer treatments, and policies for lung and colorectal cancer patients in the United
States.
Funding Source(s): NCI
Theme: Long Term Care
● An Inventory of Measurement of
Implementation Success
Carol VanDeusen Lukas, Ed.D.; Jason Silver,
M.P.H.; Mariana Gerena-Melia, Ph.D.
Presented by: Carol VanDeusen Lukas, Ed.D.,
Senior Investigator, Center for Organization,
Leadership & Management Research, Veterans
Health Administration, 150 South Huntington
Avenue, Boston, MA 02130, Phone: (857) 364-
5685; Email: carol.vandeusenlukas@va.gov
Research Objective: We conducted an inventory of measures of implementation success used in implementation studies funded by HSR&D in the
Department of Veterans Affairs (VA). Our aim was to create a typology of measures in order to advance the science of implementation -- that is, the study of moving tested innovations into routine use in actual service delivery contexts.
Implementation science is gaining prominence as we seek to identify successful strategies for integrating promising research evidence into clinical practice. One aspect that has not received consistent attention is the measurement of implementation success. By implementation success we mean the consistent use of an innovation in routine practice in an organization.
Yet this measurement is essential in identifying effective implementation strategies and contributes to understanding the success or failure of the innovation outcomes.
Study Design: For the inventory, we began with the 200 studies in two VA HSR&D portfolios most likely to contain implementation studies. Project abstracts were screened to identify implementation studies. 59 implementation studies were identified and the PIs were contacted for project documents describing the project protocol and implementation measures used. 45 study teams responded with the requested materials. After reviewing the documents, we requested telephone interviews with
PIs or designees to pursue details. 33 teams responded and were interviewed. Detailed notes from document reviews and interviews were entered into spreadsheet matrices and reviewed qualitatively. The matrices were organized by conceptual and practical considerations, including:
1) description and categorization of the innovation,
2) the scope of the study; 3) levels of organization at which implementation was targeted; 4) standards of implementation success; 5) data sources; 6) validation of measures; 7) implementation model used.
Population Studied: The inventory included implementation studies in two VA HSR&D portfolios: Implementation Service Directed Projects and Implementation and Management Research.
Principal Findings: Of the 33 studies in the final inventory, 28 (85%) formally measured implementation success. Of those 28, 50% used fidelity as the standard of implementation success and 18% used scope; 18% used multiple standards.
Data sources for the measures included: electronic records, surveys, interviews, focus groups, observations, research records. Most measures were newly created. Most studies did not measure the implementation strategy used to introduce the innovation to the organization. Measurement standards and data sources did not vary systematically by type or scope of intervention.
Conclusion: A substantial majority of studies in the inventory do measure implementation success
separate from outcome performance. However, the measures are not consistently well developed and there are no established standards across studies.
Implications for Policy, Delivery or Practice: The inventory results have immediate benefits in drawing attention to the importance of implementation success measurement and offers examples of measurement tools. In the longer term, the findings contribute to implementation science by laying the groundwork for further work in this area. Further work is needed to develop robust measures, especially those that can take into account the multiple perspectives of innovation participants at different levels of the organization and that produce valid information from less than perfect data sources.
Funding Source(s): VA
Theme: Organizational Performance and
Management
● AHRQ Patient Safety Indicator-Accidental
Puncture & Laceration: A Multi-Disciplinary
Review
Monica VanSuch, M.B.A.; James Naessens, Sc.D.;
Matthew Johnson, M.P.H.
Presented by: Monica VanSuch, M.B.A., Health
Services Analyst, Health Care Policy & Research,
Mayo Clinic, 200 First Street, Southwest,
Rochester, MN 55905, Phone: (507) 284-1166;
Email: vansuch.monica@mayo.edu
Research Objective: To conduct a validation study to determine if surgical complications identified by the AHRQ Patient Safety Indicator for Accidental
Puncture and Laceration are true performance issues.
Study Design: A multi-disciplinary project team reviewed a random sample of 50 records flagged by the AHRQ accidental puncture and laceration algorithm. The sample consisted of 30 general surgery, 10 chest surgery and 10 head and neck surgery cases. The medical records identified in the random sample were reviewed by a subgroup of the project team: two surgeons, a registered nurse and a health services analyst with an extensive coding background. To help reduce bias, the cases were independently reviewed to determine agreement with the initial assignment of the ICD-9-
CM code. The surgeons were also asked to determine the degree of preventability for the cases that were identified as being coded correctly.
Population Studied: All medical and surgical patients 18 years of age and older defined by specific DRGs. Excluded are cases with the ICD-9-
CM code 998.2 (accidental cut, puncture, perforation or laceration) denoting technical difficulty identified as present on admission as either the principal or secondary diagnosis. Also excluded are pregnancy and childbirth cases (MDC
14) and cases with an ICD-9-CM code for spine surgery. When compared to the accidental puncture and laceration benchmark data provided by
University HealthSystem Consortium (UHC), Mayo
Clinic Rochester’s performance of 15 occurrences per 1000 cases was substantially higher than the expected rate of 5 occurrences per 1000 cases; it was also higher than the UHC median of 5.3 per
1000 cases.
Principal Findings: The case reviewers disagreed with the initial code assignment of accidental puncture or laceration 78% of the time (39 of 50 cases). In most of the disagreements, the puncture or laceration was considered to be incidental to the disease, progression of the disease itself or an expected outcome of a complex surgery (e.g. serosal tear, dural tear or adhesions from prior surgery). Of the 11 cases in which the reviewers agreed with the initial code assignment, only 3 cases were identified as being clearly preventable.
The measure of agreement between the reviewers was assessed with the kappa statistic
Conclusion: The definition of an accidental puncture and laceration varied by reviewer
The assignment of the code for this condition has a substantial degree of subjectivity based on the medical knowledge of the reviewer. The PSI does not take into account the complexity of the surgery.
Implications for Policy, Delivery or Practice:
Administrative data can be an effective screening tool to identify potential patient safety and quality issues, but the sensitivity of the screening tool is influenced by many factors. Current risk adjustment approaches are not designed to account for varying levels of medical or surgical complexity.
Theme: Quality and Efficiency: Measurement
● Coordinating & Integrating Behavioral Health with Child Welfare in Florida: Process & Impact
Amy Vargo, M.A.; Patty Sharrock, M.S.W.; Patricia
Robinson, Ph.D.; Roger Boothroyd, Ph.D.
Presented by: Amy Vargo, M.A., Faculty, Child &
Family Studies, Florida Mental Health Institute,
13301 Bruce B. Downs Boulevard, MHC 2413,
Tampa, FL 33612, Phone: (813) 974-5356; Email: avargo@fmhi.usf.edu
Research Objective: Given the high rate of trauma and corresponding mental health needs among foster children, access to behavioral health services is critical. Challenges to meeting the needs of children and families involved with the child welfare system have been associated with the lack of formal policies to provide behavioral health services, the lack of timely assessment, and failure to communicate with multiple providers and caretakers. Although managed care systems have been devised to improve health care services by emphasizing primary and preventative care and system coordination, conflicts among the various child welfare and behavioral health managed care stakeholders can interfere with mental health service delivery.
Study Design: The Florida Mental Health Institute at the University of South Florida is funded by the
Agency for Health Care Administration to conduct an independent evaluation of the CW-PMHP. The overall purpose of the CW-PMHP Evaluation is to determine the extent to which the mental health service needs of children in Florida's child welfare system are being met by the CW-PMHP. In addition to examining the experiences of consumers and providers, it is important to understand the contextual and organizational features of the Child
Welfare Prepaid Mental Health Plan (CW-PMHP) through an investigation of the features of its implementation. Therefore, the evaluation consists of two components: 1) an implementation/process analysis examining the financial, structural, and clinical aspects of the carve out program; and 2) a quality of care/impact analysis utilizing semistructured interviews and focus groups with parents and service providers to examine quality of care received, as well as a mail survey sent to foster parents to obtain their perspectives regarding the mental health needs of foster children and to document their experiences in accessing mental health services.
Population Studied: Historically, children enrolled in Florida’s child welfare system have been exempt from Medicaid managed care. In 2004, Florida passed legislation requiring children in child welfare to receive behavioral health services under a managed care arrangement. In 2006, The Florida
Coalition for Children (FCC) formed a Behavioral
Health Network Board with Magellan Behavioral
Health of Florida, Inc. and was awarded the contract. Implemented February 1, 2007 and under a contract with the Agency for Health Care
Administration, the Child Welfare Prepaid Mental
Health Plan (CW-PMHP) is intended to provide a comprehensive set of behavioral health services to children who are eligible for Medicaid and involved with Florida’s child welfare system. This population previously received such services via a traditional fee-for-service arrangement.
Principal Findings: This presentation will offer detailed findings regarding process and impact collected across three years of study in geographically diverse areas of Florida. Detailed descriptive information on implementation of this unique managed care arrangement and partnership between behavioral health and child welfare service systems will be shared in combination with caregiver and provider satisfaction data.
Implications for Policy, Delivery or Practice: The findings from this study have important policy and practice ramifications for both the behavioral health and child welfare reforms taking place nationally, since entry into the child welfare system provides an opportunity for secondary prevention with regard to child mental health problems.
Funding Source(s): Florida's Agency for Health
Care Administration
Theme: Behavioral Health
● Patient Safety Outcomes in Small Urban &
Rural Hospitals
Smruti Vartak, M.P.H.; Marcia Ward, Ph.D.;
ThomasVaughn, Ph.D.
Presented by: Smruti Vartak, M.P.H., Research
Assistant, Health Management & Policy, University of Iowa, 5229 Westlawn, Iowa City, IA 50613,
Phone: (319) 384-5131; Email: smrutivartak@uiowa.edu
Research Objective: The purpose of this paper is to assess patient safety outcomes in small urban and rural hospitals and to examine the relation of hospital and patient factors to patient safety outcomes.
Study Design: Patient safety indicators (PSI) developed by the Agency for Healthcare Research and Quality (AHRQ) were used to examine nine common patient safety outcomes at small urban and rural hospitals. The unit of analysis was the patient. Associations between hospital location and patient and hospital characteristics were determined using one-way ANOVA and Pearson chi-square test. Multivariable analysis using generalized estimating equation regression models assessed the relationship between hospital location and PSIs.
Population Studied: The Nationwide Inpatient
Sample and American Hospital Association annual survey data for year 2005 were used for analyses.
The study sample was restricted to hospitals with less than 100 beds.
Principal Findings: The results showed that most of the observed rates for the 9 PSIs were higher for urban hospitals than for rural hospitals. In the multivariable analyses, after adjusting for important patient and hospital characteristics, many of these differences disappeared except for decubitus ulcer.
Urban hospitals had significantly higher odds for decubitus ulcer than rural hospitals.
Conclusion: The results from this study deviate from the findings in the literature that urban-rural differences in patient safety rates exist.
Implications for Policy, Delivery or Practice:
This study highlights the importance of understanding the factors that differ between urban and rural hospitals while developing future standards of care recommendations, identifying fair and useful hospital performance metrics for public reporting and creating hospital-relevant patient safety interventions at these hospitals.
Funding Source(s): AHRQ
● The CORE Buprenorphine Project (BUP)
Pamela Vergara-Rodriguez, M.D.; Jeffrey Watts,
M.D.; Virgilio Arenas, M.D.; Melody Bather Gardner,
M.A.; Mary Jo Tozzi, M.S.
Presented by: Pamela Vergara-Rodriguez, M.D.,
Consultant Psychiatrist, Psychiatry, The Ruth M
Rothstein CORE Center, 2020 West Harrison,
Chicago, IL 60612, Phone: (312) 572-4753; Email:
Pvergara-rodriguez@corecenter.org
Research Objective: The Drug Addiction
Treatment Act of 2000 and the approval of buprenorphine for office-based treatment of opioid addiction offered an opportunity to develop an innovative model of care for both HIV-infected and opioid-dependent patients.
Study Design: The BUP opiate addiction clinic requires one full-time clinical coordinator, a parttime psychiatrist and part-time administrative assistant. The program is highly dependent on the flexibility of the clinical coordinator who plays multiple roles in direct patient care, coordination with other providers, and linking with outside community substance abuse programs. Direct patient care includes assessment of substance abuse treatment needs, community referrals for other levels of substance abuse care, providing individual counseling, facilitating substance abuse groups, and providing comprehensive case management. Coordination with other providers includes referrals for psychiatric and medical evaluation and follow-up, as well as facilitating prescriptions for both buprenorphine and other psychotropic medications. Finally the clinical coordinator established partnerships with community substance abuse programs including residential treatment, detoxification and methadone.
The role of the part-time psychiatrist is to evaluate and provide patients with Suboxone for opioid dependence. The psychiatrist also oversees the substance abuse treatment care plan which includes behavioral treatment (outpatient groups), substance abuse counseling, and medication management. Additionally, during the follow-up period, the part-time psychiatrist evaluated patients for co-occurring psychiatric disorders and offered follow-up treatment as needed. The part-time psychiatrist and clinical coordinator work together two days per week for a two hour period with occasional back-up by other psychiatric staff. All psychiatric staff were certified to prescribe
Suboxone.
The BUP project coordinator performed administrative duties approximately 8 hrs/week.
These duties included office management, accounting requirements and payroll/human resources liaison.
The staff required one office and use of counseling space with an adjacent waiting area for the two morning clinics. Fortunately, the CORE Center has a pharmacy onsite which supplied Suboxone for
BUP-stamped prescriptions.
Population Studied: From 7/27/04 through
11/30/2007, BUP evaluated 220 unique patients for treatment. One hundred and one (101) were enrolled. The study population was 87 % African-
American, 4% Hispanic, 58% male, 71% greater than 40 years old and 58% with less than a high school education. At baseline, they were 78% unemployed, 31% self-described as homeless, 30% on probation and 6% awaiting charges, trial or sentencing.
Principal Findings: The BUP model offered HIV positive patients receiving HIV primary and specialty care at the Ruth M. Rothstein CORE
Center (CORE) a comprehensive treatment plan for opioid dependence including substance abuse treatment evaluation, substance abuse case management, substance abuse outpatient treatment groups, psychiatric medication management and opioid substitution treatment as desired. Thru Ryan White CARE funding, the
CORE offers HIV primary care and HIV specialty services including nephrology, neurology, psychiatry/individual therapy, hepatitis, hematology/oncology, and dental clinics.
Conclusion: The BUP project presents an innovative model to integrate comprehensive substance abuse treatment for opioid dependent patients in the setting of HIV medical care. BUP showed a higher retention rate during the project period in comparison to other substance abuse programs administered at CORE. The
Implications for Policy, Delivery or Practice: A model integrating behavioral and medical healthcare is necessary to effectively treat this population.
Funding Source(s): HRSA
Theme: Behavioral Health
● Practical Applications of Indirect Estimation of Race/Ethnicity Data in Health Plan Activities
Zachary Vernon; Grace Ting, M.H.A.
Presented by: Zachary Vernon, Sr. Health
Information Consultant, Programs in Clinical
Excellence, Wellpoint, Inc, 602 South Jefferson
Street, Roanoke, VA 24011, Phone: (540) 853-
3316; Email: zack.vernon@wellpoint.com
Research Objective: To overcome a common health plan challenge of the lack of self-reported race /ethnicity data and meet a number of health plan quality improvement, operations and business planning needs, WellPoint, Inc. has developed a practical methodology to derive indirect race/ethnicity information for members.
Study Design: Using an internal dataset that contains race/ethnicity data on approximately 1.3 million members; members’ addresses were geocoded and census data was appended.
Surnames and first names were matched to Census surname databases and an internal African
American first name database and associated data was appended. Prior to running the regressions,
192,096 member records were extracted from the model development database and set aside to be used for validation testing. Using the remaining
data, logistic regressions were run using the reported race / ethnicity (African American, Asian,
Hispanic, and White / Other) as the dependent variable and the Census and name data as independent variables. The results of these models were used to score the validation data set and the predicted results were compared to the reported race / ethnicity within the validation data set.
Population Studied: Logistic regressions were developed utilizing a dataset from WellPoint’s
California and Connecticut State Sponsored business.
Principal Findings: The scored data can be examined both in aggregate and as a predictor of individual race / ethnicity. By normalizing the predicted probabilities of individual members, data can be summed to provide estimated demographic detail of the aggregate population. In aggregate, the resulting model predicted within 0.1% of the actual aggregate demographics for the validation sample. When looking at individual members and assuming that the race / ethnicity with the highest probability indicated the members’ actual race
/ethnicity, the model predicted the correct race/ethnicity for 86.2% of all members.
Sensitivities ranged from 72.0% for Asians to 93.6% for Hispanics. Specificities ranged from 91.4% for
White to 99.3% for Asians. Positive predictive values ranged from 76.4% for White to 93.2% for
Hispanics. An alternative method that has been useful in interventions is to only deem an individual’s race / ethnicity as predicted when the probability of being a given race / ethnicity is greater than 80%. Using this alternative method,
93.8% of the member’s race / ethnicities were correctly identified with only 25% of the populations being classified as unknown and positive predictive values all fall above 90%.
Conclusion: Although self-reported data is the
“gold standard,” the accuracy of the logistic regression model can be used as an effective proxy for estimated race /ethnicity of members in the absence of self reported data.
Implications for Policy, Delivery or Practice: The successful indirect estimation of race / ethnicity provides a new level of detail in business decision support process for planning quality initiatives and may guide health plans in focusing limited resources for maximum impact. Using GIS tools, these predictions allow the targeting of interventions to local “hotspots.” In addition, the identification of an individual’s race / ethnicity allows for targeted culturally sensitive interventions based on the member’s predicted race / ethnicity.
Funding Source(s): Internal Funding from
Wellpoint
Theme: Disparities
● Concepts of Stress & Depression: The South
Asian Immigrant Experience
Amar Vira, B.A.; Manasi Tirodkar, Ph.D.; Namratha
Kandula, M.D., M.P.H.
Presented by: Amar Vira, B.A., Medical Student,
Division of General Internal Medicine, Feinberg
School of Medicine, 512 North McClurg Court, Unit
2408, Chicago, IL 60611, Phone: (708) 228-4933;
Email: a-vira@md.northwestern.edu
Research Objective: Mental health has not been addressed appropriately in South Asian immigrant communities and poses a great public health problem. In a recent survey of health and disease among South Asian immigrants in Chicago, stress and depression emerged as central themes in respondents’ lives and health. South Asians contextualize mental health in a social rather than in a biomedical framework. Instead of chemical imbalances being the underlying cause of stress and depression, South Asian immigrants describe mental health issues as a result of circumstances in their lives. This study aims to describe the experience of mental health in recent South Asian immigrants including the etiology and management of stress and depression.
Study Design: We conducted qualitative, semistructured interviews in English, Hindi and Urdu with 75 respondents from a federally qualified health center and at a community center for South
Asian immigrants in Chicago, Illinois.
Population Studied: Respondents’ age ranged from 20-70; 60 percent were women; 60 percent held advanced degrees; 70 percent were recent immigrants and 60 percent of the interviews were in
Hindi or Urdu.
Principal Findings: The sources of stress varied significantly by gender. While both men and women described their family as a source of stress equally,
36 percent of men versus 12 percent of women described finances as their primary source of stress and 43 percent of women and only 27 percent of men described worrying about their children as their primary source. Other sources of stress included issues of acculturation such as changing cultural values and social isolation, and lack of job security.
Religion and gender played a significant role in management practices of stress. 29 percent of
Muslims resorted to faith or prayer versus 44 percent of Hindus, 12 percent of all respondents managed their stress by thinking positively, and 29 percent of women and only versus 12 percent of men employed talking to friends and family as a mechanism to reduce stress. When asked about the effects of stress on physical health, the most significant physical manifestations of poor mental health were thought to be sleeping problems, heart attack, and increased blood pressure.
Conclusion: Interviews about health and heart disease in recent South Asian immigrants revealed that they are living with tremendous stress and depression. It is unclear if this population has the appropriate resources to cope with these issues.
Further studies must be conducted to capture the
mental health needs of the South Asian immigrant community.
Implications for Policy, Delivery or Practice:
Physicians should learn to recognize how South
Asian immigrants present with mental health as they describe their issues within a social frame work. Recognizing these signs will allow practitioners to address these important mental health issues in their patients’ lives.
Funding Source(s): National Health, Lung, and
Blood Institute
● Hospital Characteristics Predictive of Surgery for Non-Small-Cell Lung Cancer
Katherine Virgo, Ph.D., M.B.A.; Stacey Fedewa,
M.P.H.; Amy Chen, M.D., M.P.H.; Andrew Stewart,
M.A.; W. Dana Flanders, M.D.; Elizabeth Ward,
Ph.D.
Presented by: Katherine Virgo, Ph.D., M.B.A.,
Director, Health Services Research, & Professor,
Epidemiology & Surveillance Research, American
Cancer Society & Emory University, 250 Williams
Street, Northwest, Atlanta, GA 30309, Phone: (404)
417-8069; Email: katherine.virgo@cancer.org
Research Objective: Previous research suggests that black patients are less likely to undergo curative-intent surgery for early stage non-small cell lung cancer (NSCLC) compared to whites, holding all else constant. Among Medicare beneficiaries 65 and older, the likelihood of patients undergoing surgery is significantly reduced in hospitals with patient populations composed of > 30% black patients after controlling for hospital and patient characteristics. This study analyzes whether the racial composition of a hospital's patient population is associated with curative-intent surgery among early stage NSCLC patients of all ages.
Study Design: This is a retrospective analysis of secondary data from the National Cancer Data
Base (NCDB), a hospital-based cancer registry jointly sponsored by the American Cancer Society and the American College of Surgeons (ACoS).
Facility characteristics were extracted from the
NCDB and ACoS Commission on Cancer (CoC)
Facility Information Profile System (FIPS). At the facility level, the racial composition of each hospital's patient population, operationalized as the percentage of all cancer patients who were black, was initially divided into tertiles; sensitivity analyses used dichotomous definitions (10, 20, or 30% black). The characteristics of the facility at which each patient was treated were then applied to the patient-level early stage NSCLC data. Multivariate logistic generalized estimating equations & generalized linear mixed models were used to control for clustering by facility.
Population Studied: Data for all adult patients ages 19-104 diagnosed with an invasive initial primary early stage (TNM I-II) NSCLC with nonmissing treatment and demographic data during
2003-2005 were selected from the NCDB.
Principal Findings: Of 52,853 evaluable patients seen at CoC-approved hospitals, blacks were primarily (76%) seen at hospitals treating at least
10% black patients. Hispanics (69%), non-Hispanic whites (68%), and Asians or other (68%) were primarily seen at hospitals treating < 10% black patients. In additional univariate analyses, black patients seen at hospitals treating a high percentage of black patients were significantly less likely to undergo curative-intent surgery than black patients seen at hospitals treating a lower percentage of black patients, regardless of the cutpoint used. However, after controlling for patient characteristics and select hospital characteristics such as facility type and patient volume in multivariate models, being seen at a facility that treats a high percentage of black patients was no longer significant.
Conclusion: At CoC-approved facilities, being seen at a facility that treats a high percentage of black patients was not predictive of reduced likelihood of curative-intent surgery among patients
> 18 years of age with early stage NSCLC. Patientlevel predictors accounted for the majority of the variation in likelihood of curative-intent surgery.
Patient volume and facility type were significant facility-level predictors. An analysis of American
Hospital Association annual survey data is underway to determine if any additional hospital characteristics should be included in the models.
Implications for Policy, Delivery or Practice:
Controlling for patient and other hospital characteristics, including the volume of patients seen and whether the hospital is a research facility, the racial composition of a CoC-approved hospital's patient population does not impact receipt of curative-intent surgery for early stage NSCLC.
Referring patients to high volume, research facilities, irrespective of the racial composition of a hospital's patient population, increases the likelihood of patients undergoing curative-intent surgery for early stage NSCLC at CoC-approved facilities.
Funding Source(s): Intramural American Cancer
Society funds
Theme: Disparities
● Spouse and Circumstance: The Health
Insurance Status of Low-Wage Workers
Jessica Vistnes, Ph.D.; Alan Monheit, Ph.D.
Presented by: Jessica Vistnes, Ph.D., Senior
Economist, Committee For a Constructive
Tomorrow, Agency for Healthcare Research &
Quality, 540 Gaither Road, Rockville, MD 20850,
Phone: (301) 427-1671; Email:
Jessica.Vistnes@ahrq.hhs.gov
Research Objective: Numerous studies have documented the low rate of employment-based
coverage and high uninsured rate among low-wage working Americans and their families. In order to devise targeted strategies to increase coverage among these workers, policymakers need to recognize the heterogeneity in their circumstances.
For example, many low-wage workers may work for firms with a considerable proportion of high-wage workers, which increases the likelihood that they are offered coverage and face more affordable outof-pocket premium costs. Others who work in entirely low-wage firms are unlikely to be offered coverage. Additionally, the household circumstances of low wage workers are important.
Having a higher earning spouse may provide a lowwage worker with access to coverage as well as the increased income to afford it. This paper will characterize the employment and household circumstances of low-wage workers in order to identify those who are most vulnerable and likely to be candidates for targeted policy interventions.
Study Design: We use nationally representative employer data from the 2006 Medical Expenditure
Panel Survey- Insurance Component (MEPS-IC) to examine offers of coverage as well as cost-sharing requirements for premiums and medical care use.
We examine differences in these measures by the wage and firm size distributions of employers, constructing detailed wage distributions from information on the percentage of the employer’s workers in specific wage classes. Using these data, we provide estimates of how low-wage workers are distributed among employer types.
Recognizing that employer-level data provides an incomplete picture of low-wage employees’ access to coverage, we supplement this information with household-level data on insurance coverage from the MEPS-Household Component (MEPS-HC) and describe coverage status by the household circumstances of low-wage workers, focusing on the spouse’s wage level and access to offers of coverage.
Population Studied: Nationally representative samples of private sector employers in the 2006
MEPS-IC and workers in the 2006 MEPS-HC.
Principal Findings: Preliminary results reveal that small (< 10 workers) and largely low-wage employers were less likely to offer coverage in 2006
(18.6% compared to 55.8% of all employers).
Household data indicate that low-wage workers married to high-wage workers were much less likely to be uninsured than those married to low-wage workers (8.36% vs. 41.44%). However, low-wage workers were more likely to be married to another low-wage worker than to a high-wage earner (24% vs. 9.2%).
Conclusion: The wages of other employees at a low-wage worker’s firm and the wages of a worker’s spouse are important correlates of whether a lowwage worker obtains employment-based coverage or remains uninsured.
Implications for Policy, Delivery or Practice: The percentage of small-firm/low-wage employers offering coverage has declined significantly in recent years. Since these employers are likely to be exempt from any proposed employer mandate, it is important to identify the proportion of low-wage workers they employ who are increasingly in need of an alternative way to obtain access to affordable coverage. For low-wage workers who are offered coverage, it is also important to identify the high out-of-pocket premiums and large deductibles they face to determine appropriate subsidies or whether they should be allowed to enroll in public coverage.
Funding Source(s): AHRQ
Theme: Coverage and Access
● The Design, Development & Evaluation of a
Comprehensive Pediatric Asthma Project: The
Merck Childhood Asthma Network (MCAN)
Meera Viswanathan, Ph.D.; Lucia Rojas Smith,
Dr.P.H.; Carol Woodell, B.S.P.H.; Linda Lux,
M.P.A.; Yvonne Ohadika, Ph.D.; Floyd Malveaux,
M.D., Ph.D.
Presented by: Meera Viswanathan, Ph.D., Health
Services Research Analyst, RTI International, 3040
Cornwallis Road, Research Triangle Park, NC
27709, Phone: (919) 316-3930; Email: viswanathan@rti.org
Research Objective: Pediatric asthma is a multifactorial disease, requiring multi-layered interventions addressing children, families, schools, and communities. Asthma continues to result in a high burden of morbidity despite the existence of interventions proven to improve health and quality of life. The Merck Childhood Asthma Network
(MCAN) is a major new funding initiative that seeks to translate these proven interventions from research to practice. This paper highlights lessons learned in the selection, implementation, and evaluation of the effort with regard to integrating and translating evidence-based interventions across varied settings.
Study Design: The MCAN request for proposals, issued in May 2005
(http://www.mcanonline.org/sites/pdf/MCAN_call_fo r_proposal.pdf) was based primarily on a systematic review to identify pediatric asthma interventions implemented in urban, U.S. settings, with demonstrated effectiveness and materials for replication. Sites were required to implement these or other interventions with proof of efficacy, and to address multiple contexts of pediatric asthma prevention. The evaluation triangulates data from several sources: site-specific monitoring and evaluation data; site documents; qualitative assessments of families, organizational partners, and other stakeholders; and quantitative data from a common instrument on health indicators before and after the intervention.
Population Studied: MCAN grantees and organizational partners
Principal Findings: MCAN selected five sites (New
York, Puerto Rico, Chicago, Los Angeles, and
Philadelphia) with high asthma need and established asthma programs that addressed the needs of patients and families but were ready for a higher level of program integration across schools, healthcare systems, and communities. MCAN supports a community-based approach that is tailored to the specific needs of each site, allowing sites to pick from a menu of evidence-based interventions. As a result, each site is unique in its specific combination of interventions, but all sites serve common goals of integration of care, incorporation of evidence-based programs and improvement in knowledge, self-management, health, and quality of life. MCAN sites achieve service and system integration through created or enhanced linkages between health care providers, schools, community-based organizations, patients, parents, and other caregivers through care coordination, school interventions, community events, internal leadership team meetings, external coalition development, participation in external policy groups, and collaborations across MCAN sites. Sites report comparatively greater success in implementing interventions involving families and healthcare settings, and comparatively less success in involving schools in policy change; the degree of success varies by the organizational setting of the grantee.
Conclusion: Preliminary results suggest that community-based implementation of evidencebased interventions often requires significant modification from protocol to allow for translation to a wider range of settings. The organizational setting and prior partnerships of the grantee are strong facilitators of translation.
Implications for Policy, Delivery or Practice:
Lessons relevant to policy makers and funders include the importance of a planning period and the selection of sites with clear lines of authority to ensure accountability. Lessons relevant for practitioners include the need for clarifying roles and expectations with partners and for developing realistic goals despite the breadth of the stated goals. Lessons relevant for evaluators include the importance of participatory methods and the need for collaborative assessment of the balance between burden of data collection and rigor.
Funding Source(s): Merck Childhood Asthma
Network
Theme: Child Health
● Improving the Rates of Breast, Cervical &
Colorectal Cancer Screening: Are Community
Health Workers Effective?
Meera Viswanathan, Ph.D.; Jennifer Kraschnewski,
M.D.; Brett Nishikawa, M.D.; Laura Morgan, M.A.;
Patricia Thieda, M.A.; Dan Jonas, M.D., M.P.H.
Presented by: Meera Viswanathan, Ph.D., Health
Services Research Analyst, RTI International, 3040
Cornwallis Road, Research Triangle Park, NC
27709, Phone: (919) 316-3930; Email: viswanathan@rti.org
Research Objective: Community health workers
(CHWs) are defined as health workers who have a relationship with their community and are without formal professional training. These individuals can serve as a bridge to the health care system for communities with health disparities. The objective of this systematic review was to assess the effectiveness of CHW intervention for cancer screening.
Study Design: We conducted a systematic review using standard Evidence-based Practice Center methods. We searched Medline, the Cochrane
Database, and CINAHL from their inception through
October 2008 using a variety of terms for CHWs.
We included studies with a comparison group that were conducted in the United States, published in
English, with at least 40 participants. Two reviewers independently assessed each abstract and full-text article for inclusion, resolving disagreements by consensus. One reviewer abstracted data onto a standard form; a second reviewer checked for completeness and accuracy.
Trained reviewers abstracted data and assessed the methodologic quality of studies using predefined criteria from international standards.
Population Studied: Twenty-three of 992 citations, making up thirteen studies, examined outcomes of
CHW interventions for improving breast, cervical, or colorectal cancer screening met inclusion criteria for this systematic review. All studies focused on minority or underserved communities. With the exception of a single study on colorectal cancer screening that included both men and women, all other studies focus on increasing the rates of breast and cervical cancer screening among women.
Principal Findings: Among the eight randomized controlled trials and five observational cohort studies, two were of good quality, five of fair quality, and six poor. Seven studies used low-intensity
CHW models, four used moderate-intensity interventions, and two used high-intensity interventions. Six studies included more than two arms. Studies compared the CHW arm to a variety of alternatives, including no intervention (5), mail
(3), community interventions (3), CHWs in a lesser capacity (2), and CHWs in combination with other interventions (2). Studies used varied definitions of outcomes, with the greatest commonality in studies reporting on utilization of cancer screening tests such as mammography, clinical breast exam, pap smears, and colorectal cancer. Eleven of thirteen reported on changes in rates of utilization.
Preliminary results suggest variations in the consistency and volume of evidence across outcomes, with more evidence for outcomes such as mammography and pap smear rates, and either insufficient or inconsistent evidence for outcomes
such as breast self-exam, clinical breast exam, and colorectal cancer screening.
Conclusion: Preliminary analysis suggests variations in consistency, volume, and quality of evidence for cancer screening tests. The evidence also suggests that effectiveness varies by subgroups. New CHW interventions will need to take into account evidence of past effectiveness in specific
Implications for Policy, Delivery or Practice:
CHWs may serve as a means of improving some screening tests for underserved populations; the effectiveness of CHWs for other screening tests requires further research.
Funding Source(s): AHRQ
Theme: Disparities
● An Investigation of the Use of the
Environmental Scan Methodology for Rapid
Acquisition & Use of Information about Large
Health Systems
Boris Volkov, Ph.D.
Presented by: Boris Volkov, Ph.D., Assistant
Professor, Center for Rural Health, University of
North Dakota, 501 North Columbia Road, Stop
9037, Grand Forks, ND 58202-9037, Phone: (701)
777-6781; Email: bvolkov@medicine.nodak.edu
Research Objective: This paper investigates challenges and opportunities of using the environmental scan (ES) methodology to provide information about large health systems. The
Environmental Scan of Health and the Health
System in North Dakota (ND) was designed as a rapid need assessment process to help Dakota
Medical Foundation in their efforts of measurably improving health and access to high quality healthcare services, as well as enhancing knowledge about and collaboration in the health arena in North Dakota. The specific purposes of the environmental scan implemented by the Center for
Rural Health in the University of North Dakota
School of Medicine and Health Sciences were to:
1) promote collaboration of diverse stakeholders in the health arena by enhancing knowledge about health and health care issues and key health initiatives underway in North Dakota; and 2) through the use of relevant health and health care metrics, inform the development of effective interventions to address identified challenges. To date, there has been little use of environmental scanning of large health systems, with most of the past research focused on very specific areas of interest (e.g., certain medical providers, health conditions, or specific populations). There are only a few studies that have investigated ES implementation and use in the context of large health systems. Furthermore, there are only a small number of studies that have utilized the ES methodology to identify and explore strategies and vehicles for state-wide collaboration in the health arena. This paper illustrates how ES was used to assess and describe the current state, needs, and available resources of a state health system.
Study Design: To obtain a detailed, balanced, and comprehensive picture of health and health care in the state, the ES team used multiple methods and sources of data. The methods included literature review, analysis of secondary data, and semistructured interviews with a select number of key informants. The scan information was gathered from formal and well-established sources comprised of multiple databases, publications, as well as a sample of key stakeholders in the area of health and health care including health care experts, legislators, and representatives of state organizations.
Population Studied: The North Dakota population
Conclusion: The ultimate value of this research is an enhanced knowledge of the ways the
Environmental Scan approach can be applied when studying population health in the context of a large health system. While designed with a clear and realistic structure, the Scan was expected and proved to be an evolving, creative, and flexible process. It is important to have collaborative feedback of the study stakeholders throughout all stages of the Environmental Scan. The process of selection of relevant indicators has to be rigorous and comprehensive process – responsive to the needs of the stakeholders. Available time to conduct an ES is a very sensitive issue that must be taken into consideration.
Implications for Policy, Delivery or Practice: The
Environmental Scanning has the high potential for assessing diverse focus areas and factors that define and influence the health arena. There is a need for further research intended to define, develop, utilize, and evaluate effective practices for environmental scanning.
Funding Source(s): Dakota Medical Foundation
Theme: Quality and Efficiency: Measurement
● CenteringPregnancy® : Group Visit Prenatal
Care Holds Promise to Improve the Utilization &
Quality of Care & Reduce Maternal-Infant
Disparities
Susan Vonderheid, Ph.D., R.N.; Carrie Klima,
Ph.D., C.N.M.; Kathleen Norr, Ph.D.; Arden
Handler, Dr.P.H.
Presented by: Susan Vonderheid, Ph.D., R.N.,
Research Assistant Professor, Women, Children &
Family Health Science, University of Illinois at
Chicago, 845 South Damen Avenue, Chicago, IL
60612, Phone: (312) 996-7982; Email: vonde@uic.edu
Research Objective: Substantial disparities in low birth weight and infant mortality in the U.S. partly reflect a standard of prenatal care that has failed to address these and other adverse maternal-infant outcomes. Disparities are especially problematic for
African-American (AA) women who enter prenatal care later and have 2-3 times the national rates of infant mortality, prematurity, and low birth weight.
Centering Pregnancy® (Centering) is an innovative model of prenatal care that replaces the individual visit model with a group visit model and bundles support services for pregnant women. In Centering,
8-12 women having similar due dates attend ten 2hour group visits. The objectives of this study were to examine the impact of Centering on maternal and infant outcomes compared to individual prenatal care (IC); and to examine the acceptability of
Centering by women and providers.
Study Design: This descriptive study examined qualitative and quantitative data collected from focus groups with staff, self-administered questionnaires with pregnant women, and medical record review for utilization and perinatal data.
Independent T-test and Chi-square were used to analyze quantitative data. Content analysis was used to analyze qualitative data.
Population Studied: African American low-income pregnant women (Centering n = 62 and IC n= 134) receiving care prenatal care from midwives, and 16 clinical and administrative staff in a public health clinic in a large midwestern city.
Principal Findings: Women in Centering had more prenatal visits, gained more weight during pregnancy, were more likely to initiate breastfeeding at hospital discharge, and had higher satisfaction. Trends showed higher gestational age and fewer low birth weight infants among Centering women; but differences were not statistically significant. Among preterm deliveries, women in
Centering were pregnant one week longer and their infants weighed nearly 200 grams more than IC infants. Staff described Centering women as being more independent, taking the initiative to selfschedule missed visits, being more involved in their prenatal care and infants’ pediatric care, and using more available resources. Prenatal providers enjoyed group discussion of health education compared to repeating similar information to each woman individually.
Conclusion: There is widespread recognition of the need for innovative models of prenatal care to improve maternal-infant health. Findings suggest that Centering is a promising strategy to improve the utilization and quality of prenatal care, and reduce adverse maternal and infant outcomes especially among low-income African American women. Prenatal providers reported higher professional satisfaction and found group visits were more efficient.
Implications for Policy, Delivery or Practice:
Centering represents a major restructuring of prenatal care that might improve maternal-infant outcomes especially among vulnerable populations at greatest risk related to socioeconomic factors, improve the efficiency of service delivery, and increase satisfaction with care by providers and patients. Providers who are more satisfied with their work are likely more willing to remain in the workforce. While evidence suggests that Centering has favorable impacts, large randomized controlled trials are needed to more fully examine the effects of Centering on maternal and infant health-related outcomes, provider satisfaction and retention, service efficiency, and costs.
Funding Source(s): March of Dimes Illinois
Chapter
Theme: Disparities
● Coordination & Integration of Care for
Chronically Ill: Meaning & Measurement
Bert Vrijhoef, M.Sc., Ph.D.; Edward Wagner, M.D.
M.P.H.
Presented by: Bert Vrijhoef, M.Sc., Ph.D.,
Harkness Fellow, MacColl Institute for Healthcare
Innovation, Group Health Center for Health Studies,
1730 Minor Avenue, Suite 1290, Seattle, WA
98101-1448, Phone: (206) 287-2044; Email: vrijhoef.h@ghc.org
Research Objective: Innovative strategies in chronic care delivery are welcomed as responses to a set of problems that are evident to some degree in all health care services. Uncoordinated or fragmented arrangements for the delivery of care are one of such problems. Notwithstanding the theoretical conceptualization of coordination and integration, in practice variation exists in how these concepts are applied, measured and in their impact on care. Consequently, difficulties are experienced in deciding how to best resolve care fragmentation.
Study Design: First, an exploratory, qualitative design is used to make the phenomena understandable and identify the perceptions of experts in chronic care coordination and integration.
A purposive sample of 20 experts from the US,
Canada, Australia, and Europe is in-depth interviewed by telephone or face-to-face. An interview guide is being applied to initiate and guide the interview. Three central topics are addressed: care fragmentation, coordination and integration, measurement. Interviews are audio-recorded and transcribed with the consent of interviewees. A grounded theory approach is used to analyze the data. Second, based on the results of the first phase a conceptual framework to guide decision making on strategies for coordination and integration of chronic care is tested among an international sample of 100 stakeholders in chronic care. Data are collected by means of a survey and analyzed descriptively. Approval for the study is received from the Center's Human Subject Review
Office.
Population Studied: Twenty (20) experts in the field of innovative chronic care delivery from the
US, Canada, Australia, and Europe. One hundred
(100) stakeholders in chronic care from the US,
Canada, Australia, and Europe.
Principal Findings: Preliminary results indicate an urgent need to resolve care fragmentation and concepts of coordination and integration of care being interrelated and partial overlapping. By identifying a plethora of factors and stakeholders involved, the complexity of resolving care fragmentation gets revealed. Diversity is reflected by the type of indicators being identified to measure the impact of coordination or integration on care.
Conclusion: Conclusions. Final conclusions will be drawn after completion of the data analysis and available in June 2009.
Implications for Policy, Delivery or Practice:
Study results will provide guidance to healthcare providers, patients, policymakers and researchers about how to best resolve fragmentation of chronic care and what to measure in order to make meaningful comparisons among strategies to improve coordination and integration of chronic care.
Funding Source(s): CWF
Theme: Quality and Efficiency: Organized
Processes
● Does Vigilance in Decision Making Matter for
Caregivers?
Sarah Wackerbarth, Ph.D.; Yelena Tarasenko,
M.P.A.
Presented by: Sarah Wackerbarth, Ph.D.,
Associate Professor, University Kentucky, 435
Patterson Office Tower, Lexington, KY 40506-0027,
Phone: (859) 229-1949; Email: sbwack0@uky.edu
Research Objective: Family responsibility and societal expectations often put caregivers in conflict regarding the decisions they must make. Janis &
Mann’s (1977) conflict model describes vigilance as high-quality decision making resulting in optimal outcomes. The intention behind this research was to determine whether caregivers may benefit from interventions assisting them to be vigilant decision makers. Our objectives were threefold: 1) to identify decision styles of dementia family caregivers; 2) to examine their socio-economic characteristics; and
3) to determine whether vigilant caregivers report fewer caregiving problems.
Study Design: Our analysis was based on 639 survey respondents recruited from a university affiliated memory disorders clinic. We used regression methods to determine whether various outcomes could be explained by decision style, caregiver characteristics, conflicting obligations, and caregiver objective burden.
Population Studied: Our typical caregiver was
Caucasian, non-Hispanic, currently married and had 2.1 children.
Principal Findings: Approximately half of our sample used a pure vigilant decision style. Males with higher education were more likely to be vigilant decision makers. Vigilance was related to nine measures of caregiver outcome and was unrelated to the socio-demographic characteristics.
Conclusion: Approximately half (44.0%) of our dementia family caregivers’ sample used a pure vigilant decision style. Males and those with higher education were more likely to be vigilant decision makers. The vast majority (98.2%) of people who had a “pure style” were vigilant, thus, supporting the theory which states that vigilant people tend to be strictly vigilant (Janis & Mann, 1977). Most of our respondents (96.4%) agreed with at least one statement indicative of vigilance. Our results also support the theory which states that non-vigilant people will vacillate between the maladaptive styles
– perhaps, reacting to the influence of environmental factors (Janis & Mann, 1977).
Implications for Policy, Delivery or Practice:
Supporting caregivers to become vigilant decision makers would not only have major impact on the caregiving experience, it is a functionally viable intervention.
● Variations in CRC Screening Practice: Would this Patient be Screened?
Sarah Wackerbarth, Ph.D.; Yelena Tarasenko,
M.P.A.; Jennifer Joyce, M.D.; Steven Haist, M.D.,
M.S., F.A.C.P.
Presented by: Sarah Wackerbarth, Ph.D.,
Associate Professor, University of Kentucky, 435
Patterson Office Tower, Lexington, KY 40506-0027,
Phone: (859) 229-1949; Email: sbwack0@uky.edu
Research Objective: Data indicate that screening rates for colorectal cancer (CRC) in the United
States are less than the goal of 50 percent outlined in Healthy People 2010. Physician recommendation is an important predictor of patient’s compliance with CRC screening tests.
The purpose of our research was to compare colorectal screening decisions made by individual physicians and determine how variations in their decision processes impact decision outcomes.
Further, we examined whether those variations could be attributed to physician’s characteristics and guidelines’ utilization.
Study Design: We conducted semi-structured interviews with primary care physicians, developed decision trees, compared trees, used trees to predict the recommendation for eight vignette patients, and employed regression analysis to identify predictors of variation.
Population Studied: We interviewed sixty-six primary care physicians (59.1% internal medicine and 40.9% family practice).
Principal Findings: Most of the physicians (77.3%) self-reported following clinical guidelines for CRC screening. Physicians considered an average of
5.9 decision criteria (range 2-12) in making their screening recommendations. Frequently cited criteria included patient age and family history. We documented variation for three of eight vignette
patients. Regression analysis indicated that complexity of decision process, gender, age and experience of physicians contributed to recommendations on screening. In addition, the self-report adherence to guidelines did not influence whether a physician would recommend CRC screening.
Conclusion: This study supports the notion that variation in practice is a function of decision processes.
Implications for Policy, Delivery or Practice:
Studying decision processes may facilitate efforts to improve patient outcomes.
Funding Source(s): NCI
● The Earnings Penalty for Body Fat: Evidence from NHANES III
Roy Wada, Ph.D.
Presented by: Roy Wada, Ph.D., Assistant
Researcher, University of California, Los Angeles
School of Public Health, Box 951772, Los Angeles,
CA 90095-1772, Email: roywada@ucla.edu
Research Objective: Previous studies on the impact of obesity as measured by BMI have yielded mixed results, in some cases implying that obese workers are more productive than non-obese workers.
Study Design: Body composition and income data from the Third National Health and Nutrition
Examination Surveys (NHANES III) are used to examine the effect of body fat and fat-free mass on labor earnings. Interval regression analysis is used conduct the effect of body composition on labor earnings.
Population Studied: Unmarried white men and women from single households from a nationally representative survey are used in this study.
Principal Findings: Adjusting for confounding factors, this paper finds that 1 kg of body fat decreases labor earnings by up to 1%. Conversely,
1 kg of fat-free mass increases labor earnings by up to 1%.
Conclusion: Obesity as measured by body fat is unambiguously related to decreased labor earnings.
Physical fitness as measured by fat-free mass is clearly associated with increased labor earnings.
Implications for Policy, Delivery or Practice:
Obesity is clearly associated with decreased labor productivity. The social costs of rising obesity extend to labor productivity of workers.
Funding Source(s): AHRQ
Theme: Obesity Prevention and Treatment
● Provider Attitudes Associated with Adherence to Evidence-Based Clinical Guidelines in a
Managed Care Setting
Anthony Waddimba, M.B., Ch.B., M.S., D.Sc.; Mark
Meterko, Ph.D.; Howard Beckman, M.D.; Gary
Young, J.D., Ph.D.; James Burgess, Jr., Ph.D.
Presented by: Anthony Waddimba, M.B., Ch.B.,
M.S., D.Sc., Research Fellow, Health Policy &
Management, Boston University Medical Campus,
715 Albany Street, Talbot 3 West, Boston, MA
02118, Phone: (617) 414-1351; Email: waddimba@bu.edu
Research Objective: Primary care physicians
(PCPs) often exhibit unwarranted variations in clinical performance, hence the importance of studying what influences provider behavior.
Combining multiple behavior-change interventions, such as public reporting and pay-for-performance incentives, might improve practitioner adherence to clinical guidelines. We investigated PCPs' attitudes associated with incentivized adherence to clinical guidelines in a managed care setting.
Study Design: In a cross-sectional observational study of PCPs enrolled in a Rochester (New York) pay-for-performance collaboration, we linked guideline adherence rates to provider attitudes and individual, practice, and community demographics.
PCPs elected to be profiled on the clinical care pathway for one of four patient panel-representative medical conditions: diabetes, asthma, otitis media, or sinusitis. The adherence rate was defined as the percent of expected services in the clinical care pathway for the representative medical condition that were actually delivered. We derived a multivariable logistic regression model of highesttertile adherence to recommended care for the panel-representative medical condition, adjusting for pre-incentive adherence, provider demographics and practice environment.
Population Studied: From 290 physicianrespondents to an attitudes survey, we excluded those who (a) did not select a patient panelrepresentative medical condition as directed in the questionnaire (n=54), (b) named a condition ineligible for quality-based pay incentives (n=19), or
(c) lacked data on adherence rates (n=31), leaving a final sample of 186 PCPs (49 family practitioners,
84 general internists, and 53 general pediatricians).
Principal Findings: The attitudes associated with guideline adherence, independent of specialty and prior behavior, were financial salience (adjusted
Odds Ratio (OR) = 3.6; 95% Confidence Interval
(CI) = 1.7 – 8.4), cooperation from peers (OR = 2.0;
95% CI = 1.0 – 4.0), perceived control (OR = 0.5;
95% CI = 0.3 – 1.0); and perceived need for autonomy from the health plan (OR = 0.3; 95% CI =
0.1 – 0.6).
Conclusion: The providers that were most adherent to clinical guidelines perceived pay-forperformance incentives as financially salient and felt well supported by peers. PCPs with a higher need for autonomy and/or those who perceived themselves to have greater control were intrinsically demoralized by the early-stage explicit incentives and they reduced work effort.
Implications for Policy, Delivery or Practice:
Policy makers should engender “ownership” and
“buy-in” among providers so that quality incentives are not perceived as impositions from outside.
Strategies to safeguard intrinsic motivation might include early, frequent consultations with providers before instituting clinical guidelines, and inclusion of their input in ways that generate pre-commitment acceptance. Providers should verify their adherence scores before they are “profiled”, to ensure that they agree with the accuracy of the audits. Quality measures should be reliable, valid and transparent.
Providers should be convinced that these are valid measures of adherence, and not indices of documentation. A “partnership model” might be more suited than a compliance-centered approach to intrinsically motivated providers with a greater need for autonomy and self-determination.
Funding Source(s): AHRQ
● Cost-Related Nonadherence to Prescription
Medications in the Medicare Population:
Disparity by Hispanic Ethnicity?
Daniel Waldo, M.A.; Karyn Kai, Anderson, Ph.D.,
M.P.H.; Diane Frankenfield, Dr.P.H.; Iris Wei,
Dr.P.H.; Benjamin Howell, M.P.P.; Edward
Sekscenski, M.P.H.
Presented by: Daniel Waldo, M.A., Senior
Economist, Actuarial Research Corporation, 5950
Symphony Woods Road, Suite 510, Columbia, MD
21044, Phone: (410) 740-9194; Email: dwaldo@aresearch.com
Research Objective: Disparities between Hispanic and non-Hispanic populations in health care access and utilization in the United States are well documented. The objective of this research was to determine whether Hispanic Medicare enrollees were more likely to report cost-related nonadherence (CRN) to prescription medications than were non-Hispanic enrollees.
Study Design: This study employed a crosssectional research design using the Medicare
Consumer Assessment of Healthcare Providers and
Systems (CAHPS) data. In the Spring of 2007, questionnaires were sent to 695,197 Medicare beneficiaries. For those who did not return a questionnaire, the computer-assisted telephone interview (CATI) was employed to augment response rate. Of the 335,249 survey respondents
(weighted national estimate: 41,252,608), 70% were from completed and returned mail questionnaires and 30% were from CATIs. The analytic sample included 272,701 communitydwelling respondents, aged 18 or older and enrolled in Medicare, who met the inclusion criterion of responding either “Yes” or “No” to delaying or not filling a prescription medication in the last six months because of cost (roughly the last half of
2006). Hispanic enrollees comprised 6.9% of this analytic sample (unweighted n = 22,304). We assessed the odds of prescription drug CRN by
Hispanic ethnicity in adjusted logistic models controlling for race (White, Black, Asian/Pacific
Islanders [APIs], American Indian/Alaska Natives
[AIAN], and “race not reported”), Medicare Part D enrollment status, socio-demographic, health status, health care utilization, and patient rating of their personal physician. Analyses were conducted using Stata, version 10. Adjusted odds ratios
(adjOR) and 95% confidence intervals (95% CI) are presented.
Population Studied: The population consisted of adult community-dwelling Medicare beneficiaries* who reported at least one prescription during the six months preceding the survey
Principal Findings: Overall, 13.4% of respondents in the sample for analysis reported CRN. One fifth
(20.3%) of Hispanic enrollees reported CRN compared with 12.9% of non-Hispanic enrollees (p
< 0.0001). Hispanic enrollees were 70% more likely to report CRN in the past 6 months than non-
Hispanic enrollees (unadjOR [95% CI] 1.7 [1.5,
1.9]). In a logistic regression model controlling for the variables described above, Hispanic ethnicity persisted as a predictor of CRN (adjOR [95% CI]
1.25 [1.11, 1.40]).
We ran our logisitic regression separately for
Hispanic and non-Hispanic enrollees. Within the former group, API and AIAN enrollees were more likely to report CRN than Whites (adjOR [95%CI]
1.53 [1.05, 2.23] and 1.56 [1.16, 2.11], respectively). Among non-Hispanic enrollees, both
African American and AIAN enrollees were more likely to report CRN in the past 6 months than White enrollees (adjOR [95%CI] 1.44 [1.33, 1.55] and 1.55
[1.41, 1.71], respectively).
Conclusion: After controlling for a range of covariates, we found statistically significant disparity between Hispanic and non-Hispanic enrollees in reported CRN in the time period after implementation of Medicare’s Part D prescription drug benefit. Taking Hispanic ethnicity into account, disparity was also found between several racial minorities – especially AIAN – and the majority group.
Implications for Policy, Delivery or Practice:
One year after implementation of Medicare Part D,
Hispanics and racial/ethnic minorities were more likely to delay or not fill a prescription medication because of cost. Our findings suggest that targeted outreach could be helpful in reducing CRN in minority populations.
Funding Source(s): CMS
Theme: Disparities
● Alternatives to Financing Health Insurance for the Uninsured & the Effect on Government
Subsidies
Thomas Walke, M.S.P.H., Ph.D.; Charles, Ph.D.;
George Miller, Ph.D; Joe Dorris, M.A.
Presented by: Thomas Walke, M.S.P.H., Ph.D.,
Practice Area Leader, Forecasting & Economic
Analysis, Altarum Institute, 3520 Green Court, Suite
300, Ann Arbor, MI 48105, Phone: (734) 302-4720;
Email: thomas.walke@altarum.org
Research Objective: This study examines alternative methods to financing health insurance for the uninsured population and the potential effect on government subsidies. Given the current fiscal budget crisis and economic recession, the number of uninsured is likely to increase beyond the current
47 million as the ability of government to pay for increased coverage declines. Using national data on expenditures, we estimate an individual’s expected premium if that person had to purchase insurance in an individual market at an actuarially fair price (plus an administrative mark-up), and the subsidies required from government to make health insurance affordable. We also examine how premiums would vary using alternative approaches to setting premiums, such as community-based rating.
Study Design: To examine alternative methods of assessing risk, we first developed a 10-year panel dataset (1996 to 2005) on individual’s risk experience using health expenditure data from
Medical Expenditure Panel Survey (MEPS). Using
MEPS, we estimated actuarially fair premiums for individuals and families by using nonlinear regression analysis to model total healthcare spending as a function of age, gender, pre-existing conditions, insurance, and other risk factors.
Analogous to the Massachusetts health reform program approach, we developed estimates of what families can afford to pay for health insurance by using various income thresholds and affordability algorithms.
Population Studied: U.S. civilian non-institutional population
Principal Findings: Preliminary findings for the uninsured show that almost 80 percent of uninsured would have actuarially fair premiums of less than
$4,000—roughly the average annual individual premium to cover a federal employee in 2005— based on age and risk experience. The uninsured and privately insured have similar premium profiles while Medicare and Medicaid eligibles are more heavily weighted toward higher premiums, reflecting, in part, the prevalence of disabled individuals. Using the affordability algorithms, our estimates indicate that the government would have to subsidize almost three-quarters of the cost to cover the uninsured.
Conclusion: This analytic approach results in riskadjusted annual premiums for the uninsured U.S. noninsitutional population and the financial subsidies required to ensure coverage. The approach used for these subsidies could take different forms. For example, in a “voucher” model a subsidy would be provided directly to a consumer who then pays premium with the subsidy. Study findings can also be used to calculate the crosssubsidies implicit in various forms of community rating, and how the financial sustainability of the system is influenced by these factors.
Implications for Policy, Delivery or Practice: At a national and state level, this analytic framework could be used by policymakers to determine the minimum subsidy levels required to make insurance affordable. For example, these data could be used to estimate the government subsidies required under President’s Obama National Health
Insurance Exchange program.
Funding Source(s): Internally funded
Theme: Coverage and Access
● The Independent & Combined Effect of Health
Insurance and/or a Usual Source of Care on
Receipt of Diabetes Care in the U.S.
Lorraine Wallace, Ph.D.; Jennifer DeVoe, M.D.,
D.Phil.; Carrie Tillotson, M.P.H.
Presented by: Lorraine Wallace, Ph.D., Associate
Professor, Family Medicine, University of
Tennessee Graduate School of Medicine, 1924
Alcoa Highway, U-67, Knoxville, TN 37920, Phone:
(865) 305-7419; Email: lwallace@mc.utmck.edu
Research Objective: While the independent influences of both health insurance coverage (INS) and a usual source of care (USC) on receipt of preventive care among diabetics have been explored separately, no studies have examined the effects of both key elements simultaneously. We examined the independent and combined effects of
INS and/or a USC on receipt of diabetic-specific services and access to overall care for US adults with diabetes.
Study Design: Using pooled 2002-2005 data from the nationally-representative Medical Expenditure
Panel Survey-Household Component, we assessed self-reported receipt of 7 diabetes-specific yearly preventive services, including: glycosylated hemoglobin (HbA1c), lipid screening, influenza vaccination, blood pressure check, diabetic foot exam, dilated eye exam, and routine medical check-up. Seven outcome variables pertaining to compromised healthcare access were also examined, including: unmet medical need, unmet prescription need, unmet dental need, problems obtaining specialty care, delayed urgent care, and 2 composite variables for reports of =1 of 16 unmet healthcare needs and never having received =1 of
5 recommended preventive services. Diabetics were divided into 4 categories: (1) Yes USC/Yes
INS, (2) Yes USC/No INS, (3) No USC/Yes INS, and (4) No USC/No INS. Using SUDAAN statistical software, we constructed a series of multivariate logistic regression models to assess associations between USC/INS status and utilization of preventive services and reports of unmet healthcare needs, while simultaneously controlling for potential confounders.
Population Studied: Our analysis included respondents, =18 years, who reported a diagnosis
of diabetes, data about USC status, and full year
INS data (total unweighted number=6562; weighted average yearly population==15 million).
Principal Findings: Over 84% of US diabetics had full-year INS and a USC, while only 2.3% had neither one. In multivariate analyses, those with
INS and a USC were the most likely to receive recommended care, while the uninsured with no
USC received the fewest services. The uninsured with no USC had 1/5 the odds of receiving HbAlc screening (OR 0.23; 95% CI 0.14-0.38); 1/10 the odds of a blood pressure check (OR 0.08; 95% CI
0.05-0.15), compared with insured diabetics with a
USC. Similarly, being uninsured without a USC was associated with 6.5 times the likelihood of unmet needs (OR 6.48; 95% CI, 3.97-10.58) and 3 times more delayed urgent care (OR 3.13; 95% CI
1.53, 6.38), compared to being insured with a USC.
Among diabetics with either INS or a USC, those with only a USC had higher rates of diabetes services. In contrast, those with only INS had fewer overall unmet needs.
Conclusion: Diabetics with INS and a USC were better off than those with only a USC, only INS, or neither one. When comparing INS to a USC, there was no clear pattern of superiority.
Implications for Policy, Delivery or Practice: For diabetics, a USC and INS, together, are additive predictors of the likelihood that they will have optimal access to all necessary healthcare services.
Policy solutions must maximize the synergy between health insurance and USC to achieve the most optimal access to care.
Funding Source(s): AHRQ
Theme: Coverage and Access
● Uninsured Children with Insured Parents: A
Closer Look at Their Unmet Needs
Lorraine Wallace, Ph.D.; Jennifer DeVoe, M.D.,
D.Phil.; Carrie Tillotson, M.P.H.
Presented by: Lorraine Wallace, Ph.D., Associate
Professor of Family Medicine, Family Medicine,
University of Tennessee Graduate School of
Medicine, 1924 Alcoa Highwawy, U-67, Knoxville,
TN 37920, Phone: (865) 305-7419; Email: lwallace@mc.utmck.edu
Research Objective: To determine the associations of health insurance coverage, access to heath care, and receipt of preventive care among uninsured children with insured parents, as compared to insured children with insured parents.
Study Design: We conducted cross-sectional and full-year analyses of pooled 2002-2006 data from the nationally-representative Medical Expenditure
Panel Survey-Household Component. We assessed 10 outcome variables pertaining to children’s health care access and receipt of preventive care within the past year, including: any unmet medical need, no usual source of care
(USC), delayed care, no doctor visits, no dental visit, no height/weight measurement, no dental check-up advice, no healthy eating and/or exercise advice, no safety seat/booster seat/lap belt advice, and no bike helmet advice. Using SUDAAN statistical software, we constructed a series of multivariate logistic regression models, using a predictor variable linking the insurance status of children and their parents. We conducted crosssectional and full-year analyses.
Population Studied: Our analysis included all children, aged 2-17 years, with positive full-year weights who had at least 1 insured parent residing in the same household. We specifically focused on two groups of children with insured parents: uninsured children and insured children.
Principal Findings: Nationally, more than 77% of
US children were insured with at least 1 insured parent in their household, while 3% of children were uninsured with at least 1 insured parent. The remaining 20% of children had at least 1 uninsured parent. Among the subgroup with insured parents, children without insurance were more likely to have unmet medical needs (OR=1.32, 95% CI 1.11-
1.56), less likely to have a USC (OR=2.16, 95% CI
1.70-2.74), more delayed in getting needed medical care (OR=2.05, 95% CI 1.46-2.86), more likely to have fewer doctor visits (OR=1.51, 95% CI 1.27-
1.80), and more likely to have fewer dental visits
(OR=2.10, 95% CI 1.72-2.57), as compared to the group of insured children (reference= 1.00).
Uninsured children with insured parents were also less likely to have received preventive services or counseling, including: height/weight measurement
(OR=1.45, 95% CI 1.13-1.84), counseling regarding healthy eating and/or exercise (OR=1.33, 95% CI
1.11-1.59), and counseling regarding use of bike helmets (OR=1.32, 95% CI 1.08-1.62). Patterns were similar when examining full-year insurance data. For example, among children with parents covered all year, those children with coverage gaps had higher rates of unmet needs than children with no gaps.
Conclusion: Among families with insured parents, insured children were more likely to have better health care access and receipt of preventive care than were uninsured children with insured parents.
Implications for Policy, Delivery or Practice:
These findings underscore the need to pay close attention to policy solutions that keep children covered, even in families with access to insurance for parents. Ensuring that entire families have continuous health insurance coverage will help improve children’s access to needed health care and timely receipt of preventive services and counseling.
Funding Source(s): AHRQ
Theme: Coverage and Access
● Agents of Change: What Role do CAM
Providers Play in Health Behavior Change?
Edith Walsh, Ph.D.; Pamela Williams-Piehota,
Ph.D.; Fuschia Sirois, Ph.D., B.S.; Carla Bann,
Ph.D.; Karen Isenberg, M.P.H.; Edith Walsh, Ph.D.
Presented by: Edith Walsh, Ph.D., Senior
Researcher, Aging, Disability & Long-Term Care,
RTI International, 1440 Main Street, Waltham, MA
02421, Phone: (781) 434-1754; Email: ewalsh@rti.org
Research Objective: Recent investigations indicate that complementary and alternative medicine (CAM) use is associated with the practice of positive health behaviors, and that CAM use may be conducive to health behavior change. How this change occurs -- through the support of CAM providers, as part of increased responsibility for one’s health, or motivated by the success of CAM treatments -- has not been previously investigated.
Study Design: Mixed methods: Five focus groups used to identify aspects of health behavior change associated with CAM use, and a survey of a convenience sample of 216 CAM users about their
CAM therapy and health behaviors.
Population Studied: Adults reporting use of at least one complementary and alternative medicine
(CAM) therapy in the last 12 months.
Principal Findings: Over 70% of survey respondents reported making behavioral changes as a result of going to their CAM therapist, including
50% who reported improved diet and 55% who reported increased exercise. The majority of respondents who reported making behavioral changes as a result of going to their CAM therapist reported doing so because their therapist encouraged or recommended it (53%), and because feeling better due to their treatment motivated them (53%) and gave them energy to make changes (53%). The majority of respondents who improved their diet and increased exercise also reported that the treatments themselves helped
(55% and 53%, respectively). Multivariate analysis showed that increased odds of self-reported dietary change were significantly associated with increasing body awareness as a result of therapy, endorsing the statement that sustained improvement for their health conditions required self-care, using an acupuncturist, and being age 44 years or younger. We found comparable results for self-reported increase in exercise, except that using an acupuncturist was a significant negative predictor and age was not a significant predictor.
Conclusion: The findings from this initial investigation into the role CAM therapists play in health behavior change suggest that the support of
CAM providers, increased responsibility for one’s health, and CAM treatments themselves are all contributors to behavior change, although additional research in this area is warranted.
Implications for Policy, Delivery or Practice:
CAM therapists may be an underutilized resource for promoting healthy lifestyle changes including healthy eating, increased physical activity, and weight loss.
Funding Source(s): RTI International
Theme: Obesity Prevention and Treatment
● Validation of Area-Level Pediatric Prevention
Quality Indicators at the Sub-County Level in the Dallas Area
Brad Walsh, M.P.H.; Sue Pickens, M.Ed.
Presented by: Brad Walsh, M.P.H., Strategic
Planning Project Administrator, Strategic Planning,
Parkland Health & Hospital System, 5201 Harry
Hines Boulevard, Dallas, TX 75235, Phone: (214)
590-8707; Email: bwalsh@parknet.pmh.org
Research Objective: To evaluate how well hospitalization rates for AHRQ’s Pediatric
Prevention Quality Indicator (PDI) area-level categories correlate with income and insurance coverage parameters for children, and with rates of potentially avoidable pediatric ER visits per the New
York University Algorithm rules, for 28 sub-county areas in four Dallas, TX, area urban/suburban counties.
Study Design: Statistical analysis of existing datasets for hospitalizations, ER visits and censusderived economic parameters.
Population Studied: All children and all adults in four Texas counties (Dallas, Tarrant, Collin and
Denton) who had an inpatient stay or ED visit in
2007.
Principal Findings: For 28 sub-county service areas in the Dallas area, AHRQ's adult Prevention
Quality Indicators are almost all significantly positively correlated with measures of socioeconomic need (the Community Need Index), and with potentially preventable ED visits as defined by the NYU ED Algorithm. However the picture is much more mixed with AHRQ's newer area-level Pediatric Prevention Quality Indicators
(PDIs). While rates for two of the five PDIs studied are significantly positively correlated with child poverty rates and with age-adjusted avoidable pediatric ED visits, two others are not significantly correlated, and one, PDI 16 gastroenteritis, is significantly negatively correlated with child poverty rates and with avoidable pediatric ED visit rates, the only area-level PDI or PQI rate to be so.
Conclusion: PDIs 15 and 17 do not correlate with an independent measure of access to pediatric primary care (avoidable ED utilization), nor do they correlate significantly with child poverty rates, in the
Dallas area. This suggests that these two PDIs are not useful measures of pediatric access to care.
Because PDI 16 correlates negatively with child poverty and avoidable ED use, users are cautioned to interpret PDI 16 hospitalization rates with this inverse relationship in mind. One important limitation of our study is that the age categories in the databases for hospitalized children and ED
visits by children did not match the exclusion criteria in the AHRQ PDI definitions.
Implications for Policy, Delivery or Practice: We recommend that two of the area-level PDI categories (PDI 15 diabetes short-term complications and PDI 17 perforated appendix rates) should not routinely be used for policy analysis regarding adequacy of access to pediatric primary care, and PDI 16 gastroenteritis should only be used with the understanding that rates are negatively correlated with socioeconomic measures and with independent measures of access to care such as avoidable ED utilization.
Theme: Quality and Efficiency: Measurement
● Measuring Provider Support in
Complementary & Alternative Medicine
Edith Walsh, Ph.D.; Carla Bann, Ph.D.; Edith
Walsh, Ph.D.; Fuschia Sirois, Ph.D., B.Sc.
Presented by: Edith Walsh, Ph.D., Senior
Researcher, Aging, Disability & Long-Term Care,
RTI International, 1440 Main Street, Waltham, MA
02451, Phone: (781) 434-1754; Email: ewalsh@rti.org
Research Objective: To develop and psychometrically evaluate scales to measure patients’ perceptions of provider support, patientcentered care, networking, and coordination they received as a part of their CAM treatments. The quality of the patient-provider relationship is wellrecognized as having a key role in therapeutic outcomes irrespective of treatment effects. Yet there is a lack of scales to assess aspects of CAM provider support.
Study Design: Based on five focus groups with
CAM clients, we developed the following three scales: Perceived Provider Support, Patient-
Centered Care (PCC), and
Networking/Coordination. The scales were cognitively tested with six CAM users and then pilot-tested with 217 respondents to a mail survey.
Confirmatory factor analyses, IRT analyses, and
Cronbach’s alphas were conducted to evaluate their psychometric properties. To assess construct validity, we ran linear regression models comparing scale scores for various subgroups, controlling for demographic characteristics, as well as linear regression models evaluating the impact of these scales on symptom relief.
Population Studied: Adult users of complementary and alternative medicine.
Principal Findings: Results of the confirmatory factor analyses supported a one-factor solution for the Perceived Provider Support and Patient
Centered Care scales. However, rather than forming a single scale, the Networking/Coordination items split into two factors (Networking and
Coordination) (CFI=0.98, TLI=0.98, RMSEA=0.07) and were scored as separate scales. Cronbach’s alphas for the scales suggested acceptable reliability for all but the Networking scale: PCC
(0.87), Provider Support (0.90), Networking (0.63), and Coordination (0.79). Group comparisons supported the construct validity of the scales. For example, those who reported being more involved in treatment decisions had higher scores on the
Patient Centered Care scale (p < .001). Patient
Centered Care and Perceived Provider Support scores were also significant predictors of symptom relief.
Conclusion: The new scales show promise for assessing CAM clients’ experiences of provider support.
Implications for Policy, Delivery or Practice:
While these measures were developed for use in
CAM research, they include new approaches to measuring patient-provider interactions that could be applied in evaluating conventional medical care as well. The elements of the scales also provide some insight into aspects of interactions with CAM providers that consumers value- these could be used to educate medical providers about important dimensions of patient-centered care.
Funding Source(s): RTI International
Theme: Quality and Efficiency: Measurement
● Clinical & System Cost Outcomes for Patients who Graduated from an Educator-Taught
Diabetes Self-Management Class
Brad Walsh, M.P.H.; Meghan Jardine, M.S., M.B.A.,
R.D., L.D., C.D.E.; David Hoang, B.S.; Celette
Skinner, Ph.D.
Presented by: Brad Walsh, M.P.H., Strategic
Planning Project Administrator, Strategic Planning,
Parkland Health & Hospital System, 5201 Harry
Hines Boulevard, Dallas, TX 75235, Phone: (214)
590-8707; Email: bwalsh@parknet.pmh.org
Research Objective: To evaluate clinical outcomes and healthcare utilization costs for newly-diagnosed diabetics who completed a 5-session patient education class ("Healthy Living With Diabetes", an
ADA-recognized DSME course), compared with new diabetics who did not complete the class.
Patients are adults with type II diabetes, mainly lowincome and minorities.
Study Design: Retrospective cohort design comparing course completers who are patients of the Parkland Health & Hospital System with adult type II diabetics who did not take the course.
Population Studied: Low-income adult type II diabetics in Dallas, Texas.
Principal Findings: Demographically, our class completion group and non-participant group were similar. Our class completion group had slightly more females (+7.8%) and Hispanics (+9.2%) than our non-participant group. When analyzing changes in %HbA1c for the completion group from the time the patients started the program to a final value at a time 3-6 months after program completion, we found that there was an average decrease of
1.452% HbA1c. Post program completion average was 6.994% HbA1c. We noticed average decreases of over 1% HbA1c regardless of gender or ethnicity. Program completers were shown to have a greater number of encounters with our primary care clinic network but fewer encounters with the System's specialty clinics. Completion of the program was also associated with an increased incidence for a physician diagnosis of diabetesrelated co-morbidities such as hypertension, hypercholesterolemia, obesity, diabetic neuropathy, and retinopathy. We also found trends for a higher incidence of myocardial infarction, stroke, coma, and coronary artery disease for the non-participant group but no statistically significant differences.
When analyzing costs, we noticed that participation in the program was associated with increased primary care clinic-related costs, but with savings for inpatient and specialty clinic visits. There was an average savings of $3800 per year system-wide for each patient in the program completion group.
Conclusion: The Parkland System's Healthy Living
With Diabetes program shows both clinical improvements and system-wide cost savings.
Implications for Policy, Delivery or Practice:
Providers instituting or purchasing an ADArecognized diabetes self-management education course can expect both clinical improvements and overall healthcare cost reductions (when specialty,
ED and inpatient costs are included) for patients completing the course.
● Impact of Washington State Diabetes
Collaborative III on Patient Economic Outcome
Song Wang, M.A.; Douglas Conrad, Ph.D; Amira
El-Bastawissi, M.B.C.H.B., Ph.D; Eric Ossiander,
M.S.
Presented by: Song Wang, M.A., Ph.D. Candidate,
Health Services, University of Washington, 1959
Northeast Pacific Street P.O. Box 357660, Seattle,
WA 98195-7660, Phone: (206) 905-9689; Email: songwang@u.washington.edu
Research Objective: The Washington State
Department of Health (DOH) Diabetes Prevention and Control Program, with Qualis Health and
Improving Chronic Illness Care (a national program of the Robert Wood Johnson Foundation), cosponsored the first state-level series of chronic disease collaboratives, the Washington State
Diabetes Collaborative (WSDC). Thirty primary care clinics participated in WSDC III, receiving collaborative training from November 2002 to
November 2003 to improve their diabetes care management. This evaluation examines the impact of WSDC III on health care costs for patients with diabetes.
Study Design: Nine eligible Collaborative clinics participating in WSDC III and nine eligible non-
Collaborative clinics participated. Non-Collaborative clinics were matched to Collaborative clinics on number of primary care providers. Information on eligible patients with diabetes was obtained from each participating clinics’ practice management system for the period November 2001 to November
2004 (1 year pre-, 1 year during, and 1 year post-
WSDC III training). We also recruited three leading private commercial health plans in Washington
State. For eligible patients linked to those health plans, utilization was derived from claims data. To eliminate unit price-related variation in total cost, we imputed standardized prices to each unit and type of utilization. We estimated a two-way analysis-ofvariance (ANOVA): (1) Collaborative vs. Non-
Collaborative clinic; (2) Year (pre- through 2 years post-Collaborative training). A companion mixedeffects regression was estimated, clustering patients on random effects for clinic and health plan, and modeling fixed effects and interactions for
Collaborative (intervention) status and Year, adjusting for age, gender, and plan product type
(non-HMO managed care, HMO, and FFS). Clinic questionnaires supplied data on organization characteristics and chronic care management practices.
Population Studied: Adult patients with diabetes with full-year enrollment (12 months) in any one of the health plans in any of the 4 years. Number of study subjects per year ranged from 268 to 312 in non-Collaborative clinics and from 79 to 104 in
Collaborative clinics.
Principal Findings: Preliminary analyses revealed that increase in age was associated with increase in total cost (P=0.0014), and the total cost was higher in non-HMO managed care (PPO, EPO, or POS) compared to fee-for-service (P<0.0001). The total cost was lower in HMO compared to fee-for-service but the difference was not significant. There was not a significant difference in total cost between
Collaborative and non-Collaborative patients.
Conclusion: Preliminary results showed no significant difference in total cost between
Collaborative and non-Collaborative clinics.
However, cost variation potentially attributable to unobservable differences in clinic characteristics and to patient comorbidities between Collaborative and non-Collaborative clinics will be explored in subsequent analyses through instrumental variables and clinic fixed effects.
Implications for Policy, Delivery or Practice:
Careful evaluation of sources of variation in the structure and process of chronic disease management for primary care practices is vital in design and implementation of quality improvement programs.
Funding Source(s): RWJF
Theme: Quality and Efficiency: Organized
Processes
● Association of Prescription Nonsteroidal Anti-
Inflammatory Drug (NSAID) Use with Patient-
Provider Communication & Perceived
Discrimination
Janice Weatherspoon, M.D.; Sandral Hullett,
M.D.,M.P.H.; Catarina Kiefe, M.D., Ph.D.; Jeroan
Allison, M.D., M.S.
Presented by: Janice Weatherspoon, M.D., Fellow,
Preventative Medicine, University of Alabama at
Birmingham, 521 Medical Towers, 1717 11th
Avenue South, Birmingham, AL 35294, Phone:
(205) 934-6838; Email: jlwspoon@uab.edu
Research Objective: Reported racial discrimination by patients and poor patient-provider communication has been recognized as obstacles to equitable health care and ultimately best outcomes. NSAIDs are commonly prescribed medications that can be potentially nephrotoxic especially in those with chronic kidney disease
(CKD) and in those with comorbidites (i.e. diabetes mellitus, hypertension and/or congestive heart failure) that ultimately place them at risk for developing CKD. Therefore, NSAIDs should be used with caution in this at-risk population.
Study Design: Data were collected from computer assisted in-person survey and medical record review. Reported discrimination was measured using the CARDIA Discrimination Questionnaire which establishes if an individual has experienced discrimination in various settings such as in healthcare, at school, at work, when applying for employment, etc. Patient-doctor communication was assessed with the Participatory Decision
Making Scale from the Medical Outcomes Study
(MOS). The MOS questionnaire consists of three questions (responses given a score of 1 to 4) used to assess patient participation in making health care decisions. Scores of 6 or less were classified as having poor patient-provider communication.
Exposure to prescription NSAIDs and diagnoses of comorbidities were also ascertained from medical record review.
Population Studied: We studied African-American patients (N= 788) treated in the outpatient setting at an inner-city, safety-net health setting.
Principal Findings: Of the 783 participants, 71% were female, and 12% were at least 65 years of age (mean age 54 years). Within the study group,
43% were diagnosed with diabetes mellitus, 77% with uncontrolled hypertension and 5% with congestive heart failure. Although 19% were noted to have physician recognized CKD, MDRD GFR was calculated using available laboratory data and
56% were found to have KDOQI Stages 2 and 3
CKD and 7% were found to have KDOQI Stages 4 and 5 CKD. Of those surveyed, 57% reported perceived discrimination, and 24% reported poor patient-provider communication. NSAIDs were prescribed to 26% of study participants. When controlling for sex, comorbidities and socioeconomic status (education and ability to pay for basic needs), patients who reported perceived discrimination (OR=1.48, p<0.041) and poor patient-provider communication (OR=1.53, p<0.042)) were significantly more likely to be prescribed NSAIDs.
Conclusion: Poor patient-provider communication and perceived discrimination are associated with prescription NSAID use. Identifying those who are more likely to be prescribed NSAIDs could aid in heightening physician awareness of prescribing practices and recognizing comorbidites that place those at risk for the development and/or progression of CKD.
Implications for Policy, Delivery or Practice:
Detecting and understanding flaws in patientprovider communication could potentially serve as an area for intervention to implement improved, effective communication, thus promoting safer prescribing practices.
Funding Source(s): National Heart Lung and
Blood Institute
● Overlooking Contextual Information when
Planning Patient Care: A Source of Medical
Error & Avoidable Cost
Saul Weiner, M.D.; Alan Schwartz, Ph.D.; Frances
Weaver, Ph.D.; Marilyn Schapira, M.D.; Rachel
Yudkowsky, M.D., M.H.P.E.; Neil Jordan, Ph.D.
Presented by: Saul Weiner, M.D., Associate
Professor of Medicine & Pediatrics, Medical
Services, Jesse Brown VA Medical Center &
University of Illinois at Chicago, 840 South Wood
Street, M/C 856, Chicago, IL 60612, Phone: (312)
519-2697; Email: sweiner@uic.edu
Research Objective: A “contextual error” is a medical error that occurs when a physician fails to take into account information that is expressed outside of a patient´s physical boundaries – i.e. their context – that is essential to planning appropriate care. All other medical errors may be classified as “biomedical.” In this study incognito standardized patients were trained to present with a common complaint, plus additional biomedical or contextual information that must also be addressed by their physician in order to avoid a medical error.
We examined the propensity of physicians to make both biomedical and contextual errors, and measured the avoidable direct costs to patient care.
Study Design: Using coding instruments developed for 4 cases, each with 4 variants, we scored 286 notes written by 96 internal medicine attending physicians who believed they were seeing new real patients in their primary care practice. To avoid making an error, a physician had to elicit and attend to essential biomedical or contextual information planted in each case. For each error identified, we tabulated each inappropriately ordered test, medication, or other medical service that was a direct consequence of the error and, using Medicare cost-based reimbursement data summed the overall costs of each error.
Population Studied: Subjects for this study were primary care internal medicine specialists in a
variety of practice settings in two metropolitan areas.
Principal Findings: At least one error was made in
61% of cases where only a biomedical error was possible, 73% of cases where only a contextual error was possible, and 89% of cases where physicians could make both types of errors. This trend was statistically significant (p = 0.05).
Averaged across all 16 case variants, the cost of error making to patient care was $326 in unnecessary or inappropriate care. For variants in which only contextual errors were possible, error making added $725 to the per visit cost of care.
Conclusion: Inattention to contextual information, such as patients´ transportation, economic situation, or caretaker responsibilities can be even more costly than inattention to laboratory values, medication dosages, and patient identifiers when delivering care. Unannounced standardized patients provide a modality for comparing and assessing providers’ propensity to overlook essential patient context.
Implications for Policy, Delivery or Practice:
This study demonstrates the value of directly observed structured assessment of provider performance, uncovering a need for greater prioritization of contextual information in planning patients´ care to reduce medical errors and costs.
Funding Source(s): VA
Theme: Quality and Efficiency: Measurement
● Understanding Racial/Ethnic Disparities in
Patient Experiences with Care
Robin Weinick, Ph.D.; Angelo Volandes, M.D.,
M.P.H.; Lenny Lopez, M.D., M.P.H., M.Div.; Mark
Schlesinger, Ph.D.
Presented by: Robin Weinick, Ph.D., Senior Social
Scientist, RAND, 1200 South Hayes Street,
Arlington, VA 22202, Phone: (703) 413-1100 x
5151; Email: rweinick@rand.org
Research Objective: A growing body of research demonstrates that patients from different racial/ethnic groups report differing experiences with the health care system. This project assessed the extent to which these differences in ratings of patient experiences with care represent true differences as compared to differences in how the same care is perceived, attributed reasons for ratings of care, or psychometric properties of the scales for different racial/ethnic groups.
Study Design: This pilot study asked respondents to read a series of 5 health-related vignettes and respond to CAHPS questions from the communication composite regarding their content; to view a 5-minute video of a doctor-patient interaction and answer a series of related CAHPS questions; and to answer a series of questions regarding perceptions of the doctor in the video.
Prior to viewing the video, respondents received experimental priming designed to stimulate positive, negative, or neutral emotions via a short verbal and pictorial description of non-healthcare related events.
Population Studied: A random sample of white
(n=204), black (n=163), and Latino (n=200) adults selected from a nationally-representative online panel.
Principal Findings: Responses to the physician’s behavior in the written vignettes varied systematically based on their content. However, there were no significant racial/ethnic differences in responses to the vignettes. Overall, respondents from different racial/ethnic groups provided similar assessments of the video. However, black respondents who received the negative emotional priming consistently rated the doctor more positively, were more likely to recommend the doctor, and were more likely to believe that the doctor exhibited positive behaviors. These differences did not occur for white or Latino respondents receiving the negative priming.
Conclusion: Our results suggest that members of different racial/ethnic groups are likely to use the
CAHPS responses scales similarly for the communication composite questions (the doctor listening carefully, showing respect, and spending enough time with the patient). In addition, our findings suggest that medical experiences may carry a higher emotional valence for black than for white patients. Since the experimental priming materials were drawn from experiences outside of health care, the influence of negative life experiences differentially altering the evaluation of medical encounters is likely quite prevalent even among patients with limited health care exposure.
Implications for Policy, Delivery or Practice: At the same time that tools to measure patients’ experiences with care have evolved, health care plans and providers as well as the organizations that regulate and accredit them are increasingly focused on reducing racial and ethnic disparities in health care. Assessing disparities in patients’ experiences with care is likely to play a significant part in such activities in the future, but these assessments remain somewhat problematic. This study provides the first plausible explanation for previous findings that black adults report higher satisfaction despite having more problematic experiences with care. However, there may be a number of other ways in which non-medical experiences and contexts differentially shape the evaluation of medical care for minority and nonminority patients.
Funding Source(s): RWJF
Theme: Disparities
● Unmet Need Among Children with Special
Health Care Needs: Results from the 2005/6
National Survey of Children with Special Health
Care Needs
Wendy Weller, Ph.D.
Presented by: Wendy Weller, Ph.D., Assistant
Professor, Health Policy, Management, & Behavior,
University at Albany, State University of New York,
School of Public Health, One University Place,
Room 167, Rensselaer, NY 12144, Phone: (518)
402-0302; Email: wweller@albany.edu
Research Objective: This study uses data from the
2005/6 National Survey of Children with Special
Health Care Needs (2005/6 NS-CSHCN) to examine unmet need for 13 medical and health services often required by children with special health care needs (CSHCN).
Study Design: This is a cross-sectional study based on the 2005/6 NS-CSHCN.
Population Studied: The study population included
40,840 children identified as having a special health care need based on parental response to a previously validated CSHCN screener that is consistent with the MCHB definition of CSHCN.
Principal Findings: The prevalence of unmet need among CSHCN was highest for communication aids
(23.9%), substance abuse treatment (20.7%), mental health care (15.0%), physical/occupational/speech therapy services
(13.6%), and home health (10.6%). Unmet need was lowest for medical supplies (2.5%), routine preventive care (2.4%), and prescription medications (1.8%). With the exception of therapy services and home health, the prevalence of unmet need remained approximately the same or declined slightly between the 2001 and the 2005/6 NS-
CSHCN. Although mental health care was among the services with higher unmet need in 2005/6, it was the service with the most noticeable decline in unmet need between the two survey years (18.1% vs. 15.0%). Multiple regression analysis revealed that type of insurance, income, having a usual source of care, and health status are significantly associated with unmet need for multiple types of services. Compared to CSHCN in nonpoor families,
CSCHCN in poor families were significantly more likely to report an unmet need for 9 of the 13 services examined, while those in near poor families were significantly more likely to report an unmet need for 7 of the 13 services. Children who were uninsured for any of the 12 months prior to the survey were significantly more likely to report an unmet for 10 of the 13 services compared to children who were insured for the entire 12 months.
For some services, Medicaid and SCHIP provided a protective effect. CSHCN covered by either
Medicaid or SCHIP were significantly less likely to report an unmet need for therapy services and medical supplies compared to privately insured children. CSHCN covered by Medicaid were also less likely to report an unmet need for specialty care and mental health care; those covered by
SCHIP were also significantly less likely to report an unmet need for vision care. CSHCN without a usual source of care were significantly more likely to report an unmet need for routine preventive services, prescription drugs, mental health care, medical supplies, and vision care. Health related need factors were strong predictors of unmet need across multiple services in the fully-adjusted models. Overall, as severity increased the odds of unmet need increased.
Conclusion: Unmet need continues to be a concern for the most vulnerable CSHCN (e.g., lowincome, uninsured).
Implications for Policy, Delivery or Practice:
Policies are needed that address both the financial and nonfinancial barriers to care to meet the needs of CSHCN. As more children lose private insurance and states look for ways to reduce Medicaid spending, monitoring unmet need among this population is especially important during the current economic downturn.
Theme: Child Health
● Has the President’s Health Center Initiative
Improved Behavioral Health Care Access for the
Underserved?
Rebecca Wells, Ph.D., M.H.S.A.; I-Heng Lee, M.A.;
Andrea Radford, Dr.P.H.; Joseph Morrissey, Ph.D.
Presented by: Rebecca Wells, Ph.D., M.H.S.A.,
Associate Professor, Health Policy & Management,
University of North Carolina, 7411 McGavran-
Greenberg Building, Chapel Hill, NC 27514, Phone:
(919) 966-6961; Email: rwells@unc.edu
Research Objective: Community health centers have played an increasingly important role in behavioral health care for the underserved as funding for mental health centers has diminished.
This investigation compared trends in mental health and substance abuse treatment provision by federally funded community health centers before and during a 2002-2007 expansion initiative that almost doubled federal support for these facilities.
Study Design: Regressions tested associations between time and community health centers’ odds of providing routine mental health care, 24 hour crisis mental health care, and substance abuse treatment. Student’s t-tests compared trends in community health centers odds of providing each service before and during the initiative. Data on the numbers of encounters and users were not available at the center level for enough years to test the statistical significance of these trends.
Population Studied: Data from the US Health
Resources & Services Administration’s Uniform
Data System for each year between 1998 and 2007 were used to identify trends in service provision by the universe of federally funded community health centers. In 1998, there were 694 community health centers. By 2007, this number had risen to1,067.
Principal Findings: In 2007, 77% of community health centers reported on-site provision of routine mental health services and 51% provided substance abuse treatment, continuing previous upward trends. The percentage of health centers
providing crisis intervention remained steady, at
20%. The number of specialty mental health encounters provided across all community health centers increased 10% annually between 1998 and
2001, from 913,828 to 1,223,408, and 13% annually between 2001 and 2007, to 2,738,408 in 2007. The number of substance abuse encounters increased
9% annually between 1998 and 2001, from 571,496 to 745,855, and 6% annually between 2001 and
2007, to 972,857. Numbers of visits per patient with any given condition as the primary diagnosis remained generally steady, averaging 2.92 for mental health care, 4.90 for alcohol-related treatment, and 7.28 for other substance abuse treatment across the population of health center patients in 2007.
Conclusion: The recent health center growth initiative accelerated pre-existing trends toward increased on-site behavioral health service provision largely by increasing the number of federally funded community health centers.
However, on-site provision of crisis intervention services remained relatively uncommon, and the numbers of encounters per person may be inadequate.
Implications for Policy, Delivery or Practice:
Providing crisis services might be particularly difficult in areas with mental health care professional shortages, especially where travel distances are great. There are rural areas in which the only affordable solution is to keep the hospital as the crisis mental health care provider. Less remote community health centers may better serve people with behavioral health care needs by providing crisis services at their facilities. Trends in
Medicaid budgets and state and federal insurance parity requirements for private insurance are likely to affect reimbursement for services and thus the financial sustainability of on-site care. Recent
California legislation earmarking 1% of personal income taxes for the wealthy to public mental health care is a promising model for increasing both the levels and predictability of funding, which are essential to supporting adequate service provision.
Funding Source(s): NIMH, UNC School of Public
Health
Theme: Behavioral Health
● SIPP Estimates of Coverage in NY State
James Welsh, Ph.D.; Shelley Mae Drazen, Ph.D.,
M.Sc.
Presented by: James Welsh, Ph.D., Principal
Policy Specialist, Division of Coverage &
Enrollment, New York State Department of Health,
1603 Corning Tower, Albany, NY 12237, Phone:
(518) 473-7541; Email: jbw01@health.state.ny.us
Research Objective: To estimate coverage and enrollment rates and transitions into and out of public health insurance in NY State and thereby demonstrate the benefits of the SIPP compared to the CPS in state-level analysis.
Study Design: Monthly, annual and biennial health insurance estimates derived from SIPP respondents in NYS are compared to CPS estimates and to Medicaid/SCHIP administrative records for the period 2004-2005. (2006 may be added for draft presentation).
Population Studied: Children and adults in NYS, with emphasis on children. (Adult comparisons may be added for draft presentation).
Principal Findings: The SIPP estimates an average monthly uninsured rate of 10.9% for New
York State children under 19 in 2004 and 2005 and an annual ("full-year") uninsured rate of 2.2%, averaged for 2004 and 2005. The CPS estimated rate of 7.6% falls between the two, as we might expect if some respondents misinterpret the coverage questions to refer to current coverage.
Compared to the CPS, the SIPP's monthly estimates show a higher proportion of coverage in public programs (Medicaid and SCHIP) and lower enrollment in private coverage. When we compared the SIPP and CPS estimates to unduplicated caseload counts, we find the SIPP to be closer to the administrative figures. Preliminary results for calendar year 2005 show the CPS estimate is 25% lower than the number of children "ever enrolled' in the year, while the SIPP estimate is within 3% of the caseload data. Comparing selected months in
2004 and 2005, it appears that the SIPP point-intime measure underestimates the caseload by about 11% while the CPS (annual) measure interpreted as a point-in-time measure would miss about 19% of the actual enrollment. The SIPP shows that about half the state's children were enrolled in Medicaid or SCHIP at some time during
2004-2005 and preliminarty results for adults show a 25% 'reach' into the adult population in that period. Although total eligibility estimates are similar using the SIPP's monthly income data and the CPS annual income data, the SIPP estimates a larger share of the eligible children falling under
Medicaid (as opposed to SCHIP): 77% of eligibles compared to 64% of eligibles using CPS data. The
SIPP estimates that two-fifths of children in the state experienced a change in insurance status during 2004-2005, meaning they moved between or among public coverage, private coverage or no coverage. Nearly one in four experienced at least two changes over the two years.
Conclusion: Using SIPP for state-level estimates reveals greater public program "reach" into the eligible population and considerable turnover in participation. SIPP estimates are closer to the confirmed caseload.
Implications for Policy, Delivery or Practice: For large states, the SIPP may offer a refined view of eligibility, enrollment, and caseload dynamics and an alternative to ad hoc adjustment of the CPS
"undercount" of public program participation.
Theme: Coverage and Access
● “eSamples” as Direct-to-Consumer
Advertising: An Evaluation of Consumer-
Targeted Offers for Free or Discounted Drugs on the Internet
William Weppner, M.D.; Matthew Hollon, M.D.,
M.P.H.; Lisa Chew, M.D., M.P.H.
Presented by: William Weppner, M.D., Senior
Fellow, National Research Service Award Primary
Care Fellowship, Medicine, University of
Washington, 10324 13th Avenue Northwest,
Seattle, WA 98177, Phone: (206) 543-0592; Email: wweppner@u.washington.edu
Research Objective: Prescription drug samples are a traditional marketing strategy for pharmaceutical companies. However, this strategy is being used less due to concerns regarding the undue influence of drug samples on providers to select newer and more expensive brand name medications, as well as diversion for personal use and difficulty in tracking samples. At the same time, pharmaceutical companies are promoting coupons for discounted or free samples of brand name drugs, delivered directly to the consumer via the internet. This study evaluates the prevalence and characteristics of online offers for free or discounted prescription drugs by pharmaceutical companies.
Study Design: We performed a structured internet search and content evaluation of manufacturersponsored websites for the top 50 selling brand name medications of 2007. Using the Google search engine to identify websites, three independent reviewers evaluated them for the presence of an offer for free or discounted medications and for the characteristics of each offer.
Principal Findings: Of 50 brand name medications, 92% had a manufacturer-sponsored website. 60% had one or more offers of vouchers for free or discounted medications. The most common offer was a discount (24 of 35 different offers), while 11 were vouchers for free samples.
The average value of the initial discount was $20.21
(range $5-50); the average number of days of a free treatment for each voucher was 22.2 days (range 7-
30 days). 80% (28 of 35) of offers were prominently featured on the initial page of the website. On the webpage with the link to the offer, safety information was almost always presented in a less prominent fashion than the link to the drug offer
(96.7% of offers). 68.8% (24 of 35) of the websites required recipients input some amount of personal contact information prior to receiving the voucher.
The presence of a generic alternative was significantly associated with an offer for a voucher for free or discounted medications (Pearson Chi2 =
8.1, p=0.004).
Conclusion: Consumers accessing websites of popular brand name pharmaceuticals are likely to encounter prominently displayed offers for free or discounted pharmaceuticals. Safety and side effect information are not as well represented when compared to this offer. Among popular brand name medications, consumers are more likely to encounter an offer if there is a generic alternative to the medication. Little is known about potential for harm or benefit of this emerging marketing strategy.
Implications for Policy, Delivery or Practice: For health care providers, this represents a previously undescribed form of direct-to-consumer advertising, which combines internet marketing and consumerdirected incentives to use specific prescription drugs. Providers should know this practice is more common when generic alternatives may exist, and that safety information may not be prominently displayed related to these offers. Given the high prevalence of such offers on pharmaceutical websites, further investigation should explore the amount of patient use of such vouchers, consumers’ perception of risk/benefit related to them, and the effects of this practice on patient well-being.
Funding Source(s): HRSA
Theme: Health Care Markets and Competition
● Medicare Part D Medication Access Problems
& Suicidal Ideation & Behavior among
Psychiatric Patients
Joyce West, Ph.D., M.P.P.; Joshua Wilk, Ph.D.;
Donald Rae, M.S.; Eve Moscicki, Ph.D.; Maritza
Rubio Stipec, Sc.D.; Darrel Regier, M.D., M.P.H.
Presented by: Joyce West, Ph.D., M.P.P., Senior
Scientist, American Psychiatric Institute for
Research & Education, Psychiatric Research
Network, 1000 Wilson Boulevard, Suite 1825,
Arlington, VA 22209, Phone: (703) 907-8619;
Email: JWest@psych.org
Research Objective: 1) Assess the relationship between medication access/continuity problems and suicidal ideation and behavior among psychiatric patients with Medicare and Medicaid insurance during the first year of the Part D benefit; and 2) Compare rates of suicidal ideation and behavior to other adverse event rates (e.g., emergency room visits, hospitalizations, homelessness, and incarcerations) previously reported for patients with and without different types of medication access problems.
Study Design: Psychiatrists were randomly selected from the AMA Masterfile. After excluding those not practicing and with undeliverable addresses, 1,556 (62%) responded; 63% met study eligibility criteria of treating dual eligible patients and reported clinically detailed information on one systematically selected patient (N=986).
Population Studied: A nationally representative sample of psychiatric patients with Medicare and
Medicaid insurance ("dual eligibles") treated by psychiatrists in routine practice settings in the US
(N=986 patients).
Principal Findings: Overall, 16% of patients had an increase in suicidal ideation or behavior during the past year. Patients with medication access problems had four times the rates of suicidal ideation or behavior compared to patients with no access problems (22% vs. 5%, P<.0001). After adjusting for patient sociodemographics, treatment setting, diagnoses, and psychiatric symptom severity, the predicted probabilities of increases in suicidal ideation and behavior were significantly greater for patients who: were previously stable, but were required to switch medications (29%, P<.001); couldn’t access medications because they were
“off-label” (29%, P<.001); had problems accessing benzodiazepines (26%, P<.01); couldn’t access refills or new prescriptions because they were not covered (24%, P<.001); or stopped or discontinued their medications because they were not covered
(23%, P<.05), compared to patients without these access problems (5%).
Conclusion: Dual eligible psychiatric patients with medication access problems had significantly higher rates of suicidal ideation or behavior, as well as other adverse events previously reported.
These observational findings warrant investigation as they may indicate more clinically vulnerable patients with elevated rates of suicidal ideation and behavior may be at greater risk for medication access problems under Medicare Part D – or medication access problems occurring under
Medicare Part D may put these patients at greater risk for suicidal ideation or behavior and other adverse events.
Implications for Policy, Delivery or Practice: As patients with medication access problems had higher rates of suicidal ideation and behavior and other adverse events previously reported (including emergency room visits and hospitalizations), these patients should receive increased clinical attention and more effective care management to help improve outcomes of care and reduce adverse events.
Funding Source(s): Funded by a grant from the
American Psychiatric Foundation through a consortium of industry supporters, including Astra
Zeneca, Bristol Myers Squibb, Eli Lilly, Forest,
Janssen, Pfizer and Wyeth. APIRE had complete discretion and control over the design and conduct of this study and the reporting of findings.
Theme: Medicare
● High Quality Health Care for All? The Dutch
System; A Possible Model for the U.S.?
Gert Westert, M.Sc., Ph.D.; Jako Burgers, M.D.,
Ph.D.
Presented by: Gert Westert, M.Sc., Ph.D.,
Professor, PZO, National Institute of Public Health
& the Environment/ Tilburg University, A. Van
Leeuwenhoeklaan 9, Bilthoven, 3721 MA, NL,
Phone: +31302743934; Email: gert.westert@rivm.nl
Research Objective: Since the reform in 2006 the
Dutch health care system is considered a possible model for the United States and other countries.
Potentially attractive features of the system are: universal mandatory health insurance coverage from private insurers; a national risk equalization fund to compensate higher costs from predefined high-risk patients; and managed competition in curative health care services driven by quality performance and efficiency. Assessment of the system’s performance over time is needed to provide evidence of quality improvement and cost reduction.
Study Design: In 2004 the Dutch government commissioned the National Institute of Public
Health and the Environment to assess the system’s performance every two years. The Dutch Health
Care Performance Report monitors the accessibility, cost and quality of the Dutch system, using approximately 100 indicators, including international comparisons.
Population Studied: The Netherlands; population
16 million.
Principal Findings: The Netherlands has a very accessible health care system. Out-of-pocket payments in the Netherlands are low (8%).
Confidence in the affordability of the necessary care is high and very few people did not visit a doctor because of costs. Participation in the national vaccination and screening programs is excellent.
Many elements of curative care are satisfactory and have improved; however, the Scandinavian countries consistently score better. The general public and care users are positive about the care provided, but there are differences between the various types of care. Health care provided by general practitioners is highly appreciated. Mental health care, nursing homes, and residential homes have the lowest level of confidence. Care coordination and patient safety score relatively low.
Since 2004, health care expenditure has risen annually by 5%. This rate of growth is comparable to that of neighboring countries. The system reforms led to lower insurance premiums for consumers, but have still not resulted in demonstrable changes in the quality, accessibility and costs of care at the macro-level. Competition between health insurers in the care purchasing market is limited by lack of information on quality of care. A lack of such information could lead to quality losing out to competitive prices.
Conclusion: The Netherlands health care system provides health care for all and quality of care is high on most domains. The effects of the system reforms are limited, so far. The Dutch government continues to introduce incentives to encourage managed competition with the aim to further improve quality and efficiency.
Implications for Policy, Delivery or Practice: The
Dutch health care system can be considered as an example for other countries to follow. However, the
model is work in progress. It is important to continue monitoring quality, cost, equity, and outcomes over time and to compare the results with other countries to determine the pros and cons.
This is of great interest to those considering adopting bits and pieces from the Dutch system.
Funding Source(s): Dutch Ministry of Health
Theme: Coverage and Access
● Alabama School Nurses' Perceptions of the
Increase in English-As-Second-Language
Students: The Case for Cultural & Linguistic
Competence
Marilyn Whitman, Ph.D.; Jullet Davis, Ph.D.; Allison
Terry, Ph.D., M.S.N., R.N.
Presented by: Marilyn Whitman, Ph.D., Lecturer and Program Coordinator, Management &
Marketing, University of Alabama, Box 870225,
Tuscaloosa, AL 35487, Phone: (205) 348-8934;
Email: mwhitman@cba.ua.edu
Research Objective: The increasing diversification of our nation’s student population poses a number of challenges for school nurses. English-as-asecond-language (ESL) students constitute the fastest growing population in U.S. schools. In
Alabama, specifically, the ESL student population more than doubled between 2000 and 2005 representing 2.1 percent of the total student population. This increase may create unique challenges for school health services with the introduction of new cultures and languages. This paper examined the perceptions of school nurses in
Alabama schools of the increase in ESL students and the challenges they face in providing nursing care.
Study Design: School nurses working in Alabama public schools were asked to complete a survey administered by the Alabama Board of Nursing.
Population Studied: A sample of 1,211 responses was analyzed bivariately using Pearson Chi-Square and Spearman’s Rho Nonparametric Correlation.
Principal Findings: Over 43 percent of respondents indicated they had experienced an increase in ESL students in the past year. Over a third of respondents stated they had encountered difficulty communicating with ESL students and over half encountered difficulty communicating with the parents of ESL students. Although nearly 72 percent of respondents indicated that they did have access to a trained interpreter(s), over 50 percent of had relied on an ESL student to act as an interpreter when speaking to their parents.
Conclusion: As the student population continues to diversify, the need for school nurses to have the necessary training and resources in order to provide care that is culturally and linguistically congruent for ESL students and their parents can no longer be ignored. In order to provide the appropriate care to culturally diverse patients, nurses must understand how the various cultural groups they serve define health and illness, what they believe to be the acceptable forms of treatment and intervention, and their preferences for care delivery. Furthermore, it is crucial that nurses use interpreter services in order to properly communicate with ESL students and parents that may be limited English proficient.
Implications for Policy, Delivery or Practice: In many cases, school nurses may be the only source of care for ESL students. This presents school nurses with a unique opportunity to engage parents of ESL students in the health and health care of their child. It is important, therefore, that school nurses be provided with cultural facts and be encouraged to use interpreter services in order to lessen the barriers to communication.
Theme: Child Health
● Diversity Training as a Strategic Response to
Cultural Changes in the Long-Term Care
Environment
Marilyn Whitman, Ph.D.; Jullet Davis, Ph.D.; Diane
Brannon, Ph.D.
Presented by: Marilyn Whitman, Ph.D., Lecturer and Program Coordinator, Management &
Marketing, University of Alabama, Box 870225,
Tuscaloosa, AL 35487, Phone: (205) 348-8934;
Email: mwhitman@cba.ua.edu
Research Objective: As the nation’s workforce continues to diversify, organizations will increasingly be challenged to effectively manage diversity to improve working relationships. Thus far, organizations have responded to this challenge by implementing diversity management programs.
Often, these programs offer employee training designed to increase awareness and change attitudes and behaviors. Yet, research on the organizational characteristics associated with diversity training in health care is in its nascent stage. This study seeks to examine the relationship of diversity training to the organizational characteristics of long-term care providers, and whether these characteristics are associated with the percent of staff receiving diversity training.
Study Design: We utilized Strategic Human
Resource Management (SHRM) as the theoretical framework to examine diversity training and its relationship to factors associated with organizational development and performance. A two-stage model was used. Logistic regression analysis was used to determine the odds of offering diversity training given the organizational characteristics of long-term care providers. Ordinary least squares regression analysis was used to determine the relationship between the organizational characteristics of long-term care providers and the percent of staff receiving diversity training.
Population Studied: Data for this project come from a listing of Pennsylvania long-term care
providers in 2004. The total eligible sample consisted of 1,924 providers. Surveys were sent to clinical managers or administrators of these organizations. A sample of 759 providers, resulting in a response rate of 39.4 percent, was obtained.
Principal Findings: Cases with missing data were excluded from the study resulting in a final sample of 436 providers. Analysis revealed that the majority of long-term care organizations offered training, yet less than half of direct care workers (DCWs) received training. A greater number of DCWs in larger organizations received training.
Organizations that had a higher percentage of diverse residents/customers, offered training on effective communication, and linked training to performance appraisals were more likely to offer training.
Conclusion: Effectively managing a diverse workforce is indeed challenging. It requires a strategically driven process with an emphasis on increasing awareness and developing skills.
Implications for Policy, Delivery or Practice: A strategic approach to diversity management requires that organizations adopt a long-term approach, have measurable goals, enlist the support of senior management and community stakeholders, develop targeted recruitment programs, provide ongoing training programs, and incorporate a feedback system. Moreover, it is important that health leaders refrain from simply adopting diversity management policies that are compliance-oriented. Managing diversity solely for compliance purposes will never result in the organization maximizing the benefits that diversity offers. Organizations must go beyond surface changes and allocate the necessary time and resources to sustain long-term initiatives.
Theme: Health Care Workforce
● A Program of All-Inclusive Care for Elderly
[PACE] Improves Five-Year Survival Compared to Home and Community-Based Care
Darryl Wieland, Ph.D., M.P.H.; Rebecca Boland,
M.A.; Bruce Kinosian, M.D.
Presented by: Darryl Wieland, Ph.D., M.P.H.,
Research Director, Geriatrics Services, Palmetto
Health Richland, 3030 Farrow Road, #300A,
Columbia, SC 29203, Phone: (803) 434-4330;
Email: darryl.wieland@palmettohealth.org
Research Objective: To compare baseline characteristics and to assess five-year survival of
PACE v. home- and community-based waiver
[HCB] and nursing home [NH] entrants, controlling for mortality risk using a previously validated risk index.
Study Design: Longitudinal long-term care [LTC] admission cohort study. Data were derived from state client assessments, utilization and vital records. Subjects were followed until death or 5year survival; those lost to follow-up for the event, or surviving < 5 yrs on 8/8/2005, were censored.
Analyses included descriptive contrasts at admission, and survival curves tested with log-rank statistics. Mortality risk was measured using the PPI
(Carey et al. JAGS 2008; 56: 68-75).
Population Studied: Subjects (n= 2081) were residents of two counties in South Carolina admitted to Medicaid LTC between 1998 and 2003 to HCB (n=1020), PACE (562) and NH (499).
Principal Findings: Compared to HCB and NH,
PACE admissions were older (77.2+.42 v. 74.5+.32 and 75.0+.49 years respectively) and more African
American (70.5% v. 49.3 and 46.9). Mortality risk for PACE admissions (72.6% “high-tointermediate”) was significantly greater than for
HCB (58.8%), and comparable to NH admissions
(71.6%). Median survival in the PACE cohort was
4.2 years, v. 3.5 in the HCB cohort (log rank = .394; p=0.53). Median survival of the NH cohort was 2.3 years. With risk stratification, the PACE advantage over HCB is significant (log rank = 5.941; p=0.015).
Risk-stratum analyses show that intermediate and high-risk admissions are most likely to benefit
(intermediate risk: PACE median survival = 4.7 years v. HCB 3.4; high risk: PACE 3.1 years v. HCB
2.0).
Conclusion: Adjusting for admission risk, a PACE program produces a substantial long-term survival advantage compared to HCB. That the benefit seems most apparent in intermediate- to high-risk admissions suggests the particular importance of integrated, team-managed care for the older, more disabled participants who tend to be admitted to
PACE.
Implications for Policy, Delivery or Practice:
Community-based [CB] care is preferred to NH placement for Medicaid-eligible people certified for
LTC. States have widely adopted HCB programs, while PACE--prepaid, dually-capitated, communitybased comprehensive care--is available as a
Medicaid option in 30 states. States’ uptake of
PACE has been steady but slow, in part because the immediate, higher costs of PACE capitation are clear, relative to what are seen as lower cost inputs of HCB (i.e., case management and limited personal care). Questions linger regarding the longer-term costs and benefits of PACE relative to
HCB or NH placements. Our results suggest such questions need to be addressed.
Funding Source(s): SC Dept. of Health and
Human Services
● Assessing the Impact of Disability on
Achieving an Adequate Standard of Living:
Barriers, Resource Needs, and Priorities
Laura Wilkinson-Meyers, M.S., Ph.D.
Presented by: Laura Wilkinson-Meyers, M.S.,
Ph.D., Senior Tutor, Health Systems, University of
Auckland, School of Population Health, Private Bag
92019, Auckland, NZ, Phone: +0064 9 373-7599;
Email: l.wilkinson-meyers@auckland.ac.nz
Research Objective: Research suggests disabled people will not achieve a similar standard of living as non-disabled people with the same income because they require additional disability-related resources such as home modifications, equipment and support. Without careful consideration of the barriers disabled people face and the additional resources required to reduce them, benefit payments based on income thresholds for the general population will be insufficient to ensure disabled people achieve an adequate standard of living. A previous phase of this doctoral study constructed consensual budget standards for disabled New Zealanders to identify the additional equipment, support, transport and time that would be required by people with different types of impairment to achieve an adequate standard of living in the community. The purpose of this phase of the research is to assess the degree to which disabled people actually achieve this standard, the key barriers they encounter, their met and unmet disability-related resource needs, and their priorities of which resources are most useful in helping them achieve an adequate standard of living in the community.
Study Design: 200 people with physical impairment were recruited through a nationwide network of disability service providers to participate in the Impact of Physical Impairment Survey. The questionnaire was developed in partnership with members of the disabled community and was based on consensual budget standards for disabled people that were created in an earlier phase of the research. The survey tool uses a social model of disability framework to assesses barriers, resource needs and priorities across areas of life including activities of daily living, food and clothing, home modifications, participation, recreation and information. The questionnaire was offered in a web-based format as well as a paper-based version and included questions about participants’ quality of life (WHOQOL-BREF), degree of impairment
(WHODAS II), standard of living (New Zealand
Economic Living Standards Index), as well as an indication of the extent to which participants are disabled by social and environmental barriers in the community.
The methodology for this Ph.D. study was presented at last year’s AcademyHealth Disability
Interest Group Meeting.
Population Studied: 200 working aged (18-64) disabled people with a physical impairment who live across New Zealand.
Principal Findings: Data is currently being collected and analysis is due to be complete by the end of May 2009.
Conclusion: This study will provide valuable insight into the relationship between social and environmental barriers, access to disability-related resources and the standard of living achieved by disabled people in the New Zealand context. As a project developed in partnership with the disabled community, the findings will also help validate a new methodological approach for estimating the additional resources and costs associated with disability.
Implications for Policy, Delivery or Practice:
Estimating the standard of living achieved by disabled people and the extent to which their disability related resource needs are met has significant implications for assessing the adequacy of income support, the provision of public programs and services, and current legislation for disabled people.
Funding Source(s): Health Research Council of
New Zealand
● Mindful Staffing on Nursing Units: A
Qualitative Description of the Role of Charge
Nurses
Deleise Wilson, B.S.N., M.A., R.N.; AkkeNeel
Talsma, Ph.D., R.N.; Kristy Martyn, Ph.D., R.N.
Presented by: Deleise Wilson, B.S.N., M.A., R.N.,
Doctoral Student, Nursing, University of Michigan,
11911 Southwind Way, Yucaipa, CA 92399,
Phone: (909) 810-7511; Email: wilsonsh@umich.edu
Research Objective: The impact of the nurse practice environment on patient outcomes has been well documented. One of the findings is that the nurse-patient staffing ratio has influenced the quality of patient care provided in healthcare organizations. Studies have established associations among staff nurses’ professional behaviors, staffing and the nurse practice environment but there is paucity about the specific role of charge nurses. There is a lack of uniformity of the definition, functions, and employment of charge nurses across hospital units, hospitals, and regions around the nation. The purposes of this study are to: (1) describe the professional characteristics of charge nurses; (2) illustrate the organizational definitions and employment of charge nurses; (3) describe the behaviors charge nurses enact to effectively manage the intra-shift staffing needs of nursing units.
Study Design: Qualitative descriptive method utilizing semi-structured interviews focused on charge nurses’ decision-making in the role of managing the intra-shift staffing needs of medicalsurgical and intensive care units. Constant comparison was conducted with selective established literature about the work of charge nurses and other professionals in similar practice environments to verify the themes that emerged.
Population Studied: This study was conducted at a large Southeastern Michigan hospital. A purposive consented sample of 24 nurses included
11 charge nurses, 7 managers, and 6 staff nurses who did not have charge nurse experience. The nurses represented 4 medical-surgical, 1 step-
down, 2 intensive-care, 7 adult and 4 pediatric units. The nurses’ length of time on the units ranged from 4 months to over 20 years.
Principal Findings: Using qualitative descriptive method of content analysis it was revealed that the charge nurse can spend 50 to 95 % of a shift resolving staffing issues. Within the healthcare organizations there were different structures of the charge nurse role. Charge nurses who effectively made intra-staffing decisions exhibited the following behaviors: (1) resourcefulness, (2) tactful communication, (3) flexibility, (4) decisiveness, and
(5) constant awareness of the “big picture.”
Conclusion: Charge nurses are critical to improvements in the nurse practice environment and the quality of patient care because their staffing decisions affect the staff nurses’ workload as well as patient outcomes. Effective charge nurses reduced incidences of chaos on the patient care units. That is, by being attentive to subtle changes in patients’ acuity and fluctuations in census.
Charge nurses then seamlessly reassign staff nurses based on their skill level and workload capacity. Comparison with the literature on high reliability organizing concept of mindfulness suggests that charge nurses make staffing decisions by exhibiting similar behaviors of reliability professionals such as nuclear and electrical power plant dispatch operators.
Implications for Policy, Delivery or Practice: The findings of this study contribute to the improvement of the nurse practice environment by stressing the importance of the charge nurses’ role in the quality of care provided in healthcare organizations.
Targeted interventions can be developed to improve the staffing decisions of charge nurses. A better understanding of the behaviors of charge nurses contributes to the refining, and testing of high reliability concepts for adaptation to healthcare settings.
Funding Source(s): Society of Nurse Scholars,
University of Michigan
Theme: Health Care Workforce
● Racial Differences in Financial Burden
Associated with Chronic Illness
Karen Winters, Ph.D.
Presented by: Karen Winters, Ph.D., Assistant
Professor, School of Nursing, University of
Mississippi Medical Center, 2500 North State
Street, Jackson, MS 39216-4505, Phone: (601)
984-6246; Email: kwinters@son.umsmed.edu
Research Objective: To examine racial differences high financial burdens associated health care expenses and financial burden among families with and without chronic conditions
Study Design: An observational, exploratory design was used to investigate racial differences in financial burden associated with health care expenses for families with and without chronic illness. Financial burden per $1000 health care expense was calculated by dividing the total out of pocket expenses paid by the family divided by the total family income then multiplying by 1000.
Multivariable linear regression was used to examine the effect of race and chronic illness status on financial burden while controlling for sociodemographic characteristics that have been shown to influence health care utilization.
Population Studied: 78,000 families who were participants in the 2000-2005 Medical Expenditure
Panel Survey (MEPS).
Principal Findings: Pending
Conclusion: Pending
Implications for Policy, Delivery or Practice:
Pending
Funding Source(s): RWJF
● Determinants of Post-Acute Care Use &
Setting among Elderly Trauma Patients
Lok Wong, M.H.S.
Presented by: Lok Wong, M.H.S., Doctoral
Candidate, Health Policy & Management, Johns
Hopkins University, 624 North Broadway, Hampton
House Room 624, Baltimore, MD 21205, Phone:
(202) 210-3332; Email: lokwong@jhsph.edu
Research Objective: This study seeks to identify characteristics of elderly trauma patients and health-system factors such as the acute hospital and local post-acute care supply that are associated with receipt of post-acute care and the type of post-acute setting - from inpatient rehabilitation facilities, skilled nursing facilities, home health care to outpatient rehabilitation. Use of post-acute care after an injury or acute illness is important for understanding potential pathways to functional decline, disability and the need for longterm care.
Study Design: A prospective cohort study of a national sample of trauma patients hospitalized in trauma and non-trauma center hospitals across 15 metropolitan statistical areas (MSA)s in the United
States.
Population Studied: This analysis draws upon data from the National Study of Costs and
Outcomes of Trauma (NSCOT) to examine postacute care received by approximately 1000 elderly trauma patients hospitalized for a moderate-severe injury (Abbreviated Injury Score >3) between 2001 and 2002.
Principal Findings: We examined patients’ demographics and clinical characteristics including age, injury type and severity, underlying comorbidities, pre-injury health and function including neighborhood mobility and living with a spouse as predictors of post-acute care use and setting. We also examined the acute hospital characteristics as potential determinants of post-acute care, for example trauma center, teaching hospital, profit
status, size, as well as local post-acute care supply factors.
Conclusion: This study elucidates whether patient’s clinical, demographic or health-system factors determine post-acute care use and setting among elderly trauma patients.
Implications for Policy, Delivery or Practice:
Post-acute care and rehabilitation plays an important role in managing trauma-related injuries such as falls and motor-vehicle crashes to restore health and functioning of seniors, as well as prevent functional declines and long-term institutionalization. Identifying clinical and healthsystem factors that determine post-acute care use and type of setting will help policy-makers and health care providers understand current patterns of care and potential gaps in use.
Theme: Long Term Care
● Effects of Language Discordance & Limited
English Proficiency in Cancer Screening
Cindy Wong, Ph.D.; Sarita Bhalotra, Ph.D.
Presented by: Cindy Wong, Ph.D., Heller School,
Brandeis University, 334 Village Creek Road,
Aptos, CA 95003, Phone: (831) 708-2741; Email: cindyjwong@gmail.com
Research Objective: This study contributes to the literature in health disparities by exploring one of many possible streams of inquiry into the social dimensions of health, namely language access. The research question focuses on whether there are independent effects of language barriers in preventive care, and the hypothesis is that language barriers have a negative impact on cancer screening (Schenker, 2007; Jacobs, 2005; Wilson,
2005; John-Baptiste, 2004; Yeo, 2004; Bischoff,
2003; Ngo-Metzger, 2003; Centers for Medicare &
Medicaid Services, 2002; Jacobs, 2001; Pitkin,
2000; Sarver, 2000; Carrasquillo, 1999; Hampers,
1999; Crane, 1997; Baker, 1996; Seijo, 1991;
Manson, 1988).
Study Design: Statistical analysis was performed on data from the 2005 California Health Interview
Survey (CHIS) in Stata v9.2. CHIS 2005 is a computer assisted telephone interview survey, and it has a two-stage, geographically stratified randomdigit-dial sample that includes data from interviews with one randomly selected adult in 43,020
California households. The Survey is conducted in five languages, English, Spanish, Chinese
(Mandarin and Cantonese dialects), Vietnamese and Korean.
Population Studied: Two study populations were examined in this study: adult California residents who completed the 2005 CHIS survey who report that they have 1) limited English proficiency and those who report that they have experienced 2) language discordance with their physician.
Principal Findings: Multivariate logistic regression reveals significant associations between language barriers and screenings. However, the direction is positive, rather than negative, as other research suggests. The findings show California residents who experience language discordance with their physicians have a greater positive likelihood of receiving breast cancer screening, cervical cancer screening and colorectal cancer screening compared with residents who have language concordance with their physicians. Additionally,
California residents with limited English proficiency show a greater positive likelihood of receiving cervical cancer screening compared with English proficient residents. California residents who experience language discordance with their physicians have a 33% greater positive likelihood of receiving breast cancer screening [OR 1.33*, CI
1.01, 1.75], a 61% greater positive likelihood of receiving cervical cancer screening [OR 1.61*, CI
1.17, 2.22] and a 37% greater positive likelihood of receiving colorectal cancer screening [OR 1.37*, CI
1.06, 1.77] compared with residents who are language concordant with their physicians.
Additionally, California residents with limited English proficiency show a 67% greater positive likelihood of receiving cervical cancer screening [OR 1.67*, CI
1.35, 2.07] compared with English proficient residents. (*P<.05)
Conclusion: Theoretical foundations and contextual research suggest multilevel social policy and targeted public health initiatives are factors in the positive results with respect to cancer screening in California.
Implications for Policy, Delivery or Practice:
Therefore, in efforts to reduce health disparities, policymakers should consider the approaches found in California, which include securing the right to health information, enforcing culturally and linguistically appropriate standards of care, strengthening local public health programs, and supporting population-centered health initiatives.
Theme: Disparities
● Training Future Leaders in Healthcare: An
Interdisciplinary Graduate Program in
Healthcare Quality & Patient Safety
Donna Woods, Ed.M., Ph.D.; Kevin Weiss, M.D.,
M.P.H.; Stephen Persell, M.D.; Cndy Barnard,
M.B.A.; John Vozenilek, M.D.; Meetal Acharya
Presented by: Donna Woods, Ed.M., Ph.D.,
Assistant Professor, Institute for Healthcare
Studies, Feinberg School of Medicine,
Northwestern University, 750 North Lake Shore
Drive, 10th Floor, Chicago, IL 60611, Phone: (312)
503-5550; Email: woods@northwestern.edu
Research Objective: Healthcare quality and patient safety have been shown to represent significant challenges in healthcare. Knowledge and skills in these fields are central to the delivery of safe, high-quality healthcare. Effective graduatelevel education is necessary to train a workforce in
the application of key knowledge and skills to improve the quality and safety of healthcare.
Study Design: An interdisciplinary graduate-level curriculum for healthcare professionals to achieve mastery in healthcare quality and patient safety competencies was constructed, approved, and began student enrollment in 2006. The curriculum includes reading of seminal papers and original sources, lecture and discussion, interactive individual and group exercises and self-directed projects through which student must demonstrate knowledge acquisition and skill mastery. The curriculum broadly covers quality and safety history, theory, measurement development, improvement science and organizational change, evidence-based interventions, risk assessment, teamwork, culture, leadership and governance and healthcare organization. Student experience in the program and educational outcomes related to program participation were evaluated. Students rated the core courses on a scale of 1-4 (1 = excellent and 4
= poor), for seven dimensions: quality and relevance, depth of material, course readings, session exercises, faculty, and opportunities for discussion. To assess the overall effectiveness of the program, we tracked: 1) students’ knowledge and skill acquisition, 2) their healthcare quality and patient safety improvement activities, and 3) their professional development.
Population Studied: Thirty-eight students enrolled in a multi-disciplinary graduate program across three years: Physicians (29%), healthcare administrators (29%), nurses (24%), allied health
(8%), communication professionals (8%), medical students (2%).
Principal Findings: The average student assessment of the core courses was excellent
(1.15). Exams, demonstration exercises, written assignments and competency assessments demonstrated student acquisition of core knowledge and skills. Students rated highly the positive contribution of the multi-disciplinary learning context, which enriched their understanding of the perspectives of multiple stakeholders in healthcare. Students demonstrated facility in the application of healthcare quality and patient safety knowledge and skills through successful implementation of Capstone improvement projects. A few examples of student capstone improvement projects include: application of a framework to improve adoption of computerized clinical decision support, application of hospital credentialing processes to improve physician engagement in quality/safety activities, implementation of interdisciplinary rounds to improve teamwork. After a year of enrollment in the program 90% of students matriculated into this educational program received professional recognition of their knowledge and skill acquisition in the form of professional promotions and attainment of quality/safety leadership positions. By program completion, this was true for all the students (100%) graduating from the program.
Conclusion: This evaluation provides results of a successful graduate training program in healthcare quality and patient safety. The program was rated highly by learners, resulted in demonstrated knowledge and skill acquisition, successful completion of meaningful improvement activities, and leadership development in healthcare quality and patient safety.
Implications for Policy, Delivery or Practice:
Graduate education in these fields provides training to build a capable workforce equipped to address the current quality and safety challenges, has developed graduates who are well received in the workplace and have been promoted into leadership positions in the healthcare market.
Funding Source(s): Tuition
Theme: Health Care Workforce
● Developing a Communication Sensitive Chart
Review Tool for Pediatric Patient Safety
Donna Woods, Ed.M., Ph.D.; Jane Holl, M.D.,
M.P.H.; Olivia Ross, M.P.H.; Michelle Leasure,
R.N.; Anna Torricelli
Presented by: Donna Woods, Ed.M., Ph.D.,
Institute for Healthcare Studies, Northwestern
University, 750 North Lake Shore Drive, 10th Floor,
Chicago, IL 60611, Phone: (312) 503-5550; Email: woods@northwestern.edu
Research Objective: Communication has been found to be the most frequent root cause for sentinel events reported to the Joint Commission.
Effective methods to assess of communication related to patient safety risk has remained a challenge. There is a need for medical chart review tool that will identify medical errors and adverse events and is sensitive to communication problems and improvement. The aim of the Chart Review project is to develop a medical chart review tool to measure problematic clinical communication and safety-related events in pediatric patient care.
Study Design: In order to develop an evidencebased understanding of the types of communication and safety errors occurring in pediatric care, a qualitative review of pediatric medical charts is essential. Research nurses with pediatric expertise were given the directive to read through a specific admission and note any safety issues. Reviews included documentation of pediatric care problems related to problematic clinician communication and adverse events. Reviews also included documentation of similar types of pediatric care problems related to problematic clinician communication and adverse events, although the reviews were shorter and fewer specific problems were noted. From this review, the types of documentation reliably identified were compiled into a chart abstraction tool. Additionally, the chart abstraction tool was tested by four pediatric
research nurses. Seventeen charts including eight case and nine control charts were tested (four charts were part of matched sets). The tool was tested by nurses who had worked on the original free-form reviews as well as nurses who were new to the project. This allowed for a comparison of the findings from each type of review and helped to gain insight about the tool’s ability to capture all the potential communication and safety-related errors.
Population Studied: Case medical charts were selected from records for patients who have experienced a high risk event (e.g. transfer from another facility, Critical Assessment Team (CAT) activated, Safety Event Reporting System cases, transfer to the PICU within 24 hrs of being admitted to the floor). Control medical charts were identified from records in which none of the above nine experiences are known to have occurred, and were then matched with case charts based on age, gender, admitting unit, and date of admission.
Principal Findings: The chart abstraction tool categorizes the safety issues found in a medical chart into eleven general types of problems/incidents: Code/CAT team activated,
Death, Forms, Imaging, Labs, Medication
Administration, Nosocomial Infections, Operating
Room, Orders, Plan of Care and Procedures. In case where a safety issue or error is found in the chart that does not “fall” into one of those categories, there is a twelfth category titled “Other.”
These eleven general types of problems/incidents are then broken down into specific types of problems/incidents. These specific problems/incidents are listed as drop-down menu options in the chart review tool. Additionally, nurses always have the option to choose “Other/See
Comments” and describe the issue in their own words in a free text area.
Conclusion: By collecting and organizing information into the chart review tool, the medical record can be used to 1) identify general and specific patient safety events, medical errors and problems and 2) identify communication-related patient safety events, medical errors and problems.
Implications for Policy, Delivery or Practice: The final chart abstraction tool will be tested for external validity at other pediatric healthcare institutions. The
Illinois Department of Healthcare and Family
Services (IDHFS) Quality Control Contractor will identify medical charts from the 5 Chicago hospitals with large pediatric services called the Pediatric
Patient Safety Consortium hospitals. The chart abstraction tool may be revised further or be shown to be effective for use.
Funding Source(s): Michael Reese Health Trust
Theme: Quality and Efficiency: Measurement
● Identifying Safety Risks in Live & Deceased
Donor Liver Transplantation: A Failure Modes
Effects Analysis
Donna Woods, Ed.M., Ph.D.; Anton Skaro, M.D.,
Ph.D.; Olivia Ross, M.P.H.; Daniela Ladner, M.D.,
M.P.H.; Gwen McNatt, M.A., R.N.; Jane Holl, M.D.,
M.P.H.
Presented by: Donna Woods, Ed.M., Ph.D.,
Assistant Professor & Co-Director of the Graduate
Programs in Healthcare Quality & Patient Safety,
Institute for Healthcare Studies, Northwestern
University Feinberg School of Medicine, 750 North
Lake Shore Drive, 10th Floor, Chicago, IL 60611,
Phone: (312) 503-5550; Email: woods@northwestern.edu
Research Objective: Liver transplantation is a high-risk, complex process involving numerous clinicians in varied clinical settings. Several high profile transplantation safety events demonstrate the existence of safety risks and the need to develop strategies for improvement. Nonetheless, a systematic assessment of the patient safety risks of liver transplantation has not been completed.
Study Design: Failure Modes Effects Analysis
(FMEA) is a prospective risk assessment methodology applied to identify risks and direct safety improvement. FMEA evaluates each step in a process, identifying the potential failure modes, causes, frequency, and consequences. A FMEA was conducted to analyze the patient safety risks for adults regarding the key step of donor evaluation and selection in live donor liver transplantation (LDTL) and deceased donor liver transplantation (DDLT). Donor evaluation is a critical process for both LDLT and DDLT involving the compilation of key information about the donor for decision-making.
Population Studied: A multi-disciplinary team of clinicians and staff involved with and knowledgeable about the liver transplantation process.
Principal Findings: A complete process map of the steps involved in the stage of donor evaluation and selection of LDLT and DDLT was created by a multidisciplinary transplantation team (physicians, nurses, physician assistants, coordinators, social workers, and transplantation administrators). A risk priority number, based on the frequency and consequences of each failure mode and the failure mode causes related to the mapped steps was assessed. Through aligning comparable steps in the LDLT and DDLT donor evaluation and selection processes, these two procedures were compared.
For example a key set of failure modes for donor evaluation and selection in both LDLT and DDLT involved the availability of accurate information about the donor. Donor evaluation and selection for
LDLT involves a planned and scheduled set of processes. The donor can report on their own medical and social history. The recording and reporting of the information on which the evaluation and selection is based involves significant clinical redundancies enabling more reliable availability of accurate clinical information necessary for final donor eligibility decision-making. In contrast,
notification of the availability of a potential deceased donor can come at any time of the day, with no previous preparation or warning, and in a remote location. The information on the quality of the potential organ is not standardized, medical and social history must be pieced together from frequently incomplete records and often distant relatives. The information of the donor is collected and presented by one individual and is presented to clinicians remotely for decision.
The results of these analyses identified safety risks in donor evaluation and selection in liver transplantation that could become targets for improvement and revealed that many of the failure modes are more frequent in DDLT than LDLT, with more severe consequences in DDLT.
Conclusion: This FMEA systematically identifies the failure modes of LDLT and DDLT for the critical step of donor evaluation and selection. These
FMEA risk results can inform the development of interventions to improve patient safety.
Implications for Policy, Delivery or Practice:
Patient safety vulnerabilities are of major interest and significance in liver transplantation. This type of analysis of the processes in liver transplantation is extremely important for the understanding of safety risks. Additionally, this process has elucidated the significant differences in patient safety risks between LDLT and DDLT in a critical aspect of the process.
Theme: Quality and Efficiency: Organized
Processes
● Leveraging Existing Assessments of Risk
Now (LEARN) for Patient Safety: A Meta-
Analysis of Risk Results
Donna Woods, Ed.M., Ph.D.; Jane Holl, M.D.,
M.P.H.; Jonathan Young; Sally Reynolds, M.D.;
Olivia Ross, M.P.H.; Anna Torricelli
Presented by: Donna Woods, Ed.M., Ph.D.,
Assistant Professor, Institute for Healthcare
Studies, Northwestern University, 750 North Lake
Shore Drive, 10th Floor, Chicago, IL 60611, Phone:
(312) 503-5550; Email: woods@northwestern.edu
Research Objective: Prospective identification of specific paths and mechanisms of medical care risks is a challenge to patient safety improvement.
While institutional differences do exist, frequently safety vulnerabilities and risks that exist at one institution also exist at other institutions. However, when specific vulnerabilities leading to preventable risks are uncovered through a risk assessment at one institution these results are rarely shared to help other institutions address potential latent risks.
The objective of this study was to identify generic risks across institutions.
Study Design: The LEARN method consists of conducting a meta-analysis of the medium and high risk fail-points from the risk assessment results.
Risk assessments were collected from institutions across the country. Risk assessment process steps and failure modes from all the collected studies were entered into database. Medium to high risk failure modes and failure mode causes were cataloged across risk assessments. Review across the failure modes and failure mode causes revealed common failure modes and common causes. In addition, an assessment of the quality of risk assessments was conducted
Population Studied: Risk assessment results of healthcare processes from multiple institutions across the United States.
Principal Findings: Fifteen risk assessments were collected from institutions across the country. The topics of the collected risk assessments included:
Blood Transfusion, Digital Imaging – Sending,
Digital Imaging – Receiving, Pediatric Emergency
Transport, Crash Cart Standardization, Emergency
Department Patient Flow, Patient Identification -
Blood Bands, Specimen Labeling, ECHMO,
Laboratory Test Delays, and Interventional
Radiology. Across these risk assessments, 559 failure modes are described. Of these, 296 had a risk priority number designating them medium to high risk. Meta-analysis of medium and high failpoints identified several generic underlying systemic issues which lead to harmful errors and patient safety risk with relative frequency.
Examples of prominent generic risks included:
Unreliable hand-off of clinical information regardless of the process or context; Missing information; Lack of mechanisms of systems for verification of the stage of completion of clinical tasks; Lack of feedback in medical care systems; and Patient identification. Assessment of the quality of cross institutional risk assessments revealed that these assessments could have been more comprehensive. For example, they did not represent failure mode possibilities for both omission and co-mission events in most process steps.
Conclusion: New and existing foci for patient safety improvement were revealed. Hand-offs and patient identification are known contexts of risk.
Three newly described pervasive contexts of risk: 1) verification of the stage of task progress, 2) the need for feedback systems in healthcare, and 3) frequent missing information described through this meta-analysis of risk assessments will require significant and fundamental approaches to address and represent new directions for research and improvement. Improving the comprehensiveness of risk assessments will offer further opportunities for improved safety and reliability.
Implications for Policy, Delivery or Practice:
Sharing risk results from FMEA and other risk assessments across institutions is possible and will enable fundamental learning about existing risk to establish the necessary understanding to direct significant safety improvement.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Organized
Processes
● Organization and Logistics in Facilities that
Perform Percutaneous Coronary Intervention
Without Cardiac Surgery Backup On-Site
Albert Woodward, Ph.D, M.B.A.;.; Gregory Dehmer,
M.D.; Lloyd Klein, M.D.; Thomas Wharton, M.D.;
Michael Kutcher, M.D.
Presented by: Albert Woodward, Ph.D, M.B.A.,
Director, Science & Research Services, Division of
Science & Quality, American College of Cardiology,
2400 N Street, Northwest, Washington, DC 20037,
Phone: (202) 375-6629; Email: awoodwar@acc.org
Research Objective: There is controversy in the
United States concerning the safety of percutaneous coronary intervention (PCI) at facilities without cardiac surgery on-site. In addition, there is little information regarding the organization and logistics of such centers.
Study Design: The National Cardiovascular Data
Registry (NCDR) is a large multi-center database that offers contemporary clinical information on this issue. During initial analysis of submitted registry data, there was an opportunity to gather additional information via a Capabilities Survey to assess the organization, staffing, and logistics of the centers that did not have surgery on-site.
Population Studied: Clinical characteristics and inhospital outcomes were assessed in consecutive
PCI cases reported to the NCDR from January 1,
2004 to March 30, 2006. Of this cohort, 8,736 patients who had PCI performed in 60 centers without on-site cardiac surgery (Off-Site) were compared with 299,425 patients at 405 centers that had on-site surgery (On-Site). Of the 60 Off-Site centers invited to fill out the Capabilities Survey, 53 centers (88%) completed the questionnaire.
Principal Findings: Clinical results have been presented elsewhere and show no statistically significant differences in clinical outcomes between
Off-Site and On-Site centers. The organization and structure of these Off-Site centers were highlighted in the Capabilities Survey. Fifty-four percent of the centers had average travel distances to the on-site surgical center of < 20 miles; 72% reported an average transport time < 30 minutes. Alternatively,
26% of the sites travel distances > 40 miles and an average transport time > 30 minutes. Ground transport was used by 52% of the Off-Site centers, with helicopter transport used in 22% and a combination of ground and air in 24%. Staff and facility coverage 24/7 was provided in 92% of Off-
Site centers. Twenty-two per cent of the Off-Site centers only provided primary PCI; the other 78% performed both primary and elective procedures.
There were no programs that reported performing only elective PCI. The mean number of interventional cardiologists at each Off-Site program was 5 ± 4. Only 10% of the sites had fewer than 2 interventional cardiologists who performed procedures at the Off-Site center. Rotation between
Off-Site and On-Site centers was reported by 82% of the interventional cardiologists with only 6% performing PCI solely at the Off-Site center.
Conclusion: Off-site centers face unique challenges to provide safe and effective PCI services for the communities they serve.
Organization, staffing, and travel logistics at Off-Site facilities are important elements to provide structure and process to assure acceptable outcomes measurements in PCI that are comparable to On-
Site centers.
Implications for Policy, Delivery or Practice:
This study adds information regarding appropriate cardiovascular organizational guidelines for Off-Site
PCI centers. In addition, the study raises issues regarding staffing numbers, rotation of operators, and transportation modalities, particularly in rural or underserved areas.
Funding Source(s): National Cardiovascular Data
Registry
● The Proportion of Deaths in Virginia
Associated with Reduced Household Income
Steven Woolf, M.D., M.P.H.; Resa Jones, M.P.H.,
Ph.D.; Robert Johnson, Ph.D.; Robert Phillips, Jr.,
M.D., M.S.P.H.; Andrew Bazemore, M.D., M.P.H.;
M. Norman Oliver, M.D., M.A.
Presented by: Steven Woolf, M.D., M.P.H.,
Professor, Virginia Commonwealth University
Center on Human Needs, Virginia Commonwealth
University, 1200 East Broad Street, Richmond, VA
23298, Phone: (804) 828-9625; Email: swoolf@vcu.edu
Research Objective: The magnitude of the relationship between income/wealth and mortality is difficult to quantify in the United States, because income is not captured in vital statistics (death certificates). Area-based measures of household income provide an alternative, acting as a proxy not only for income and other individual characteristics
(e.g., education) with which income is associated but also for community and environmental factors that vary by area income levels. Knowing how this
“package” of factors relates to mortality is highly relevant now, a time of severe economic stress in the United States in which policymakers are weighing the merits of social programs. Virginia, which is home to the nation’s most affluent counties but also to areas of deep poverty, provides an interesting setting for exploring this question.
Study Design: For the period of 1990-2006, we measured the proportion of deaths that would be averted in Virginia if the state population experienced the mortality rates of the reference population (RP)—the five counties/cities with the highest median household income. Vital statistics were obtained from the state registrar. Population
counts and income data were obtained from the
U.S. Census Bureau.
Population Studied: Individuals in Virginia who died between 1990 and 2006.
Principal Findings: If the mortality rates of the RP had applied throughout Virginia in 1990-2006,
21.8%-28.1% (mean = 24.3%) of the state’s deaths would not have occurred. In absolute numbers,
10,548 to 14,569 (mean = 12,954) deaths would have been averted each year—a total of 220,211 deaths between 1990 and 2006. More than 70% of the averted deaths would involve older adults (age
65 and older), non-Hispanic whites, and adults with no more than a high school education. In some areas of Virginia, nearly half of deaths would have been averted if the RP mortality rates had applied.
These areas—which include cities, and depressed rural counties in southwestern Virginia, the border area with North Carolina (Southside), and
Chesapeake Bay area (Middle Peninsula and
Eastern Shore)—have severe poverty levels, low graduation rates, and a population overrepresented by minorities.
Conclusion: Approximately 25% of deaths in
Virginia between 1990 and 2006 would have been averted if the state had experienced the mortality rates of the five most affluent counties/cities.
Income is interrelated with other social determinants of health (e.g., education, race, and ethnicity) and personal behaviors (e.g., smoking) that affect mortality rates. Moreover, residents of affluent areas may have lower mortality rates because of community characteristics, such as healthier neighborhoods, better schools, access to health care, and social services.
Implications for Policy, Delivery or Practice:
Policies aimed at economic development and other initiatives to improve socioeconomic conditions, which have gained urgency amid the current economic crisis, have powerful health implications.
Economic conditions may have greater effect on health than health care itself; the projected 25% reduction in mortality reported here could not be achieved by any known medical intervention.
Further research is needed to know which aspects of affluent areas—e.g., household income and education, health care, the built environment, legislation—are worth emulating or prioritizing.
Funding Source(s): RWJF
Theme: Disparities
● Assessing a Practice Coach Intervention for
Improving Chronic Care in Safety Net
Organizations
Shinyi Wu, Ph.D.; Marjorie Pearson, Ph.D.; Katie
Coleman, M.S.P.H.; Brian Austin, B.S.; Ed Wagner,
M.D., M.P.H.
Presented by: Shinyi Wu, Ph.D., Assistant
Professor, Epstein Department of Industrial &
Systems Engineering, University of Southern
California, 3715 McClintock Avenue, GER 240C,
Los Angeles, CA 90089, Phone: (213) 740-5073;
Email: shinyiwu@usc.edu
Research Objective: To evaluate the impact on the delivery of chronic care, patient outcomes, and business outcomes in safety net organizations of an intervention that utilized a comprehensive toolkit and practice coaching.
Study Design: The intervention included: 1) development of a comprehensive toolkit containing guidance and tools for implementing the Chronic
Care Model (CCM), and 2) low-intensity, timelimited facilitation by two external coaches to provide implementation support. The coaching, conducted via email, telephone, and a limited number of in-person visits, was provided in two phases. The coaches first laid the foundation by soliciting leadership support, learning about the organizational context, and getting acquainted with team members. They then provided technical assistance on CCM and associated business change strategies for a six-month period. All teams were asked to try rapid cycle (PDSA) changes in a pilot population consisting of their adult diabetic patients (and were encouraged to broaden successful strategies to other patients with related co-morbidities). The evaluation used a block randomized experimental design with randomization at the physician practice level within study clinics, as well as external control groups.
Intervention implementation was assessed via document review and site visits. To evaluate impact, a difference-in-differences method is being used to analyze change in business measures (no show rate and patient encounter volume, costs, and revenues) and HEDIS diabetes measures of process, outcomes, and utilization.
Population Studied: Two California public health and hospital systems participated in the study. The intervention group included nine primary care teams from two clinics with low-income, diverse patient populations (78% with a payor source of Medi-Cal or self-pay/uninsured). The control group consisted of nine other practice teams within the same clinics, and six teams from two external clinics with similar populations.
Principal Findings: Almost all practices implemented some PDSA changes, including action plans, reminder letters, group visits, and registry reports. More than 50 changes in total were tested, with the number and type varying by practice. The toolkit itself was used minimally due to its size and the clinicians’ lack of time. Coaching, however, was perceived favorably. In the semi-structured interviews, the participants indicated that the motivation, guidance, and prompts provided by the coaches were particularly useful. They valued the coaches’ sharing of outside experience and information on other clinics’ efforts, as well as the regular meeting structure and multidisciplinary interaction that accompanied the coaching.
Suggestions for improving coaching were proposed,
including meeting mode, intensity, and duration of coaching. The intervention costs approximately
$41,000 in coaching and $51,000 per clinic site for implementation. The presentation will provide the first report on quantitative results – including diabetes care process and outcome measures, healthcare utilization, and business outcomes.
Conclusion: An intervention organized around a practical toolkit and external, low-intensity coaching was feasible to facilitate practice level tests of change based on the CCM. Participants in safetynet ambulatory care settings evaluated the coaching component as a particularly valuable support.
Implications for Policy, Delivery or Practice:
Low-intensity, time-limited facilitation by external coaches offers a feasible alternative to quality improvement collaboratives for disseminating the
CCM and improving chronic illness care in safetynet organizations.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Organized
Processes
● Implementing Process Redesign Strategies for Improving Hospital Care
Shinyi Wu, Ph.D.; Marjorie Pearson, Ph.D.; Lisa
Smith, R.N., B.S.N., B.S.; Raj Behal, M.D., M.P.H.;
Julie Cerese, R.N., M.S.N.; Joanne Cuny, R.N.,
M.B.A.
Presented by: Shinyi Wu, Ph.D., Assistant
Professor, Epstein Department of Industrial &
Systems Engineering, University of Southern
California, 3715 McClintock Avenue, GER 240C,
Los Angeles, CA 90089, Phone: (213) 740-5073;
Email: shinyiwu@usc.edu
Research Objective: Recent research has shown that hospital care at night is not as safe or patientcentered as care provided during weekdays. The goal of the “24/7 Care Delivery Model” project was to deliver and evaluate an intervention designed to improve delivery efficiency and care coordination around the clock, without compromising clinical quality and financial health.
Study Design: The intervention focused on detailed process redesign strategies designed to modify hospital workload demand or adjust staffing models to match supply to optimized demand.
These 24/7 strategies were developed from the best practices demonstrated in the National Health
Service “Hospital at Night” pilot project, published literature, and recommendation of an advisory group. They consist of five areas: structured handoffs, paging policy, common complains protocol, discharge planning, and staffing. A rigorous and uniform performance improvement (PI) process was employed to facilitate implementation in all intervention hospitals. This Commit-to-ACTion
(CTA) improvement process used a telephone collaborative and toolkit, along with improvement teams and designated support roles at the hospitals. A prospective, quasi-experimental study design was used to assess intervention implementation and effects on care processes and outcomes. Implementation assessment used data from telephone interviews with key informant(s) at each hospital, observation of the facilitation process, document review, and a strategy penetration survey.
Population Studied: Eleven hospitals across the country participated in the intervention, including 9 medical and 6 surgical service units, while 5 additional hospitals served as the comparison group.
Principal Findings: Our data suggest that the PI facilitation was much more demanding than originally planned due to the complexity of the intervention and the study design. The hospitals’ participation in the CTA processes (e.g., calls, use of tools) was generally high, and they spent on average $25,000 on the implementation. All sites implemented some of the redesign strategies, but few implemented at least one element in all four demand strategies. Most had at least one of the redesign elements in place when they started the
CTA collaborative. Implementation rate for elements not already in place varied among elements (as well as among the sites). Structured handoffs and multidisciplinary discharge rounds by residents, nurses, and social workers were among the most frequently implemented strategies. A number of sites reported that they were challenged by the simultaneous focus on multiple strategies.
Other implementation barriers included lack of geographic overlap between services and nursing units, clinician buy-in, time commitment requirements, and staff turnover. We will provide a first report of intervention effects at the meeting – including length-of-stay, bed utilization, average discharge time, risk-adjusted mortality, readmission-to-hospital, and cost of care.
Conclusion: Hospital participation was high and resulted in redesign of some care delivery processes at all hospitals. Each of the redesign strategies was implemented in some hospitals.
Implications for Policy, Delivery or Practice:
These 24/7 strategies offer viable roadmaps for redesigning hospital care delivery processes.
Simultaneous facilitation of multiple strategies allows hospitals to focus on those that best adapt to their workplace context.
Funding Source(s): AHRQ
Theme: Quality and Efficiency: Organized
Processes
● Multi-Criteria Profiling of Diabetic
Populations: What’s Driving your Cost?
Donghui Wu, Ph.D.; Ognian Asprouhov, Ph.D.;
Anton Berisha, M.D.; Krasimir Latinski, M.S.
Presented by: Donghui Wu, Ph.D., Principal
Scientist, MEDai, inc., 4901 Vineland Road, Suite
450, Orlando, FL 32811, Phone: (321) 281-4561;
Email: dwu@medai.com
Research Objective: Applying multi-criteria predictive modeling to Diabetic population to increases the chances of proactively identifying and treating members with the greatest opportunity of being positively impacted
Study Design: This is retrospective cohort study based on administrative claims data for 29,149 members with diagnoses of Diabetes, from managed care organization with mixed Medicaid and Commercial population. Members enrolled for full 12 months during two consecutive years (ending on Sept. 2005) were randomly assigned to either training or validation set (2/3 vs.1/3). Five composite complication outcomes were identified as dependent variable4, 5, 6 and additional 11 utilization outcomes were predicted. For each of the dependent variables we built a separate model using logistic or multivariate linear regression. As many as 911 predictors from demographic, medical and prescription data were utilized. No laboratory data were available. All measures are either MEDai owned or derived from ETG® Grouper (Ingenix).
Population Studied: 29,149 members with diagnoses of Diabetes, from managed care organization with mixed Medicaid and Commercial population.
Principal Findings: Highest increase in next year costs is expected for members; with new onset diabetes complications (92.74% in this data set);
Complication models had sensitivity ranging from
10.93 – 59.64% with ALL Complications model scoring the highest;
Utilization models had Top 10% sensitivity ranging from 13.9 – 73.1% with highest sensitivity for Total
Rx costs. Top 10 predictors are identified within each model.
Conclusion: Predictive modeling offers flexibility and accuracy in targeting specific populations by providing multi-criteria risk stratification approach.
Integration of these models into well designed webapplications enables disease management professionals to focus their efforts toward improving health and financial outcomes. Similar strategies should be evaluated for other chronic diseases.
Implications for Policy, Delivery or Practice:
Improve diabetic pouplation care management and treatment.
● Predictive Modeling for Identification of
Asthma Patients for High Risk Complications &
Saving Opportunities for Care Management
Donghui Wu, Ph.D.; Ognian Asparouhov, Ph.D.
Presented by: Donghui Wu, Ph.D., Principal
Scientist, MEDai Inc., 4901 Vineland Road, Suite
450, Orlando, FL 32811, Phone: (321) 281-4561;
Email: dwu@medai.com
Research Objective: Present an Asthma case study conducted after analysis of three-years of medical and pharmacy claims data. Explore the trends and timeline for High Risk Complications of
Asthma. Demonstrate the power and value of multiple predictive modeling techniques for profiling and identifying high risk complications of Asthma patients and projecting future costs.
Study Design: It is a NEW study with three years of insurance claims data with broad inclusion of
Asthma-related complications, and innovative study settings and techniques.
Population Studied: The case study is conducted through three years of medical and pharmacy claims data for both commercial and Medicaid members. Members with Asthma-related claims in the first year are selected from about 2 million commercial members and more than half a million
Medicaid members. These members (23,874 commercial, 24,293 Medicaid) are subjects in this case study.
Principal Findings: The patterns and trends for
Asthma conditions, complications and costs were studied for the first, second, third, and a combined two-year period (second and third year). The following high-cost complications of asthma were evaluated: pneumonia, GERD, acute bronchitis and bronchiolitis, acute and chronic respiratory failure, pulmonary congestion and hypostasis, acute edema of the lung, pulmonary collapse/atelectasis, pleural effusion, etc.
Conclusion: Demonstrated the power and value of multiple predictive modeling techniques for profiling and identifying high-risk complications of Asthma patients and projecting future costs. Find the trends and timelines for high risk complications of Asthma.
Explore potential saving opportunities/returns for
Asthma Disease Management based on analysis of three-years of medical and pharmacy claims data.
Implications for Policy, Delivery or Practice:
Improve the effectiveness and ROI on Asthma Care
Management Programs.
Theme: Prevention and Treatment of Chronic
Illness
● Validation of a Toolkit to Assess
Organizational Performance in Supporting
Patient-Centered Communication
Matthew Wynia, M.D., M.P.H.; Megan Johnson,
M.A.; Thomas McCoy, M.A.; Leah Passmore Griffin,
M.A.
Presented by: Matthew Wynia, M.D., M.P.H.,
Director, The Institute for Ethics, Institute for Ethics,
American Medical Association, 515 North State
Street, Chicago, IL 60610, Phone: (312) 464-4980;
Email: matthew.wynia@ama-assn.org
Research Objective: To help organizations monitor and improve health care communication by validating an organizational assessment toolkit on patient-centered communication. The specific goals
of this research were to assess whether:
(1)communication performance can be reliably measured at the organizational level;
(2)communication performance varies between organizations; and (3) communication performance in an organization can be evaluated in specific content areas (domains), which might then be targets for quality improvement (QI) interventions.
Study Design: Prospective validation study of an organizational assessment toolkit, developed intially through a national, multistakeholder expert consensus process and refined through pilot testing in 13 organizations. In this validation study, 9 organizations (5 hospitals and 4 clinics) were assessed using the toolkit, which comprises a set of matched surveys of leaders, staff and patients, as well as an internal, team-based assessment of organizational resources and structures that affect communication. A standardized scoring system was created (scores: 0-100, with 100 indicating perfect performance in a domain according to both staff and patients).
Population Studied: Completed surveys were received from a diverse group of 1,763 patients
(13% African American, 39% hispanic/latino, 29% with limited English proficiency), 651 staff members(44% clinical and 56% non-clinical), and
29 organizational executives. Each organization also had a team complete an organizational assessment workbook.
Principal Findings: Survey reliability was assessed by measuring internal consistency in each of 9 domains; reliability was acceptable for both patient survey measures (range of Cronbach’s alpha: 0.59-0.90) and staff survey measures (range:
0.69-0.96). Scores demonstrated considerable within-site and between-site variability and all sites showed strong results (scores>80) in some communication domains but weaker results (scores
<60) in others. No domain received consistently low or high scores across all sites; the lowest range observed was in the domain "Understanding
Organizational Commitment" where the lowest scoring organization was 67.2 and the highest was
76.6.
Conclusion: Organizations can undertake a valid,
360-degree assessment of organizational support for effective communication in 9 distinct domains using matched patient, staff and executive surveys.
Assessment results may be useful for tracking organizational performance, benchmarking and to inform QI.
Implications for Policy, Delivery or Practice:
Effective communication is the foundation for high quality health care. Most measures of health communication focus on the patient-clinician interaction, yet communication is often affected by organizational factors, such as a hospital or clinic’s internal climate and infrastructure. We have validated a toolkit for measuring the communication environment in health care organizations across 9 key domains. The sites involved in this study are now undertaking tailored interventions based on their initial assesment results and are using iterative assessments to determine if communication performance can be improved through organizationlevel changes in targeted domains.
Funding Source(s): The California Endowment
Theme: Organizational Performance and
Management
● The Impact of Comparative Quality
Information on Physician-Patient Relationship
Bin Xie, Ph.D.
Presented by: Bin Xie, Ph.D., Assistant Professor,
Obstetrics & Gynecology, University of Western
Ontario, 454 Platt's Lane, London, CA, Phone:
(519) 661-2111 x 55174; Email: bin.xie.1@gmail.com
Research Objective: Current theoretical models of physician-patient interaction do not provide a framework for comparative quality information that is available to the public to impact the physicianpatient interaction and patient outcomes.
The objective of this theoretical paper is to provide a framework in which comparative quality information directly impact the physician-patient relationship.
Study Design: A game theoretical model has been developed. In the model, the patient has access to limited comparative quality information, whereas the physician has more accurate quality information known only to herself.
Principal Findings: The availability of comparative quality information, even with limited quality or accuracy, improves patient welfare in the individual and aggregate levels. The higher the quality of such information, the bigger the improvement in patient welfare, although the relationship is not linear.
Conclusion: Comparative quality information improves patient welfare.
Implications for Policy, Delivery or Practice: It should be a public policy to have comparative quality information available to the public whenever and wherever possible. The form of such information should be as detailed and accurate as possible to achieve the maximum benefit.
Theme: Consumer Choices in Health Care
● How Close Are We to Eliminate Health
Professional Shortage Areas?
Imam Xierali, Ph.D.; Robert Phillips, M.D.; Stephen
Petterson, Ph.D.; Andrew Bazemore, M.D.; Martey
Dodoo, Ph.D.
Presented by: Imam Xierali, Ph.D., Research
Scientist & Health Geographer, The Robert Graham
Center, American Academy of Family Physicians,
1350 Connecticut Avenue, Washington, DC 20036,
Phone: (202) 331-3360; Email: ixierali@aafp.org
Research Objective: This study examines the geographic distribution of primary care physician workforce in the U.S. Mal-distribution of the health workforce and of health care resources contributes to health outcome disparities. Primary care Health
Professional Shortage Areas (HPSA) designation requires an application by state and population to primary care physician ratio of 3500:1 or 3000:1 for special population. There are 2783 primary care
HPSAs with a total of 99.4 million residents across the U.S. We assess the status of the population to physician ratio in existing HPSAs, identify shortage areas that are undesignated, and the shift required to eliminate them.
Study Design: Physicians practice locations are geocoded from the 2007 American Medical
Association (AMA) Master File. Their locations are then overlaid on primary care HPSA areas
(obtained from HRSA Data Warehouse, March 14,
2008). Population estimates are based on 2007 estimates. We use the common indicator of population to physician ratio to measure shortages.
We use 3000:1, 2000:1, and 1500:1 ratios to assess the status of physician shortages. The first one the most liberal threshold used for HPSA designation.
Population Studied: Direct-patient care primary care physicians, 2007. The U.S. population, 2007 estimates.
Principal Findings: We found substantial variation in ratios for current primary care HPSAs: the primary care physician to population ratio suggests within HPSAs, 933 (17.2 million population) of the
2783 HPSAs have ratios at or above 3000:1 and would require a shift of only 2158 extra primary care physicians to eliminate shortage areas; 1493 (35 million population) HPSAs have ratios at or above
2000:1 and would require only 6,478 extra physicians; and 1956 (55.3 million population)
HPSAs have ratios at or above 1500:1 ratios and would require a shift of 14,053 extra physicians. Of areas not designated as HPSAs we identified areas of primary care physician shortages: 156 counties
(and/or partial counties) with 2.2 million residents are in need of 242 extra primary care physicians at
3000:1 ratio. This worsens to 276 counties (or partial counties) with 7.7 million residents that are in need of 948 extra primary care physicians at 2000:1 ratio. This further increases to 540 counties (and/or partial counties) with 29.9 million residents that are in need of 3,916 extra primary care physicians at
1500:1 ratio.
Conclusion: Many current HPSAs do not meet the liberal 3000:1 designation threshold; however, many are one physician away from it. Other proposed ratios would capture a larger proportion of current HPSAs and would also capture a number of undesignated areas.
Implications for Policy, Delivery or Practice:
Shifts of relatively small number of physicians could eliminate most HPSAs and focused policy options like quadrupling the National Health Service Corp could achieve this.
Theme: Health Care Workforce
● The Health Workforce Tapestry: A
Comparative Study of the National Provider
Identifier Database
Imam Xierali, Ph.D.; Andrew Bazemore, M.D.;
Robert Phillips, M.D.; Martey Dodoo, Ph.D.;
Stephen Petterson, Ph.D.
Presented by: Imam Xierali, Ph.D., Research
Scientist & Health Geographer, The Robert Graham
Center, American Academy of Family Physicians,
1350 Connecticut Avenue Suite 201, Washington,
DC 20036, Phone: (202) 331-3360; Email: ixierali@aafp.org
Research Objective: This study examines the publicly available free of charge National Provider
Identifier (NPI) database from Centers for Medicare
& Medicaid Services (CMS). In 2004, the US
Department of Health & Human Services published the Final Rule adopting the NPI as the standard unique health identifier for health care providers, with an effective date of implementation occurring on May 23, 2005 and a compliance date for all covered entities of May 23, 2007, except for small health plans (which had to comply by May 23,
2008). Health workforce distribution should be approached from a multidisciplinary perspective not only by looking at the physician workforce distribution but also by examining the availability of other service providers especially mid-level primary care providers. This is because the health service workforce tends to agglomerate to take advantage of health resource infrastructure sharing. The NPI data may be able to shed some light on the distribution of clinicians in the U.S.
Study Design: We used 2008 American Medical
Association Master File (AMA) and 2007 Area
Resource File (ARF) to correlate to November 2008
NPI data to test correlations of area-based clinician counts across the U.S. We geocoded AMA master file and NPI to the census block group level and then aggregated the physician counts in the specified geographies: state, county, census tract,
Health Professional Shortage Areas, Medically
Underserved Areas / Populations, and rural areas.
For mid-level providers we aggregated NPI to state and county. We then compare the health workforce structure across state.
Population Studied: Health care providers and physicians in the United States
Principal Findings: Preliminary result shows that
NPI is close to AMA Master File for account of physicians and to ARF for account of mid-level providers: NPI has 749,187 physicians, 119,772
Family Physicians (FP), 78,087 Physician
Assistants (PA), and 90,791 Nurse Practitioners
(NP), whereas there are 680,530 direct-patient care physicians, 99,423 FPs (AMA 2008), 78,087
projected PAs (ARF 2007), and 70,135 NPs (ARF
2001). NPI physician count shows significantly high correlation with AMA Master File with a coefficient of 0.995 at 0.0001 level for all types of direct patient care physicians. The correlation is also significant for FPs with a coefficient of 0.990 at 0.0001. At the county level enumeration, NPI physician count similarly shows significantly high correlation with
AMA Master File with coefficient of 0.982 at 0.0001 level for all types of direct patient care physicians; whereas for family practice the correlation is 0.982 at the same level. At the state level NPI and Area
Resource File shows a significant correlation of
Physician Assistant with a coefficient of 0.994 at
0.0001 level, Nurse Practitioners coefficient of
0.936 at 0.0001.
Conclusion: NPI shows significant consistency in describing the health workforce landscape in the
U.S. It provides more precise locational information for mid-level health care providers. NPI could be particularly useful to studies of access disparity to health service providers by smartly pinpointing areas most in need.
Theme: Health Care Workforce
● Spatial Modeling of Primary Care Physician in the U.S.
Imam Xierali, Ph.D.
Presented by: Imam Xierali, Ph.D., Research
Scientist & Health Geographer, The Robert Graham
Center, American Academy of Family Physicians,
1350 Connecticut Avenue, Suite 201, Washington,
DC 20036, Phone: (202) 331-3360; Email: ixierali@aafp.org
Research Objective: This study examines the neighborhood factors affecting the distribution of primary care physicians in the US. Access to primary care is important for the health of populations. The primary care workforce is geographically mal-distributed resulting in unequal access to primary care. It shows that primary care physician distribution is affected by neighborhood attributes – not only within neighborhoods but also in areas adjoining.
Study Design: Physician distribution is predicted by the socioeconomic attributes of and spatial relationship among census tracts in a spatial regression model. Spatial relationship is explicitly measured as a. distance from a census tract to the nearest urban center, b. spatial dependence among census tracts, and c. regional effect. The research utilizes a mixed-order spatial weight matrix to understand the spatial effects in primary care physician distribution. Physician data come from
AMA Master File 2008. The tract socioeconomic data come from Census 2000 and Census
Population Estimate 2007.
Population Studied: The US physicians and the
US population
Principal Findings: The preliminary results show that at the census tract level, large numbers of primary care physicians are associated with high proportions of population with Higher Education
Degree (coefficient= 2.603; p-value= 0.0000), the elderly 65 or over (coefficient=0.927, p-value=
0.0000), non-HPSA status (HPSA coefficient=-
1.9140, p-value= 0.0004), and proximity to urban centers (distance coefficient=-0.001, p-value=
0.110). There is also a statistically significant positive spatial effect (lambda coefficient=0.1525, p-value= 0.0000).
Conclusion: In considering risk of poor physician location probability in an area, the attributes of surrounding areas should also be considered. The findings can contribute to efforts to resolve poor physician distribution and may help identify areas at greatest need for new clinical safety net resources.
Implications for Policy, Delivery or Practice:
Location factor is important for health access issues. Physician distribution is affected by neighborhood attributes not only within neighborhoods but also in areas adjoining.
● Is There a Difference in Financial Performance
Between Rural & Urban Hospitals?
Haichang Xin, M.Sc.; Niccie Lee Mckay, Ph.D.; Lori
Bilello, M.H.A.; Matthew Kukla, B.A.
Presented by: Haichang Xin, M.Sc., Ph.D. Student,
Health Service Research, Policy & Management,
Univeristy of Florida, 101SouthNewell Drive, P.O.
Box 100195, Gainesville, FL 32610, Phone: (352)
273-6627; Email: hsppyxhc@ufl.edu
Research Objective: Comparing financial performance between urban and rural hospitals, and the sources of any differences will allow policymakers to offer more targeted suggestions for improving profitability. Numerous studies have examined the differences in financial performance among urban hospitals, yet very few have examined how financial performance differs between urban and rural hospitals. The goals of this study are to examine financial performance between rural and urban hospitals for the state of
Florida in 2006 and to determine the sources of any differences.
Study Design: The 2006 Florida Hospital
Association Hospital Financial Data files are a cross-sectional study design, however the exogenous nature of the independent variable makes the results convicing. Using the multiple regression analysis, we used the following the dependent variables to measure financial performance: operating margin, net revenue per admission, total cost per admission, patient care related labor cost per admission and case mix adjusted average length of stay (ALOS). The independent variable of primary interest is hospital location (Rural vs. Urban), and control variables
include hospital ownership, size, critical access status, and payer mix.
Population Studied: The analysis used the 2006
Florida Hospital Association Hospital Financial Data files, which are derived from the financial reports reported to the Florida Agency for Health Care
Administration by all licensed Florida hospitals. The database was sorted to eliminate non-acute care hospitals, teaching hospitals, children’s hospitals and specialty hospitals from the analysis. The analyzing unit is each individual rural and urban hospital.
Principal Findings: The final data set included 172 acute care hospitals, 26 rural and 146 urban hospitals. After controlling other factors, rural hospitals earned $499.99 less in net revenues per case-mix adjusted admission than urban hospitals, but the difference was not significantly different.
Rural hospitals had a marginally significant lower total cost per case-mix adjusted admission and total patient care labor cost per admission than their urban counterparts ($747 and $194, respectively), p<0.06. Rural hospitals had marginally significant higher profit than urban hospitals (5.5%), p<0.06.
There was no statistically significant difference in case-mix adjusted average LOS between rural and urban hospitals.
Conclusion: Overall, although not different in revenue, rural hospitals in the study were less costly and more profitable than their urban counterparts.
Implications for Policy, Delivery or Practice: The results shed light on the worries for payment policies. They relieve the policymakers’ concerns about rural hospitals’ exarcerbating finacial performance due to decreased payment brought by adjusting wage rates to match labor costs in combination with those from the Balanced Budget
Act of 1997, which results in decreased payment to a level below hospital costs. The results imply that current payment policies can be safely conducted to certain extent.
Theme: Organizational Performance and
Management
● Primary Care Physicians' Pap Test Screening
Recommendations & Practices: Findings from a
National Survey
K. Robin Yabroff, Ph.D.; Mona Saraiya, M.D.,
M.P.H.; Helen Meissner, Ph.D.; David Haggstrom,
M.D.; Louise Wideroff, Ph.D.; Gigi Yuan, M.S.
Presented by: K. Robin Yabroff, Ph.D.,
Epidemiologist, Division of Cancer Control &
Population Sciences, National Cancer Institute,
6130 Executive Boulevard, Bethesda, MD 20892-
7344, Phone: (301) 594 1723; Email: yabroffr@mail.nih.gov
Research Objective: Cervical cancer screening guidelines were substantially revised in 2002 and
2003. The objective of this study was to assess primary care physicians’ current Papanicolaou
(Pap) test screening practices, including screening initiation, frequency, and stopping.
Study Design: In 2006-2007, we surveyed a nationally representative sample of practicing primary care physicians stratified by specialty, including general practice (GP), family practice
(FP), internal medicine (IM), and obstetrics/gynecology (OB/GYN). The survey included questions about physician and practice characteristics, and recommendations for Pap test screening in clinical vignettes describing women by age, and sexual and screening history.
Characteristics associated with guideline-consistent recommendations for starting and stopping screening across multiple vignettes were evaluated with logistic regression.
Population Studied: Nationally representative sample of 1212 practicing primary care physicians stratified by specialty.
Principal Findings: The majority of primary care physicians reported providing Pap testing as a service to their eligible patients (91.0%; 95% CI:
89.0%, 92.6%). Among Pap test providers
(N=1,114), Pap screening practices, including number of tests ordered or performed, use of patient reminder systems, and cytology method used, varied by physician specialty (p<0.001).
More than 80% of all physicians reported that at least one set of screening guidelines was very influential, however, few physicians had guidelineconsistent recommendations for starting and stopping Pap test screening across multiple vignettes (22.3%; 95% CI: 19.9%, 25.0%). Most physicians whose screening recommendations were not guideline-consistent recommended initiating screening in women younger than 21 who have not yet had sexual intercourse or continuing screening in their patients who had a hysterectomy for benign disease or in well-screened patients with unresectable non-small cell lung cancer. Compared to OB/GYNs, IMs and FP/GPs were more likely to have guideline-consistent screening recommendations ((OR = 1.96; 95% CI: 1.22, 3.14) and (OR = 1.43; 95% CI: 0.97, 2.09), respectively) in multivariate analyses.
Conclusion: Primary care physicians’ recommendations for Pap test screening are not consistent with screening guidelines.
Implications for Policy, Delivery or Practice: We observed overuse of Pap test screening, even in clinical situations where major guidelines are consistent, among physicians who describe professional society screening guidelines as being very influential. These findings suggest that strategies to improve guideline-consistent screening may need to consider more comprehensive models of patient, physician, and system level influences on screening, including multi-level theoretic models of factors that influence and explain overuse of health services and incorporation of reimbursement policies.
Funding Source(s): NCI, Centers for Disease
Control and Prevention
Theme: Public Health
● Peer Effects in Weight Gain Among College
Students: A Natural Experiment
Olga Yakusheva, Ph.D.; Marianne Weiss, D.N.S.,
R.N.
Presented by: Olga Yakusheva, Ph.D., Assistant
Professor, Economics, Marquette University, 606
North 13th Street, Milwaukee, WI 53201, Phone:
(414) 288-3409; Email: olga.yakusheva@marquette.edu
Research Objective: This study examines peer effects and other environmental influences on weight management behaviors and outcomes among freshman college students.
Study Design: The study design is a natural experiment. On the college campus that is the study site, incoming students are required to live on campus during the freshman year, and the dormitory/room assignments are based on a randomized lottery. Taking advantage of the randomized living assignment, the study examines correlations in weight gain/loss among students on three levels: 1) between roommates, 2) among students living on the same floor, and 3) among students living in the same dorm. Behavioral factors
(i.e. eating habits, exercise, smoking, alcohol consumption, sleeping habits) and their commonalities within each of the three levels are also examined. The information is being collected through a series of two online interviews: one conducted in the beginning of the freshman year
(September, 2008) and one to be conducted at the end (May, 2009). The respondents will be matched based on their dormitory, floor, and room number, and the data will be examined using hierarchical regression analysis.
Population Studied: The study examines 860
Marquette University students, ages 18-19, who entered in the fall of 2008. This represents a 47% response rate. 62% of the sample are female, and
95% are Caucasian.
Principal Findings: Data analysis is currently in progress and preliminary findings will be presented.
Conclusion: Data analysis is currently in progress and preliminary findings will be presented.
Implications for Policy, Delivery or Practice:
Higher education administrators will be interested to know what type of living circumstances lead to optimal weight management behaviors and outcomes among students. More broadly, the study has a potential of informing health policy about the importance of peer and environmental influences in weight gain and provide scientific ground for designing an effective evidence-based obesity prevention policy.
Funding Source(s): Marquette University College of Business Research Grant
● Potentially Avoidable Hospitalizations among
Medicare Beneficiaries with Diabetes
Yi Yang, M.D., Ph.D.; Vennela Thumula, B.S.;
Patrick Pace, Ph.D.; Benjamin Banahan, Ph.D.;
Noel Wilkin, R.Ph., Ph.D.; William Lobb, R.Ph.,
Ph.D.
Presented by: Yi Yang, M.D., Ph.D., Assistant
Professor, Department of Pharmacy Administration,
University of Mississippi, 234 Faser Hall, University,
MS 38677, Phone: (662) 915-1062; Email: yiyang@olemiss.edu
Research Objective: In 2007, the Agency for
Healthcare Research and Quality (AHRQ) developed Prevention Quality Indicators (PQIs) for ambulatory care-sensitive conditions (ACSCs), such as diabetes, asthma, and congestive heart failure, for which timely and adequate ambulatory care can potentially prevent the need for hospitalizations. The PQIs were designed to be used with hospital inpatient discharge data to identify potentially avoidable hospitalizations
(PAHs) among patients with ACSCs. The objectives of this study were to describe the prevalence of
PAHs among Medicare Part D enrollees with diabetes and to identify patient characteristics that can predict the risk for PAHs.
Study Design: Retrospective cohort study. Using
Medicare Part A claims data from 07/01/2006 to
03/31/2007, we examined the prevalence of PAHs due to diabetes short-term complications, long-term complications, uncontrolled diabetes, and diabetes lower-extremity amputations. Using multivariate logistic regression to adjust for covariates, we examined the association between patient demographics (age, gender, and race), clinical risk factors and PAHs.
Population Studied: Medicare beneficiaries with diabetes from six states (Alabama, California,
Florida, Mississippi, New York, and Ohio) in 2006 who had enrolled in Medicare Part D plans and who had continuous Medicare coverage through
03/31/2007.
Principal Findings: A total of 1,888,682 diabetic patients met our inclusion criteria. Mean age (±SD) was 71.6 (±11.6) years, 59.5% were female, and
66.4% were white. Approximately 58.9%, 5.8%,
0.5% had coexisting hypertension, nephropathy, and hypertension + nephropathy. Overall, 92,201
(4.88%) patients had at least one PAHs, 0.02% had hospital admissions due to diabetes short-term complications, 3.12% due to diabetes long-term complications, 1.96% due to uncontrolled diabetes, and only 94 cases of PAHs were due to diabetesrelated lower-extremity amputation. Multivariate regression models show that the after adjusting for covariates, compared with patients without the diagnosis of hypertension and/or nephropathy, patients with coexisting hypertension and nephropathy had the highest risk of having PAHs
(odds ratio: 6.950, 95% confidence interval: 6.742-
7.165), followed by patients with isolated nephropathy (odds ratio: 5.953, 95% confidence interval: 5.574-6.357), and hypertension (odds ratio:
2.607, 95% confidence interval: 2.541-2.675). Other statistically significant independent predictors of
PAHs include age younger than 65 years, female gender, black or Hispanic race, and other cardiovascular comorbidities, including congestive heart failure, retinopathy, peripheral artery disease, coronary artery disease, myocardial infarction, and cerebrovascular artery disease.
Conclusion: The results suggest that the risk for
PAHs among inpatient stays were strongly associated with clinical risk factors such as comorbid hypertension and/or nephropathy. Age younger than 65 and black or Hispanic races were also significant predictors of PAHs.
Implications for Policy, Delivery or Practice: Our finding of the extremely high risks for PAHs among patients with coexisting hypertension and/or nephropathy deserves greater attention and further investigation, because hypertension and nephropathy are important risk factors for diabetes morbidity and mortality. AHRQ’s PQIs for diabetes are valuable tools in assessing the adequacy and effectiveness of quality of diabetes care.
Funding Source(s): CMS
Theme: Medicare
● Primary Care Providers’ Attitudes Toward
VA’s Computerized Patient Record System
Elizabeth Yano, Ph.D., M.S.P.H.; Bradley
Doebbeling, M.D.; Ismelda Canelo, M.P.A.; And
Lanto, M.A.; Steven Asch, M.D.
Presented by: Elizabeth Yano, Ph.D., M.S.P.H.,
Research Career Scientist, VA Greater Los Angeles
Health Services Research & Development Center of Excellence, Greater Los Angeles VA Healthcare
System, 16111 Plummer Street (152), Sepulveda,
CA 91343, Phone: (818) 895-9449; Email: elizabeth.yano@va.gov
Research Objective: Recent research and media attention have raised concerns about the potential pitfalls of electronic medical records (EMRs), pointing to safety issues and unintended consequences surrounding their implementation.
We surveyed practicing primary care providers in
VA health care settings nationwide to determine their perceptions of the functionality, safety, and quality of care they are able to provide using the
VA’s Computerized Patient Record System
(CPRS).
Study Design: We analyzed results from the 2007
VHA Clinical Practice Organizational Survey
(CPOS), a national key informant survey assessing the organization and management of primary care at geographically distinct VA health care facilities.
We used tailored design methods and administered surveys through email and express-mailed hard copy to optimize response rates (RR). Information technology (IT) questions covered perceptions of the benefits of a range of CPRS functions (e.g., decision support) and performance (e.g., medical errors), as well as sufficiency of IT resources, such as equipment, training and available informatics expertise.
Population Studied: We surveyed VA primary care program directors at all VA facilities serving 4,000 or more veterans and delivering 20,000 or more outpatient visits (N=250), and obtained a 90% RR
(n=235 primary care programs) representing all VA regions and networks.
Principal Findings: Primary care providers’ ascribed substantial benefits to CPRS: better access to clinical information (96% agree-tostrongly agree), improves quality of care/services
(95%), improves patient safety (95%), reduces medical errors (92%), improves coordination between physicians (88%), provides data for making better managerial/operational decisions
(74%), improves dissemination of new information
(71%), reduces costs of providing services (57%) and improves clinical decision support (49%). In sharp contrast, only 20% felt that CPRS reduced the time required for a patient visit.
Overall, 88% of practices had sufficient hardware.
However, neither the amount (24%) or types (30%) of health IT training were considered adequate, with
71% of practices reporting fewer than 5 hours/year and 20% with 5-10 hours/year. Over 90% of practices indicated use of computerized reminders for many routine screening and management activities (e.g., for depression or diabetes), from 30-
59% used computerized performance profiling, and from 30-40% had specialized CPRS templates for guiding care for the same conditions. One-third of practices (33%) reported that providers sometimes turn off guideline prompts. Overall, providers described access to health IT technical staff as somewhat-to-not at all sufficient (61%), comparable to their ratings of access to medical informatics support for developing needed clinical applications
(59%). Seventy percent of practices indicated that
CPRS training time in advanced functions (e.g., registries, tracking labs over time) was insufficient.
Conclusion: Practicing primary care physicians widely perceive substantial benefits of current
CPRS functionality. However, recurring resources are needed to support ongoing provider training and ensure local access to sufficient medical informatics and technical expertise.
Implications for Policy, Delivery or Practice:
Many EMR implementation efforts face challenges with hardware and software applications, as well as provider skepticism and reluctance to change daily practice. VA’s extensive experience with EMRs clearly demonstrates that these challenges are surmountable, however, sustaining and continually improving care supported by EMRs will require ongoing investment in dedicated time for health IT training and accessible technical expertise.
Funding Source(s): VA
Theme: Health Information Technology
● Affordability of Premiums for Children’s
Public Health Insurance: How do States Decide?
Ambeshie Yesus, M.D.; Jose Escarce, M.D., Ph.D.;
Miriam Laugesen, Ph.D.; Carole Gresenz, Ph.D.;
Nickie Lurie, M.D., M.S.P.H.
Presented by: Ambeshie Yesus, M.D., National
Research Service Award Primary Care Research
Fellow, Department of Internal Medicine, University of California, Los Angeles, 1935 North Kenmore
Avenue, #4, Los Angeles, CA 90027, Phone: (323)
252-9201; Email: ayesus@mednet.ucla.edu
Research Objective: To understand how
Pennsylvania, Illinois, and Washington- three states that expanded their children’s public insurance programs to higher income groups- made determinations about the affordability of coverage under the expansions, and to analyze how the strategies used by these states compare with approaches to assessing affordability and affordability thresholds suggested in the literature.
Study Design: We reviewed state documents including but not limited to legislation, budgets, and program reports, and conducted semi-structured interviews with state program officials, gubernatorial and legislative staff, legislators, and advocacy organizations from September-December 2008.
We also conducted a literature review of journal articles and reports discussing affordability of health insurance.
Population Studied: Children’s health insurance coverage expansions in Illinois, Pennsylvania, and
Washington.
Principal Findings: Affordability thresholds in the literature were, for the most part, based on the proportions of income spent by families on premiums for health insurance purchased through the private market. Another common type of affordability analysis in the literature uses family budget calculations to determine how much income is needed to cover necessary living expenses and health insurance in particular geographic regions.
However, these methods were not explicitly employed by the three states when setting recipients’ premium contributions so as to ensure the affordability of children’s coverage. Premium contributions in Pennsylvania were influenced by the SCHIP statute limiting the proportion of family income spent on health insurance to 5%, actuarial estimates and enrollment projections, and political negotiation. Policymakers in Illinois looked to state employee premiums for children to define a starting point for their premium schedule, and gradually increased this amount for higher incomes. The premium schedule for higher income groups was also influenced by crowd-out considerations and political negotiation. Affordability determinations in
Washington were greatly influenced by premium levels in other states that have expanded eligibility to similar income levels. In our analysis, we compare the current premium schedules from each state to the affordability thresholds in the literature.
Conclusion: States set premiums under public insurance expansions using a range of information from existing programs rather than approaches suggested in the literature. Nonetheless, there is significant variability between states in how health insurance affordability determinations are made.
This variability is likely due in part to economic and political factors internal to each state, but the lack of widely accepted affordability thresholds derived for public insurance may also contribute, as well as difficulty in applying family thresholds to a portion of the family unit (i.e. the children).
Implications for Policy, Delivery or Practice:
Post-implementation research from these states, analyzing the affordability of children’s insurance plans in the context of the family unit, will be essential to understand what is affordable in the public insurance market. Such data will be helpful to the states studied in modifying their existing premium schedules, if necessary, and in guiding other states that embark on similar expansions. A model set of guidelines on affordability could help states make more informed decisions. However, the research on affordability will need to be widely disseminated to state government policymakers in order to have an impact.
Funding Source(s): RWJF
Theme: Medicaid, SCHIP and State Health Reform
● Working to Improve Language Services in
California Public Hospitals: An Assessment of
Organizational Readiness to Engage
Juliet Yonek, M.P.H.; Alan Cohen, Sc.D.; Joseph
Restuccia, Dr.P.H.; Melissa Biel, Dr.P.A.; Kevin Van
Dyke, M.P.P.; Romana Hasnain-Wynia, Ph.D.
Presented by: Juliet Yonek, M.P.H., Director,
Program Evaluation, Health Research &
Educational Trust, 1 North Franklin Street, 30th
Floor, Chicago, IL 60625, Phone: (312) 422-2644;
Email: jyonek@aha.org
Research Objective: (1) To evaluate the impact of the work by California public hospitals to improve language access, (2) identify strategies to engage hospitals in facilitating system change to address the provision of language services, and (3) develop an instrument to measure hospitals’ readiness to engage in language service improvement.
Study Design: Site visits to hospitals in 2007.
Researchers collected qualitative data through semi-structured key informant interviews with hospital leaders, clinicians, and administrative staff exploring the following domains focusing on the provision of language services: commitment, hospitals’ policies and procedures, self-evaluation, strategies for improvement, and strengths and weaknesses. To examine prior, present, and future
engagement, we developed a Readiness to Engage
(RTE) instrument with 17 specific measures based upon 7 key elements: 1.Commitment of top leaders;
2.Sense of urgency/importance among staff members; 3.Defined goals and strategies;
4.Methods, frequency and extent of communication to staff; 5.Adequate infrastructure;
6.Spread/diffusion of services; and 7.Sustainability of services. Each measure was scored on a scale of 0 – 3, with 0 indicating the absence and 3 indicating the presence of an element in its most well-developed form. Scores were summed across all 17 measures to produce a total score for each hospital (maximum attainable score = 51).
Population Studied: Twelve California public hospitals
Principal Findings: Group mean for the RTE scores was 35.2 out of 51. Hospitals clustered into three groups of four hospitals each: high-scoring
(mean = 43.8; range: 43 – 44); medium-scoring
(mean = 38.3; range: 36.5 – 39.5); and low-scoring
(mean = 23.5; range: 20 – 26). High- and moderatescoring hospitals demonstrated a strong commitment to language services (LS) and strong staff perceptions of the importance of language services. They invested in infrastructure over time, especially in remote interpretation technologies such as videoconferencing, and had allocated funds to improve LS. Low-scoring hospitals exhibited deficiencies and experienced multiple barriers to improvement. LS were important but not a high priority of the hospital, goals and strategies were not consistently communicated to staff, and the infrastructure for providing timely and effective LS was generally perceived as inadequate.
Conclusion: These findings show strong systemlevel support for language services among many
California public hospitals, but there are clear areas for improvement.
Implications for Policy, Delivery or Practice:
Further analysis and validation of the RTE instrument in other hospitals is necessary to determine its long-term value. However, it has the potential to serve as a valuable evaluation tool for hospitals to employ in assessing their readiness to engage in LS provision and improvement.
Funding Source(s): The California Endowment
Theme: Quality and Efficiency: Measurement
● Effects of Ongoing Epidemic on the Annual
Influenza Vaccination Rate & Vaccination
Timing among the Medicare Elderly: 1999-2005
Byung-Kwang Yoo, M.D., Ph.D.; Megumi Kasajima,
B.S.; Charles Phelps, Ph.D.; Peter Szilagyi, M.D.,
M.P.H.
Presented by: Byung-Kwang Yoo, M.D., Ph.D.,
Assistant Professor, Community & Preventive
Medicine, University of Rochester, 601 Elmwood
Avenue, Box 644, Rochester, NY 14642, Phone:
(585) 275-3276; Email: Byung-
Kwang_Yoo@urmc.rochester.edu
Research Objective: Influenza vaccination coverage among the elderly has not improved since the late 1990s. Prior studies have documented influenza vaccination to be positively associated with the prior year’s influenza epidemic level. We hypothesize the similar positive association with the ongoing influenza epidemic’s weekly changes.
Study Design: Cross-sectional survival analyses during each of six influenza seasons from the 1999-
2000 season to the 2004-2005 season, using the
Medicare Current Beneficiary Survey (MCBS) with associated claims. The outcome variable was daily vaccine receipt based on claims data, available only among Medicare fee-for-service (FFS) enrollees.
Explanatory variables included the weekly influenza epidemic change rates based on viral surveillance data at the nine census region level, the weekly vaccine supply at the nine census region level, and fifteen controlling variables such as individual health status, demographic and socioeconomic characteristics.
Population Studied: Community-dwelling elderly aged 65 or older who were unvaccinated at onset of an influenza epidemic, and continuously enrolled in a Medicare FFS plan from September to March during each season (weighted N= 8-10 million per season).
Principal Findings: In all six influenza seasons, the ongoing epidemic change was positively associated with annual as well as daily and weekly influenza vaccine receipt, i.e., earlier vaccination timing (p<.05). Similar positive associations were observed between the vaccine supply and the vaccine receipts in all seasons examined (p<.001).
For example, during the 2003-2004 season, out of an 18.3% increase in annual vaccination coverage after the start of the epidemic, 2.9% (p<.001) was attributed to weekly epidemic changes and 5.8%
(p<.001) to the weekly incremental vaccine supply.
In this season, 100% increase in the weekly epidemic level was associated with a 2% (p<.001) increase in the annual vaccination as well as a 13%
(p<.001) increase in the daily vaccination during the subsequent weeks. The positive effect of vaccine supply on annual and daily vaccination tended to be greater during a season with a vaccine supply problem (p<.05). For instance, during the 2000-
2001 season with a sever vaccine supply delay, out of a 9.8% increase in annual vaccination coverage after the start of the epidemic, 8.4% (p<.001) was attributed to the weekly incremental vaccine supply.
Conclusion: Influenza vaccine receipt is positively associated with weekly changes in ongoing epidemic levels and vaccine supply at the regional level.
Implications for Policy, Delivery or Practice:
Health care leaders should consider individual perception of influenza risk being modified at short intervals (i.e., weekly), and “demand” for influenza vaccine being responsive to ongoing epidemic and vaccine shortages/delays. One potential public
health policy implication is to monitor areas with epidemics and consider redistribution as part of the government vaccine supply chain. The wellcoordination of vaccine supply with epidemicresponsive demand for vaccination is expected to improve the current influenza vaccination rate among the elderly and approach the Healthy
People 2010 goal of 90%.
Funding Source(s): NIH/National Institute of
Allergy and Infectious Diseases
Theme: Public Health
● The Impact of Higher Drug Cost-Sharing on
Short-Term Outcomes for Hypertension
Jean Yoon, Ph.D., M.H.S.
Presented by: Jean Yoon, Ph.D., M.H.S., Health
Economist, Higher Education Recruitment consortium, Palo Alto VA, 795 Willow Road, (152
MPD), Menlo Park, CA 94025, Phone: (650) 493-
5000; Email: jean.yoon@va.gov
Research Objective: In order to manage chronic conditions such as hypertension, better evidence linking drug adherence to health outcomes is critical to inform clinicians and policymakers. Cost-sharing for drugs has been shown to be one of the strongest predictors of drug adherence. To limit omitted variables bias due to unobserved characteristics of high adhering patients with better health outcomes, the direct impact of higher drug cost-sharing (leading to lower drug adherence) on short-term outcomes is estimated for a sample of working-age adults with hypertension.
Study Design: This study uses data from the 2000-
2001 Medstat MarketScan Database, a claims database of privately insured patients for a short panel design. Patients’ cost-sharing for drugs was captured from benefits information in 2000 and categorized as <=$5 copayment, $6-$12 copayment, >=$15 copayment, and 10% or 20% coinsurance. Acute events were measured the following year as any hospitalization or ER visit with a diagnosis of uncontrolled hypertension, acute myocardial infarction, acute ischemic heart disease, arterial thrombosis and embolism, angina pectoris, or other hypertensive crisis. Analyses adjusted for patient age, sex, comorbidity, health plan type, and county-level sociodemographic characteristics. A probit regression was conducted in order to test a reduced-form model predicting any adverse health event with the exogenous determinants of drug adherence including drug cost-sharing. Predicted probabilities with bootstrapped confidence intervals were obtained by drug cost-sharing category.
Population Studied: The sample population was all working-age adults aged 18 to 64 with hypertension and refilling antihypertensive drugs in
2000, a total of 83,893 patients.
Principal Findings: Overall, 835 patients or 1% of patients with hypertension experienced an acute event during the nine-month study period.
Unadjusted analyses showed that risk for acute events was higher for patients who were older, had more comorbidities, had chronic heart failure, diabetes, or ischemic heart disease, and higher drug cost-sharing. Results from a multivariable model yielded negligible differences in risk for acute events between drug cost-sharing groups and ranged from 0.6% (95% CI, 0.3, 0.9) for the lowest and 1.1% (95% CI, 0.8, 1.5) for the highest costsharing group. The most significant predictor of acute events was comorbidity.
Conclusion: There was no evidence that increased cost-sharing for antihypertensive drugs, leading to lower drug adherence, significantly worsened acute outcomes. While management of hypertension is an important aspect of preventing poor cardiovascular outcomes, greater adherence to these drugs may not have much influence over outcomes and costs in the short term.
Implications for Policy, Delivery or Practice:
Strategies to reduce cardiovascular disease should continue to stress multiple risk factors.
Funding Source(s): AHRQ, VA HSR&D
Theme: Prevention and Treatment of Chronic
Illness
● How well do Pregnant Patients Understand
Preeclampsia?
Whitney You, M.D.; Michael Wolf, Ph.D., M.P.H.;
William Grobman, M.D., M.B.A.
Presented by: Whitney You, M.D., Fellow,
Obstetrics & Gynecology, Northwestern University,
250 East Superior Street, Suite 05-2175, Chicago,
IL 60611, Phone: (312) 472-4685; Email: wyou@md.northwestern.edu
Research Objective: Preeclampsia is a pregnancy specific syndrome that affects up to 8% of the gravid population and is a leading cause of maternal mortality in the United States. Despite its prevalence, predicting which patients are at greatest risk for morbidity related to preeclampsia remains a challenge. Therefore, it is important that all patients have at least a basic understanding of the disease. Currently, little is known about what patients know or understand about preeclampsia.
The objective of this study is to explore the extent to which pregnant women understand the symptoms and potential complications related to preeclampsia.
Study Design: This was a cross sectional study where pregnant patients were interviewed to obtain information about their preeclampsia knowledge.
Additional information about demographics, medical and obstetrical history, and health literacy was obtained. Health literacy was assessed utilizing the
Test of Functional Health Literacy in Adults
(TOFHLA). Preeclampsia knowledge was evaluated utilizing a questionnaire that asked series of questions regarding the symptoms, consequences, and proper responses associated with the preeclampsia. Patients were also asked in an open-
ended question to give a definition of preeclampsia and all of their responses were independently rated by 3 obstetricians, each blinded to patient information. When a response received a discordant rating the final rating was determined by consensus.
Population Studied: Women between 18 and 40 weeks of gestation who were at least 18 years of age and received care in a university indigent clinic.
Principal Findings: Patients correctly responded to only 43% of the questions assessing preeclampsia knowledge correctly. Moreover, eighty six percent of patients were not able to provide an adequate definition of the syndrome.
Factors associated with a greater proportion of correct answers were: higher health literacy, multiparity, history of preeclampsia in a prior pregnancy, receiving provider counseling addressing preeclampsia, and hearing about preeclampsia from a source other than their provider (i.e., internet, TV, book, friend).
Conclusion: Pregnant patients have a poor understanding of preeclampsia. Those with prior exposure to the term through experience, provider counseling, or lay media seem to demonstrate more knowledge about the disease.
Implications for Policy, Delivery or Practice: As preeclampsia can have devastating maternal and neonatal consequences, it is imperative that women ultimately diagnosed with the syndrome seek medical care. In our study women who received information about preeclampsia from their provider or alternative source scored better on our preeclampsia questionnaire. This suggests that programs educating women about preeclampsia may improve their understanding of the syndrome.
Initiatives for offering such education should be developed so that they are both accessible and comprehensible to patients of all literacy levels as women with lower literacy levels demonstrated the greatest deficit in understanding.
Funding Source(s): AHRQ
● Validating a Severity Score to Identify Women with Near Miss Obstetric Morbidity
Whitney You, M.D.; Suchitra Chandrasekaran,
M.D.; John Sullivan, M.D.; William Grobman, M.D.,
M.B.A.
Presented by: Whitney You, M.D., Fellow,
Obstetrics & Gynecology, Northwestern University,
250 East Superior Street, Suite 05-2175, Chicago,
IL 60611, Phone: (312) 472-4685; Email: wyou@md.northwestern.edu
Research Objective: Maternal mortality is a rare occurrence in the United States and consequently is not an adequate marker for the quality of maternal health care. To evaluate the impact of various obstetric interventions across health care systems it is imperative that a universal and reproducible measure of maternal health is utilized.
Near miss morbidity is defined as severe morbidity that puts a patient at considerable risk of death and is more common than maternal mortality. As such, near miss morbidity may provide a useful outcome for case review and a measure for epidemiologic data collection. An objective definition for near miss morbidity in obstetrics has not been adequately researched or validated. Several methods for identifying high risk morbidity have been proposed in the surgical literature but are not appropriate for a peripartum population. Geller et al proposed a weighted 5 factor scoring system to identify women with near miss morbidity. The accuracy of this system has not been described in a patient population other than the one in which the system was established. The objective of this study is to validate the Geller scoring system in a patient population other than the one in which it was created.
Study Design: This was a retrospective cohort study conducted over a 2 year period. Women with a high potential for significant obstetric morbidity were identified through ICD-9 codes, their medical records were obtained and a narrative summary of their hospital course was prepared. The summary was reviewed by an obstetrician and the degree of morbidity was characterized into one of the following 4 categories: none, minor morbidity, severe morbidity, and near miss morbidity. The same cases were then scored utilizing the five factor scoring system developed by Geller et al. to identify cases that would be classified as near miss morbidity. Sensitivity, specificity, and positive and negative predictive values were calculated.
Population Studied: Pregnant women receiving care from a high volume, urban, tertiary care center.
Principal Findings: Eight hundred and fifteen women with a high potential for significant morbidity were identified. Clinical review of these cases identified 23.2% with no morbidity, 52.1% with minimal morbidity, 19.2% with severe morbidity, and 4.6% with near-miss morbidity. When these cases were instead categorized based on the Geller scoring system 4.2% of the cases where categorized as near-miss morbidity. Thus, in our population the sensitivity, specificity, and positive and negative predictive value of the 5 factor scoring system were 63.2%, 98.7%, 70.5% and 98.2% respectively.
Conclusion: In our population, the Geller scoring system had excellent specificity and reasonable sensitivity for identifying near-miss morbidity.
Further research may determine if additional factors will help capture those cases of near-miss morbidity that were not identified by the 5 factor system.
Implications for Policy, Delivery or Practice:
Near miss morbidity that is accurately identified through administrative data using a validated scoring system could be used in epidemiologic surveillance of maternal health and to assess the impact of quality control measures to improve obstetrical practice.
Funding Source(s): AHRQ
● Should Medicaid Enrollees Health Beliefs be an Important Factor in the Consumerism
Debate?
Gail Young, M.A., Ph.Dc.; Keva Thompson, M.P.H.;
Allyson Hall, Ph.D.; Kimberly Elliott, B.S.; Lilly Bell,
M.H.A.; Rada Dahger, Ph.D.
Presented by: Gail Young, M.A., Ph.Dc., Research
Assistant, Health Education & Behavior, University of Florida, P.O. Box 118210, Gainesville, FL 32611,
Email: gaildale@phhp.ufl.edu
Research Objective: Health care consumerism advocates for patients to be more active in their health care decision-making. This is an explicit goal of the Florida’s Medicaid Reform demonstration program. However, a gap in the literature exists regarding consumers’ general health beliefs and regarding controlling their health and experiences. This study explores enrollee health beliefs in light of Medicaid Reform policy changes. Highlighting Medicaid enrollees health beliefs in specific reform counties in Florida will help ensure that newly designed policy is in accordance with the health beliefs of consumers.
Study Design: This qualitative study used a combination of individual telephone interviews and in-person focus groups with Medicaid enrollees in 5 counties. There were 85 male and female participants in this study. The study period was from 2006-2008. Participants were recruited using community liaisons, Agency for Health Care
Administration eligibility/enrollment file data pulls, and advocacy groups. Data analysis was facilitated by Atlas.ti 5.0 software. A coding algorithm was developed and overarching themes were identified and verified by the study team.
Population Studied: The sample consisted of adult and child Medicaid enrollees’ eligible to participate in Medicaid Reform in Florida. Participant’s ages ranged between 12 and 73. There were 31 participants for individual interviews and 54 participants for the 7 focus groups. Females made up approximately two thirds of the sample.
Principal Findings: Enrollees health beliefs were divided into subthemes including: general understanding of health; health as a life experience; health as a function or action; health as a value; health defined based on the health care system; dimensions of health; and the ability to control health. In general, enrollees were concerned about their health. Health behaviors were affected by their ability to control their health. Ability to control their health was based on external factors such as the lack of money and the health care system. Other factors included individual ability, God and prayer, and the role of others. Good health was defined in various ways based on their quality of life, ability to complete certain activities of living, and the value of being in good health. A strong appreciation was expressed for Medicaid in facilitating access to health care and in ensuring good health. Factors within their control were based on general definitions of health such as quality of life. The health care system was regarded as being outside of their control, but various components of the system such as the clinician’s role, had inherent value both within and outside their realm of control.
Conclusion: Control of health and behaviors depended on enrollees’ heath beliefs. Health beliefs were based on both factors inside and outside of the individual’s control.
Implications for Policy, Delivery or Practice:
There is value in understanding the role of health beliefs in consumerism as a way to manage enrollee directed care and potentially reduce costs.
The study suggests that Medicaid enrollees’ beliefs may have some impact on their ability to control, manage and participate in their health care decision making and behaviors. Further, the consumerism shift also demands that policymakers design programs that increasingly concur with consumers’ health beliefs.
Funding Source(s): Agency for Health Care
Administration
● A Cost-Benefit Analysis of Higher Medication
Copayments in Veterans with Schizophrenia
John Zeber, Ph.D.; Luci Leykum, M.D.; Laurel
Copeland, Ph.D.; Alexander L Miller, M.D.; Marcia
Valenstein, M.D.
Presented by: John Zeber, Ph.D., Investigator,
VERDICT, VA Health Services Research &
Development, 7400 Merton Minter Boulevard
(11c6), San Antonio, TX 78229, Phone: (210) 617-
5300; Email: zeber@uthscsa.edu
Research Objective: Medication non-adherence represents a significant problem for patients with schizophrenia, resulting in deteriorating symptomology and functional status, along with a substantially increased risk of psychiatric admission. The 2002 Veterans Health Care Act
(Millennium Bill) raised outpatient medication copayments from $2 to $7 in patients with minimal financial or other healthcare resources. From the
VA’s perspective, such major health policy decisions should balance anticipated financial benefits with potential costs associated with unintended clinical consequences. Expanding upon our prior work, the current study documents the cost-offset of copayment revenue versus higher inpatient and emergency department (ER) costs.
Study Design: Pharmacy prescriptions, health services utilization, and treatment costs were analyzed throughout the 66 months surrounding this health policy change (11 quarters both Pre and
Post February 1, 2002). This observational study calculated additional copayment revenue versus changes in health services utilization costs (1999 adjusted dollars), contrasting veterans subject to
copayment increases with a natural control group of exempt patients.
Population Studied: A national population of all veterans diagnosed with schizophrenia that were treated by the VA health system (N=69,986).
Principal Findings: Although Exempt veterans
(36,555) were slightly sicker in terms of higher baseline pharmacy and inpatient utilization, these trends for both groups prior to the copayment increase were quite similar. However, during the
Post period, compared to no change observed in
Exempt patients, total overall prescriptions for
Copayment veterans (N=33,431) continued increasing slightly while psychotropic fills dropped substantially, nearly 20%. During the same period, mean psychiatric admissions and hospital days for this group rose 4%, reversing steady downward trends observed over the past decade. The total prescriptions increase yielded $17.3 million in new copayment revenue, but with higher pharmacy costs of $5.5 million. Furthermore, inpatient and ER costs increased $13.3 million and $0.6 million, respectively. Therefore, the net cost-benefit revenue change from the VA perspective was a negative $2.1 million, or $745,000 annualized losses. Sensitivity analyses altering predicted utilization costs and the proportion of post-policy changes due to higher copayments produced annualized cost-benefits ranging from -$1.4 million to $0.3 million (gain).
Conclusion: This descriptive study suggests that the policy change translated into greater VA copayment revenue while possibly dampening overall pharmacy cost increases. However, unanticipated clinical consequences included sharply reduced psychotropic fills leading to poorer adherence and higher utilization. Recognizing blunt calculations and complex causal assumptions, the
VA nevertheless appeared to experience financial losses while simultaneously reducing veterans’ quality of life.
Implications for Policy, Delivery or Practice:
Policy changes targeting essential pharmacy benefits for vulnerable veterans with schizophrenia should be implemented carefully, recognizing tradeoffs between immediate financial gains and the potential costs associated with clinical deterioration.
Longer term studies are needed to gauge the sustained effect over time as veterans reconcile their behaviors with higher medication expenses.
Funding Source(s): VA
Theme: Behavioral Health
● Does Good Academic Performance Prevent
Suicide? Evidence from the Accountability
Laws
Ning Zhang, Ph.D.; Feng Liu, Ph.D.; Helen
Schneider, Ph.D.
Presented by: Ning Zhang, Ph.D., Assistant
Professor, Community & Preventive Medicine,
Univerity of Rochester School of Medicine, 601
Elmwood Avenue, Box 644, Rochester, NY 14642,
Phone: (585) 275-0165; Email: ning_zhang@urmc.rochester.edu
Research Objective: Suicide has become the third leading cause of death among American youths
Many social factors may be associated with these behaviors, academic performance that causes academic pressure, is one such factor. However, the causal impact of these factors is difficult to identify because adolescents who have mental disorders may have poorer academic achievement than those who do not.
The objective of this paper is to investigate whether good academic performance prevents suicides. It has been shown that students’ scores improved as the consequences of the accountability laws introduced by the No Child Left Behind (NCLB) Act
2001. Academic performance can affect suicidal behaviors in two ways. On one hand, the improvement of scores at school could ease the mental pressures and hence, decrease the risk for mental disorders and for suicides. On the other hand, being under pressure to succeed could increases the risk for mental disorders. Which mechanism dominates is uncertain.
Study Design: The study is based on the data from the Youth Risk Behavior Surveillance System
(YRBSS) (1999 -2005). Suicidal behaviors are reported as attempts to commit suicide and any suicide committed during the previous 12 months.
Two analyses are conducted: one estimates the total effect of the implementation of accountability laws on suicide attempts and the suicide. The second examines the differences in changes in suicidal behavior with the strictness of laws, measured by the number of law components in each state.
Population Studied: All school students were selected for the study if they answered the questions regarding the suicide attempts and behaviors.
Principal Findings: During these years, 14% of adolescents attempted to commit a suicide and 3% of them actually did. The preliminary results show that students in states which pioneered to adopt accountable system were 4 % more likely to have planned suicide compared to others. Among whites, laws increased the suicide plans by 7% while little effects on real suicide. For blacks, these laws predicted 10 percentage points increase in suicide plan. Even worse, the laws increased the real suicide by 4 percentage points. Moreover, results presented that on average, students living in states with more strict accountability laws were 1% more likely to report to have a suicide plan or actually commit such plan, though little effects were detected among whites. The laws increased the suicide rates among blacks by 7%.
Conclusion: The NCLB aimed to improve the educational achievement for students and ultimately, accumulation of their human capital.
However, our research suggests that such progress comes with higher academic pressure and higher risk for mental disorders. The evidence from this study shows that although black students experienced a greater improvement in academic scores as a result of these laws, they also demonstrated greater improvement in in suicide behaviors. Overall, the better school performance was associated with higher suicide rate among blacks.
Implications for Policy, Delivery or Practice: The policy makers should carefully consider psychological pressure brought about by the accountability system.
Theme: Child Health
● The Impact of Medicare Part D on Medication
Adherence & Use of Medical Care Services
Yuting Zhang, Ph.D.; Judith Lave, Ph.D.; Julie
Donohue, Ph.D.; Michael Fischer, M.D.; Michael
Chernew, Ph.D.; Joseph Newhouse, Ph.D.
Presented by: Yuting Zhang, Ph.D., Assistant
Professor, Health Policy & Management, University of Pittsburgh, 130 DeSoto Street, A664 Crabtree,
Pittsburgh, PA 15261, Phone: (412) 383-5340;
Email: ytzhang@pitt.edu
Research Objective: Medicare Part D was enacted to protect beneficiaries against catastrophic drug expenditures and reduce cost-related medication underuse. Its actual effects on adherence are largely unknown. This study evaluates Part D’s impact on adherence and subsequent medical care use among those with hyperlipidemia, hypertension, and/or diabetes.
Study Design: Using a pre-post-with-comparisongroup design, we examined adherence and medical use pre- and post- Part D among four groups of elderly Medicare beneficiaries continuously enrolled between 1/1/2003-12/31/2007 with a Pennsylvania insurer. Three intervention groups, totaling 14,965, had no or limited drug benefits prior to Part D; the other 5,924 (comparison group) had stable coverage through former employers throughout. We assessed three measures of medication adherence in all groups: medication possession ratios
(proportion of days covered of any medication used to treat the illness), an indicator of good adherence, and treatment intensity (multidrug therapy), as well as five measures of medical care use: total medical spending; spending on and number of outpatient visits; and likelihood of emergency department and hospital admissions.
Population Studied: We obtained enrollment, benefits, and pharmacy and medical claims information on a 40 percent random sample of individuals who were continuously enrolled with the insurer between January 2003 and December
2007. We identified our study population by selecting individuals who had at least two claims with a diagnosis for one or more of the selected chronic diseases in 2003 (hypertension (ICD-9 401,
402, 403, 404), hyperlipidemia (ICD-9 272.0-272.4), and diabetes (ICD-9 250)) and who filled at least one prescription in 2003 for a medication for the diagnosed condition
Principal Findings: Part D improved medication possession ratios most in the group with no prior drug coverage, by 0.14 (95% CI 0.11-0.17), 0.18
(95% CI, 0.14-0.22), and 0.14 (95% CI, 012-0.16) for those with hyperlipdemia, diabetes, and hypertension, respectively. Adherence improved by lesser amounts in the other two groups with limited drug benefits prior to Part D. Results for treatment intensity were similar. Improved adherence reduced the number of outpatient visits (-3, 95% CI
-4.84 - -1.17) and medical spending (-$741, Pvalue<0.051) among a group with a $150 quarterly cap with hypertension, but not in other groups.
Conclusion: Part D improved adherence most among those with no prior drug coverage. Whether there were any cost offsets from this improvement is ambiguous.
Implications for Policy, Delivery or Practice:
Nevertheless these results strengthen the case that
Part D gives the elderly better protection against financial risk and is improving the health of the elderly.
Theme: Medicare
● Choosing to Work With Underserved Patients:
A Multiple Case Study
Wanqing Zhang, M.D., M.Ed.; Mark Christiansen,
M.S., MPAS
Presented by: Wanqing Zhang, M.D., M.Ed.,
Research Associate, Health Service Research &
Administration, University of Nebraska Medical
Center, 984350 University of Nebraska Med Center,
Omaha, NE 68198-4350, Phone: (402) 559-8441;
Email: wazhang@unmc.edu
Research Objective: Little is known about how and why of health care practitioners vary in their propensity to care for underserved patient populations. The purpose of this study is to describe the reasons that health care practitioners choose to work with underserved patients. By gaining an understanding of these factors we hope to obtain insights that might lead others to choose this type of practice in their careers.
Study Design: A multiple case study approach to qualitative inquiry was chosen for this study.
Collective case study data were obtained through one-on-one interviews with a purposeful sample of six individuals. A case study approach was appropriate to obtain in-depth understanding of the issues involved in health care delivery to underserved populations and to make comparisons between different types of health care practitioners and different types of health care settings.
Population Studied: Six individuals were chosen based upon knowledge of their practice settings
and health care discipline. The intention was to obtain maximal variation by selecting two individuals from each of the three disciplines of interest: physicians, physician assistants and nurse practitioners, and by choosing a variety of health care settings to include both urban and rural underserved populations.
Principal Findings: The type and degree of involvement with underserved patients could be different for each of these cases. The health care practitioners we interviewed reflect their experiences and attitude toward working with underserved patient populations. The comprehensive cross-case theme analysis reveals the similarities and the differences. The similarities include: (1) the problems with US health care systems; (2) choosing to work with underserved patients due to some background experiences; and
(3) language and culture barriers existed for underserved populations. The differences include:
(1) various primary care settings including a mobile center, a rural health clinic, a SHARING clinic, and a homeless shelter clinic; and (2) the type and degree of involvement with underserved patients.
Conclusion: This multiple case qualitative investigation indicates that dedication and compassion are key individual level factors for the health care practitioners who have chosen to work with underserved patient populations or communities. Cultural and language barriers can affect access to good health care among underserved patient populations. From the system level, issues such as resources, funding, universal health coverage, and recruitment of primary health care practitioners in underserved areas emerged during the data analysis.
Implications for Policy, Delivery or Practice: We feel that among the most complex and challenging issues facing the U. S. health care system is improving access to health care among patients in need. The focus is on the populations in need, including the uninsured or underinsured, racial and ethnic minorities, and geographically isolated population in remote rural area. Many small steps will need to be taken to have an impact on the much larger problem. As we have seen from the participants in this study, a few individuals who are committed to this concept can make a significant difference.
Theme: Disparities
● Value-Based Cost-Sharing for
Antihypertensive Medications
Yuting Zhang, Ph.D.; Julie Donohue, Ph.D.; Judith
Lave, Ph.D.; Joseph Newhouse, Ph.D.
Presented by: Yuting Zhang, Ph.D., Assistant
Professor, Health Policy & Management, University of Pittsburgh, 130 DeSoto Street, A664 Crabtree,
Pittsburgh, PA 15261, Phone: (412) 383-5340;
Email: ytzhang@pitt.edu
Research Objective: Half of those 60-69 years of age and three-fourths of those 70 years of age and older have hypertension, but only about 60 percent of Americans with hypertension are receiving treatment and only one-third have their blood pressure appropriately controlled. Results from trials have showed anti-hypertensive medications are effective to control blood pressure and prevent more severe heart diseases. Medicare Part D was enacted to protect beneficiaries against catastrophic drug expenditures and reduce costrelated medication underuse. Its actual effects on adherence are largely unknown. This study evaluates Part D’s impact on adherence for major drug classes used to treat hypertension and prevent heart related diseases.
Study Design: Using a pre-post-with-comparisongroup design, we examined adherence and medical use pre- and post- Part D among four groups of elderly Medicare beneficiaries continuously enrolled between 1/1/2003-12/31/2007 with a Pennsylvania insurer. Three intervention groups, totaling 14,965, had no or limited drug benefits prior to Part D; the other 5,924 (comparison group) had stable coverage through former employers throughout. We assessed three measures of medication adherence in all groups: medication possession ratios (MPR, proportion of days covered of any medication used to treat the illness, ranging from 0 to 1), an indicator of good adherence, and treatment intensity
(multidrug therapy), as well as five measures of medical care use: total medical spending; spending on and number of outpatient visits; and likelihood of emergency department and hospital admissions.
Population Studied: We obtained enrollment, benefits, and pharmacy and medical claims information on a 40 percent random sample of individuals who were continuously enrolled with the insurer between January 2003 and December
2007. We identified our study population by selecting individuals who had at least two claims with a diagnosis for hypertension (ICD-9 401, 402,
403, and 404) and filled at least one antihypertensive prescription in 2003. The major drug classes in this study include beta blockers, diuretics, angiotensin converting enzyme inhibitors
(ACEI), angiotensin II receptor blocker (ARB) and calcium channel blockers (CCB).
Principal Findings: Part D improved MPRs most in the group with no prior drug coverage, by 0.14
(95% CI, 012-0.16) to a level of 0.75 post Part D, or
23% increase, for any anti-hypertensive use. The percentage increase in the pre- and post- Part D
MPRs for ACEI, CCB and diuretics were similar to the overall increase in MPR for any antihypertensive drug, but use of ARB and beta blockers increased notably larger, by 41% and 35% respectively. Adherence improved by lesser amounts in the other two groups with limited drug benefits prior to Part D. Results for treatment intensity were similar. Improved adherence reduced the number of outpatient visits (-3, 95% CI -4.84 - -
1.17) and medical spending (-$741, P-value<0.051) among a group with a $150 quarterly cap with hypertension, but not in other groups.
Conclusion: Part D improved adherence most among those with no prior drug coverage. Use of
ARB and beta blockers responds to the price change more than other drug classes. Improvement in adherence to antihypertensive medication appears to reduce use of other non-drug medical services.
Implications for Policy, Delivery or Practice: Part
D improves adherence for antihypertensive drugs and improves the health of the elderly.
Policymakers could differentiate cost-sharing for drugs with larger responsiveness.
Theme: Prevention and Treatment of Chronic
Illness
● Health Benefits & Medical Cost Savings
Attributable to Calorie, Sodium & Saturated Fat
Reductions in the American Diet
Yiduo Zhang, Ph.D.; Timothy, Dall, M.S.; Victor
Fulgoni, Ph.D.; Kristin Reimers, Ph.D.; Patty
Packard, Ph.D.; James Astwood, Ph.D.
Presented by: Yiduo Zhang, Ph.D., Senior
Research Associate, The Lewin Group, 3130
Fairview Park Drive, Suite 800, Falls Church, VA
22042, Phone: (703) 269-5603; Email: yiduo.zhang@lewin.com
Research Objective: Model the potential long term national health benefits and medical savings from reduced daily intake of calories, sodium, and saturated fat from the American diet
Study Design: We developed a Nutrition Impact
Model to estimate the current prevalence, health care use, and national medical costs for chronic conditions attributed to three risk factors—excess weight, uncontrolled hypertension, and elevated cholesterol. These chronic conditions include arthritis, asthma, cancer, cerebrovascular disease, congestive heart failure, coronary heart disease, diabetes mellitus, esophagus/stomach disease, gallbladder diseases, gynecological abnormalities, hypertension, kidney and urinary disease, other cardiovascular disease, respiratory problems, and sleep apnea. Then, we simulate the potential reduction in these three risk factors through hypothetical changes in diet. We compare current health care use and cost attributed to these risk factors with predicted utilization and costs if the nation reached a new health equilibrium through improved diet. The adult population was divided into 480 unique risk groups across 5 dimensions: age (18-44, 45-54, 55-64, 65-74, 75+); gender; four weight categories defined using body mass index
(BMI)—normal weight (18.5<BMI<25), overweight
(25<=BMI<30), obese class I (30<=BMI<35), and obese classes II & III (BMI<=35); three blood pressure categories defined using systolic (SBP) and diastolic (DBP) blood pressure—normal pressure (SBP<120 and DBP<80 mm Hg), prehypertension (120<=SBP<140 and 80<=DBP<90), and hypertension (SBP<=140 or DBP<=90); and four cholesterol levels defined using LDL-C level— optimal (LDL-C<100 mg/dL), near optimal
(100<=LDL-C<130), borderline high (130<=LDL-
C<160), and high (LDL-C>=160)
Population Studied: Using data from 1999–2004
National Health and Nutrition Examination Survey
(N=17,061) and 2000-2004 Medical Expenditure
Panel Survey (MEPS), we estimated the prevalence rates for an array of chronic conditions to establish the health risk profile for each unique risk group at baseline. Authoritative scientific reports and peerreviewed literature (described below) were used to quantify the relationships between dietary change and BMI, SBP/DBP, and LDL-C and between these same factors and disease risk. Estimates are based on 224 million adults in the U.S.
Principal Findings: Permanent 100 kcal daily energy deficits within 3-5 years could eliminate approximately 71.2 million cases of overweight and obesity. A permanent shift in prevalence rates would lead to a new equilibrium in prevalence of coronary heart disease (1.7 million fewer cases), type 2 diabetes (1.5 million fewer cases), and other chronic conditions. Under this new equilibrium, national medical expenditures would be approximately $58 billion lower than current spending levels. Long term sodium reductions of
400 mg in uncontrolled hypertensives would eliminate about 1.5 million cases, saving $2.3 billion; and decreasing 5 grams of saturated fat in those with elevated cholesterol would eliminate 3.9 million cases, saving $2.0 billion
Conclusion: A modest 100 calorie reduction and more aggressive sodium and saturated fat reductions can improve health and reduce annual national medical expenditures by $60 billion.
Maximal dietary reductions are estimated to reach
$120 billion savings.
Implications for Policy, Delivery or Practice:
One use of the model is to estimate the impact of dietary change related to setting public health priorities for dietary guidance. The findings here argue that emphasis on reduction in caloric intake should be the highest priority
Funding Source(s): ConAgra Food, Inc
Theme: Prevention and Treatment of Chronic
Illness
● The Impact of Medicare Prescription Drug
Benefit on Antibiotic Medication Use
Yuting Zhang, Ph.D.; Bruce Lee, M.D.; Julie
Donohue, Ph.D.; Judith Lave, Ph.D.
Presented by: Yuting Zhang, Ph.D., Assistant
Professor, Health Policy & Management, University of Pittsburgh, 130 DeSoto Street, A664 Crabtree,
Pittsburgh, PA 15261, Phone: (412) 383-5340;
Email: ytzhang@pitt.edu
Research Objective: Evidence suggests that there continues to be significant antibiotic overuse, which has promoted the spread of antibiotic resistant bacteria. To curb antibiotic overuse, hospital epidemiologists and policymakers have focused on educational programs and formulary restrictions.
However, few studies have looked at the effects of reimbursement policy on antibiotic use. Since antibiotics are designed to treat infections, appropriate antibiotic use should be relatively price inelastic. The 2006 implementation of Medicare
Part D, a US federal program to subsidize outpatient prescription drug for the elderly, presents an opportunity to determine how a major prescription drug reimbursement change may have affected antibiotic use.
Study Design: Using a pre-post-with-comparisongroup design, we examined adherence and medical use pre- and post- Part D among three groups of elderly Medicare beneficiaries continuously enrolled between 1/1/2003-12/31/2007 with a Pennsylvania insurer. One intervention group had no drug coverage and the other had a quarterly cap of $150 drug coverage prior Part D and switched to more generous Part D coverage in 2006; the comparison group had stable coverage through former employers before and after Part D. We used propensity score weighting to enhance comparability. We assessed Part D’s impact on antibiotic use, stratified by major antibiotic classes: penicillins, cephalosporins, macrolides, tetracyclines, and others.
Population Studied: We obtained enrollment, benefits, and pharmacy and medical claims information on a 40 percent random sample of elderly Medicare beneficiaries who were continuously enrolled with a Medicare Advantage insurer between January 2003 and December
2007.
Principal Findings: Between December 2005 and
December 2007, compared with spending in the comparison group of enrollees with drug coverage without caps before and after Part D, total antibiotic prescriptions increased 4% (95% CI, 0-8%) among those with no prior drug coverage, and 5% (95% CI,
0-12%) among those with prior quarterly caps of
$150. Use of new antibiotic class (macrolides) did not change. Use of penicillins declined post Part D.
The biggest increase in use came from tetracyclines: those without drug coverage increased number of antibiotic prescriptions by 0.5 per year, while as those with $150 cap increased number of prescriptions by 0.2 per year on average.
Conclusion: In our study, antibiotic use increased modestly among elderly patients whose drug coverage improved post Part D. This suggests that some antibiotic use may be price-elastic and may be inappropriate.
Implications for Policy, Delivery or Practice:
Reimbursement policy may have a role in antibiotic use control. Further studies on appropriate and inappropriate use of antibiotics at the diseasespecific level are warranted.
Theme: Public Health
● Children’s Mental Health: Does Economy
Matter?
Ning Zhang, Ph.D.; Feng Liu, Ph.D.
Presented by: Ning Zhang, Ph.D., Assistant
Professor, Community & Preventive Medicine,
University of Rochester School of Medicine, 601
Elmwood Avenue, Box 644, Rochester, NY 14642,
Phone: (585) 275-0165; Email: ning_zhang@urmc.rochester.edu
Research Objective: Losses of employment and income are two sources of stressful life events, insecure attachment, unhealthy habits among parents, which are highly related to child mental disorders. However, the causal impact of these factors on onset and development of child mental problems is not yet identified because the treatment of child mental problems may affect parents’ decision of labor market performance and family income. The objective of this paper is to investigate the relationship between the unemployment rate at a child’s conception year and at pre-kindergarten year and the onset of Attention Deficit /
Hypertention Deficit (ADHD), the most common mental disorders that affect millions of American children.
Study Design: The primary data are from the restricted files of Early Child Longitudinal Survey
Kindergarten – fifth grade. These data were merged with the state-wide unemployment rate in parents' conception year, survey year and lags of unemployment rate between those two years. Two analyses will be conducted. One is to investigate whether and to what extend risks of job loss and income instability at birth, measured by statespecific unemployment rate, affect the onset of
ADHD. The other one is to explore whether the transitory downturns (the differences in unemployment rate between two adjacent years) and intermediate recession (the number of years when unemployment rate increases) have different affects.
Population Studied: Children were selected for analysis if their state of residence was identified and if their status of ADHD were validly recorded.
Principal Findings: On average, the onset of
ADHD was 1.3 age years earlier among children living in states with average higher unemployment rate before kindergarten years than those with lower rates. Morever, any 1% increase in unemployment rate was associated with 1.1% increase in probability of ADHD in early childhood, such effects were larger among black children (3%).
In addition, intermediate recession has large and significant effects on onset and diagnosis of ADHD.
In particular, the persistence of high unemployment
rate over 3 years increased the probability of ADHD by 2.1%.
Conclusion: The pressure of parental employment and income may increase the indidence of children's mental illnesses such as ADHD.
Implications for Policy, Delivery or Practice: The economy faces one of the serious recession currently. The policy makers should not only be alert to the negative effects of economy downturns on adults physical and mental health status, but also to the effects on children's mental health.
Theme: Public Health
● Utilization, Cost, Payment & Patient
Satisfaction of Rehabilitative Services in
Shandong, China
Mei Zhao, Ph.D.; D. Rob Haley, Ph.D.; JoAnn Nolin,
J.D.; Kerry Dunning, M.H.A.; Jian Wang, Ph.D.;
Qiangsan Sun, M.D.
Presented by: Mei Zhao, Ph.D., Assistant
Professor, Public Health, University of North
Florida, 1 UNF Drive, Jacksonville, FL 32224,
Phone: (904) 620-1444; Email: mzhao@unf.edu
Research Objective: Chinese health policy has focused on increasing integration of Traditional
Chinese Medicine (TCM) and western medicine as
China has moved to a market-based economy.
However, little research has been done to examine patient response to issues related to those services.
This study looks at utilization, cost, payment and patient satisfaction with rehabilitative services in
Shandong province.
Study Design: A structured questionnaire was designed to collect information from patients about rehabilitative services identified by the type of TCM and Western Medicine they received, their associated cost, their payer, as well as an indication of patient satisfaction with administered treatments.
Respondents were also provided with the opportunity to record open-ended comments at the end of the structured questionnaire. The questionnaire was translated into Mandarin and was pre-tested and refined with patients for clarity and validity. Pilot respondents were also asked to suggest treatments that were not listed on the questionnaire.
Population Studied: A sample of 192 patients receiving rehabilitative treatments in four hospitals located in Shandong, China in July 2007 were surveyed by the researchers.
Principal Findings: Most of the patients in this study came from urban areas and the majority of the participant had a minimum of a high school education and almost 85% had some form of health insurance. The most frequently ordered TCM therapies were acupuncture (14.1%) and massage therapy (15.6%) and western therapies were physical therapy (62.5%) and occupational therapy
(6.3%). Physical therapy was the most cost effective service at almost half the cost of acupuncture and massage therapy. Patient satisfaction with service was high at 90%.
Conclusion: Our analysis indicated that rehabilitative patients received both TCM and western medicine treatments in urban areas in
Shandong, China. Patients were generally satisfied with the services obtained. Additional studies are needed to assess difference in use and acceptance of rehabilitation services by rural populations.
Implications for Policy, Delivery or Practice:
This research provides insight to suggest the following policy changes: Providing access to rehabilitative services within urban hospitals for those who live in rural areas; Subsidizing and training rural clinicians and paraprofessionals to cost-effectively provide rehabilitative services to
China’s rural populations; Encouraging clinicians to allow patients to participate in their treatment decisions; Developing new insurance models that will encourage healthcare coverage for China’s rural population and will include benefits for chronic conditions prevalent in aging populations;
Mandating the inclusion of rehabilitative service benefits in existing health plans; and Encouraging the use of both TCM and Western Medicine when providing rehabilitative services.
Funding Source(s): UNF Brooks Foundation
Theme: Global Health
● Predictors of Pain Medication Selection among Patients Diagnosed with Diabetic
Peripheral Neuropathic Pain
Yang Zhao, Ph.D.; Luke Boulanger, M.A., M.B.A.;
Ning Wu, Ph.D.; Kimberly Fraser, B.A.
Presented by: Yang Zhao, Ph.D., Global Health
Outcomes, Eli Lilly & Company, 1400 West
Raymond Street, Indianapolis, IN 46221, Phone:
(317) 651-3154; Email: zhao_yang_yz@lilly.com
Research Objective: Multiple pharmacologic therapies have been recommended for treating diabetic peripheral neuropathic pain (DPNP).
However, the factors associated with each treatment initiation have not been well established.
This study assessed demographic and clinical predictors of duloxetine versus other pain medications dispensed among patients with DPNP.
Study Design: Employing a retrospective cohort design and data from commercial and Medicare supplemental insurance separately, we examined predictors of treatment among DPNP patients who initiated duloxetine versus tricyclic antidepressants
(TCAs), venlafaxine, gabapentin, pregabalin, or narcotics between 7/1/2005 and 12/31/2006.
Treatment initiation was defined as no access to the same medication over the previous 90 days. All patients selected had at least 1 diabetic neuropathy diagnosis and 1 pain medication dispensed in the
12 months prior to initiation of each study therapy.
Multiple logistic regression models were estimated
to assess predictors of initiating duloxetine versus each of the other DPNP therapies.
Principal Findings: Of the identified patients, average age was 55 years for the commercially insured (n=8,965) and 75 years for the Medicare insured (n=7,528); cardiovascular disease (81% and 70%, respectively) and cerebrovascular/peripheral vascular disease (47% and 31%, respectively) were the most common diabetic-related complications. After controlling for demographic and clinical characteristics, commercially-insured patients who received TCAs, venlafaxine or tramadol in 12-month prior to index date were less likely, whereas patients who received pregabalin were more likely to initiate duloxetine. Other predictors of duloxetine initiation included physician specialty, and history of cerebrovascular disease and chronic lower back pain. Similar patterns were found in Medicareinsured patients except that medication prescribed by specialty physician was not significantly associated with duloxetine initiation.
Conclusion: Findings from this analysis indicate that physician specialty, presence of selected comorbidities, and prior use of certain medications to treat pain were significant predictors of duloxetine initiation. Demographic characteristics were not significantly associated with the initiation of duloxetine.
Implications for Policy, Delivery or Practice:
Funding Source(s): Eli Lilly and Company
Theme: Prevention and Treatment of Chronic
Illness
● Effect of C.N.A. Communication &
Coordination on End-of-Life Care Processes
Nan Zheng; Helena Temkin-Greener, Ph.D.,
M.P.H., M.S.
Presented by: Nan Zheng, Department of
Community & Preventive Medicine, University of
Rochester, 601 Elmwood Avenue, Box 644,
Rochester, NY 14642, Phone: (585) 273-2425;
Email: nan_zheng@URMC.rochester.edu
Research Objective: Today more than 20% of
Americans die in nursing homes, where quality of end-of-life (EOL) care is reportedly woefully inadequate. Several empirical studies addressing the reasons for the inadequate quality of care have demonstrated the impact of staff communication/coordination on nursing home performance. However, these studies have not specifically focused on the communication/coordination of the CNAs—the backbone of nursing home workforce. The objective of this study is to examine the relationship between
CNA communication/coordination and nursing home performance measured by EOL assessment of symptoms and care needs, and EOL care delivery.
Study Design: This study used data from two prior
(parent) nursing home studies. The first parent study measured the attributes of the work environment (i.e. leadership, communication/coordination, conflict management and cohesion) by surveying direct care workers
(including CNAs, nurses and others). This study was done in 2006-2007.The reliability and validity of the measurement tool among CNAs were demonstrated. CNA communication/coordination was assessed based on responses to a 5-point
Likert scale consisting of 15 items (Cronbach’s
Alpha = 0.83). The second parent study evaluated
EOL care processes in nursing homes by surveying directors of nursing (DONs). This study was done in
2007. The reliability and validity of the measurement tool were also demonstrated. A nursing home’s performance in EOL assessment and EOL care delivery was rated based on10-item
(Cronbach’s Alphas = 0.82) and 6-item (Cronbach’s
Alphas =0.72) scales respectively. Additional variables of interest were obtained from the Online
Survey, Certification and Reporting (OSCAR) dataset and from the Minimum Data Set (MDS).
The analysis is based on facility-level data.
Multivariate regression analysis, with probability weights to adjust for higher than expected proportion of non-profit facilities in the sample, was used to test the relationship between CNA communication/coordination and EOL assessment and care delivery, controlling for facility characteristics and staffing.
Population Studied: 108 nursing homes in New
York State participated in both parent studies. Two nursing homes did not have CNA respondents and therefore were excluded. Overall, 3,694 CNAs from
106 nursing homes composed the study sample.
Principal Findings: Better CNA communication/coordination is significantly associated with higher (better) EOL assessment scores (p = 0.003) and higher (better) EOL care delivery scores (p= 0.018). Facilities with greater racial/ethnic concordance between staff and residents demonstrated better EOL assessment scores (p = 0.063) and EOL care delivery scores (P
= 0.027). In addition, facilities with greater reliance on hospice are associated with better EOL care delivery (p = 0.012).
Conclusion: Our findings support the hypotheses that nursing homes perform better in EOL assessment and EOL care delivery if the CNAs experience better communication and coordination among their co-workers. Furthermore, we have identified several modifiable measures of nursing home structure, which are associated with better
EOL care processes.
Implications for Policy, Delivery or Practice: Our findings provide specific insights for nursing home leaders striving to improve EOL care processes and ultimately the quality of care for dying residents.
Funding Source(s): NIA, Foundation for Healthy
Living, Inc.
● Changes in Rural Health Insurance Coverage,
1997-2005
Erika Ziller, M.S.; Andrew Coburn, Ph.D.; Jennifer
Lenardson, M.A.; Nathaniel Anderson, M.P.H.;
Melanie Race, B.A.
Presented by: Erika Ziller, M.S., Research
Associate, Muskie School of Public Service,
University of Southern Maine, P.O. Box 9300,
Portland, ME 04104, Phone: (207) 780-4615;
Email: eziller@usm.maine.edu
Research Objective: Ample evidence from the
1980’s through late 1990’s indicates that rural residents were at greater risk of being uninsured than their urban counterparts. However, estimates of how this may have changed since then have been limited. The purpose of this study is to compare rural and urban changes in health insurance coverage between 1997 and 2005, to assess how public expansions (namely SCHIP) may have affected uninsured rates and sources of coverage.
Study Design: We use the 1997 and 2005 panels of the Medical Expenditure Panel Survey (MEPS), a nationally representative survey of U.S. residents conducted by the Agency for Healthcare Research and Quality (AHRQ). Employing a combination of bivariate and multivariate methods, including difference-in-difference modeling, we address two key research questions: 1) how have uninsured rates and sources of coverage changed for rural and urban residents? And, 2) whether the characteristics of the uninsured have changed over the eight-year span and if residence is a factor in any identified changes.
Population Studied: We studied the health insurance status of children and non-elderly adults in rural and urban areas. Residence is defined at the county level by modified rural-urban continuum codes (RUCCs) where metropolitan statistical areas
(MSAs) are classified as urban and non-MSA counties are divided into “rural, adjacent to an urban area” and “rural, non-adjacent” based on their proximity to urban counties.
Principal Findings: Our descriptive analyses revealed that the overall disparity in coverage between rural and urban residents was somewhat reduced between 1997 and 2005, although higher uninsured rates in rural areas persisted. However, separate analyses for adults and children yielded vastly different results. Among adults, both rural and urban residents experienced a shift from private coverage to greater public coverage, although changes in uninsured rates were largely unchanged. Among children, on the other hand, enrollment in public health insurance increased so much that the uninsured rate declined by approximately half in non-adjacent counties (versus about a third in urban counties). At the same time, private coverage erosion was more limited among rural than urban children (although in absolute percentages, rural children remain less likely to have private coverage) meaning that rural children actually experienced lower uninsured rates than their urban counterparts in 2005.
Conclusion: Our findings suggest that the expansion of public health insurance through
SCHIP has been effective at reducing rural-urban disparities in coverage rates, and even reversing them among children. To the extent that “crowd out” of private coverage occurred, this appears to be relatively balanced in rural and urban areas.
However, take-up of public coverage among uninsured children in rural areas appears to have been particularly high, suggesting that despite concerns about rural outreach, rural families are willing and able to participate in public programs when they are available.
Implications for Policy, Delivery or Practice:
These findings have important implications for the renewed efforts at health reform that are likely to be forthcoming. Given the apparent response to prior public health insurance expansions by rural residents, further expansions should be considered as a possible tool for increasing coverage. In addition, the continuing analyses planned for this study will identify the populations that did or did not benefit most from public expansions to aid in identifying where future efforts should be targeted.
Funding Source(s): HRSA
Theme: Coverage and Access
● Antidepressant Adherence after Psychiatric
Hospitalization
Kara Zivin, Ph.D.; Dara Ganoczy, M.P.H.; Paul
Pfeiffer, M.D.; Erin Miller, M.S.; Marcia Valenstein,
M.D.
Presented by: Kara Zivin, Ph.D., Assistant
Professor, Psychiatry, University of Michigan/ VA
Health Services Research & Development, 4250
Plymouth Road, Ann Arbor, MI 48105, Phone:
(734) 232-0538; Email: kzivin@umich.edu
Research Objective: Antidepressants are an important and effective component in the treatment of depressed patients, but research finds that antidepressant adherence is suboptimal. Poor adherence increases the likelihood of depression relapse or recurrence, and occurs for a number of reasons including medication side effects and patient beliefs. Depressed patients discharged from psychiatric hospitalizations face increased risks for adverse outcomes including suicide, yet adherence rates during this high-risk period are unknown. We assessed antidepressant adherence three and six months following psychiatric hospitalization and factors associated with poorer adherence.
Study Design: Eligible Veterans Affairs patients included those who: 1) received at least two depression diagnoses or one diagnosis followed by
an antidepressant fill between April 1, 1999 and
September 30, 2003; 2) were hospitalized for psychiatric reasons during that time; 3) received an antidepressant fill at discharge; 4) had an outpatient visit within three or six months post-discharge; and
5) lived at least 90 days following discharge. Using outpatient pharmacy data, we calculated medication possession ratios (MPRs) to examine whether patients had good adherence (MPRs = 0.8) or poor adherence (MPRs < 0.8) three and six months postdischarge, and used logistic regression analysis to identify predictors of poor adherence among patients who took only one type of antidepressant.
Population Studied: We used data from the VA
National Registry for Depression (NARDEP) to study patients who were recently discharged from a psychiatric hospitalization and had been prescribed an antidepressant post-hospitalization.
Principal Findings: 21,289 patients met criteria for three month MPRs and 23,582 met criteria for six month MPRs. The mean three month MPR was
0.79 (0.37); 49% had poor adherence. The mean six month MPR was 0.66 (0.40); 60% had poor adherence. Patients with poorer adherence were male, younger, non-white, Hispanic, and had a substance abuse disorder, but were less likely to have PTSD or other anxiety disorders.
Conclusion: Among depressed patients, poor antidepressant adherence is common during the post-hospitalization period. Additional efforts to promote adherence during this high-risk period may prevent adverse outcomes.
Implications for Policy, Delivery or Practice:
This study extends prior work on antidepressant adherence, and examines a high risk population and factors associated with their poor adherence.
More research is needed to determine why these patients, who are presumably in greater need of treatment, have low rates of antidepressant adherence following hospitalization. Our findings suggest that these patients would benefit from interventions aimed at improving antidepressant adherence.
Funding Source(s): VA HSR&D, NIMH
● The Impact of Depression & Cognitive
Impairment on Enrollment in Medicare Part D.
Kara Zivin, Ph.D.; Mohammed Kabeto, M.S.; Helen
Kales, M.D.; Kenneth Langa, M.D., Ph.D.
Presented by: Kara Zivin, Ph.D., Assistant
Professor, Psychiatry, University of Michigan/ VA
Health Services Research & Development, 4250
Plymouth Road, Ann Arbor, MI 48105, Phone:
(734) 232-0538; Email: kzivin@umich.edu
Research Objective: While the creation of
Medicare Part D enabled millions of beneficiaries to receive prescription drug benefits, we sought to examine concerns that vulnerable populations, such as depressed and/or cognitively impaired (CI) beneficiaries would have challenges accessing Part
D coverage.
Study Design: We used logistic regression analysis to assess whether elderly Medicare beneficiaries with depression and/or CI differentially planned to and actually signed up for Part D. Our outcome variables of interest were planned and actual enrollment in Part D. Our key independent variable was depression and/or CI status. We adjusted our analyses using clinical and demographic predictors including age, gender, race/ethnicity, educational attainment, net worth, marital status, health status, number of health conditions being treated with prescription medications, and presence of a caregiver.
Population Studied: We used 2004 and 2006 data from the Health and Retirement Study (HRS) including a subsample that completed the
Prescription Drug Study (PDS) in 2005.
Principal Findings: In unadjusted analyses, compared to those beneficiaries who did not plan to enroll, those with depression and CI were 9.08
(95% CI; 2.63, 31.30) times more likely to say they didn’t know if they would enroll and 5.94 (95% CI;
1.54, 22.89) times more likely to enroll than those without both depression and CI. In unadjusted analyses, people with both depression and CI were
2.61 (95% CI: 1.62, 4.20) times more likely to have signed up for Part D compared to those without either disorder. Those respondents who had depression only were also more likely to have signed up for Part D than those without either disorder (OR: 1.34, 95% CI: 1.17, 1.52), but those with CI alone were no more likely to sign up for Part
D than those without either disorder. However, in our adjusted model, respondents with depression and/or CI were no more likely to have enrolled in
Part D than respondents without either disorder.
Conclusion: Vulnerable Medicare beneficiaries with depression and/or cognitive impairment were as likely as beneficiaries without these disorders to have obtained Medicare Part D coverage when it became available in 2006.
Implications for Policy, Delivery or Practice:
More research is needed to determine how well
Part D meets the needs of these populations and whether they also have similar access to medications as their healthier counterparts.
Funding Source(s): NIA
● The Association of Provider Factors with
Quality Routine Pain Management
Lisa Zubkoff, Ph.D.; Karl Lorenz, M.D., M.S.HS;
Andy Lanto, M.A.; Cathy Sherbourne, Ph.D.; Lisa
Shugarman, Ph.D.; Steven Asch, M.D., M.P.H.
Presented by: Lisa Zubkoff, Ph.D., Post-Doctoral
Fellow, Veterans Administration Greater Los
Angeles Healthcare System, Department of
Veterans Affairs, 16111 Plummer Street (152),
Sepulveda, CA 91343, Phone: (818) 891-7711 x
7169; Email: lisa.zubkoff@va.gov
Research Objective: Research has shown suboptimal pain control in general medical populations. It is unclear, however, to what extent that shortfall is due to poor antecedent processes of care and/or other provider factors. The goal of this research was to examine the associations between provider characteristics and quality indicators for routine pain management in the Department of
Veterans Affairs.
Study Design: We developed four process quality indicators of good pain management for patients with moderate to severe pain: Q1) Is pain noted in progress notes? Q2) Is the character of pain described? Q3) Is the degree of pain control assessed? Q4) Do providers initiate or intensify treatment in response to moderate to severe pain?
Population Studied: We randomly sampled 140 patients with a numeric pain rating (NPR) of >= 4 on a 1-10 scale in clinics at two VA medical centers and 6 affiliated community sites from March 2006 to
March 2007. Trained nurse abstractors reviewed medical records to ascertain adherence with the indicators and other patient demographic and clinical factors. We also administered surveys to providers to ascertain provider attitudes toward pain treatment.
Principal Findings: Average (NPR) in this population with moderate to severe pain was 6.7.
Seventy-three percent of progress notes acknowledged the presence the pain (Q1), 13.9% mentioned the character (Q2), 23.6% noted about the degree of pain control (Q3), and 15.3% received an increase (versus no change or a decrease) in pain medication. Hierarchical logistic regressions were used to examine determinants of the quality indicators. None of the models were statistically significant when all variables were in the model.
However, each model had significant predictors. In the model predicting pain notation, providers agreeing that patients wanting them to ask about pain (OR: 4.5, 95% CI: 1.17-17.30) and pain having a negative consequence on functioning (OR: 6.71,
95% CI: 1.23-36.74) were significant. In predicting the character of pain being noted, having musculoskeletal pain (OR: .22, 95% CI: .05-.97) was significant. Being black, compared to white
(OR: 3.52, 95% CI: 1.08-11.50), was a significant variable in predicting the degree of pain control being noted. Providers agreeing that the 0-10 ratings by nurses accurately reflects patient pain
(OR: .40, 95% CI:.17-.94) was the only significant predictor of the provider initiating or intensifying treatment.
Conclusion: Patients do not receive basic pain processes of care, even when they suffer from moderate to severe pain. Our analysis of limited number of patients, and inability to assess individual effects because of power limitations, revealed that provider attitude may affect pain identification and subsequent treatment.
Implications for Policy, Delivery or Practice:
Despite the implementation of the nurse pain rating scale and the 5th Vital Sign, patients are not receiving quality care with regards to their pain documentation and treatment. Further studies are needed to guide provider-based quality improvement efforts.
Funding Source(s): VA
Theme: Military and Veterans Health Care