Call for Panels
Consumer Direction in Personal Care: Lessons from the
Cash & Counseling Demonstration
Chair: Randall Brown, Mathematica Policy Research, Inc.
Sunday, June 26 • 8:30 am- 10:00 am
● Cash and Counseling's Effects on Medicaid Costs
Stacy Dale, MS, Randall Brown, Ph.D.
Presented By: Randall Brown, Ph.D., Senior Fellow,
Mathematica Policy Research, Inc., PO box 2393, Princeton, NJ
08543-2393; Tel: (609) 275-2393; Fax: (609) 799-0005; Email: rbrown@mathematica-mpr.com
Research Objective: Determine the effect of the Cash and
Counseling model of consumer direction on costs of personal care and other Medicaid costs in 3 states.
Study Design: Medicaid beneficiaries in Arkansas, Florida, and New Jersey who were eligible for the covered services and volunteered for the study were randomly assigned either to receive a spending account they could use to hire personal care and related goods and services or to receive agencysupplied services in the usual fashion. Costs of the two groups were compared using regression models.
Population Studied: Medicaid beneficiaries eligible for personal care benefit or home and community based service waiver who voluntarily enrolled in the demonstration.
Principal Findings: Personal care costs are higher for the treatment group than the control group in each of the 3 states, but for different reasons in each state. In Arkansas, costs were higher for the treatment group than controls because many control group members received no services they qualified for and recipients received only 2/3 of the hours for which they were authorized. In Florida, the treatment group requested and received higher allowances than were initially authorized. In New Jersey, costs were similar in year 1 but higher for the treatment group in year 2 for reasons similar to those in Arkansas. Other Medicaid nursing home costs were significantly lower for the treatment than the control group in both Arkansas and New Jersey, partially offsetting their higher personal care costs.
Conclusions: The higher costs for those managing their own care under Cash and Counseling arises because of failures of the traditional program to provide authorized services to
Medicaid beneficiaries, and correctable design problems in how the spending allowance amounts were adjusted over time.
Implications for Policy, Delivery, or Practice: The huge benefits of this program for beneficiaries and their caregivers can be achieved at no greater cost per month of service than the existing system if the state draws on lessons from the demonstration states to develop a system for setting and monitoring spending allowance amounts that is independent of whether consumers self-direct their care. In some cases,
Cash and Counseling may generate savings in other Medicaid costs, including nursing home costs.
Primary Funding Source: RWJF, DHHS/ASPE
● Cash and Counseling's Effects on the Amount and
Quality of Care Received
Barbara Carlson, MS, Leslie Foster, MPP, Stacy Dale, MS,
Randall Brown, Ph.D., Barbara Phillips, Ph.D.
Presented By: Barbara Carlson, MS, Survey Researcher,
Mathematica Policy Research, Inc., PO box 2393, Princeton, NJ
08543-2393; Tel: (609) 799-3535; Fax: (609) 799-0005; Email: bcarlson@mathematica-mpr.com
Research Objective: Estimate effects of self-direction on consumers' hours of care received, satisfaction with care, unmet needs, and adverse health effects.
Study Design: Medicaid beneficiaries wishing to participate in
Cash and Counseling were randomly assigned to the program or to a control group. Treatment and control group consumers' responses to survey conducted 9 months after enrollment were compared to estimate the effects of the program.
Population Studied: Adult Medicaid beneficiaries in
Arkansas, Florida, and New Jersey who were aged or had disabilities; were eligible for personal care benefit or home and community based service waiver; and expressed interest in managing their own care by volunteering to participate in the
Cash and Counseling demonstration. Enrollment took place between December 1998 and June 2002.
Principal Findings: Treatment group members--who almost always hired family or friends to provide the care they needed-were much more likely than controls to receive paid assistance in Arkansas, but no more likely to do so in the other 2 states.
Treatment group members also received slightly fewer hours of unpaid care in 2 of the states. However, in each state they were much more satisfied with the care received, their relationship with their caregiver, their control over their care, and the quality of their lives. They were much less likely than control group members to report having remaining unmet needs. For none of the eight measures of accidents and adverse health events examined that could arise due to inadequate personal care was the treatment group more likely to report experiencing such events, and was significantly less likely to have such problems on some measures in one or more states.
Conclusions: The Cash and Counseling program improved participants' lives substantially, while increasing the hours of paid help received, but reducing slightly the amount of unpaid care received.
Implications for Policy, Delivery, or Practice: States can improve the well-being of Medicaid beneficiaries who wish to manage their own personal care by offering a Cash and
Counseling type of program. Evidence suggests there is no basis for concerns that consumers are at greater risk of falls and other adverse events if cared for by workers they hire themselves instead of by agency workers.
Primary Funding Source: RWJF, DHHS/ASPE
● The Experience of Paid and Unpaid Caregivers Under
Cash and Counseling
Leslie Foster, MPA, Stacy Dale, MS, Randall Brown, Ph.D.,
Barbara Lepidus Carlson, MS, Barbara Phillips, Ph.D.
Presented By: Leslie Foster, MPA, Researcher, Mathematica
Policy Research, Inc., PO Box 2393, Princeton, NJ 08543-2393;
Tel: (609) 936-3265; Fax: (609) 799-0005; Email: lfoster@mathematica-mpr.com
Research Objective: Assess the effects of Cash and
Counseling on paid and unpaid caregivers.
Study Design: At their baseline interview, all enrollees in the demonstration identified the person providing the most unpaid care to them at that time. These primary informal caregivers were interviewed about 10 months later about the physical, financial, and emotional strain they experienced as a result of caregiving. At 9 months after enrollment, study participants were interviewed and asked to name the person who provided the most paid care to them at that time. Those individuals were also interviewed about one month after being identified and were asked about training received, wages, satisfaction with their work, relationship with the care recipient, and physical and emotional strain experienced as a result of caregiving. For both the primary informal caregivers at baseline and the primary paid worker at 10 months, outcomes for those providing care to treatment group members were compared to outcomes for those serving control group members.
Population Studied: Two populations in Arkansas, Florida and New Jersey: First, the people providing the most unpaid care to Medicaid beneficiaries at the time the beneficiaries voluntarily enrolled in the Cash and Counseling demonstration. Second, paid workers providing home care to
Medicaid beneficiaries 10 months after the beneficiary enrolled in the demonstration.
Principal Findings: In all 3 states, primary informal caregivers for treatment group consumers reported much less physical, emotional, and financial stress than primary informal caregivers for control group members, and were much more satisfied with their lives overall. Paid workers for the treatment group, who were almost always family members or close friends, received hourly wages about 4 percent lower than agency workers, but were significantly more satisfied with their pay and working conditions than agency workers.
Workers hired by and related to treatment group members receiving care were more likely than agency workers to report high emotional stress, and more likely to say that the care recipients' family did not show them enough respect. For hired workers not related to the consumer, no such differences from agency workers were observed. The agency and directly hired workers were equally likely to suffer injuries on the job and to report high physical strain.
Conclusions: Medicaid beneficiaries' self-direction of their care through participation in Cash and Counseling consistently improves the emotional and physical well-being of their primary unpaid caregiver. Workers hired under the program received similar wages, and overall were more satisfied with their jobs than agency workers, but experienced more emotional strain as a result of their family ties to the consumer.
Implications for Policy, Delivery, or Practice: The well-being of caregivers, paid and unpaid, under consumer direction is critical to their ability to continue providing care. States concerned about caregiver "burnout" and possible attendant effects on nursing home use should consider adopting a Cash and Counseling option for Medicaid. Workers hired under the program appear to be more satisfied with their jobs than agency workers, and therefore willing to continue. However, states may want to refer consumers and their hired workers to community resources available to help them deal with their higher emotional stress and feelings of being unappreciated by other family members of the consumer.
Primary Funding Source: RWJF, DHHS/ASPE
● Lessons for Consumer Directed Programs from the Cash and Counseling Experience
Barbara Phillips, Ph.D., Kevin Mahoney, Ph.D., Lori Simon-
Rusinowitz, Ph.D., Jennifer Schore, MS, MSW, Sandra Barrett,
WIlliam Ditto, Tom Reimers, Pamela Doty,
Presented By: Kevin Mahoney, Ph.D., Associate Professor,
Graduate School of Social Work, Boston College, Room 306
McGuinn Hall, Chestnut Hill, MA 02467; Tel: (617) 552-4039;
Fax: (617) 552-1975; Email: kevin.mahoney@bc.edu
Research Objective: Identify important design and operational issues for implementing a successful, budget neutral Cash and Counseling model of consumer direction.
Study Design: Visits were made to each of the 3 Cash and
Counseling states about 18 months year after they began operations to collect information on how key program design decisions were made, including setting the allowance amount and how it can be used, conducting assessments and reassessments, monitoring consumers' spending, determining who will provide the counseling services to participants, identifying bookeeping services, training the counselors, and many others. Periodic telephone calls were also made to programs to obtain updates and information on new operational issues that arose.
Population Studied: The program operators in the 3 states in the Cash and Counseling demonstration.
Principal Findings: Combining counseling and fiscal services in one organization is an effective structure. Full-time counselors are more efficient than part-time ones. Consumers generally need counselor assistance to develop the initial spending plan but successfully manage their day to day responsibilities as employers. Nearly all consumers will rely on the fiscal agent if fiscal services are provided at little direct expense to them. To be budget neutral, programs need to determine a method for setting the spending allowance that is based on an independent assessment of consumer need and recognizes that consumers will typically receive less than their authorized care amounts.
Conclusions: The 3 cash and counseling demonstration states learned numerous valuable lessons that have enhanced their own programs and can help other states avoid some of the problems they encountered.
Implications for Policy, Delivery, or Practice: The 11 new
Cash and Counseling states should review carefully the lessons learned by the first 3 states, and incorporate these in their program design and in their operational procedures.
Primary Funding Source: RWJF, DHHS/ASPE, Univ of MD
Call for Panels
Evaluating the Impact of Parity Mental Health &
Substance Abuse Benefits in the Federal Employees’
Health Benefits Program
Chair: Howard Goldman University of Maryland, Bal imo e
Sunday, June 26 • 10:30 am - 12:00 pm
● Does Quality of Behavioral Health Care Change with
Parity Benefit Expansion?
M. Audrey Burnam, Ph.D., Alisa B. Busch, M.D., MPH, David
Dausey, Ph.D.
Presented by: M. Audrey Burnam, Ph.D., Senior Behavioral
Scientist, RAND
Research Objective: To determine whether a parity policy that mandated expansion of nominal insurance coverage for mental health and substance abuse treatment for federal employees was associated with changes in the quality of behavioral health care.
Study Design: Indicators of quality of care for major depression and substance abuse were examined using archival health plan enrollment and claims/encounter data from seven large health plans. Analyses tracked quality indicators for two years before (1998-1999) and two years after
(2000-2001) parity mandates for federal employees were implemented. Quality indicators included rates of identifying adults with major depression and substance abuse disorders, since these disorders are often undetected. Using existing guidelines and recommendations, we defined seven indicators of treatment quality among those with major depression, and six indicators of quality care among those with substance abuse disorders.
Population Studied: Adults (ages 18-64) insured as part of the Federal Employees’ Health Benefits Program (FEHBP) and enrolled in 7 large health plans serving diverse regions of the country were studied.
Principal Findings: There was no change in major depression identification rates from the pre- to post-parity time period. Four plans improved from pre- to post-parity in the likelihood that those with major depression received any antidepressants and/or therapy. However, among more nuanced measures that examined adequacy of psychotherapy and pharmacotherapy for major depression during acute phase treatment episodes, the greatest improvement postparity was seen in the likelihood of follow-up for at least four months duration, but few improvements were observed in the remaining measures. For substance abuse, increases in identification rates were found for six of the seven plans.
Among those with substance abuse disorders, there were higher rates of initiation of active treatment post-parity in one plan that had the lowest initiation rates during the pre-parity period. Across all plans, there was little evidence that intensity of substance abuse treatment improved among those who initiated treatment.
Conclusions: Implementation of parity was not associated with declines in quality of behavioral health care, even though health plans had the opportunity to respond to parity mandates by instituting tighter controls over utilization through managed care practices. Parity was associated with improvements in quality of care for some indicators and some plans. These improvements were consistent with secular trends observed in the literature and therefore may be unrelated to parity implementation.
Implications for Policy: Policies that mandate parity of insurance benefits for mental health and substance abuse conditions are not likely to adversely affect quality of care for these conditions. Some quality improvements may be seen after parity implementation, but these may be related to secular trends.
● The Impact of Parity Mental Health and Substance Abuse
Benefits on Spending and Utilization in the Federal
Employees Health Benefits Program
Richard G. Frank, Ph.D, Haiden A. Huskamp, Ph.D., Sharon-
Lise Normand, Ph.D., Vanessa Azzone, Ph.D., Howard
Goldman, M.D., Ph.D.
Presented by: Richard G. Frank, Ph.D. Professor of Health
Economics, Harvard Medical School, Department of Health
Care Policy
Research Objective: To assess the impact on mental health and substance abuse (MH/SA) utilization and spending patterns of the implementation of full MH/SA parity for enrollees of the Federal Employees’ Health Benefits Program
(FEHBP).
Study Design: We use a quasi-experimental design where the
FEHBP experience before and after the parity policy is compared to a matched set of private insurance control plans.
We use archival health plan enrollment and claims/encounter data from nine large plans from the FEHBP for the period
1999-2002 (2 years before and 2 years after parity implementation). Using a difference-in-differences approach, we estimate logistic regression models of the probability of
MH/SA service use and generalized linear models of MH/SA spending conditional on use for the nine FEHBP plans and a matched set of non-FEHBP comparison plans.
Population Studied: Non-elderly individuals (under age 65) insured as part of the FEHBP and enrolled in one of nine large plans serving diverse regions of the country.
Principal Findings: For adults, only one of the nine plans had a significant post-parity increase in the probability of
MH/SA use after taking into account secular trends. The estimated probability of use for the remaining eight plans was either not significantly different from zero or was negative and significant. The difference-in-differences analysis for total spending showed no significant increases for any of the plans.
The estimate of the effect on MH/SA out-of-pocket spending for adults showed a significant decline in out-of-pocket spending for five plans and a non-significant decline for a sixth plan. Out-of-pocket spending increased in three plans. The results for use and spending for children parallel those for adults.
Conclusions: Overall, the parity policy was implemented with little or no significant adverse impact on rate of MH/SA use or spending, while providing users of MH/SA care improved financial protection in most instances.
Implications for Policy: Mandating parity in insurance benefits for MH/SA and general medical services in the context of managed care is unlikely to result in large in increases in total
MH/SA spending and will likely reduce financial burdens borne by enrollees with MH/SA conditions.
● How did Health Plans Implement Parity for Federal
Employees?
M. Susan Ridgely, JD, M. Audrey Burnam, Ph.D., Colleen
Barry, Ph.D., Howard Goldman, M.D., Ph.D., Kevin
Hennessey, Ph.D.
Presented by: M. Susan Ridgely, JD, Senior Policy Analyst,
RAND
Research Objective: To track the implementation of mental health and substance abuse (MH/SA) parity in the Federal
Employee Health Benefits (FEHB) Program.
Study Design: In collaboration with OPM, we developed a
Parity Reporting Requirement (PRR) questionnaire to collect information about parity implementation from FEHB health plans. The PRR questionnaire was composed of closed-ended, fixed choice items on changes from pre to post-parity in: amount, scope or duration of MH/SA benefits; deductible, copay and co-insurance limits; use of MH/SA carve outs; use of risk-based contacting with vendors; composition of provider networks; financial relationships with providers; and number and mix of utilization management controls.
Population Studied: Individual FEHB health plans with more then 500 enrollees (n=156) and the government-wide Blue
Cross Blue Shield Service Benefit Plan (n=62 individual plans).
Principal Findings: Health plans expanded MH/SA benefits consistent with parity. While 7.7% of FEHB health plans
(n=12) reported carving out MH/SA benefits in response to
FEHB parity, these health plans represented about half of all
FEHB enrollment. Among fee-for-service plans in the BCBS
Service Benefit Plan, 25% of plans (n=41) reported carving out in response to parity, increasing the share of BCBS enrollees served by carve outs from 33% to 67% from pre to post-parity.
Health plans reported increasing provider networks, but did not report changes in financial relationships with providers.
There was little parity effect on the use of risk-based contracting. The effect of parity on the use of utilization management was mixed.
Conclusions: For many of the FEHB health plans we studied,
MH/SA benefits were delivered under managed care arrangements before the implementation of parity. Among those plans that did not previously manage MH/SA benefits, a number responded by carving out MH/SA benefits to specialty vendors. When viewed from the vantage point of number of
FEHB enrollees affected, rather than number of health plans, we observed a strong effect of parity on the decision to carve out MH/SA benefits. Other effects were mixed. Our data suggest that additional changes to the effective benefit design may be made by health plans over time as they gain experience with the post-parity benefit design.
Implications for Policy: Our study adds to the literature suggesting that health plans, especially fee-for-service plans, are likely to respond to the implementation of parity by carving out MH/SA benefits for management by specialty vendors.
We did not find a strong parity effect for other supply-side techniques such as putting vendors at risk, closing provider panels, or modifying provider behavior through the use of financial incentives, and we found only a mixed effect on utilization management. Some will question whether the
FEHB experience with the implementation of parity will generalize more broadly because of the strong influence of
OPM as an especially active purchaser. However, the FEHB experience may generalize to large employers willing to use their leverage to influence how health plans with whom they contract deliver MH/SA services to their employees.
Panel Discussants: Chapin White, Ph.D. and Henry Harbin,
M.D. (former Chairman, Magellan Health Systems)
Call for Panels
Patient Preferences & Participation in Health Care
Decision Making
Monday, June 27 • 11:00 am - 12:30 pm
● Patient Preferences for Shared Medical Decisions
Betty Chewning, Ph.D., Carma Bylund, Ph.D., Bupendra Shah,
RPh, MS, Sarah Karmin, MA, MPH, Neeraj Arora, Ph.D., Greg
Makoul, Ph.D.
Presented By: Betty Chewning, Ph.D., Associate Professor,
Pharmacy, University of Wisconsin, 777 Highland Avenue,
2523 Rennebohm Hall, Madison, WI 53705; Tel: (608)263-
4878; Fax: (608)262-5262; Email: bachewning@pharmacy.wisc.edu
Research Objective: This paper: 1) summarizes published studies of patient role preferences for participating in medical decisions; 2) describes factors associated with patient preferences for participation and 3) discusses limitations in the measures.
Study Design: An existing bank of articles along with searches using PsycINFO, Medline and ISI Web of Knowledge resulted in 676 studies published between 1980 and 2004 examining patient preferences for participation.
Population Studied: The body of articles was divided into three groups 1) Cancer; 2) Other Chronic Illnesses and
Invasive Procedures; 3) General Health. From the total body of studies, 70 quantitative studies had sufficient detail about the measures, analyses and findings to meet our criteria for inclusion.
Principal Findings: In more than three-forths of the Cancer
Studies (n=27), more than half of the patient sample preferred shared decision making. In slightly less than half of studies addressing Other Chronic Illnesses and Invasive Procedures
(n=25), patients preferred shared decision making. In twothirds of studies addressing the General Population (n=18) more than half the study patients preferred sharing decisions.
Across the three groups of studies, two characteristics were associated with a desire for greater shared decision making.
Younger patients and better educated patients were more likely to prefer shared decisions. In addition, although some have argued that seriously ill patients may not wish to participate in shared decisions, more Cancer studies revealed the opposite trend. Some methodological issues remain to be addressed in this research. The studies reviewed are atheoretical and cross-sectional. Further, the studies have not evaluated how role preference could vary across time and range of care decisions sufficiently. In the Cancer area, the temporal focus of the measures is on the initial treatment
decisions rather than palliative and/or follow up treatment decisions. In the other two groups of studies, similar crosssectional studies have been conducted posing largely hypothetical situations or vignettes to identify patient preferences.
Conclusions: 1) Although study patients varied, a substantial number of patients wanted to share medical decisions. 2)
Younger and better educated patients were more likely to prefer shared decision making. 3) There is a need to expand measures to include a range of decisions following the initial tests and treatment selection and to operationalize decision making as a process rather than an event. 4) Lastly, a need for standardized methods of reporting results is clearly evident.
Implications for Policy, Delivery, or Practice: We need to reframe measures away from labeling patients as “active” versus “passive”. Rather we need to assess better: 1) what aspects of medical decisions people want to share in the full trajectory of caring for a condition or issue and 2) identify a process for clinicians to involve patients effectively who prefer sharing decisions. Clinicians need sufficient skill development and time to offer a shared decision process to patients who desire participation.
Primary Funding Source: AHRQ, National Institute on Aging
● Factors Associated with Children’s Participation in
Shared Decision Making
Elizabeth Cox, M.D., MS, Maureen Smith, M.D., MPH, Ph.D.,
Roger Brown, Ph.D., Mary Anne Fitzpatrick, Ph.D.
Presented By: Elizabeth Cox, M.D., MS, Clinical Assistant
Professor, Population Health Sciences and The Center for
Women's Health & Research, University of Wisconsin-
Madison Medical School, 610 Walnut Street, 634 WARF
Building, Madison, WI 53726; Tel: (608)263-9104; Fax:
(608)263-2820; Email: ecox@wisc.edu
Research Objective: Due to its ability to produce improved health outcomes in adults, shared decision-making (SDM) has become a preferred method for reaching treatment decisions with adults. Among adults, males, minorities, and those less educated are less likely to desire SDM and less likely to participate in their healthcare. No work has examined how such characteristics impact children’s participation in key
SDM tasks. We examine how the child, parent and physician share visit talk with children of differing ages and how child and parent factors influence child participation in key SDM tasks (relationship building and information exchange).
Study Design: We analyzed videotapes of children’s medical visits as well as physician/participant surveys of demographics, practice characteristics and healthcare utilization. Using the Roter Interaction Analysis System, utterances were coded and aggregated to reflect relationship building and information exchange (information giving and information gathering). Negative binomial models were used to analyze the association between the child’s participation in key SDM tasks and parent factors (education and gender of the accompanying parent) or child factors (age, gender, and race), adjusting for visit length and physician factors, as well as clustering by physician.
Population Studied: 100 children and their parent(s) visiting
1 of 15 family physicians or pediatricians for acute concerns.
Principal Findings: Children’s mean age was 5.4 years (sd
5.0, range 0-18). Physicians spoke 63% of visit talk, parents spoke 31% of visit talk and children spoke the remaining 6% of visit talk. The amount of physician talk did not change with child age, but parent talk decreased with child age, p<0.001, while child talk tended to increase, p<0.06. For relationship building talk, children of college graduates spoke 5 times more than children of less educated parents, while girls spoke 3 times more than boys, both p<0.02. Children gave 8% more information for each year of child age, p<0.05. Children of college graduates did twice as much information giving than children of less educated parents, p<0.05. Only 15 of the children gathered any information during the visit.
Conclusions: In pediatric visits, children talked at the expense of parent talk while the amount of physician talk remained unchanged across child age. Increased child participation in key tasks of SDM is seen with higher parental education and among female children.
Implications for Policy, Delivery, or Practice: Patterns of participation in key SDM tasks among children appear similar to those of adults, suggesting that such patterns begin early in life. Future work should examine how these patterns are established as a step toward identifying mutable factors that encourage patient participation in SDM.
Primary Funding Source: AHRQ
● Older Patients' Preferences for Participation in Decision
Making
Kathryn Flynn, MS, Maureen Smith, M.D., MPH, Ph.D., David
Vanness, Ph.D.
Presented By: Kathryn Flynn, MS, Ph.D. Candidate, AHRQ
Trainee, Population Health Sciences, University of Wisconsin-
Madison Medical School, 610 Walnut Street, 632 WARF
Building, Madison, WI 53726; Tel: (608)263-4416; Fax:
(608)263-2820; Email: keflynn@wisc.edu
Research Objective: Older patients have traditionally been characterized as less likely to want to participate in health care visits, but that view has been challenged recently. In order to promote optimal patient-centered care for older patients, it is essential that we understand their preferences for participation. Our objective is to categorize older adults into types based on their preferences for information exchange, deliberation, and decisional control based the Charles model of health care decision making.
Study Design: The Wisconsin Longitudinal Study, WLS, is a random sample of men and women who graduated from
Wisconsin high schools in 1957. Respondents have completed extensive telephone and mail surveys with sample retention greater than 80%. We included four items on preferences for information exchange to and from providers, deliberation, and control over important medical decisions on the 2003-04 mail survey, coded on a five-point scale from strongly agree to strongly disagree. We characterized preference types using cluster analysis.
Population Studied: 3910 respondents aged 63-65 from the first 7 random replicates of the WLS mail survey; each replicate represents a 1/10th random sample of the original study.
Principal Findings: The four largest cluster solutions represented 96% of respondents. These cluster solutions were given names starting with the word informed because they were characterized by high preference for information to and from providers. We labeled preference for low deliberation
classical and preference for high deliberation modern. We labeled preference for provider control over important medical decisions paternalist and preference for personal control autonomist. 48% of respondents were informed modern autonomists, preferring personal control over decisions with deliberation. Informed modern autonomists tended to be female and have higher education and excellent self-rated health. 18% of respondents were informed classical autonomists, preferring personal control over decisions without deliberation. Informed classical autonomists tended to have lower education and longer median relationships with a usual place of care. 15% of respondents were informed modern paternalists, preferring provider control over important decisions with deliberation. Informed modern paternalists tended to have good self-rated health. 15% of respondents were informed classical paternalists, preferring provider control over important decisions without deliberation. Informed classical paternalists tended to be male and have lower education and fair or poor self-rated health.
The remaining 4% of respondents were represented by types that were either neutral on all items or preferred deliberation and personal control over decisions without first exchanging high amounts of information with providers.
Conclusions: Older adults prefer to exchange a high level of information with providers almost universally. Contrary to conventional wisdom, most also prefer to retain control over important medical decisions. Preferring doctor control over decisions does not mean that older patients do not want to exchange information or deliberate with providers.
Implications for Policy, Delivery, or Practice: Our health care delivery system needs to be able to accommodate individual patient preferences while encouraging participation.
Recognizing that preferences for information and deliberation are not explicitly linked to preferences for decisional control is an essential step towards promoting patient-centered care that respects individual patients’ preferences by responding to them particularly.
Primary Funding Source: AHRQ, NIA
● Continuity of Care and Trust Among Individuals with
Chronic Conditions
Maureen Smith, M.D., MPH, Ph.D., Kathryn Flynn, MS,
Jessica Bartell, M.D., MS
Presented By: Maureen Smith, M.D., MPH, Ph.D., Assistant
Professor, Population Health Sciences, University of
Wisconsin-Madison Medical School, 610 Walnut Street, 603
WARF Building, Madison, WI 53726; Tel: (608)262-4802; Fax:
(608)263-2820; Email: maureensmith@wisc.edu
Research Objective: Chronic conditions affect millions of
U.S. adults, but fewer than half of patients receive appropriate treatment for conditions that require significant selfmanagement. The objective of this analysis is to determine how patients with chronic conditions that require significant self-management differ in key elements of a sustained patientprovider partnership (i.e., continuity of care and trust in provider).
Study Design: Nationally-representative data from the 2000-
2001 Community Tracking Household Survey.
Population Studied: Adults who had visited a provider in the past 12 months and reported either no serious medical problems or chronic conditions (N=11,869), or having diabetes, asthma, chronic obstructive pulmonary disease
(COPD), or coronary heart disease, with or without a doctor visit during the last two years for their condition(s) (N=5,624 and N=1,495, respectively).
Principal Findings: Individuals with diabetes, asthma/COPD, or heart disease but no visits for those conditions during the last two years did not differ in continuity of care or trust compared to individuals without chronic conditions.
Individuals with visits for diabetes had greater odds of a change in usual provider or place (versus a continuous usual place and provider) and lower odds of having a usual place of care only or no usual place compared to individuals without chronic conditions. Individuals with visits for asthma/COPD, heart disease, or multiple conditions had lower odds of lacking a usual place. All patients with visits for chronic conditions had greater odds of highest trust in their provider compared to those without chronic conditions.
Conclusions: Key elements of a sustained patient-provider partnership differ significantly by chronic condition.
Implications for Policy, Delivery, or Practice: Future research should continue to evaluate differences in the relative importance of a sustained patient-provider partnership for different diagnoses and for patients who have multiple chronic conditions that require significant self-management.
Primary Funding Source: AHRQ
● My Own Benevolent Dictator: Merging Rank Dependent
Utility with Social Welfare Functions to Improve Health
Policy and Individual Treatment Choice
David Vanness, Ph.D., Maureen Smith, M.D., MPH, Ph.D.
Presented By: David Vanness, Ph.D., Assistant Professor,
Population Health Sciences, University of Wisconsin-Madison
Medical School, 610 Walnut Street, 785 WARF Building,
Madison, WI 53726; Tel: (608)265-8600; Fax: (608)263-2820;
Email: dvanness@wisc.edu
Research Objective: Health policy evaluations and treatment guidelines based on expected differences in mean outcomes ignore heterogeneity of treatment effects (Kravitz et al., 2004).
Policy-makers and individual decision-makers are all assumed to have the same "utilitarian" values, which ignore the variability in possible outcomes. This is essentially another skirmish in the longstanding economic contest between utilitarianism (which considers only total or average utility in assessing the benefit of a policy) and Bergson-Samuelson welfarism (which considers the utility of each individual).
Welfarists colorfully describe certain social welfare functions
(SWFs) as being like a "benevolent dictator" whose own utility is the weighted sum of the utilities of each individual affected by a policy decision. The SWF has been proposed as an alternative model for evaluating health programs from the societal perspective (Dolan, 1998; Nord et al., 1999). The expected utility model of decision-making has undergone a parallel attack as a poor descriptor of how individuals evaluate risky choices, as individuals tend to overweight the probability of some outcomes and underweight others (see Machina,
1995 for review). An alternative to expected utility, called rank dependent utility (RDU) (Quiggin, 1981; Yaari, 1987), weights probabilities of events according to the relative desirability of each outcome. RDU has been proposed as an axiomatic foundation for health care program evaluation (Bleicherodt et al., 2004). We argue that each individual facing a treatment
decision with uncertain outcomes is essentially their "own benevolent dictator" – allocating health to different possible realizations of their future self. The probability of becoming any one such future realization can be re-weighted according to its rank in the distribution of all realizations from best-off to worst-off. We seek to develop a system of decision weights that both satisfies the axioms of RDU and has the properties of SWFs.
Study Design: We use analytic methods to bridge social choice theory based on SWF with individual decisions based on RDU. Specifically, we develop a method of assigning decision weights to the probabilities of outcomes based on the position of that outcome in the rank-ordering of all possible outcomes. The decision weights are calculated using different parameterizations of the Beta distribution.
Population Studied: Not applicable.
Principal Findings: We demonstrate that a weighting function based on the Beta distribution successfully implements RDU and can flexibly represent a number of
SWFs. We explore in detail one parameterization that arguably yields minimal prior information about an individual's future status (Jeffrey, 1965), and hence is a natural implementation of Rawls' "veil of ignorance" (Rawls, 1973).
Conclusions: A simple, flexible system of decision weights consistent with RDU and reflecting various SWFs can be derived from the Beta distribution. Treatment guidelines incorporating such decision weights may potentially increase both social and individual welfare.
Implications for Policy, Delivery, or Practice: Health policy makers, clinicians and patients concerned with heterogeneity of outcomes should be cautious in relying on guidelines that depend only on expected differences in mean outcomes.
Those responsible for developing treatment guidelines and evaluating health policy might consider incorporating RDU and SWFs into their analyses.
Primary Funding Source: No Funding
Call for Panels
The Costs & Consequences of Unstable Health Insurance
Coverage
Chair: Jennifer Edwards, The Commonwealth Fund
Monday, June 27 • 11:00 am - 12:30 pm
● Extent of Churning for Children in Medicaid Managed
Care and Its Attendant Costs
Gerry Fairbrother, Ph.D.
Presented By: Gerry Fairbrother, Ph.D., Senior Scientist, New
York Academy of Medicine, 1216 Fifth Avenue, New York, NY
10029; Tel: (212)822-7398; Fax: (212)822-7369; Email: gfairbro@NYAM.ORG
Research Objective: Recent studies have documented high levels of churning in health insurance coverage, especially for low income populations. Policies for enrollment and renewal in public insurance programs contribute to churning.
Churning has implications for cost, as children are processed and re-processed into coverage, and may have implications for quality, as well. The purpose of this project is to determine both the level of churning and the cost to enroll children in
Medicaid and Medicaid managed care.
Study Design: The study was carried out in five states chosen to be geographically representative: California, Michigan,
Ohio, Oregon, and Pennsylvania, Medicaid enrollment files were analyzed to determine levels of churning over a threeyear period. Length of time in the program and numbers of breaks in coverage were calculated. Costs associated with the enrollment and renewal process were determined through examination of budgets and interviews with state officials.
Population Studied: The population studied was children enrolled in Medicaid as of December 2004 in the five states.
Principal Findings: Preliminary findings from these states show that the costs to enroll in Medicaid and Medicaid managed care is substantial and varied, ranging from $160 to almost $300 per child. Preliminary analysis of level of churning indicates that approximately half of the children had enrollments that spanned the three-year period, but some of these had breaks in enrollment. Approximately one quarter of the children had enrollments spanning a more than one year, some of whom had breaks. Analysis is underway to determine number of breaks in enrollment for these children.
Implications for Policy, Delivery, or Practice: Considerable cost appears to be associated with enrolling children in
Medicaid. Churning not only disrupts care, but also adds to cost of processing and re-processing eligible children. State policies that make it difficult for eligible children to remain enrolled contribute to churning and add to administrative costs.
Primary Funding Source: CWF, The California Endowment,
Blue Cross/Blue Shield of Michigan, and the Jewish
HealthCare Foundation
● Transitions in Health Insurance Coverage 1998-2000
Sherry Glied, Ph.D., Kathryn Klein, BA, Danielle Ferry, BA
Presented By: Sherry Glied, Ph.D., Department Chair,
Department of Health Policy and Management, Joseph L
Mailman School of Public Health, 600 West 168th Street,
Room 611, New York, NY 10032; Tel: (212)305-0295; Fax:
(212)305-3405; Email: sag1@columbia.edu
Research Objective: This project uses MEPS data to 1) describe the demographic characteristics of those who are currently at greatest risk for insurance churning; 2) understand how type of coverage affects insurance stability, and 3) understand how demographics and coverage type interact to place some individuals at higher risk for experiencing instability in insurance coverage.
Study Design: We use panels 3 and 4 of the Medical
Expenditure Panel Survey - MEPS - which covers the years
1998-1999 and 1999-2000, respectively, to examine both overall churning across different insured states -- publicly insured, private group coverage, and private non-group coverage -- and churning that includes spells of uninsurance.
Population Studied: This project studies children and adults up to age 65, focusing on young adults, racial and ethnic minorities, and less-educated and low-income populations.
Principal Findings: This analysis shows large numbers of people with unstable health insurance coverage. Young adults, Hispanics, less-educated people, those who transition into and out of poverty, and those with private non-group insurance are most likely to have unstable coverage. There are
also important interactions between demographic factors and insurance type in determining coverage stability. Young adults and Hispanics with Medicaid or private insurance were likely to lose their coverage. Less than half of those who transitioned into and out of poverty and were initially uninsured were able to obtain coverage, and almost 40% of these people who started off with Medicaid lost their coverage.
Conclusions: The problem of the uninsured is larger and more complex than is suggested by a simple cross-sectional observation. There are a number of groups that are extremely vulnerable to insurance churning who may require different types of interventions than the chronically uninsured. This more recent data provided by MEPS highlights some of the most current high-risk populations affected by this problem and suggests those in need of specially targeted policies and interventions.
Implications for Policy, Delivery, or Practice: Policies to protect these high risk groups include simplifying the procedure for renewing Medicaid coverage and raising the age for claiming dependency status to 23.
Primary Funding Source: CWF
● Eligibility Process Improvement: Change at the
Operational Level
Vicki Grant, Ph.D., MSW
Presented By: Vicki Grant, Ph.D., MSW, Vice President for
Policy and Program Strategies, Southern Institute on Children and Families, 500 Taylor Street, Suite 202, Columbia, SC
29201; Tel: (803)779-2607; Fax: (803)254-6301; Email: vgrant@thesoutherninstitute.org
Research Objective: States and counties have been working to create better application and renewal systems so that children and families eligible for Medicaid and the State
Children’s Health Insurance Program (SCHIP) are able to get coverage and keep coverage as long as they are eligible. Local offices recognize that stopping coverage of eligible children and families generates applications the next month, the following month and in future months from those who were eligible when coverage was stopped. Increasing the effectiveness of the renewal process can prevent the churning of eligible children and families on and off of public health coverage programs. The Eligibility Process Improvement
Center (EPIC) at the Southern Institute on Children and
Families focuses on improving the accuracy, efficiency and effectiveness of public programs. EPIC uses quality improvement theory and principles with Medicaid and SCHIP officials to bring about improvements at the operational level of the eligibility system, including improvements related to minimizing the effects of churning on state and local agency staff, providers and families. Over the last two years, EPIC has conducted two eligibility process improvement collaboratives with 24 teams composed of state and county eligibility officials and advocates. Teams have learned the value of becoming customer centered and have learned new skills to achieve improvement rapidly. Much of their work has focused on preventing unnecessary churning by increasing the effectiveness of the renewal process. A Process Improvement
Collaborative is a 12-month engagement between the EPIC leadership and teams of public program and related private business leaders who share a common goal of rapidly achieving Medicaid and SCHIP eligibility process improvements. A disciplined approach is utilized to impart knowledge about process improvement, to develop process improvement and results-testing skill sets and to test and implement actual process improvements. The teams work individually and collectively to achieve rapid, effective improvements in their real work situations.
Primary Funding Source: RWJF
● Instability of Public Health Insurance Coverage for
Children and their Families:
Laura Summer, MPH, Cindy Mann, JD
Presented By: Laura Summer, MPH Senior Researcher,
Health Policy Institute, Georgetown University, 2233
Wisconsin Ave NW, Suite 525, Washington, DC 20007; Tel:
(202)687-3595; Fax: (202)687-3110; Email: lls6@georgetown.edu
Research Objective: The project examines the extent to which children and families participating in the SCHIP and Medicaid programs have unstable insurance coverage, for example gaps in coverage or patterns of cycling on and off of coverage. It also examines the causes of insurance instability, particularly program policies and practices that may have an impact on the length or consistency of coverage and it seeks to show how unstable coverage in Medicaid and SCHIP affects costs incurred by states and local administering agencies, managed care plans, health centers, other providers and families.
Study Design: The project involves three activities: 1) a literature review, 2) in-depth information gathering in four states to assess the extent and impact of churning in those states, to examine recent changes in the state that may have affected insurance coverage, and to learn from their data and insight on these issues. 3) “key informant” roundtables with program administrators, health providers and health plans to identify and, when possible, document the consequences
(financial and non-financial) of instability in coverage from their different perspectives.
Population Studied: The focus of the project is on lowincome children and families.
Principal Findings: Research is ongoing. At this point we can report that there is evidence of instability in each of the states with which we are working (Louisiana, Rhode Island, Virginia,
Washington) and that changes in policies and practices in at least three of the states appear to have had an impact on insurance coverage patterns. In Rhode Island, for example it appears that decreases in renewal rates are associated with the use of program premiums and in Washington, changes in rules related to eligibility periods and verification requirements for Medicaid and SCHIP are affecting program enrollment.
Results from roundtable discussions indicate that administrative costs for programs and health plans are higher when coverage lapses and individuals re-enroll in programs later and that turnover among program or health plan enrollees presents difficulties in terms of managing care and monitoring quality of care.
Implications for Policy, Delivery, or Practice: The study draws on a number of sources of information to provide a better understanding of how Medicaid and SCHIP policies affect program participants’ ability to get and keep insurance coverage. The findings to date suggest that reductions in insurance instability could be accomplished by using better methods to coordinate coverage among public programs and
between public and private insurers, providing year-long continuous eligibility for public programs, allowing more flexibility with regard to the timing for coverage renewal, and implementing simpler procedures for renewing coverage.
Primary Funding Source: CWF
Call for Panels
Evaluating Prospects for PPOs Under Medicare
Chair: Leslie Greenwald, RTI International
Monday, June 27 • 2:30 pm – 4:00 pm
● Beneficiary Perspectives on Medicare PPOs
Shulamit Bernard, Ph.D., Gregory Pope, MA, Leslie
Greenwald, Ph.D., Wayne Anderson, Ph.D., Judy Lynch, John
Kautter, Ph.D.
Presented By: Shulamit Bernard, Ph.D., Program Director,
Division for Health Services and Social Policy Research, RTI,
International, P.O. Box 12194, Research Triangle Park, NC
27709-2194; Tel: (919)485-2790; Email: sbernard@rti.org
Research Objective: To examine beneficiary reasons for plan choice, awareness of plan options, and plan experience among enrollees in preferred provider organizations (PPOs) as compared to beneficiaries in Medicare HMOs or Original
Medicare.
Study Design: Mail survey with telephone follow-up of
Medicare beneficiaries conducted in 2004. The stratified random sample of 35 PPO demonstration service areas includes HMO and Original Medicare comparison groups.
Overall response rate was 67%, with 20,304 respondents.
Population Studied: Medicare beneficiaries
Principal Findings: The most important reasons for choosing a PPO were reducing healthcare costs, followed by greater freedom of provider choice. Half of PPO enrollees paid monthly premiums ranging from $51 to $100, double the proportion of HMO enrollees. On the other hand, approximately 45% of beneficiaries in Original Medicare reported paying monthly premiums in excess of $150 for supplemental insurance. 71% of PPO enrollees had some prescription drug costs covered as opposed to 60% of HMO enrollees and beneficiaries in Original Medicare with supplemental insurance. Almost twice the proportion of PPO enrollees (18%) as HMO enrollees went out of plan for services, and 40% of these PPO enrollees paid more for those services, double the proportion of HMO enrollees. About 15% of PPO enrollees received services from a provider that did not accept their insurance plan. PPO enrollees were more likely than HMO enrollees to have had supplemental insurance prior to enrolling in the PPO and to have had at least some prior coverage for prescription drugs. PPO enrollees have higher current out of pocket expenses. Recognition of PPO or
POS type plans by beneficiaries in Original Medicare was low
(30%), and almost half of PPO enrollees did not recognize those terms, suggesting that beneficiaries have a difficult time distinguishing abstract insurance designations. Among beneficiaries having heard of a 'Medicare PPO', slightly more than one-third of PPO enrollees knew that a PPO provides more freedom of choice of providers, but two-thirds of HMO enrollees and three-quarters of beneficiaries in Original
Medicare responded either incorrectly or that they did not know. Regarding issues of choice and satisfaction, there was little variation in responses between the three comparison groups on measures of getting a desired doctor or nurse or problems in seeing a specialist or obtaining desired care.
PPO enrollees reported lower ratings of their overall health insurance coverage as compared with HMO enrollees and beneficiaries in Original Medicare.
Implications for Policy, Delivery, or Practice: More work is needed to help Medicare beneficiaries understand differences between PPO and HMO options. Since these differences affect consumer ability to choose providers as well as out of pocket expenditures, it is important for beneficiaries to understand the implications of their choice at the time of enrollment. Understanding the experiences of these beneficiaries will inform planning for CMS’ future PPO initiative under the MMA.
Primary Funding Source: CMS
● From Demonstration to the Medicare Advantage (MA)
Program: Factors That Will Influence the Future of
Medicare PPOs
Leslie Greenwald, MPA, Ph.D., Greg Pope, M.A., Thomas
Hoerger, Ph.D., Lee Mobley, Ph.D., Elisabeth Root, M.A.,
Nathan West, M.A.
Presented By: Leslie Greenwald, MPA, Ph.D., Senior Scientist,
Division for Health Services and Social Policy Research, RTI,
International, 5104 Wetheredsville Road, Baltimore City, MD
21207; Tel: (410)448-2611; Email: lgreenwald@rti.org
Research Objective: Through a combination of findings from
PPO demonstration case studies and public comments on the
MA regions, identify factors that may influence the future of
Medicare PPOs.
Study Design: This presentation will use findings from two sources: the Medicare PPO demonstration case study analysis and the market analysis/ public comment process undertaken to support the Secretary of HHS in the development of MA
PPO regions. In the Medicare PPO case studies, RTI visited each of the 17 PPO demonstration parent companies that sponsored plans, conducting full day interviews with plan executives. The interview protocol included questions on the factors that influenced plans in offering a Medicare PPOs, and perceptions of the issues encountered by PPO plans during implementation. The market analysis conducted for CMS was used as a way to define initial options for regions under the
MA program, and used a wide range of data including commercial PPO market participation data from InterStudy, population data from the U.S. Census, and Medicare payment and beneficiary characteristics data from CMS. Potential MA regional PPO plans submitted detailed comments on this market analysis and the resulting options for MA regions.
These numerous comments identified the key issues that potential regional PPO plans will face in offering Medicare regional PPOs.
Population Studied: Potential MA PPO plans
Principal Findings: Based on these two studies, a number of common issues were raised by managed care plans with respect to the future of Medicare PPOs. Plans were consistently concerned about the difficulties they face in developing provide networks, particularly in rural areas. Some plans believed that the scarcity of providers in rural areas,
their lack of interest in participating with managed care, and the current universal standards Medicare uses for provider network adequacy will make penetration of PPOs in currently underserved rural areas difficult – even under the new MA program. Plans were also concerned about Medicare reimbursement levels and how payment rates influence the attractiveness of benefits and premiums offered by PPOs.
Plans continue to perceive that monthly premiums may be a limiting factor in attracting Medicare beneficiaries to PPOs.
We heard many plans express issues about the low level of
Medicare beneficiary knowledge of PPOs, and how much the burden and cost of educating Medicare beneficiaries about
PPOs will fall to the plans. Finally, plans raised other issues including PPO licensure and competition with Medicare Part
D.
Conclusions: Managed care plans appear guardedly optimistic about Medicare PPOs. But they also continue to raise a range of policy issues such as provider networks in rural areas, payment rates and levels of beneficiary knowledge that may limit how widely PPOs are available to beneficiaries in some areas.
Implications for Policy, Delivery, or Practice: How the major policy issues raised by potential PPO plan are addressed under the final MA regional PPO plan program will be an important factor in the success of regional PPOs – a key part of the MA program.
Primary Funding Source: CMS
● Medicare Preferred Provider Organization
Demonstration: Plan Offerings and Beneficiary
Enrollment
Gregory Pope, MA,Leslie M. Greenwald, Ph.D., John Kautter,
Ph.D., Eric Olmstead, Ph.D., Lee Mobley, Ph.D.
Presented By: Gregory Pope, MA, Program Director, Division for Health Services and Social Policy Research, RTI,
International, 411 Waverley Oaks Road, Suite 330, Waltham,
MA 02452; Tel: (781)788-8100 x134; Email: gpope@rti.org
Research Objective: To examine how PPOs have operated so far under the Medicare PPO demonstration, including PPO availability and market entry; premiums, benefits and beneficiary cost sharing; and enrollment, market share, enrollee characteristics, and disenrollment to date.
Study Design: This research analyzed descriptive data from counties in which the 35 Medicare PPO demonstration plans currently operate. The primary purpose of the analysis was to compare PPO and competing coordinated care plans (CCPs).
Findings were generated primarily from the Medicare Health
Plan Management System (HPMS) maintained by CMS.
HPMS collects service area, premium, benefit, cost sharing, out-of-pocket cost, physician network size, and other information for most Medicare private plans. In addition,
Medicare's Enrollment Database, which records monthly health plan enrollment status and various demographic characteristics for all Medicare beneficiaries, was used.
Finally, we obtained and analyzed CMS’ Hierarchical
Coexisting Conditions (CMS-HCC) risk scores measuring beneficiary health status.
Population Studied: Medicare beneficiaries
Principal Findings: Demonstration PPOs are widely, but unevenly, available. They are less likely than coordinated care plans to be available in rural areas. PPOs have located mostly where other CCPs are offered, but have increased the choice of such plans for Medicare beneficiaries. Higher demonstration county payment rates did not increase demonstration PPO availability; the rate of PPO entry in counties where demonstration payment was higher than the regular M+C amount was almost the same as in counties where it was not—about 7 percent in both types of counties. When comparing the benefit packages and out of pocket costs, we found that demonstration PPO monthly premiums are generally higher than competing CCP options, but lower than the most popular Medigap plan. On average, PPOs charge monthly premiums more than twice as much as competing
CCPs, $76 versus $29. Demonstration PPOs are more likely than competing CCPs to provide some coverage for prescription drugs; but among plans with a drug benefit, PPO coverage is less generous on average. Finally, enrollment in the demonstration PPOs have been somewhat lower than plans’ expectations, but steadily increasing. Of PPO enrollees,
42 percent were previously enrolled in FFS Medicare, 43 percent in a CCP, and 15 percent are recent enrollees in the
Medicare program. The demographic and health status characteristics of PPO enrollees are similar to those of recent enrollees in competing CCPs.
Conclusions: The Medicare PPO demonstration has succeeded in making a new managed care option available to a significant proportion of Medicare beneficiaries, though mostly in urban or near-urban areas where other managed care options already exist. The premiums of plans offered by the demonstration PPOs tend to be more costly than competing CCPs, but less costly than popular Medigap options. PPOs may represent a reasonable mid-point product for beneficiaries. But higher monthly premium cost of PPOs seems to be an important factor limiting PPO enrollment relative to HMOs. The new regional PPOs authorized by MMA beginning in 2006 may address some of the limitations of the demonstration PPOs.
Implications for Policy, Delivery, or Practice: PPOs are a key part of the MMA MA program and will therefore be important in future delivery of Medicare services.
Primary Funding Source: CMS
Call for Panels
International Comparisons of Primary Care: An
Opportunity for Learning from Patients’ Experiences &
Country Approaches to Health Care Delivery
Chair: Robin Osborn, The Commonwealth Fund
Monday, June 27 • 2:30 pm – 4:00 pm
● Cross-National Comparison of Primary Care Practice in
Australia, New Zealand, and the United States
Christopher Forrest, M.D., Ph.D., Andrew Bindman, M.D.,
Helena Britt, Ph.D., Peter Crampton, MBChB, Ph.D., Azeem
Majeed, MB BCh, M.D.
Presented By: Christopher Forrest, M.D., Ph.D., Associate
Professor, Department of Health Policy and Management,
Johns Hopkins Bloomberg School of Public Health, Hampton
House Room 689, 615 North Wolfe Street, Baltimore, MD
21205-2179; Tel: (410) 614-1932; Fax: (410) 614-7189; Email: cforrest@jhsph.edu
Research Objective: Few cross-national comparisons of healthcare systems have been done to explicitly examine primary care practice. These types of analyses may provide new insights into strategies for primary care reform. We used nationally representative surveys to contrast patient mix, scope-of-practice, visit duration, and specialty referral rates among primary care physicians in Australia (AU), New
Zealand (NZ), and the United States (US).
Study Design: Cross-sectional analysis of 3 nationally representative surveys of primary care visits: the BEACH program (AU), the NATMEDCA (NZ), and the NAMCS (US).
Each survey used a similar questionnaire that was completed by the physician after an office visit. Diagnostic codes (ICPC in AU, READ in NZ, and ICD-9-CM in US) were mapped to the
Johns Hopkins Expanded Diagnostic Clusters (EDCs) to create a uniform taxonomy for problems managed.
Population Studied: Office (“surgery”) visits made to general practitioners in AU (n=79,850), general practitioners in NZ
(10,064), and generalist physicians in the US (n=25,838).
Data were collected between 2000-2002.
Principal Findings: About 20-23% of visits were made by elders in each country, but a higher share of visits was accounted for by children in the US (35% versus 17% AU and
26% NZ). Females constituted 56-58% of visits. An average of 1.38 problems/visit was managed in AU, 1.41 in NZ, and
1.55 in the US. The average number of minutes spent per problem managed was 9.7 in NZ, 10.8 in AU, and 11.9 in US.
As a measure of scope of practice, we determined the number of problems (EDCs) accounting for 75% of all problems managed (higher numbers = greater breadth): 46 in the US,
47 in NZ, and 52 in AU. Australian GPs were more likely than
NZ and US counterparts to manage symptoms during a visit:
20.0% AU, 17.2% NZ, and 15.1% US. Correlations in the frequencies of managed health problems between countries were high (.89-.96 for pairwise comparisons). Overall, the referral rates were 10.6% in AU, 8.5% in NZ, and 7.4% in the
US; differences persisted after controlling for age. In all three countries, less frequently managed problems were referred at higher rates (correlations ranged from –0.29 to –0.33).
Conclusions: On average, US primary care physicians spend more time managing a health problem and refer less frequently than their Australian and New Zealand counterparts. Australian primary care appears to have greater breadth in terms of diagnostic scope of practice than the US or New Zealand. The influence of clinical cues, such as type of morbidity managed, on referral to specialists is similar across the countries.
Implications for Policy, Delivery, or Practice: Despite different national populations and healthcare systems, the healthcare needs that present to primary care practice in
Australia, New Zealand, and United States are similar. The salient differences in primary care practice are found in physicians’ responses to these needs. Greater direct access to specialists may explain the lower US referral rates. The high intensity of service provision in US primary care may result from consumer expectations or a more resource intensive practice style of physicians.
Primary Funding Source: CWF
● Comparison of Primary Care Delivery and Patients’
Experiences in Nine European Countries
Richard Grol, Ph.D.
Presented By: Richard Grol, Ph.D., Professor, Centre for
Quality of Care Research (WOK), Radboud University
Nijimegen Medical Centre, Code 229 KWAZO, PO Box 9101,
6500 HB, Nijmegen; Tel: 31 (0) 24 3540166; Email:
R.Grol@kwazo.umcn.nl
Research Objective: This paper will focus on primary care in
European countries. The objective was to study health care provision and patient experiences with primary health care, to determine differences between countries, differences between single handed and group practices and differences between practices in rural and urban areas.
Study Design: Practices in 9 countries were recruited
(stratified sampling). Physicians and staff completed questionnaires developed by an international panel (EPA). Per practice a minimum of 30 patients visiting the practice completed validated written questionnaires (e.g. EUROPEP).
A trained observer visited the practices to collect additional data.
Population Studied: 270 practices in nine countries (UK,
Netherlands, Germany, Belgium, France, Switzerland, Austria,
Slovenia and Israel) participated; over 8000 patients contributed.
Principal Findings: On average 87% of the patients were positive about their regular doctor (range 80-93% between countries), while 80% were positive about the organization of services (range 67-91%). There was wide variation found in most aspects of care delivery between practices in different countries. Patients in single-handed practice were more positive about both the physicians and the organization of services than patients in group practices.
Conclusions: This project was a first large scale test to see if good international, comparative data on practice performance and patient experiences can be collected in a reliable and practical way. It showed that the instruments and methodologies used were feasible and acceptable. Data relevant for European policy making were gathered. The project was the start of a European data base (TOPAS-Europe) with comparative data on different aspects of practice
performance and patient experiences in primary health care in a large number of European countries (15-20), launched in the beginning of 2005 in Berlin.
Implications for Policy, Delivery, or Practice: The data collected by the different validated European instruments are valuable for individual practices (e.g. feedback, plans for improvement), for policy making at a national level as well as for policy making and comparison at an international level.
Data will be linked to OECD data. New instruments (e.g.
Commonwealth Fund 2004 international survey on primary care) will be added in the future.
Primary Funding Source: Other Foundation
● International Comparison of Primary Care: Patients
Experiences and Health Care Delivery Models
Richard Grol, Ph.D., Robin Osborne, John Bindham
Presented By: Richard Grol, Ph.D., Professor, Centre for
Quality of Care Research (WOK), Radboud University
Nijmegen Medical Centre, P.O. Box 9101, Nijmegen, 0
6500HB; Tel: 31-(243)615-302; Email: r.grol@kwazo.umcn.nl
Research Objective: This presentation is part of a panel with
3 papers on international comparison of primary care delivery.
This paper will focus on European countries. Objective was to study health care provision and patient experiences with primary health care, to determine differences between countries, differences between single handed and group practices and differences between practices in rural and urban areas.
Study Design: Practices in 9 countries were recruited
(stratified sampling). Physicians and staff completed questionnaires developed by an international panel (EPA). Per practice a minimum of 30 patients visiting the practice completed validated written questionnaires (e.g. EUROPEP). A trained observer visited the practices to collect addtional data.
Population Studied: 270 practices in 9 countries (U.K.,
Netherlands, Germany, Belgium, France, Switzerland, Austria,
Slovenia and Israel) participated; over 8000 patients contributed.
Principal Findings: On average 87% of the patients were positive about their regular doctor (range 80-93% between countries), while 80% were positive about organisation of services (range 67-91%). There was a wide variation found in most aspects of care delivery between practices in different countries. Patients in single handed practice were more positive about both the physicians and the organisation of services than patients in group practices.
Conclusions: This project was a first large scale test to see if good international, comparitive data on practice performance and patient experiences can be collected in a reliable and practical way. It showed that the instruments and methodologies used were feasible and acceptable. Data interesting for European policy making were gathered. The project was the start of a European data base (TOPAS-
Europe)with comparitive data on different aspects of practice performance and patient experiences in primary health care in a large number of European countries (15-20), launched in the beginning of 2005 in Berlin.
Implications for Policy, Delivery, or Practice: The data collected by the different validated European instruments are valuable for individual practices (feedback, plans for improvement), for policy making at a national level as well as for policy making and comparison at an international level.
Data will be linked to OECD data. New instruments (e.g.
Commonwealth Fund study) will be added in the future.
Primary Funding Source: Other Foundation
● Primary Care and Health System Performance: Adults’
Experiences in Five Countries
Robin Osborn, MBA, Cathy Schoen (co-presenter), MS,
Phuong Trang Huynh, Ph.D., MPH, Michelle Doty, Ph.D.,
MPH, Karen Davis, Ph.D., Kinga Zapert, Ph.D., Jordan Peugh,
MA
Presented By: Robin Osborn, MBA, Vice President and
Director, International Program in Health Policy and Practice,
The Commonwealth Fund, 1 East 75th Street, New York, NY
10021; Tel: (212)606-3809; Fax: (212)606-3875; Email: ro@cmwf.org
Research Objective: The seventh in a series of annual international surveys, this study compares adults’ experiences with access to care, emergency care, safety, coordination and continuity of care, and doctor-patient interactions in five countries: Australia, Canada, New Zealand, the U.K., and the
U.S. By comparing countries from the patient’s perspective, it seeks to inform policy and use country variations to identify opportunities for cross-national learning.
Study Design: The study draws from a 2004 telephone survey with random, representative samples of people age 18 and older in each of the five countries that included over 8,600 adults. (Samples included 1,400 in Australia, Canada, New
Zealand and the U.S. and 3,000 in the U.K.)
Population Studied: Random, representative samples of people age 18 and older in each of the five countries that included over 8,600 adults. (Samples included 1,400 in
Australia, Canada, New Zealand and the U.S. and 3,000 in the
U.K.)
Principal Findings: In all five countries, majorities call for major reforms in their health systems, with fewer than onethird “very confident” they will get safe, quality medical care when needed. Access to care varied widely, with U.S. and
Canadian adults the most likely to report long waits to see the doctor and the least likely to get same day appointments when sick. Emergency room use was also much higher in the U.S. and Canada. Getting after-hours care was particularly difficult for Australia, Canada and the US. As in past surveys, US adults were most likely to forgo needed medical care or medications because of costs and reported the highest out of pocket costs. Raising safety concerns, 8-15 percent of patients said they were given incorrect laboratory or x-ray test results or had experienced delays in being notified about abnormal test results. In all five countries patients reported gaps in patient centered care, including doctors not giving patients treatment choices or asking for their opinions, and patients leaving the doctor’s office without getting important questions answered.
On measures of clinical preventive care, the U.S. tended to lead the five countries. Yet, one-half or more doctors across all countries failed to provide advice on weight, nutrition, or exercise. In several countries, but particularly the United
States, disparities by income emerged on multiple indicators of care.
Conclusions: In all five countries the study found shortfalls in delivering safe, patient-centered, timely, efficient and equitable care. Although performance varied among countries, the U.S.
often ranked low or lowest on multiple dimensions of primary care. The lack of prompt access to a regular doctor was linked to higher emergency department use in Canada and the
U.S. Cost-sharing also undermined timely care. While patient safety concerns have focused on the hospitals, risks in the ambulatory care setting were identified around prescription drugs and diagnostic tests. In all countries, doctors are missing opportunities to engage patients in their own care and to promote health and preventive care.
Implications for Policy, Delivery, or Practice: Wide variations among the countries suggest that individual countries’ policies make a difference. Promising policy initiatives underway in each country warrant further study and tracking over time.
Primary Funding Source: CWF
Call for Panels
The Non-Group Health Insurance Marketplace
Monday, June 27 • 2:30 pm – 4:00 pm
● How Much Risk Pooling Is There in the Individual
Insurance Market?
Melinda Beeukwes-Buntin, Ph.D., M. Susan Marquis
Presented By: Melinda Beeukwes-Buntin, Ph.D., Health
Economist, RAND, 1200 South Hayes Street, Arlington, VA
22202; Tel: (703)413 1100 x5581; Email: buntin@rand.org
Research Objective: To examine how much pooling of risks occurs among potential purchasers in the individual market, how much pooling occurs among those who purchase coverage, and whether there is greater pooling among longerterm enrollees.
Study Design: Logit models were fit to estimate the difference in 5 health outcome measures between the insured and uninsured after adjusting for other factors that affect insurance status and health. Multinominal models were also fit to explore the relationship between health and three measures of pooling in the market: plan type, pricing tier, and the actuarially adjusted premium paid by the enrollee.
Population Studied: Enrollees in individual insurance plans in California in 2001, and surveyed samples of individually insured and uninsured Californians.
Principal Findings: Those who purchase individual health insurance are in better health than those who remain uninsured. On the other hand, a large share of people with health problems do obtain individual insurance. The distribution of subscribers across plan type and pricing tier varies with their health status. Those in poor health are less likely to purchase low benefit plans. There is less separation of risks for those who become sick after enrollment based on the measure of pricing tier. The distribution of subscribers across plan type for those who have health problems at enrollment and those who become sick differs, but so does the distribution of those who become sick and those who remain healthy.
Conclusions: Despite small differences among the healthy and sick, our results support the conclusion that there is considerable risk pooling in the individual market. To some extent, this pooling occurs because underwriting happens at the time people enroll and there is greater pooling among those who become sick than those who enroll sick.
Implications for Policy, Delivery, or Practice: Results point to the importance of guaranteed renewal, required by the
Health Insurance Portability and Accountability Act (HIPAA), for risk pooling in the individual market. Our results also suggest, however, that Health Savings Accounts may further fragment the market.
Primary Funding Source: Other Foundation
● Health Risk and the Purchase of Private Health Insurance
M. Kate Bundorf, MBA, MPH, Ph.D., Bradley Herring, Ph.D.,
Mark Pauly, Ph.D.
Presented By: M. Kate Bundorf, MBA, MPH, Ph.D., Assistant
Professor of Health Research and Policy, Health Research and
Policy, Stanford University, Redwood Building T152, Stanford,
CA 94305-5405; Tel: (650) 725-0067; Fax: (650) 725-6951;
Email: bundorf@stanford.edu
Research Objective: One of the most persistent problems facing health policy makers in the U.S. is the existence of a large number of uninsured Americans. While economic theory suggests that health risk may be an important factor determining the purchase of health insurance in private markets, relatively little empirical evidence exists of differences in health risk between the insured and uninsured. Yet, both the implications of being uninsured and the types of policies likely to reduce the number of uninsured depend at least in part on the health status of those without coverage. The objectives of this project are to determine how health risk affects the purchase of health insurance and to identify differences in its effect in individual and group markets.
Study Design: The data source is the panel component of the
1996 through 2001 Medical Expenditure Panel Survey (MEPS).
We measure health risk based on an individual’s prior period medical expenditures, isolating expected expenditures attributable to age and gender from those attributable to the presence of chronic conditions, and we examine the relationship between health risk and the purchase of private health insurance using multivariate regression models. We also estimate models on a sub-sample of individuals more likely to purchase coverage in the individual market based on their family employment status and compare the results to models estimated on a sub-sample of individuals more likely to have access to group coverage.
Population Studied: A nationally representative sample of individuals 25-64 not covered by public health insurance.
Principal Findings: We find that high risk individuals are more likely to purchase private coverage than low risk individuals. 82% of individuals in the top decile of expected health expenditures had private health insurance coverage for the entire year compared to 60% of individuals in the lowest decile. Differences in coverage rates related to health risk are attributable to both age and gender and the presence of chronic conditions. The effects of health risk, however, vary between the group and individual markets.
Conclusions: Health risk is a determinant of the purchase of private health insurance. Those more likely to use medical services are likely to place greater value on coverage, particularly when premiums are not adjusted to reflect differences in expected utilization.
Implications for Policy, Delivery, or Practice: While policy often focuses on the difficulties high risk individuals, particularly those without access to employer-sponsored coverage, face in obtaining health insurance coverage, our findings point to issues of access among low risk, particularly low income, individuals with access to group coverage.
Primary Funding Source: RWJF,
● Risk Segmentation in the Individual Health Insurance
Market
Bradley Herring, Ph.D., Mark Pauly, Ph.D.
Presented By: Bradley Herring, Ph.D., Assistant Professor,
Department of Health Policy & Management, Emory
University’s Rollins School of Public Health, 1518 Clifton Road,
NE, Atlanta, GA 30322; Tel: (404)727-3738; Fax: (404)727-9198;
Email: bradley.herring@emory.edu
Research Objective: There is much uncertainty regarding how the individual health insurance market functions, particularly in regards to those in poor health status due to chronic health conditions. The conventional wisdom is that while those in poor health have a higher demand for coverage, medical underwriting by insurers screens them out and/or causes relatively-higher premiums for those who do obtain coverage. As a result, many states have implemented rating regulations, hoping to make coverage more affordable and obtainable for persons in poor health. Our objective is to examine the manner in which premiums in this market actually do vary with health status.
Study Design: We test the relationship between expected medical expenses, individual insurance premiums, and the likelihood of obtaining individual insurance using data from both the National Health Interview Survey (NHIS) and the
Community Tracking Study’s Household Survey (CTS-HS); we generate expected expenses for persons in these data sets using the Medical Expenditure Panel Survey (MEPS).
Specifically, we estimate selection-corrected models for individual insurance premiums as a function of expected expense, using the local-level availability of charity care as an instrument for obtaining coverage. Moreover, we distinguish between expected expenses based on age and gender versus those based on health status, and we test for differences in these relationships based on the level of rating regulations.
Population Studied: A nationally representative sample of both potential and actual purchasers of individual health insurance.
Principal Findings: We find that premiums in the individual market vary significantly with age and gender-related expenses, but in a manner that is less than proportionate. We also find that families with high levels of condition-related expenses are less likely to obtain individual insurance and pay higher premiums—particularly in less-regulated states.
However, the magnitude of these effects for health status is small.
Conclusions: Contrary to conventional wisdom, there is a high level of pooling in the individual health insurance market.
While state-level rating regulations do make those in poor health relatively more likely to obtain coverage and reduce variation between premiums and health status, the extent to which health status impacts health insurance premiums in non-regulated individual health insurance markets appears to be small—suggesting effective guaranteed renewability provisions.
Implications for Policy, Delivery, or Practice: Concerns about risk segmentation in the individual health insurance market may be somewhat overstated, as we find substantial pooling of risks, even in states with low levels of rating regulations. However, the extent to which premiums do vary in the individual market is indeed reduced by these regulations. However, states with low levels of rating regulations may benefit more from implementing high-risk pools, since the overall effect on coverage rates due to rating regulations are unclear.
Primary Funding Source: Other Government
● Community Rating and Sustainable Individual Coverage in New Jersey: Assessing Policy Options for the Nongroup Insurance Market
Alan Monheit, Ph.D., Joel C. Cantor, Sc.D., Piu Banerjee, Ph.D.
Candidate, Margaret Koller, MS
Presented By: Alan Monheit, Ph.D., Professor, Department of
Health Systems and Policy, School of Public Health, UMDNJ,
683 Hoes Lane West, Piscataway, NJ 08854; Tel: (732)235-
2865; Email: monheiac@umdnj.edu
Research Objective: The New Jersey Individual Health
Coverage Program (IHCP) was established in 1993 with the intent of providing access to a broad choice of private health insurance products to individuals regardless of health status.
The IHCP introduced a number of sweeping regulatory provisions, including pure community rating and guaranteed issue and renewal of health plans. However, after the mid-
1990s, the IHCP market place experienced a precipitous decline in enrollment, a corresponding increase in premiums, and a change in enrollment composition toward older and potentially more expensive enrollees. These trends have threatened the stability of the IHCP market, the prospects that the IHCP will achieve its stated goals, and have led policymakers and stakeholders to discuss alternative regulatory interventions. In this paper, we review the experience of the IHCP and apply a micro-simulation model to evaluate the probable impacts on enrollment, premiums, and market risk composition of shifting from pure community rating of premiums. The alternative market interventions that we assess include modified community rating and premium rating bands, merging the IHCP with New Jersey’s small employer market, and implementing reinsurance that would ease the costs associated with the most expensive market enrollees.
Study Design: Our study includes a descriptive analysis of trends in IHCP health plan enrollment, enrollee risk composition, and premiums between 1994 and 2001. These trends are discussed in the context of economic and institutional realities that confronted the IHCP over our study period. We apply a micro-simulation model to assess policy alternatives to pure community rating, including modified community rating and premium rate bands, merging the
IHCP and small employer market, and reinsuring the health care expenditures of high cost enrollees. Our model is driven by a decision rule whereby individuals compare their reservation prices for health insurance to market premiums and we assess how changes from pure community rating affect decisions by enrolled individuals to remain or withdraw
from the IHCP and for uninsured persons to obtain individual coverage. We assess the sensitivity of our findings to alternative assumptions regarding demand elasticities, rate bands, and affordability criteria.
Population Studied: Our study population consists of persons aged 21 to 64 who were enrolled in the IHCP and those who were uninsured during 2001. These data are obtained from the 2001 New Jersey Family Health Survey and from its supplemental survey of IHCP enrollees. Additional data used in our study are from the Medical Expenditure Panel
Survey and the Current Population Survey.
Principal Findings: Our study reveals that the IHCP has experienced a striking change in its risk composition and a shift in plan enrollment from fairly generous to more restrictive coverage. We find that the age composition of the
IHCP market place has shifted markedly to older and potentially more expensive enrollees, that this change is independent of national trends in non-group enrollment, and is associated with changes in economic activity and institutional changes in the IHCP market. We cannot definitively attribute causality to the regulatory provisions of the IHCP. Our micro-simulation model reveals that a change from pure community rating to provisions such as modified community rating and premium rating bands is associated with a large increase in overall market enrollment. This change is driven by a sizeable decline in premiums for young adults, large increase in their enrollment, but at the cost of a moderate increase in premiums for older enrollees with potentially greater health care needs and a corresponding moderate decline in their enrollment.
Conclusions: Alternatives to pure community rating are likely to stimulate enrollment in the New Jersey IHCP with modest harm to current enrollees. The nongroup insurance market is very fragile and its stability is linked to changes in economic activity and opportunities for alternative health insurance coverage.
Implications for Policy, Delivery, or Practice: Our analysis raises the issue of whether a private insurance market such as the IHCP can successfully serve a social insurance function.
Primary Funding Source: CWF, RWJF
Call for Panels
Strategies to Control Prescription Drug Expenditures:
Implications for Costs & Health Outcomes
Chair: Jennifer Haas, Brigham and Women’s Hospital
Monday, June 27 • 4:30 pm – 6:00 pm
● Out-Of-Pocket Drug Costs Before and After the Medicare
Modernization Act: Are There Differences in How People
Fare?
Walid Gellad, M.D., Haiden A. Huskamp, Ph.D., Kathryn A.
Phillips, Ph.D., Jennifer S. Haas, M.D., MSPH
Presented By: Walid Gellad, M.D., Medicine, Brigham and
Women's Hospital, 75 Francis Street, Boston, MA 02115;
Email: wgellad@partners.org
Research Objective: The Medicare Modernization Act of
2003 (MMA) includes a prescription drug benefit that will take effect on January 1, 2006. Because this benefit contains gaps in coverage, many beneficiaries will have significant out-ofpocket drug costs. The purpose of this study is to evaluate whether an individual’s sociodemographic characteristics and number of chronic illnesses are associated with projected drug expenditures in 2006, and to estimate how out-of-pocket drug costs may change with the MMA drug plan.
Study Design: This analysis is based on data from the 1996-
2000 Medical Expenditure Panel Survey (MEPS). The total and out-of-pocket drug expenditures for respondents were inflated by a factor to approximate their 2006 total and out-ofpocket expenditures. This inflation factor was based on data from the Centers for Medicare and Medicaid Services to account for the yearly rise in drug prices and utilization. Once the estimation of 2006 total drug expenditures was made, that value was inserted into the generally accepted formula for cost-sharing that most seniors will face once the MMA takes effect in 2006. Out-of-pocket drug costs prior to the MMA and projected out-of-pocket costs after the MMA benefit, both in 2006 dollars, are then compared using multivariate analyses.
Population Studied: Cross-sectional analysis of adults over the age of 64 with an income at least 150% of the poverty level included in the nationally representative MEPS, 1996-2000
(n=6,085).
Principal Findings: In 2006, 30.5% of Medicare beneficiaries are estimated to have total drug expenditures over $2,250, a level that will require significant cost-sharing under the MMA.
35% of individuals with three or more chronic conditions fall into the $2,251-$5,100 category, meaning if an additional medicine is prescribed for them, they will likely pay the full cost of that new medicine. Mean annual out-of-pocket drug expenditures will decline from $1,219 per person prior to the
MMA to $974 after the MMA (i.e. average savings of $245).
In multivariate linear regression models of the difference in out-of-pocket costs before and after the MMA, African
American and Hispanic seniors had smaller out-of-pocket savings on drugs under the MMA compared to whites
(average of $150 and $152 less per year, respectively), after controlling for sociodemographic characteristics.
Conclusions: Many Medicare beneficiaries will have total drug expenditures that will require substantial cost-sharing under the MMA. Overall, the MMA may be associated with modest declines in out-of-pocket drug expenditures for seniors, although the savings in out-of-pocket drug expenditures may not be shared equitably for some vulnerable populations.
Implications for Policy, Delivery, or Practice: There has been tremendous national discussion about providing prescription drug coverage to seniors through the MMA. This study adds to that debate by trying to identify how different demographic groups will be affected. Several studies have shown that as out-of-pocket drugs costs increase, adherence to medication decreases and in some cases may be associated with adverse events. The government and the research community will have to pay very close attention once the
Medicare drug bill is in place to ensure that seniors benefit equitably from this government program.
Primary Funding Source: AHRQ
● The Effect of Three-Tier Formulary Adoption on
Medication Continuation and Spending Among Elderly
Retirees
Haiden Huskamp, Ph.D., Patricia A. Deverka, M.D., MS, Mary
Beth Landrum, Ph.D., Robert S. Epstein, M.D., MS, Kimberly
A. McGuigan, Ph.D.
Presented By: Haiden Huskamp, Ph.D., Department of
Health Care Policy, Harvard Medical School, 180 Longwood
Avenue, Boston, MA 02115; Email: huskamp@hcp.med.harvard.edu
Research Objective: Many retiree health plans have adopted three-tier formularies in an attempt to control rising prescription drug costs among the elderly. Plans participating in the new Medicare Part D drug benefit are also permitted to use three-tier and other types of formularies. Our goal is to assess the impact of copayment increases associated with three-tier formulary adoption on medication continuation and spending among the elderly.
Study Design: This is an observational study of linked eligibility data and pharmacy claims for elderly enrollees of four large retiree plans that switched from a two-tier to a three-tier formulary during the period 1999-2002. We compare their experience to that of enrollees from two large retiree plans that maintained a two-tier formulary during the same period by creating matched samples of users of each of seven drug classes: angiotensin-converting-enzyme (ACE) inhibitors, angiotensin-receptor blockers (ARBs), calcium channel blockers (CCBs), proton pump inhibitors (PPIs), selective serotonin reuptake inhibitors (SSRIs), non-steroidal anti-inflammatory drugs (NSAIDs) and 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins). Our main outcome measures are: medication continuation measured using medication possession ratio (MPR); non-persistent medication use (i.e., gaps in use); medication discontinuation; medication changes; and enrollee, plan and total monthly spending.
Population Studied: 137,211 intervention group and 119,669 comparison group retired adults aged 65 and older who were enrolled continuously from July 1999-December 2002.
Principal Findings: Intervention group users were significantly more likely to change to a lower-tier drug, to have gaps in use and to discontinue use entirely relative to comparison group users for many plan/class combinations, although discontinuation rates were low overall. Three-tier adoption resulted in small or no effects on continuation as measured by MPR, depending on the plan and the drug class in this population of relatively compliant seniors. In most cases, three-tier adoption resulted in a shift in spending from plan to enrollee and a decrease in total spending, relative to the comparison group.
Conclusions: Copayment increases associated with three-tier formulary implementation by four large retiree plans resulted in some shifting of costs from the plan to the enrollee with somewhat mixed results on medication continuation.
Implications for Policy, Delivery, or Practice: Our findings have important implications for employers and retirees. The costs of retiree health coverage have increased rapidly in recent years and many employers have cut retiree health benefits, including increasing cost sharing requirements for prescription drugs. These results are also important for understanding the potential effects of three-tier formulary use in the new Medicare Part D drug benefit, which allows plans considerable flexibility in how they structure and manage incentive formularies. The results suggest that the Centers for
Medicare and Medicaid Services, which are responsible for overseeing implementation and management of the Part D, should monitor incentive formulary use closely in the new Part
D plans to ensure beneficiaries have access to needed medications.
Primary Funding Source: No Funding
● Understanding how High Cost Patients Respond to
Pharmacy Benefit Changes
Geoffrey Joyce, Ph.D., Dana P. Goldman, Ph.D.
Presented By: Geoffrey Joyce, Ph.D., Economist, RAND, 1776
Main Street, Santa Monica, CA 90407; Tel: (310)393 0411 x6779; Email: gjoyce@rand.org
Research Objective: Recent studies demonstrate that rising co-payments substantially reduce use of drugs by the chronically-ill. An outstanding issue is how the high-risk population responds to co-payment increases. This population has higher prevalence of co-morbid conditions, more likely to have multiple treatment regimens, and is more likely to be hospitalized or visit an emergency room.
Study Design: Retrospective study from 1997 to 2003 examining linked pharmacy and medical claims data with health plan benefit designs from 75 different plans across multiple employers. Exploiting both time-series and crosssectional variation in plan benefits, we simulate the effects of various benefit packages on the outcomes of interest.
Population Studied: We identify patients in the upper quantiles of the distribution of medical and pharmacy expenses using claims data and examine how they respond to changes in cost sharing for prescription drugs. All analyses are conducted by age group (ages 18-39, 40-64, and 65 and older).
Principal Findings: For the upper quantiles of spenders, we describe the diseases they have, their ages, and other characteristics. Multivariate analyses examine how high cost patients respond to changes in cost sharing by looking at service use: pharmaceutical days supplied, hospitalizations,
emergency room visits, physician office visits, and ancillary services.
Implications for Policy, Delivery, or Practice: This research is highly relevant to health plans, employers and policymakers seeking to design drug benefits that induce more rationale consumption of pharmaceuticals without restricting use of appropriate medications.
Primary Funding Source: No Funding
● Prescription Drug Dispensing Limits
Kathryn Phillips, Ph.D., Su-Ying Liang, Ph.D., Jennifer S. Haas,
M.D., MSPH, Marilyn Stebbins, PharmD, Brian K. Alldredge,
PharmD
Presented By: Kathryn Phillips, Ph.D., Professor, Pharmacy,
University of California, San Francisco, 3333 California Street
#420 Box 0613, SF, CA 94143; Tel: (415) 502-8271; Fax: (415)
502-0792; Email: kathryn@itsa.ucsf.edu
Research Objective: Among the strategies used to control costs are prescription drug dispensing limits, which limit prescriptions to 30-day supplies when filled at community pharmacies. Yet, little is known about dispensing limits. This issue is particularly timely since the new Medicare prescription drug benefit will reimburse for up to 90-day supplies at community pharmacies instead of restricting enrollees to 30day supplies.
Study Design: To examine dispensing patterns for five drug classes commonly prescribed for chronic conditions.
Population Studied: Cross-sectional data from the 1996 and
2000 Medical Expenditure Panel Surveys, which are nationally representative samples.
Principal Findings: In 2000, 64% of individuals taking chronic medications for five commonly used drugs obtained
30-day supplies. Individuals taking oral contraceptives, antiepileptic drugs, or antidepressants were more likely to obtain 30-day instead of longer supplies, while individuals taking hormone replacement therapy or beta-blockers were less likely to obtain 30-day supplies. Individuals who were younger, male, lower income, less educated, uninsured or
Medicaid insured, or who were less satisfied with care were more likely to obtain 30-day instead of longer supplies. The majority of Medicare enrollees obtained 30-day instead of more than 30-day supplies. Use of mail order services was low in both 1996 (12%) and 2000 (14%). Mean per unit expenditures for 30-day supplies were significantly higher than for longer supplies. We estimated the aggregate potential savings in prescription drug expenditures for these drug classes could be $3.4 Billion if individuals currently obtaining
30-day supplies would obtain the same supplies at the lower unit cost for longer than 30-day supplies.
Conclusions: A majority of individuals taking medications for the drug classes examined obtain 30-day instead of more than
30-day supplies. Disadvantaged individuals are more likely to obtain 30-day supplies, and per unit expenditures for 30-day supplies are higher.
Implications for Policy, Delivery, or Practice: This is the first study to our knowledge to examine prescription drug dispensing patterns using nationally representative samples.
Although we could not examine the appropriateness of dispensing limits or whether dispensing limits cause unfavorable outcomes, our findings suggest that further research on dispensing patterns is critically needed. Our results are relevant to policy debates on the Medicare prescription drug benefit, mail order services, and costcontainment for prescription drugs, and suggest that the impact of the new Medicare prescription drug benefit on dispensing patterns should be closely monitored.
Primary Funding Source: NCI
Call for Panels
Using Performance Indicators to Assess & Improve
Cancer Care in Three Nations
Chair: Eric Schneider, Harvard University
Monday, June 27 • 4:30 pm – 6:00 pm
● The National Initiative on Cancer Care Quality (NICCQ):
Assessing Cancer Care in the United States
Jennifer Malin, M.D., Eric Schneider, M.D., Katherine Kahn,
M.D., John Adams, Ph.D., Ezekial Emmanuel, M.D., Ph.D.,
Arnold Epstein, M.D.
Presented By: Jennifer Malin, M.D., Assistant Professor,
RAND Health, 1776 Main Street, Santa Monica, CA 90407;
Email: JMalin@mednet.ucla.edu
Research Objective: The U.S. National Initiative on Cancer
Care Quality (NICCQ) is a professionally led initiative involving multiple professional societies, academic experts, and patient advocates to develop and test performance indicators of care for patients with a cancer diagnosis. NICCQ evaluated the feasibility of a non-governmental national cancer care quality monitoring system.
Study Design: Performance indicators were developed with assistance of a task force of clinical specialists, academic experts, and patient representatives. This retrospective cohort study of patients with an incident diagnosis of cancer included detailed medical record review of the diagnostic and treatment phase and a survey of patient experiences after diagnosis and treatment. The data were aggregated in a variety of ways to shed light on the quality of the multiple processes of care received by many patients with cancer.
Population Studied: Patients with a recent diagnosis of breast cancer or colorectal cancer in five major metropolitan areas of the United States.
Principal Findings: A total of 108 indicators were specified including general cancer treatment indicators and diseasespecific indicators. Performance was generally high for treatment-related indicators. Other indicators demonstrated opportunities for improvement. In this voluntary project, participation rates were high, but challenges to feasibility of national monitoring included biased samples of patients with cancer, small sample sizes for some indicators, limited data availability, and the impact of protocols for protecting the confidentiality of patient data.
Conclusions: In the U.S., a collaboration between professional societies, academic experts, and consumers enabled an assessment of the quality of care for patients with cancer and identified areas of care that may benefit from improvement.
Implications for Policy, Delivery, or Practice: There are important opportunities in the U.S. to improve care for patients with cancer. The NICCQ project has highlighted
significant feasibility challenges and the changes that state and national policy makers should consider to prepare the
U.S. for national performance monitoring and improvement of cancer care.
Primary Funding Source: Other Foundation, ASCO
● Development and Reporting Of System-Level Cancer
Service Performance Indicators in Ontario (Canada)
Terrence Sullivan, Ph.D.
Presented By: Terrence Sullivan, Ph.D., Chief Executive
Officer, Cancer Care Ontario, 620 University Avenue, Toronto;
Email: Terry.Sullivan@cancercare.on.ca
Research Objective: Cancer services in Ontario have recently moved from a provincial agency delivering cancer services to a purchasing and quality authority for cancer services coordinated through 12 regional cancer programs. This presentation will report on the development, production and reporting on population level cancer system indicators, sketching the purposes and the process of indicator selection.
Study Design: The indicators have been mapped around 5 broad goals for the Ontario Cancer System. Using a modified
Delphi panel method, comprising a systematic literature review and multiple rounds of structured feedback from 34 experts, we developed a set of quality indicators spanning cancer prevention through to end-of-life care. To be useful to decision maker and providers, indicator selection criteria included a clear focus on the cancer system, relevance to a diversity of cancer providers, a strong link to the mission and strategic objectives of the cancer system, clear directionality of indicator results, presence of targets and/or benchmarks, feasibility of populating the indicator, and credibility of the measure as an indicator of quality. To ensure that the selected indicators would measure progress over time against specific and widely accepted goals, we created a strategy map based on the five strategic objectives of the Ontario cancer system:
(1) improve the measurement & reporting of cancer quality (2) increase the use of evidence & innovation in decision-making
(3) improve access to cancer services & reduce wait times (4) increase efficiency across the system (5) reduce the burden of cancer.
Population Studied: The population of Ontario, the largest provincial jurisdiction in Canada.
Principal Findings: An analysis of the mean indicator ratings by experts, and the strategy mapping exercise resulted in the identification of 36 indicators deemed suitable for routine performance measurement of the Ontario cancer system. The presentation will present highlights from the final product, which was developed with a focus group methodology to assess the preferences of the public, medical leaders, regional administrators and managers within the Ontario Cancer
System on the presentation and functionality of public reporting on cancer system performance.
Conclusions: The resulting instrument incorporates a credible evidence-basis for performance measurement aligned to the 5 strategic goals for the Ontario cancer system. It represents the integrating of a management culture, focused on the implementation of new strategic direction for the cancer system, with the underlying evidence-based culture of clinicians. Initial response by the three audiences will also be highlighted including the utility of such indicators for regional priority setting and performance improvement.
Implications for Policy, Delivery, or Practice: A rigorous process for mapping and developing evidence-based systemlevel performance indicators with active involvement from multiple stakeholder groups can produce a selection of performance indicators suitable for public disclosure and also guide quality improvement efforts among providers delivering cancer care.
Primary Funding Source: Province of Ontario
● The United Kingdom's National Health Service Cancer
Plan
Lindsay Wilkinson
Presented By: Lindsay Wilkinson, Head of Cancer Policy,
Department of Health, England, PO Box 777, London; Email:
Lindsay.Wilkinson@doh.gsi.gov.uk
Research Objective: England’s National Health Service
Cancer Plan was published in 2000, reflecting increased prioritisation of cancer and cancer services. It is a long term, national strategy to prevent, diagnose and treat cancer; to reform the way cancer services are delivered; to standardise care and improve patient experience; to co-ordinate research and to invest in equipment and the cancer workforce.
Study Design: The presentation summarizes the work under way to help the National Health Service achieve a reduction in cancer mortality and inequalities in mortality. A key issue for
England in 2005 is a reduction in waiting times from urgent referral to treatment and from diagnosis to treatment with specific targets to be achieved by the end of the year. The presentation describes progress against the four key strands of the recently established National Cancer Waits Project: 1) increased focus on waiting time reductions; 2) implementing best practice; 3) robust performance management and 4) targeted support. In addition to the overall mortality reduction and waiting time reduction targets, the Government has set a new target to reduce the inequalities gap by at least
6% between the fifth of areas with the worst health and deprivation indicators and the population as a whole. The presentation will also touch upon the future direction of the cancer programme for England, including the introduction of a new national screening programme for bowel cancer.
Population Studied: The population of England.
Principal Findings: In 2000 the UK Government undertook to reduce mortality in the under-75 age group by a minimum of 20% by 2010 compared with a 1995-7 baseline. In England, there has been a rise in cancer incidence but mortality is falling and survival is improving. Figures for 2001-3 show that the death rate has fallen by 12.2%. Cancer incidence and mortality rates are higher in areas of social deprivation mainly due to higher smoking levels and poorer diet. Later stage at diagnosis due to lower uptake of screening and later presentation of patients with symptoms also plays a part.
Conclusions: A targeted national program is demonstrating improvement on key indicators of cancer care including the outcomes of improved prevention, access to care, and treatment for patients with cancer.
Implications for Policy, Delivery, or Practice: This presentation of the National Cancer Plan highlights key steps in the change process, the components of a national cancer control programme, and the ways in which change is being delivered and monitored in the National Health Service.
Primary Funding Source: United Kingdom
Call for Panels
Local Efforts to Expand Access to Care for the Uninsured:
A Tale of Different Capacities
Chair: Beth Stevens, Mathematica Policy Research, Inc.
Monday, June 27 • 4:30 pm – 6:00 pm
● Financial Preconditions for Successful Medical
Community Initiatives for the Uninsured
Dean Smith, Ph.D., Paula Song, MHSA
Presented By: Paula Song, MHSA, Doctoral Student, Health
Management and Policy, University of Michigan, 555 South
Forest Street, 3rd floor, Ann Arbor, MI 48104; Tel: (734) 647-
9604; Fax: (734) 764-4338; Email: phsong@umich.edu
Research Objective: The purpose of this research is to identify financial preconditions or common characteristics among medical care organizations that are conducive to successful community initiatives for the uninsured.
Study Design: This study is a component of an evaluation of the Robert Wood Johnson Foundation (RWJF) sponsored
Community in Charge (CIC) program of 15 sites across the
US. We collected financial data from multiple sources including hospital cost reports and annual statements for fiscal year 2000/01. We also conducted interviews with key leaders from hospitals and other providers at four sites. The interviews added insight on the financial burdens of uncompensated care and how providers have responded to these burdens.
Population Studied: Hospitals, community health centers, insurers, providers, and community leaders in Alameda,
Austin, Birmingham, Spokane and Wichita.
Principal Findings: There are four financial preconditions associated with successful CIC program initiation.
The first precondition is the perception of substantial costs due to uncompensated care. This precondition garners the attention of community leaders and providers and motivates participation in the program. The second precondition is relative financial stability among providers, enabling them to dedicate resources to the program. We find that levels of profitability vary by site and among providers within a site.
However, if the overall provider market is financially stable, providers can afford to participate in the program. The third precondition involves the financial position of third party payers, particularly for programs that follow the health insurance expansion model. Similarly, the stability of state funding for Medicaid and other government programs for the uninsured affects program success. The final precondition is the ability to create new sources of funding -- critical for the sustainability and long-term viability of the programs. Shortrun grant support can help organize a community, but longerrun sources of funding are necessary to sustain a program.
For example, data on select hospitals in Wichita’s donated care model indicate uncompensated care approaching 9% of overall operating expenses, yet positive net incomes of approximately 4% in a stable health insurance market.
Providers were willing to donate their services, and Wichita also obtained new sources of funding for prescription drug coverage enabling them to be a successful program.
Conclusions: We find that “successful” CIC sites that satisfy these financial preconditions are able to initiate a program with promising levels of activity. Good starts are observed where providers are somewhat financially strained by uncompensated care, but are not so strained that they cannot afford to participate. Minimum levels of financial stability are necessary for organizations to actively participate in the program. Even with good starts, the long-run viability of these programs depends upon sustained sources of funding.
Implications for Policy, Delivery, or Practice: Communities interested in starting initiatives like Communities in Charge should evaluate their own financial environment to determine if it meets any of the above preconditions. Communities that share similar financial characteristics may position themselves for greater participation from providers, payers, and community leaders and increase the likelihood of a successful program.
Primary Funding Source: RWJF
● Political and Organizational Preconditions for Successful
Community-based Initiatives for the Uninsured
Beth Stevens, Ph.D., Lawrence D. Brown, Ph.D.
Presented By: Beth Stevens, Ph.D., Senior Health Researcher,
Mathematica Policy Research, 600 Alexander Park, Princeton,
NJ 08540; Tel: (609) 716-4542; Fax: (609) 799-0005; Email: bstevens@mathematica-mpr.com
Research Objective: In light of the difficulties encountered by recent national and state health care reform efforts, reformers have turned to community-based programs as the solution to the problem of the uninsured. This research sought to identify the political and organizational conditions necessary for community coalitions to mount an effective program to increase access to care for the uninsured.
Study Design: This study is one component of the evaluation of the Communities in Charge (CIC) Program of the Robert
Wood Johnson Foundation. We undertook case studies of seven of the fifteen sites, interviewing stakeholders in each site at least twice over four years, with intermittent phone interviews, as well as documentary analyses. Interviews focused on the motives for participation in the coalition, implementation of the reform model, attitudes towards the model, and its impact on their own interests.
Population Studied: CIC staff and coalition members, public officials, health care providers, business leaders, communitybased organizations, and insurers in Alameda, CA.;
Birmingham, AL.; Brooklyn, N.Y.; Jackson, MS.; Jacksonville,
FL.; Portland, ME.; and Portland, OR.
Principal Findings: Not all CIC communities had institutional arrangements that could sustain reform efforts. The types of political and organizational arrangements necessary vary with the strategy for expanding access to the uninsured. In communities where expanded access would result from new forms of insurance or expansions of insurance programs, communities needed: a continuing source of surplus funds, control of those funding sources or else ties to those that do
(be they local or state officials, providers, or foundations), and good access to regulators who could reduce regulatory requirements. In communities where expanded access would result from a re-organization of the delivery system that would free resources for use in expanding access, communities needed: cohesive organizations representing each type of
provider, a forum for negotiating the new division of responsibilities, low levels of competition among providers, strong ties between mainstream and safety net providers, and the capacity to link data systems.
Conclusions: CIC communities that had political and organizational arrangements that matched their model of change were more successful in mounting an effort to expand access to the uninsured. Those communities where the arrangements were lacking did not fare as well, despite any advantages inherent in their program model.
Implications for Policy, Delivery, or Practice: Communitybased health care reform sponsors, whether foundation or government, need to assess whether their applicant communities already have the political and organizational capacities appropriate to their reform plans. When communities do not have these resources, program sponsors could, in some cases, provide technical assistance to help would-be reformers secure them. Without such capacities, communities cannot hope to successfully implement programs that expand access to the uninsured.
Primary Funding Source: RWJF
● Health Care Market Preconditions for Successful
Enrollment in a Community-based Initiative: The Effect on
Enrollment and Utilization Change
Catherine McLaughlin, Ph.D., Erin Taylor, Ph.D., Paula Song,
MHSA, Anne Warren, MHSA
Presented By: Erin Taylor, Ph.D., Health Researcher,
Mathematica Policy Research, 600 Maryland Ave. SW Suite
550, Washington, DC 20034; Tel: (202) 484-9220; Fax: (202)
863-1763; Email: etaylor@mathematica-mpr.com
Research Objective: Various conditions must exist for community coalitions to mount a program that enrolls significant numbers of uninsured adults. In addition, these conditions influence the type of program developed and which uninsured adults enroll. The purpose of this research is to describe 3 different community-based initiatives, who enrolled, and how the characteristics of the community, the program, and the enrollees affect changes in utilization.
Study Design: The study, part of the evaluation of the Robert
Wood Johnson Foundation's Communities in Charge
(CIC)Program, is based on a multi-wave telephone survey of participants in three different CIC programs. We present descriptive statistics on enrollees and their health care utilization for the year before enrollment and for the first year of enrollment.
Population Studied: Over 1,100 enrollees in three CIC programs: the Family Care coverage product in Alameda
County,California; the CarePartners program in southern
Maine; and the Seton Care Plus program in Austin, Texas.
Survey resonse rates were over 80% in each wave in each site.
Principal Findings: Each program exploited an existing infrastructure already treating large numbers of uninsured adults. The Family Care program offered subsidized coverage through a local MediCal HMP to the parents of MediCal enrollees. Enrollees typically learned about the product through one of the two clinics devoted to providing care to low-income Asian and Latino immigrant populations. As a result, the majority of program participants are Asian and
Latino immigrants who are in their mid-30s, in relatively good health, with low levels of health care utilization prior to enrollment. CarePartners enrollees are in their mid-40s, with the majority reporting having a limiting or chronic, nonlimiting condition and above average levels of previous health care utilization. Seton Care Plus targeted low-income workers already using Seton Health System's clinics, providing enrollees with additional services like pharmaceutical benefits.
The system's clinics are used predominantly by Latinos, and over 70 percent of program enrollees are Hispanic, with 22 percent reported a limiting or chronic, non-limiting condition at enrollment. Hospital utilization did not change significantly in any of the three programs. Access to preventive care increased significantly for all three programs, though Family
Care (which had the lowest baseline utilization levels) exhibited the largest increase. CarePartners, which had the highest baseline utilization levels, was the only group that exhibited a (marginally significant) decline in emergency department use.
Conclusions: These case studies illustrate the variety in community-based programs, highlighting differences in target populations and suggesting how program structure can influence enrollment, programmatic effects, and retention.
Each of these programs had a distinct set of objectives and faced a unique set of challenges and constraints.
Implications for Policy, Delivery, or Practice: Policymakers considering local programs should carefully consider how local conditions will determine what would be a viable program structure, which in turn will shape both the population of enrollees and the program's effect on their health care utilization patterns.
Primary Funding Source: RWJF
Call for Panels
Application of Predictive Modeling for Care Management
Chair: Soyal Momin, Blue Cross Blue Shield of Tennessee
Tuesday, June 28 • 8:30 am – 10:00 am
● Predictive modeling of diabetic inpatient hospitalizations for the Tennessee diabetic Medicaid population
Avery Ashby, MS, Soyal Momin, MS, MBA, Raymond Phillippi,
Ph.D., Allen Naidoo, Ph.D., Judy Slagle, RN, MPA, Steven
Coulter, M.D.
Presented By: Avery Ashby, MS, Research Analyst, Research
Development & Consulting, Health Services Research,
BlueCross BlueShield of Tennessee, 801 Pine Street 3E,
Chattanooga, TN 37402; Tel: (423) 763-7482; Fax: (423) 755-
5100; Email: Avery_Ashby@bcbst.com
Research Objective: The primary objective of this retrospective cross-sectional study was to develop a predictive model to project future diabetic hospitalizations for the diabetic Medicaid population in Tennessee.
Study Design: Information on medical and pharmacy claims was extracted from the clinical data mart called MCSource
(Version 5.2) for the two-year time period of July 1, 2001 through June 30, 2003. SAS (Version 9.1) was used for primary data analysis. Data in the first year of the time period were used to model whether a diabetic hospitalization occurred during the second year. Logistic regression and
rational artificial intelligence (artificial neural networks) was employed for model building.
Population Studied: The study group consisted of 11,002
(mean age 47 years, 65% female) diabetic Medicaid members who were continuously enrolled for the study period, and administered by a MCO in Tennessee. Dual-eligible members were excluded from the study population.
Principal Findings: Stepwise logistic regression showed that the number of first year diabetic hospitalizations (OR = 2.836), insulin not prescribed (OR = 0.258), age (OR = 0.962),
Diagnostic Cost Grouper (DCG) risk score (OR = 1.073), and miscellaneous anti-diabetics not prescribed (OR = 1.510) were all related to the risk of a diabetic hospitalization in the ensuing year. Chi-square statistics for these variables had pvalues of less than 0.01. The receiver operating characteristic
(ROC) curve showed a good classification rate (area = 0.824).
Artificial neural networks produced similar results to logistic regression.
Conclusions: Information about specific co-morbidities and demographic variables such as location of member residence and gender were not useful in prediction. Predictors, such as number of diabetic hospitalizations in the previous year, a member’s overall risk, and non-drug usage were significant indicators for prospective diabetic hospital admissions.
Implications for Policy, Delivery, or Practice: Traditional statistical methods and emerging technologies like rational artificial intelligence (or machine learning) provide some tools for predictive modeling of healthcare data. The results in this research allow for improvement, however. This improvement may be found in the use of other predictors not currently available or perhaps, the combination of modeling methods.
Primary Funding Source: BlueCross BlueShield of Tennessee
● Application of Predictive Modeling to Identify, Stratify, and Triage Members in Care Management Programs: A
Health Plan Case Study
Soyal Momin, MS, MBA, Sylvia Sherrill, RN, MS, Lelis Welch,
RNC, CCM, Judy Slagle, RN, MPA, Terence Shea, PharmD,
Steven Coulter, M.D.
Presented By: Soyal Momin, MS, MBA, Manager, Research
Development & Consulting, Health Services Research,
BlueCross BlueShield of Tennessee, 801 Pine Street - 3E,
Chattanooga, TN 37402; Tel: (423) 763-7352; Fax: (423) 755-
5100; Email: Soyal_Momin@bcbst.com
Research Objective: The primary objective of this study was to develop, implement, and evaluate a data driven care management model. The secondary objectives were to identify/stratify potential high-risk members through predictive modeling, provide members with evidence-based care, empower members to make informed healthcare decisions, increase member compliance with therapy, reduce catastrophic case managers’ (CM) caseloads and staff turnover along with improving member/provider satisfaction rates.
Study Design: A more collaborative and integrated model was developed using Diagnosis Cost Groups (DCGs) and Episode
Treatment Groups (ETGs). Using Year-2 predicted amount, members were triaged into one of the three levels
(Lifestyle/Health Counseling, Care Coordination (CC), and
Catastrophic Case Management (CCM)) of the care management model. CC included Pharmacy Care
Management programs, ER Management program, Center of
Excellence Program, Transition of Care Program, and Disease
Management Programs. As per the care need on the followup, members were re-triaged in the appropriate level of the model. The model was implemented in April 2003 and has been evaluated quantitatively using pre and post and longitudinal study designs. Preliminary evaluation has also been carried out for qualitative aspects such as member and provider satisfaction, CM staff turnover, and triaging efficiencies.
Population Studied: Commercial population (nearly 1.7 million PPO members) administered by a large, single-state health insurance carrier in the Southeastern US.
Principal Findings: The number of referrals for CCM has reduced from 17,000 in 2002 to 10,000 in 2003 with almost the same number of total referrals for case management. The case load/CM/Month has reduced from 103 to 76 cases/CM/Month. CC touched 7,233 lives in the first year, which would have otherwise not received this service. These efficiencies coupled with clinically involved predictive modeling have improved case managers’ job satisfaction levels and resulted into a 10% drop in staff turnover from
2002 to 2003. The average member and provider satisfaction score on a scale of 0-5 has improved from 4.1 to 4.8 and 4.4 to
4.8 respectively. The Hepatitis-C program approved therapy extension for 73% of members resulting in medication cost avoidance of $1.5 million. AMI-Beta Blocker longitudinal evaluation showed 94.8% Beta Blocker and 73.9% Aspirin utilization rate post AMI with an estimated cost avoidance of
$1.3 million per year. Physician compliance with the treatment guidelines, as well as member compliance with the medication regimen, has increased.
Conclusions: By appropriately identifying members who would benefit from CCM has resulted in more efficient care delivery. The model has allowed a comprehensive approach for care management by incorporating various interventions under one umbrella and managing the member health across the care continuum. Use of the model has enabled touching more lives proactively.
Implications for Policy, Delivery, or Practice: Predictive modeling is rapidly gaining acceptance as a tool to control costs and disease progression. Considering the model scalability/replicability/adaptability, it can be applied to any line of business including government-sponsored programs.
The model also addresses shortfalls of current health care delivery system such as fragmented health care and more emphasis on acute conditions.
Primary Funding Source: BlueCross BlueShield of Tennessee
● Using Predictive Modeling to Evaluate Disease
Management Financial Impact
Terry Whitlock, MBA, Kenton Johnston, MPH, MSIR, Joyce
Stickles, RN, MBA, Daniel Barnett, M.D.
Presented By: Terry Whitlock, MBA, Bio-Statistical Research
Scientist, Health Services Research, BlueCross BlueShield of
Tennessee, 801 Pine Street - 3E, Chattanooga, TN 37402; Tel:
(423) 763-3690; Fax: (423) 755-5100; Email:
Terry_Whitlock@bcbst.com
Research Objective: The objective is to use predictive modeling (PM) to evaluate the impact of a disease management (DM) program on a chronically ill population.
Specifically, Diagnostic Cost Grouping (DCG) predictive modeling is utilized to measure the financial impact of DM in populations of individuals with chronic heart failure (CHF) and coronary artery disease (CAD).
Study Design: The literature of current practices regarding
DM financial impact measurement is reviewed and critiqued—especially with reference to the DM industry favorite population based pre-post methodology. The time period covered for the present study is three years and the variables of interest are financial metrics. Claims data and DM program specific data covering the 24-month period of 2001-
2002 and the 24-month period of 2002-2003 are analyzed.
DCG predictive modeling is used to generate predicted claims costs for the year 2002 based on actual data for the year 2001 and likewise for 2003 based on actual data for the year 2002.
The mean differences between DCG predicted and actual total claims costs in 2002 and in 2003 are computed in order to determine DM program financial impact. Finally, inflation factors based on actual health plan population experience for the populations in question are developed and applied in order to evaluate financial impact.
Population Studied: The diseased populations considered are those with CHF and CAD from two large self-insured employers who purchased DM. In order to be selected into the study group an individual must be identified as having
CHF or CAD (using an algorithm based on ICD-9 and NDC codes), be enrolled in the DM program during the 2001-2002 and/or the 2002-2003 24-month study periods, and be continuously enrolled in the insurance claims database for one or both of the 24-month study periods.
Principal Findings: Our preliminary findings suggest that a study design utilizing DCG predictive modeling in conducting
DM program financial impact evaluations shows more accurate results than the population based pre-post methodology currently favored by DM companies does.
Conclusions: A study design using predictive modeling is an effective tool to assist investigators in computing more accurate DM impact and return on investment (ROI) than the pre-post methodology. Such a design also controls for regression to the mean (RTM), which is one of the principal threats to the internal validity of findings from studies using the pre-post design. The use of the population based pre-post methodology can lead to overestimations of DM program savings.
Implications for Policy, Delivery, or Practice: Health services researchers as well as potential DM purchasers should be skeptical of studies based on the pre-post methodology conducted by DM companies that show a large program ROI. In the future, study designs that use predictive modeling can be a viable option to make valid determinations of true DM impact.
Primary Funding Source: BlueCross BlueShield of Tennessee
Call for Panels
Medicare as a Leader in Reducing Racial & Ethnic Health
Disparities
Chair: June Eichner, National Academy of Social Insurance
Tuesday, June 28 • 8:30 am – 10:00 am
● An Organized Process Approach to Reduce Clinical
Disparities in Medicare
Lawrence Casalino, M.D., Ph.D.
Presented By: Lawrence Casalino, M.D., Ph.D., Assistant
Professor, Health Studies, University of Chicago, 5841 South
Maryland Avenue, MC 2007, Chicago, IL 60637; Tel: (773)834-
4865; Fax: (773)702-1979; Email: lcasalin@uchicago.edu
Research Objective: To develop a clinical care framework that will support a reduction in racial and ethnic health disparities though improvements in quality of care.
Study Design: The framework is developed by combining insights from literature reviews from 1993 to 2004, using
Medline searches as the primary identifier. Our search focused on studies that evaluated health disparities and quality improvement.
Population Studied: The general U.S. population, with an emphasis on Medicare beneficiaries. From the provider perspective, we focused on private practice physicians and group practices in outpatient settings.
Principal Findings: The individual physician view of quality, where quality is what a physician does for a patient while that patient happens to be in front of him or her, is necessary but not sufficient to reduce racial and ethnic health disparities. An organized process model, where quality is also what a group of physicians does for the population of patients for whom they provide care, is also necessary. Such a model should also include cultural competence, as it is as important for health care organizations as it is for physicians. Organizations also have more capacity than individual providers to develop processes necessary to reduce disparities, with examples being the Chronic Care Model and disease management.
Conclusions: Contrary to the IOM Unequal Treatment recommendations, focusing change on individual physicians will not effectively reduce racial and ethnic disparities in clinical care. Combining outreach to individuals with use of organized processes by medical groups is necessary to make a significant effect towards reducing disparities in clinical care.
We propose a simple organized process model, the 4P model for reducing disparities, for patients, physicians, physician organizations, and processes, with processes to include the use of disease registries, guideline prompts at the point of care, feedback to physicians on their performance, care management of the sickest patients, and ongoing, systematic efforts to improve patients’ ability to self-manage their chronic diseases. Also, incentives to encourage such processes will be necessary but must be carefully devised to avoid increasing disparities.
Implications for Policy, Delivery, or Practice: Medicare has the ability to enact great change within the healthcare system.
Through its administrative structure, Medicare can pursue the kinds of changes necessary to reduce racial and ethnic disparities in clinical care, and eventually to the general health system. This paper examines the role that changes to clinical care, through large purchasers such as Medicare, could have on reducing racial and ethnic disparities. It proposes and evaluates an organized process 4P model to improve care provided by patients, physicians, physician organizations and processes.
Primary Funding Source: RWJF, National Academy of Social
Insurance
● Racial and Ethnic Disparities in Medicare: What HHS and CMS Can and Should Do
Timothy Jost, JD
Presented By: Timothy Jost, JD, Professor, School of Law,
Washington and Lee University, 1370 Lincolnshire Drive,
Harrisonburg, VA 22802; Tel: (540)458-8510; Fax: (540)458-
8488; Email: jostt@wlu.edu
Research Objective: To identify what the Department of
Health and Human Services HHS and more specifically, the
Centers for Medicare and Medicaid Services CMS, could do without further statutory authority to address racial and ethnic disparities in Medicare.
Study Design: The study involved a literature search on
Medicare and health disparities; review of relevant statutes and regulations; and interviews with current and former HHS and CMS staff.
Population Studied: CMS and its present and former contractor staff who are responsible for the Medicare program, with particular emphasis on those working on programs and projects targeting racial and ethnic health disparities.
Principal Findings: CMS has not made the reduction of racial and ethnic health disparities a priority. It does not, for example, list the elimination of racial and ethnic disparities as a priority in its performance plan, have an office of minority health like other HHS agencies, nor a dedicated budget for addressing disparities issues. It has also not successfully addressed four major problems that result in racial and ethnic disparities in Medicare: financial barriers; logistical, organizational, or systemic barriers; provider attitudes and behavior; and patient attitudes and behavior.
Conclusions: HHS and CMS could do much more within the scope of their current statutory authority to reduce or eliminate racial and ethnic disparities in the Medicare program. It has within its authority the ability to tackle minority health care as both a quality and compliance problem.
Implications for Policy, Delivery, or Practice: CMS should take a number of steps to remove financial barriers that minorities face in gaining access to Medicare services. This should include ensuring that minority Medicare beneficiaries receive all Medicaid benefits to which they are entitled, requiring that state Medicaid programs have adequate providers available to meet the needs of Medicaid-dependent minority beneficiaries, and use conditions of participation for
Medicare Advantage plans to safeguard racial and ethnic minorities’ access to Medicare Advantage plans. In addition,
CMS should make sure that racial and ethnic minority beneficiaries, and in particular poor minority beneficiaries, have access to Medicare Part D prescription drug plans.
Medicare should also decrease or eliminate institutional barriers to care by helping to increase the availability of minority health professionals; enforcing the civil rights laws against professionals and providers who discriminate against minorities; and requiring and assisting providers to improve their cultural and linguistic competency. CMS should, in particular, ensure that Medicare Advantage plans serve minorities and their communities. CMS should continue its efforts to change professional attitudes and behavior that result in disparate treatment of minorities. The QIO program, and possibly financial incentives, have a role to play here.
CMS currently has a number of initiatives underway to educate beneficiaries as to Medicare benefits and how to use them, and these should be encouraged and expanded.
Primary Funding Source: RWJF, National Academy of Social
Insurance
● Medicare Race and Ethnicity Data
Marshall McBean, M.D., MSc
Presented By: Marshall McBean, M.D., MSc, Professor,
Health Services Research and Policy, University of Minnesota,
3407 West 28th Street, Minneapolis, MN 55416; Tel: (612)625-
6175; Fax: (612)378-4866; Email: mcbea002@umn.edu
Research Objective: Eliminating racial disparities in health and health care is one of the two overarching goals of Healthy
People 2010. In order to achieve that goal, accurate and complete race/ethnicity information is essential. The Medicare program has played a major leadership role in documenting existing disparities and implementing activities to eliminate them. A major reason for this has been the quality of the
Medicare data including race/ethnicity information. The specific focus of the study was to review the origins and current quality of the Medicare administrative data information on race/ethnicity.
Study Design: Review of literature published regarding the history and accuracy of race/ethnicity information; update of validation of the race/ethnicity information using recent, 2001 and 2002, Medicare Denominator file and Medicare Current
Beneficiary Survey data.
Population Studied: U.S. Medicare population
Principal Findings: Several improvements in the collection of
Medicare race/ethnicity information have occurred during the past decade. However, deficiencies remain, particularly with respect to the smaller minority groups, specifically Asians,
Hispanics and American Indians/Alaskan Natives. The primary issues affecting the Medicare data are 1. the use of race/ethnicity information from the Social Security
Administration (SSA) as the primary source for Medicare data, and 2. the lack of a two-item designation method for race/ethnicity at SSA and Centers for Medicare and Medicaid
Services. In addition, the current quality of the race/ethnicity information available to the agency and researchers has plateaued. For example, for 2002, the most recent year that can be analyzed, only 52% of Asian, and 33% of Hispanic, and
33% of American Indian/Alaskan Native Medicare beneficiaries could be correctly identified, i.e., sensitivity, in the Medicare data. While this was approximately a 3-fold increase since 1997 for Native Americans/Alaskan Natives, it
was a slight decrease for Asians and Hispanics. However, those persons identified as belonging to a particular race/ethnicity are highly likely to be of that race/ethnicity.
Thus, in 2002, the positive predictive values for Whites, Blacks and Hispanics were > 94%; 76% for Asians; and 88% for
American Indians/Alaskan Natives. There are certain longstanding impediments to correcting these deficiencies in the near-term: the use of only three, specifically White, Black,
Other race designations on the SSA’s SS-5 form prior to 1980; the failure of SSA to capture ethnicity as a separate item on the SS-5 form when the race code options were expanded in
1980; the difficulty of systematically correcting the existing information by Medicare or SSA by contacting beneficiaries for that purpose; and the voluntary nature of the reporting of race.
Conclusions: Some improvement in the quantity and quality of race/ethnicity information could be made through the addition to the Medicare databases of race/ethnicity information collected by surveys, disease registries, managed care plans, administrative databases of other organizations, or the imputation of race/ethnicity using surname imputation.
Implications for Policy, Delivery, or Practice: While improving the Medicare race/ethnicity information is very important, the absence of perfect race/ethnicity data must not prevent the use of Medicare data in studies and interventions designed to understand and alleviate existing disparities.
Primary Funding Source: RWJF
● CMS’ Programs and Initiatives to Reduce Racial and
Ethnic Disparities
Ellen O'Brien, Ph.D.
Presented By: Ellen O'Brien, Ph.D., Assistant Research
Professor, Health Policy Institute, Georgetown University, 2233
Wisconsin Avenue, NW, Suite 525, Washington, DC, DC
20007; Tel: (202)687-3867; Fax: (202)687-3110; Email: obriene@georgetown.edu
Research Objective: Medicare has substantially narrowed racial and ethnic disparities in health care, most dramatically in its first year when Medicare effectively desegregated the nation’s hospitals. Over nearly four decades, racial and ethnic gaps in health care use and quality have been reduced, but not eliminated. The objective of this paper is to describe and assess current efforts within the Centers for Medicare and
Medicaid Services or CMS to narrow peristent racial and ethnic barriers in access to and the delivery of clinically appropriate care. The paper provides an inventory of programs and initiatives implemented by CMS, including efforts to: 1. develop culturally appropriate education and outreach; 2. measure and reduce clinical health care disparities through beneficiary and provider-focused interventions; and 3. enhance providers’ cultural competency.
The paper describes the initiatives, the disparities they seek to reduce, efforts to evaluate them, and their actual and potential impact to reduce disparities both locally and nationally.
Study Design: The study relies on a literature review and interviews with key informants. Through a review of the policy research and health services literature, several strategies for narrowing disparities in Medicare and the health care system generally were identified. With this framework as a guide, more than 20 expert interviews were conducted over a threemonth period in 2004 to identify and qualitatively assess the agency’s activities and to highlight options for enhancing or redirecting these activities.
Population Studied: The paper includes an examination of policies that affect Medicare beneficiaries of all racial and ethnic minority groups, including African Americans, Asian and Pacific Islanders, Native Americans and Alaska Natives, and Hispanics.
Principal Findings: The central finding is that, despite successes in some areas, ongoing efforts within CMS have had limited reach and limited impact on racial disparities.
Efforts to improve communications with minority populations, through the HORIZONS effort and the State Health Insurance
Counseling and Assistance Program have failed to achieve intended goals. Efforts to identify and reduce disparities in clinical practice have been undertaken by Medicare’s Quality
Improvement Organizations with some successes in certain communities, but little measurable impact across the nation.
Data on the efforts of managed care plans to improve their organizational cultural competency are not yet available, but early assessments suggest that plans’ efforts have been relatively small in scale and impact. The paper identifies several ways to enhance these ongoing efforts to make more significant progress toward eliminating racial and ethnic disparities in Medicare, and identifies several key considerations for the development of future initiatives.
Conclusions: Current efforts within CMS to reduce racial and ethnic disparities in Medicare fall short of stated goals for the interventions. More concerted efforts are needed to reduce racial and ethnic disparities in Medicare. CMS has considerable opportunities within its current regulatory structure redirect and strengthen these activities to make more substantial progress toward narrowing racial and ethnic disparities in Medicare and the health system generally.
Implications for Policy, Delivery, or Practice: As a health care payer for 40 million individuals and a key regulator of the health care marketplace, Medicare has opportunities within its current administrative structure to narrow disparities in access to care and the delivery of clinically appropriate care. This paper lays a foundation for a discussion of what Medicare’s role could and should be by identifying and assessing the current policies and practices.
Primary Funding Source: Joint Center for Political and
Economic Studies
Call for Panels
Informatics Systems Supporting Collaborative Care of
Chronic Illness
Chair: Alexander Young, UCLA Neuropsychiartric Institute
Tuesday, June 28 • 8:30 am – 10:00 am
● Optimizing Informatics Support for Collaborative Care:
Examples from Veterans Administration Smoking and
Depression Treatment Programs
Edmund Chaney, Ph.D., Laura Bonner, Ph.D., Susan Vivell,
Ph.D., Colletta Austin, MA, Scott Sherman, M.D.
Presented By: Edmund Chaney, Ph.D., Psychologist, Health
Services Research & Development 151, Veterans
Administration & University of Washington, 1100 Olive Way,
Suite 1400, Seattle, WA 98101; Tel: (206) 764-2815; Fax: (206)
764-2572; Email: Edmund.Chaney@med.va.gov
Research Objective: To study the optimization of informatics support for collaborative care for chronic illnesses in the context of planned overall redesign of a health care system’s electronic health record.
Study Design: Collaborative care functions not well supported by the current VHA VistA electronic health record were identified. Local incremental modifications were made to
VistA to improve support for these functions. Modifications were tested for usability in a Test-Operate-Test-Execute model and results were used to guide additional modifications as necessary.
Population Studied: Greater Los Angeles VA started a pilot
Smoking Cessation Telephone Counseling Program in April
2003. The program took consults from primary care providers, and case managers were to complete four telephone interviews per patient over an 8-week period. A file was created in Vista to capture panel data, augmented by a graphical user interface (written in Delphi) to allow the counselors to enter data as they conduct their telephone interviews. Patient data objects were created for selected fields in the file and a Smoking Cessation progress note template was created using the patient data objects. A summary of the interview automatically uploads into the VistA progress note template. Having the data in VistA enabled us to run reports needed to track clinical outcomes and administrative and cost data. Entering scheduling data allowed us to produce tracking reports showing when all patients need to be called again, thus supporting crosscoverage among the case managers. VA Mental Health
QUERI funded a project to disseminate collaborative care for depression in seven community based outpatient clinics.
Remotely located care managers accepted electronic consults from primary care physicians, assessed patients by telephone, made treatment plan suggestions and completed followup calls to support adherence and necessary treatment plan modifications. With the support of Information Resource
Management personnel at each site, user-modifiable consult and progress note templates were created to assist care manager/primary care provider communication and to allow tracking of care manager activities to support depression performance measure auditing and a cost analysis. Templates were optimized through feedback from primary care providers to display necessary information for treatment planning and rudimentary decision support rules.
Principal Findings: Usability assessment of the modifications by care managers, primary care providers and program managers were positive and were useful in optimization.
Utilization tracking made possible by the modifications pointed to additional care management activity issues that might be made more efficient by additional optimization.
Conclusions: Two collaborative care projects were successfully supported by local informatics optimization while the health care record system as a whole was undergoing planned revision.
Implications for Policy, Delivery, or Practice: It is unnecessary (and often clinically infeasible) for informatics developments to support improved collaborative care to wait on global system redesign. Studying incremental optimization of current system capabilities can help inform global redesign.
Primary Funding Source: VA
● Evaluating an Intervention to Improve the Quality of Care for Schizophrenia
Amy Cohen, Ph.D., Alexander Young, M.D., MSHS, Jim Mintz,
Ph.D., David Dorr, M.D., MS
Presented By: Amy Cohen, Ph.D., Psychologist, Mental Illness
Research, Education, and Clinical Center (MIRECC), Greater
Los Angeles VA Healthcare Center, 11301 Wilshire Boulevard
(210A), Los Angeles, CA 90073; Tel: (310) 478-3711 x40770;
Fax: (310) 268-4056; Email: ancohen@ucla.edu
Research Objective: Schizophrenia occurs in about 1% of the population and accounts for 2.5% of all healthcare expenditures. Practice guidelines have been promulgated, yet patients often do not receive effective treatments. Principals of collaborative care and chronic disease management have been used to improve care for chronic disorders. The VA
EQUIP project (“Enhancing Quality Utilization in Psychosis”) applied these principals to schizophrenia.
Study Design: EQUIP was a randomized, controlled trial of a collaborative care model at two medical centers. Psychiatrists were randomized to the intervention or usual care. Interviews were conducted with patients and clinicians before and after the intervention, and data obtained from medical records.
The intervention focused on proactive care, caregiver involvement, and treatment response to patients’ needs. An informatics system, called the “Medical Informatics Network
Tool” (MINT), was developed to support the project. MINT includes an internet browser interface for data entry and management, and PC-based applications that interact with providers and the VA electronic medical record. At each site, a nurse briefly assessed symptoms, side effects and other problems before each psychiatrist visit. These “psychiatric vital signs” were entered into a secure database using a web browser. When the psychiatrist subsequently opened the patient’s medical record, MINT displayed a “PopUp” window.
This provided assessment data, one-click access to guidelines, and a messaging system for team communication. Clinicians used MINT reports to identify patients who were not receiving necessary services. Managers used reports to monitor performance and identify pervasive problems that required reorganization of services.
Population Studied: 68 psychiatrists, 8 other clinicians, and
375 patients with schizophrenia were enrolled.
Principal Findings: The care model was sustained for 15 months. Clinician use of MINT was steady throughout implementation. PopUps were displayed more than 3,000 times, and viewed by psychiatrists for a median duration of 7 seconds. Providers stated that the intervention provided important new information about their patients, especially in psychosocial domains, and allowed them to monitor needs and treatment of their patients. Providers reported that team communication was substantially enhanced regarding patients, facilitating provision of appropriate services. The intervention’s effect on medication prescribing varied by psychiatrist and site. MINT identified treatment refractory patients which led to the establishment of a clozapine clinic at one site. MINT identified a pervasive problem with weight gain which led to the establishment of wellness programs at both sites. Provision of caregiver services did not increase.
Conclusions: It has not generally been possible to manage the quality of care for schizophrenia, and multiple barriers exist to improving care. MINT established an infrastructure that allowed clinicians and managers to monitor and improve care.
Implications for Policy, Delivery, or Practice: Informatics systems can be useful in a variety of chronic illnesses. At the patient level, these systems can identify specific clinical problems that need to be addresed. At the provider level, they identify when a provider should consider new treatment strategies. At the practice level, they identify pervasive problems that require changes in the organization of care.
Primary Funding Source: VA
● Clinician Interactions with an Automated Clinical
Decision Support System for Managing Hypertension in
Primary Care Clinics
Mary Goldstein, M.D., Robert Coleman, MS Pharm, Samson
Tu, MS, Susana Martins, M.D., MS Lond, Hayden Bosworth,
Ph.D., Albert Chan, M.D.
Presented By: Mary Goldstein, M.D., Associate Director,
Clinical Services, Associate Professor of Medicine, Geriatrics
Research Education and Clinical Center (GRECC), VA Palo
Alto and Stanford University, 3801 Miranda Avenue 182B, Palo
Alto, CA 94304; Tel: (650) 858-3933; Fax: (650) 496-2505;
Email: goldstein@stanford.edu
Research Objective: Information technology (IT) systems that are technically excellent have sometimes failed in clinical practice because of insufficient attention to organizational factors in the design and implementation. Furthermore, while automated clinical reminders for screening and preventive care have been extensively used by clinicians, many clinicians have declined to use automated decision support systems for chronic diseases, with rates of interaction with the system often less than 5% of available opportunities. We used a
“sociotechnical” approach to implement an automated decision support system, ATHENA DSS, for hypertension management in primary care clinics. The overall project aim was to improve quality of care by implementing evidencebased guidelines for management of primary hypertension.
Our aim in this analysis is to assess the success of the IT implementation as measured by clinician use of the system.
Study Design: Design and implementation of the ATHENA
DSS included explicit steps to meet institutional goals and to address other organizational factors, with an iterative technical design process that allowed for re-design to address organizational needs (JAMIA 2004; 11:368-376). In a clusterrandomized trial, we assigned clinicians to receive advisories about hypertension management generated by ATHENA DSS in a popup window in the electronic medical record for appropriate patients, or to receive a control popup window.
The system processed individual patient’s clinical data with a hypertension knowledge-based to generate patient-specific recommendations for clinicians about next steps in drug therapy of hypertension. Clinicians could ignore the ATHENA advisory window or elect to interact with it in a variety of ways.
We report here the rates of interaction with the system (a measure of usefulness and usability) by the active intervention group, focusing on staff clinicians (not trainees) and patients with previously diagnosed hypertension.
Population Studied: 53 primary care clinicians (physicians and nurse practitioners/physician assistants) in 47 clusters at three large geographically separated medical centers who received popup windows with ATHENA DSS recommendations for 5088 eligible patients with hypertension.
Principal Findings: When clinicians received a popup window advisory from the ATHENA system, they entered new blood pressure measurements and updated the advisories for 1907
(20%) of patients. They clicked a checkbox to indicate that they had considered the recommendations for 2425 (48%) of patients. Overall, clinicians interacted actively in some way with the advisory window for 68% of patients.
Conclusions: Clinicians interacted with the clinical decision support system extensively, demonstrating that they found it useful and usable.
Implications for Policy, Delivery, or Practice: Explicit attention to sociotechnical integration of a clinical decision support system may enhance its acceptability and usefulness for clinicians. With increasing use of electronic medical records, new opportunities are opening up for provision of sophisticated decision support for clinicians tied to specific patient data. Implementation of such systems should include an iterative design process that allows for technical design and re-design in light of organizational input, and should track clinician use of the system as a measure of effectiveness of the sociotechnical implementation.
Primary Funding Source: VA
● Implementing Routine Outcome Assessment to Improve
Care for Mental Illness
Alexander Young, M.D., MSHS, Matthew Chinman, Ph.D., Jim
Mintz, Ph.D., Joseph Hassell, MA, Jennifer Magnabosco,
Ph.D., Amy Cohen, Ph.D.
Presented By: Alexander Young, M.D., MSHS, Director,
Health Services Unit, Mental Illness Research, Education, and
Clinical Center (MIRECC), Greater Los Angeles VA Healthcare
Center, 11301 Wilshire Boulevard. (210A), Los Angeles, CA
90073; Tel: (310) 268-3416; Fax: (310) 268-4056; Email: a1young@earthlink.net
Research Objective: The VA has made a national commitment to improve access to effective, recovery-oriented mental health services. Previously, implementing these services has been very difficult. One major barrier has been a
lack of routine psychiatric outcome data. Using existing medical records, it is often impossible to identify which patients would benefit from specific services. Computerized patient self-assessment could routinely measure outcomes at minimal cost. The objective of this project was to determine whether such a system can accurately administer researchquality outcome instruments to people with severe, persistent mental illness; and to evaluate a pilot project to implement routine outcome monitoring at mental health clinics.
Study Design: We developed an audio computer-assisted patient self-assessment system (PAS) specifically to meet the needs of people with cognitive impairments. Patients access this internet-based system using a touch-screen monitor, and complete a survey of symptoms, drug and alcohol use, medication side-effects, and adherence. Accuracy of assessment was evaluated in patients with schizophrenia or bipolar disorder who completed duplicate surveys 20 minutes apart: once face-to-face, and once using the computer.
Reliability was assessed with correlations, and bias with mode x order repeated measures ANOVAs. The PAS was then implemented at two large mental health clinics, one at a VA medical center and one in a non-VA public system. Patients use the PAS before their routine visits, a report is printed, and the patient brings the report to their psychiatrist. Surveys, interviews and focus groups have been conducted with patients, clinicians, and managers to evaluate system usability, usefulness, and its effect on treatment provision and care organization.
Population Studied: 330 patients with schizophrenia, schizoaffective disorder or bipolar disorder, and 8 physicians.
Principal Findings: Outcome scales assessed by computerized and face-to-face interviews had similar internal reliability, high correlations (r=.78-1.00), and similar means (at p<.05). Most patients rated the PAS as easier, more enjoyable, preferable and more private. 97-99% perfectly answered questions about how to use it. Patients completed the PAS faster. The PAS cost less than $2000 per kiosk and was easily integrated into care processes at the clinics.
Patients reported that the PAS was easy to understand and use, facilitated talking about feelings and difficulties, and helped their psychiatrists understand them better.
Psychiatrists stated that PAS reports were easy to use, helped cue them to ask about information they could otherwise overlook, and allowed them to narrow their questioning and thereby save time during short visits. The perceived value of outcome information varied by clinician and domain, ranging from somewhat to extremely valuable.
Conclusions: Computerized self-assessment is a reliable and valid method for evaluating outcomes in people with chronic mental illness. It is relatively easy and affordable to implement and sustain at typical mental health clinics.
Implications for Policy, Delivery, or Practice: Routine patient self-assessment has the potential to support outcome monitoring broadly. Controlled research is needed to evaluate interventions that use this approach as part of efforts to improve care.
Primary Funding Source: VA
Call for Panels
Money Isn't Everything: Why Long-Term Care Frontline
Workers Enter and Stay on the Job
Tuesday, June 28 • 10:30 am – 12:00 pm
● Caregiver Job Quality and Turnover Intentions in Long
Term Care Organizations
Diane Brannon, Ph.D, Teta Barry, Ph.D., Peter Kemper, Ph.D.,
Joseph Angelelli, Ph.D.
Presented By: Diane Brannon, Ph.D., Professor, Department. of Health Policy and Administration, Pennsylvania State
University, 116 Henderson Building, University Park, PA
16802-6500; Tel: (814)863-8514; Fax: (814)-863-2905; Email: f8z@psu.edu
Research Objective: To identify the job rewards and concerns of providing personal care in diverse long term care settings and to assess how those rewards and concerns relate to organizational commitment, operationalized as turnover intention.
Study Design: A paper and pencil survey was administered to direct care workers -DCWs- currently employed at long term care provider organizations. The Job Role Quality questionnaire, Marshall et. al. 1991, and two items measuring turnover intention were administered to two different samples in a cross sectional survey. The factor structure of the JRQ scales was assessed and minor modifications made to the apriori scales. Variations by provider organizational type were identified. Regression models controlling for provider type were tested to assess relationships among rewards and concerns, respectively, and intent to turnover.
Population Studied: Surveyed were direct care workers at nursing homes, assisted living facilities, home care agencies, and adult day health centers participating in the Better Jobs
Better Care Demonstration n=200~ and those from the same array of providers in a 2004 survey of long term care workers in Pennsylvania DCWs n=600~. These are individuals who provide hands-on personal care, e.g. assistance with bathing, toileting, dressing, transferring, and feeding, and are not LPNs or RNs.
Principal Findings: In analysis of the PA sample, a significant provider type effect was observed, p=.006. Nursing and personal care facilities’ workers were more likely to express an intent to leave than were comparable workers in home care or adult day centers. Two reward factors, one measuring how challenging and rewarding the work itself is, p.=.12, and the other measuring supervisor support, p.=000. were inversely related to intent to leave. The factors helping others, decision authority, and pay satisfaction were not significant in the multivariate model. In terms of the relationship between concern factors and intent to leave, provider type was also significant, p=.000, with the same pattern as above.
Controlling for provider type, work overload,p=.000, the perception of a dead end job, p=.000, and poor supervision,p=.011, were positively related to intent to leave.
Hazards related to illness and injury and racial or gender discrimination were not significant factors. Additional analyses
are in progress, including incorporation the BJBC sample as it is completed.
Conclusions: Turnover intentions, which are predictive of turnover, are influenced by factors derived from the job itself.
These include the extent to which rewarding aspects of the work are perceived, including supervisor support and to a lesser extent challenge and recognition. Controlling for provider effects, the more overload, perceptions of a dead end job and poor supervision a direct care worker reports, the more likely she is to report turnover intentions.
Implications for Policy, Delivery, or Practice: Long term managers need resources to improve supervision of direct care workers. Policy-makers need to address the impact of caregiver overload in order to stabilize this important workforce.
Primary Funding Source: RWJF, Atlantic Philanthropies,
Penn Dept. of Aging
● Flexible work with benefits as a strategy for increasing the LTC Workforce
Candace Howes, Ph.D.
Presented By: Candace Howes, Ph.D., Associate Professor,
Chair, Department of Economics, Connecticut College, 270
Mohegan Avenue, New London, CT 06320; Tel: (860)439-
5447; Fax: (860)430-5332; Email: candace.howes@conncoll.edu
Research Objective: To evaluate why homecare workers working in the consumer-directed mode of LTC delivery choose to enter and remain in the job.
Study Design: A survey was administered by mail and phone to a sample of 5000 individual home care providers living in 8
California counties that varied by wage and benefit level and population density. Survey data was then linked to an administrative database that contained demographic and workforce participation information on those surveyed as well as on an additional 300,000 providers over a 4 year period.
Administrative data was used to corroborate self-reported reasons for entering and remaining in the workforce.
Population Studied: 300,000 California In-Home Supportive
Services (IHSS) consumer-directed homecare workers.
Principal Findings: Providers report that principle reasons for entering and staying in the job, beyond their commitment to their client, are that they are seeking a job that is part-time or has flexible hours and that has benefits. While 53 percent reported that the most important reason they entered the job and the most important reason they stayed in the job was a personal commitment to their client, 16 percent reported that the most important reason for entering and staying was that it was either part-time or flexible hours. Overall 11 percent reported that health benefits were the most important reason for entering and 14 percent for staying. When providers who ranked personal connection as the most important reason for entering or staying were asked what the second most important reason was, 46 percent cited either flexible or parttime work, 20 percent said it was the job they knew best how to do, and 12 percent cited health benefits. However, in counties where the eligibility requirements for health benefits are low 22 percent cited health or dental benefits as the most important reason for entering and 32 percent cited benefits as the second most important reason after personal connection.
Finally, among non-family providers in counties with low eligibility requirement, health benefits were the most important reason for entering the workforce.
Conclusions: Self-reported reasons why providers enter and remain in the workforce corroborate earlier results (Howes
2005) based on statistical analysis of an administrative database. While providers are attracted to consumer-directed homecare work because they want to care for their friends or family and they want a part time and flexible job, they are more likely to take and remain in the job if it offers health benefits.
Implications for Policy, Delivery, or Practice: Offering health insurance to part-time workers appears to be one of the most important factors that will attract and retain providers in consumer-directed homecare.
Primary Funding Source: RWJF, The Atlantic Philanthropies
● Improving Jobs of Frontline Workers in Long Term Care:
What Do Workers in the Better Jobs Better Care
Demonstration Say?
Peter Kemper, Ph.D., Joe Angelelli, Ph.D., Teta Barry, Ph.D.,
Diane Brannon, Ph.D., Haiyan Wang
Presented By: Peter Kemper, Ph.D., Professor, Department of
Health Policy and Administration, Pennsylvania State
University, 116 Hendeson Building., University Park, PA 16802-
6500; Tel: (814)865-6899; Fax: (814)863-2905; Email: pkemper@psu.edu
Research Objective: To learn what frontline workers in long term care say about how to best to improve their jobs; how their recommendations vary across type of provider; and whether workers recommending one type of improvement are more likely to say they will quit than workers recommending other types of improvements.
Study Design: A paper and pencil survey was administered to all DCWs currently employed at long term care provider organizations. Respondents were asked the open-ended question, “What is the single most important thing your employer could do to improve your job as a direct care worker?” Text responses were coded into one of four categories. After weighting to adjust for differences in nonresponse, coded responses were tabulated by provider type and reports of intent to quit.
Population Studied: Direct care workers (DCWs) at nursing homes, assisted living facilities, home care agencies, and adult day health centers participating in the Better Jobs, Better Care
Demonstration (n = approx 200). DCWs are individuals who provide hands-on personal care (e.g. assistance with bathing, toileting, dressing, transferring, and feeding) and are not
LPNs or RNs.
Principal Findings: First, about a third of the DCWs said increased compensation was the single most important thing that employers could do to improve their job. Almost an equal number said changes in management process or organizational culture were most important. Another 12% called for increased staffing, and the rest indicated that they were satisfied or did not respond. Second, the recommendations differed greatly across provider types.
DCWs at home care agencies were more likely to suggest management process changes and less likely to suggest improvements in organizational culture than DCWs at nursing and assisted living facilities. Reflecting their work setting, almost no home care workers suggested increasing staffing.
Third, DCWs who suggested that improving organizational culture was the most important change were more likely to say they were likely to quit their job in the next year than DCWs who recommended improved compensation, changes in management process, increased staffing, or made no recommendation.
Conclusions: According to DCWs themselves, the most important things employers could do to improve their jobs fall into two categories: those that principally require spending more money—increasing compensation and increasing staffing—and those that require changing the way workers are managed—improving the management process and improving organizational culture. Thus, money is not the only concern of frontline workers. However, the strategies DCWs recommend to improve jobs in home care agencies differ from those needed in nursing or assisted living facilities.
Implications for Policy, Delivery, or Practice: The aging of the population will greatly increase the number of people with disabilities who need personal care and the need for frontline workers who provide the bulk of long term care. Policymakers who seek to increase the supply of workers and long term care provider organizations that seek to recruit and retain workers should heed the recommendations of the workers themselves: improve pay, increase staffing, improve management processes, and especially work to improve organizational culture.
Primary Funding Source: RWJF, The Atlantic Philanthropies
Call for Panels
The Impact of Drug Coverage & Benefit Design on the Use and Cost of Prescription Medications among the Elderly
Tuesday, June 28 • 10:30 am – 12:00 pm
● Generosity of Retiree Drug Benefits and Essential
Medication Use among Aged Medicare Beneficiaries with
Employer-Sponsored Health Insurance
Jalpa Doshi, Ph.D., Daniel Polsky, Ph.D.
Presented By: Jalpa Doshi, Ph.D., Health Services Research
Scientist, Division of General Internal Medicine, University of
Pennsylvania, 1214 Blockley Hall, Philadelphia, PA 19104; Tel:
(215)898-7989; Fax: (215)898-0611; Email: jdoshi@mail.med.upenn.edu
Research Objective: The new Medicare drug benefit (Part D) has created unintended incentives for employers to drop or scale-back retiree drug benefits. The CBO has projected that
2.7 million beneficiaries will be dropped from retiree drug plans and shift into Part D plans. The 2004 Kaiser/Hewitt
Survey suggests that some beneficiaries will also have their retiree drug benefit scaled back to match the standard Part D benefit. Our objective is to examine how the potential net reduction in drug coverage for those with currently generous retiree drug benefits would affect their medication use. The study takes advantage of the existing variation in retiree drug benefits to examine the association between generosity of drug coverage and essential medication use among retired seniors.
Study Design: The study used the 1997-2000 Medicare
Current Beneficiary Survey, a nationally-representative survey of the Medicare population linked with Medicare claims. The essential medications selected were (i) statin use in retirees with coronary artery disease (CAD) and hyperlipidemia and (ii)
ACE-inhibitor/ARB use in retirees with diabetes and a cardiovascular risk factor (CVD). Generosity of retiree benefits was defined based on the percentage of the beneficiary’s annual drug expenditure paid by the employer (0%, 1-50%, 51-
75%, and 76-100%). Multivariate logistic regressions with survey estimators were used to estimate the impact of different drug coverage generosity levels on odds of receiving the study medications controlling for socioeconomic, demographic, and disease-related characteristics.
Population Studied: Aged community-dwelling fee-for-service
Medicare beneficiaries from 1997 to 2000 with supplemental health insurance from own/spouse’s former employer, and one or more claims in the survey year with diagnoses of (i)
CAD and hyperlipidemia [N=1,220] and (ii) diabetes and CVD
[N=1,141]. Both samples exclude Medicare retirees with drug coverage from sources other than an employer.
Principal Findings: More than half the retirees with CAD and hyperlipidemia had 76-100% and another quarter had 51-75% of their annual drug expenditures paid by their employer.
Only 10 percent retirees were in the 1-50% generosity group and about 8 percent received no retiree drug benefits (i.e. 0% group). Prevalence of statin use significantly decreased with decreasing generosity of drug coverage [76-100% group:
72.6%; 51-75% group: 55.5%; 1-50% group:54.5%; and 0% group: 45.6%]. These results held even after multivariate adjustment. Retirees in the 51-75% group [OR:0.48; 95% CI:
0.35-0.64] and the 1-50% group [OR:0.44; 95% CI: 0.29-0.67)] had less than half the odds of receiving statins than those in the 76-100% group. Similar results were observed for ACEinhibitor/ARB use in retirees with diabetes and CVD.
Conclusions: Less generous retiree drug benefits are associated with lower use of recommended medications among Medicare retirees with selected disease conditions.
Implications for Policy, Delivery, or Practice: Our findings suggest that an elimination or reduction of retiree drug coverage due to the Medicare drug benefit may lead to a decrease in the use of essential medications for retirees who were otherwise receiving generous employer drug benefits. It will be critical to monitor employer response to Part D over the coming years and determine the extent to which access to effective medications and potentially the health of the nation’s retired seniors may be adversely affected.
Primary Funding Source: Leonard Davis Institute of Health
Economics
● The Impact of Tiered Copayments on the Use and Cost of Prescription Drugs among the Elderly in Employer-
Sponsored Health Plans
Boyd Gilman, Ph.D., John Kautter, Ph.D.
Presented By: Boyd Gilman, Ph.D., Health Economist, RTI
International, 411 Waverley Oaks Road, Suite 330, Waltham,
MA 02452; Tel: (781)788-8100 x187; Fax: (781)7881-8101; Email:
BGilman@rti.org
Research Objective: Medicare beneficiaries will soon face a range of tiered copayment options under the new Medicare drug benefit. Tiered copayments are designed to limit plan exposure, curtail excessive use of medications, and encourage use of generic alternatives without sacrificing the benefits of drug therapies. This study assesses the impact of tiered copayments for Medicare retirees with employer-sponsored drug coverage on use of prescription medications, substitution of generic for brand named drugs, enrollee out-ofpocket spending, and plan financial risk. The differential effects of tiered copayments for beneficiaries with chronic conditions are evaluated separately.
Study Design: The study uses data from the 2002 and 2003
MarketScan Medicare Supplemental and Benefit Plan Design
Databases. Pharmacy claims for one million retirees were merged with drug benefit design information for over 50 employer-sponsored health plans. Plans were classified into 1-
, 2- and 3-tiered copayment groups and a coinsurance group.
Regression analyses were used to assess the impact of copayment tiers on the probability of filling a prescription; the proportion of prescriptions dedicated to single source and multisource brand named drugs and generic drugs; annual plan payments; and annual enrollee out-of-pocket spending.
Covariates include enrollee characteristics (age, gender, geographic region); employment characteristics (industry, employment status, union status); and plan characteristics
(plan type, drug formulary, generic incentive, pre-certification, preventive care benefit, utilization review). Models were estimated over all enrollees and enrollees with acute and chronic conditions separately.
Population Studied: Medicare beneficiaries and their dependents with employer-sponsored drug coverage drawn primarily from large unionized firms in the manufacturing and durable goods industry.
Principal Findings: Enrollees in 2-tiered and 3-tiered plans were less likely to fill a prescription than those in 1-tiered plans, while beneficiaries in coinsurance plans were less likely to submit a pharmacy claim than enrollees in copayment plans. Number of prescriptions filled and proportion of brand named drugs used were negatively correlated with the number of tiers. Mean plan payments were $250 less for 2-tiered plans and $639 less for 3-tiered plans compared with 1-tiered plans. Average payments among coinsurance plans were
$580 lower than 1-tiered plans. Conversely, enrollee out-ofpocket spending was positively correlated with number of tiers. Enrollee copayments were $238 higher for those in 3tiered plans compared with those in single tiered plans. All results were highly significant. While enrollees with chronic conditions were more responsive than those with chronic conditions, the differences were not significant.
Conclusions: Multi-tiered copayments limit the likelihood of submitting a pharmacy claim and the number of claims submitted, and increase the likelihood of using generic substitutes. Higher enrollee copayments reduce plan financial risk, but increase beneficiary out-of-pocket spending.
Beneficiaries with chronic conditions were more price sensitive than those with chronic conditions.
Implications for Policy, Delivery, or Practice: Medicare retirees are sensitive to enrollee copayments in terms of quantity and type of medications used. While copayment tiers are effective for limiting plan exposure and encouraging use of generic alternatives, the impact of benefit design on health outcomes, particularly among beneficiaries with chronic conditions reliant on medications to manage their disease and maintain health, should be monitored under the Medicare drug benefit.
Primary Funding Source: No Funding
● Estimating the Moral Hazard Effect of Prescription Drug
Coverage
Dennis Shea, Ph.D, Joseph Terza, Ph.D., Bruce Stuart, Ph.D.,
Becky Briesacher, Ph.D.
Presented By: Dennis Shea, Ph.D., Professor, Health Policy and Administration, Penn State, 116 Henderson Building,
University Park, PA 16802; Tel: (814)863-5421; Fax: (814)863-
2905; Email: dshea@psu.edu
Research Objective: To estimate the moral hazard impact of prescription drug coverage among the elderly
Study Design: The design is a retrospective cross-sectional correlational design.
Population Studied: The sample studied is communitydwelling respondents to the 1999 Medicare Current
Beneficiary Survey who had full-year Medicare coverage and either full-year or no prescription drug coverage.
Principal Findings: Using a new econometric method and these more recent data, we find that while selection is present in prescription drug coverage, its magnitude is small. As a result, our findings suggests that moral hazard in prescription drug coverage among the elderly may be twice as large as prior estimates.
Conclusions: We conclude that prior estimates of the effects for prescription drug coverage may not have controlled for selection as well as our methods do or that changes in the market for prescription drugs have increased the price sensitivity among Medicare beneficiaries.
Implications for Policy, Delivery, or Practice: If our estimates are correct, then estimates of the cost effects of adding prescription drug coverage to Medicare significantly underestimate the costs of the coverage provided.
Primary Funding Source: CWF
● Gaps in Drug Benefits: The Impact on Utilization and
Spending for Drugs Used by Medicare Beneficiaries with
Mental Illness
Linda Simoni-Wastila, BSPharm, Ph.D., Bruce Stuart, Ph.D.,
Christopher Blanchette, MS, Thomas Shaffer, MHS
Presented By: Linda Simoni-Wastila, BSPharm, Ph.D.,
Associate Professor, Pharmaceutical Health Services
Research, University of Maryland Baltimore School of
Pharmacy, 515 West Lombard Street, Room 162, Baltimore,
MD 21201; Tel: (410)706-4352; Fax: (410)706-1488; Email: lsimoniw@rx.umaryland.edu
Research Objective: Medicare beneficiaries have experienced unstable drug coverage in the past; with the new Part D drug benefit, it is likely that many beneficiaries will continue to experience gaps in their drug coverage. In prior work, we found evidence that Medicare beneficiaries with mental health conditions were more likely than those without such problems to experience drug coverage gaps. The objectives of this study are to: 1) quantify the number and duration of gaps in drug coverage among seriously mentally-ill Medicare beneficiaries between 1997 and 2001; and 2) describe drug use and spending patterns by these individuals as a function of their experience with coverage gaps.
Study Design: This study is based on data from the 1997-
2001 Medicare Current Beneficiary Study (MCBS). The primary outcome measures are: 1) drug coverage gap count;
2) drug coverage gap duration; and 3) total and out-of-pocket spending for prescription drugs used by those with mental illness, including psychotherapeutic drugs by brand/generic and by “breakthrough”/ traditional drug status. Covariates include age, gender, race/ethnicity, poverty level, geographic region, health status, and mortality. We control for burden of illness through the CMS-HCC risk adjuster. We use OLS and logistic regression to model our outcome measures.
Population Studied: We pooled 3 three-year cohorts of community-dwelling Medicare beneficiaries (1997-98, 1998-99, and 1999-00) for a total sample of 9,465 unique individuals.
Of these, 1,603 (16.9%) had at least one serious mental illness
(major depression, bipolar disorder, schizophrenia, and other psychotic disorders) as assessed through ICD-9 diagnoses on
Part A and B billing records.
Principal Findings: Among seriously mentally ill Medicare beneficiaries, 43.4% had continuous drug coverage throughout their three-year observation period, 36.7% had at least one gap in drug coverage, and 19.9% had no drug coverage. Most beneficiaries with gaps had a single spell without coverage (62.4%); however, among these individuals the mean annual gap duration was 7.3 months, one month longer than beneficiaries without serious mental illness.
Mentally ill beneficiaries with continuous coverage spent
$1,777 annually on prescription drugs, compared to $1,507 spent by those with at least one gap, and $948 by those with no drug coverage. Out-of-pocket spending as a proportion of total drug spending was higher among those with coverage gaps (42.0%) and those with no drug coverage (87.8%), compared to those with continuous coverage (15.2%).
Conclusions: Interruptions in prescription drug coverage result in reductions in drug use and spending compared to those with continuous drug benefits.
Implications for Policy, Delivery, or Practice: The
“doughnut-hole” design of the Medicare Part D places individuals with serious, chronic mental illness at high risk for gaps in prescription drug coverage. These gaps may translate into clinical consequences – reduced access to, adherence with, and substitution of less-appropriate medications – that may result in increased morbidity, mortality and increased
Medicare costs. Researchers and policy-makers should consider the disproportionate impact of the Part D provisions on these particularly vulnerable Medicare beneficiaries.
Primary Funding Source: RWJF
Call for Panels
A Comparison of Integrated vs. Enhanced Referral Models l
Health or Substance Use Problems
Tuesday, June 28 • 10:30 am – 12:00 pm
● The Influence of Integration on the Expenditures and
Costs of Mental Health and Substance Use Care: Results from the PRISM-E Study
Marisa Domino, Ph.D., James Maxwell, Ph.D., Karen Cheal,
MPH, Eugenie Coakley, MPH, Sue Kim, Saminaz Zaman,
Presented By: Marisa Domino, Ph.D., Assistant Professor of
Health Economics, Health Policy and Administration,
University of North Carolina at Chapel Hill, 1104G McGavran-
Greenberg Hall, Chapel Hill, NC 27599-7411; Tel: (919) 966-
3891; Fax: (919) 966-6961; Email: domino@unc.edu
Research Objective: To determine differences in costs of care to older adult patients in integrated versus enhanced referral care.
Study Design: Older adults aged 65 and above were screened at primary care clinics from 10 study sites throughout the U.S.
Those who met diagnostic criteria for depression, anxiety, and/or at-risk alcohol consumption were randomly assigned to either IC or ES models. The PRISM-E cost study relied on a combination of claims and self-reported data on health care utilization. Because of the reliance on different claims systems, analyses for the non-VA clinics and the VA clinics were run separately.
Population Studied: Older adult (over 65 years old) users of primary care services at eleven study sites throughout the country. Study settings included community health centers,
VA facilities, and other community primary care clinics.
Over 25,000 older adults were screened for mental health and alcohol drinking problems in primary care settings. More than
2,200 elderly primary care patients with depression, anxiety, and at-risk drinking participated in this study.
Principal Findings: This study found no significant difference in outpatient MH/SA costs between integrated and enhanced referral care for non VA or VA participants. Integrated MH/SA interventions relying on mid-level mental health professionals are thus equivalent in terms of costs to care delivered by highly trained mental health specialists working in specialty settings. Although the costs were not significantly different, we did find higher levels of outpatient visits in the integrated arm. Moreover the types of outpatient MH/SA visits differ with the integrated intervention having shorter visits and less medication than enhanced referral. When examining the
relationship between the intervention and total health care costs, we found that non-VA participants of the study had lower total health care expenditures. A treatment assignment in the integrated arm resulted in a decrease of $245 in total expenditures in the non-VA sample. Total health care costs were not significantly different in the VA sample. In an analysis of at risk drinkers, we found higher rates of outpatient utilization and costs on the integrated arm for both the non-
VA and the VA among less severe patients. Outpatient SA utilization was significantly associated with the severity of alcohol conditions as measured by the GMAST among VA patients (P <.0001). The brief intervention on a per-visit basis was less costly than alternative forms of interventions provided to at risk drinkers. Total health care costs were related to the intervention arm and to the severity of the condition. The brief intervention on a per-visit basis was less costly than alternative forms of interventions provided to at risk drinkers. Total health care costs were related to the intervention arm and to the severity of the condition.
Conclusions: The cost of outpatient MH/SA care is equivalent for IC and ESR models. For at-risk drinkers, IC is associated with higher utilization and higher costs of outpatient SA treatment. Total health care costs were associated with the severity of the alcohol condition in both arms of the study.
Implications for Policy, Delivery, or Practice: From a cost perspective, neither care model is more favorable for depressed older adults. Therefore, Medicare policy should be designed to be consistent in the incentives it sets for each model.
Primary Funding Source: SAMHSA
● Depression and At-Risk Alcohol Use Outcomes for Older
Primary Care Patients in Integrated Care and Enhanced
Specialty Referral
Dean Krahn, M.D, Steve Bartels, M.D., Eugenie Coakley, MPH,
David Oslin, Ph.D., Henry Chung, M.D., Sue Levkoff, Sc.D.
Presented By: Dean Krahn, M.D., William S. Middleton
Memorial Veterans Hospital, 2504 Overlook Terrace,
Madison, WI 53705; Tel: (608) 280-7015; Email: dean.krahn@med.va.gov
Research Objective: Although studies have shown that integrated mental health care delivery in primary care results in better outcomes for older patients than usual care, no research has examined whether this collaborative model results in better outcomes than enhanced referral to specialty mental health clinics. This paper compares clinical outcomes for older adults with depression and at-risk drinking in integrated versus enhanced referral care.
Study Design: Older adults aged 65 and above were screened at primary care clinics from 10 study sites throughout the
U.S.A. Those who met diagnostic criteria for depression, anxiety, and/or at-risk alcohol consumption were randomly assigned to either IC or ES models.
Population Studied: Older adults (over 65 years old) users of primary care services at eleven study sites throughout the country. Study settings included community health centers,
VA facilities, and other community primary care clinics.
Over 25,000 older adults were screened for mental health and alcohol drinking problems in primary care settings. more than
2,200 elderly primary care patients with depression, anxiety, and at-risk drinking participated in this study.
Principal Findings: Overall depression remission rates and symptom reduction were similar for both models at 6 month follow-up. However, ESR was associated with a greater reduction in depression severity compared to IC for major depression (95% CI 1.0, 4.5, p=.003). Six months outcomes demonstrate greater engagement in care in the collaborative care model (p<0.001) and significant time effects with reductions in both weekly drinking (p<0.001) and binge drinking (p<0.001),
Conclusions: These findings suggest that outcomes for older persons with minor depression are equivalent to outcomes in referral. For major depression, outcomes may be superior in a system in which referral is optimized. The study shows that a significant number of older at-risk drinkers can substantially modify their drinking over time, regardless of treatment models.
Implications for Policy, Delivery, or Practice: The study results support the New Freedom Commission's recommendations to let patients choose the system that best meets their needs.
Primary Funding Source: SAMHSA
● Overview of the PRISM-E Study: Methodology and
Access to Care
Sue Levkoff, Sc.D., Eugenie Coakley, MPH, Steve Bartels,
M.D., Hongtu Chen, Ph.D., Susan Grantham, Ph.D., Dean
Krahn, M.D.
Presented By: Sue Levkoff, Sc.D., Associate Professor,
Psychiatry, Brigham & Women's Hospital, Harvard Medical
School, 1249 Boylston Street, 3rd Floor, Boston, MA 02215;
Tel: (617) 525-6122; Email: sue_levkoff@hms.harvard.edu
Research Objective: The PRISM-E Study is a multisite randomized trial comparing two types of care models for delivery of mental health services to elderly people through primary care, i.e., the integrated care (IC) model and an enhanced specialty referral model (ESR). The paper will describe the PRISM-E research methodology and findings with regard to access to care.
Study Design: Older adults aged 65 and above were screened at primary care clinics from 10 study sites throughout the
U.S.A. Those who met diagnostic criteria for depression, anxiety, and/or at-risk alcohol consumption were randomly assigned to either IC or ESR models.
Population Studied: Older adult (over 65 years old) users of primary care services at eleven study sites throughout the country. Study settings included community health centers,
VA facilities, and other community primary care clinics. Over
25,000 older adults were screened for mental health and alcohol drinking problems in primary care settings. More than
2,200 elderly primary care patients with depression, anxiety, and at-risk drinking participated in this study.
Principal Findings: Among those randomized, seventy-one percent of patients engaged in treatment (defined as at least one visit) in the integrated model compared with 49% in the enhanced referral model. Integrated care was associated with more mental health and substance abuse visits per patient
(mean=3.04) relative to enhanced referral (mean=1.91).
Overall, greater engagement was predicted by integrated care and higher mental distress. For depression, greater
engagement was predicted by integrated care and more severe depression. For at-risk alcohol users, greater engagement was predicted by integrated care and more sever problem drinking.
For all conditions, greater engagement was associated with closer proximity of mental health/substance abuse services to primary care.
Conclusions: Older adult patients are more likely to accept mental health treatment within a primary care setting than in mental health/substance abuse clinics.
Implications for Policy, Delivery, or Practice: As older adults under-use the mental health system, this study provides empirical support for policy reform aimed at improving access to coordinated behavioral health services by integrating the mental health and primary care sectors.
Primary Funding Source: SAMHSA
● PRISM-E findings and the Evolution of Federal
Behavioral Health Policy
Elizabeth McDonnell Herr, Ph.D., Bill Van Stone, Ph.D., James
Maxwell, Ph.D.
Presented By: Elizabeth McDonnell Herr, Ph.D., Social
Science Analyst, Center for Mental Health Services, U.S.
Substance Abuse and Mental Health Services Administration,
Room 11C-22 Parklawn Building, Rockville, MD 20857; Tel:
(301)594-2197; Email: bmcdonel@samhsa.gov
Research Objective: The research explores the implications of
PRISM-E findings for behavioral health policy. Findings will be discussed in the context of changing behavioral health policy at SAMHSA, the VA, and CMS.
Study Design: Older adults aged 65 and above were screened at primary care clinics from 10 study sites throughout the U.S.
Those who met diagnostic criteria for depression, anxiety, and/or at-risk alcohol consumption were randomly assigned to either IC or ES models.
Population Studied: Older adult (over 65 years old) users of primary care services at eleven study sites throughout the country. Study settings included community health centers,
VA facilities, and other community primary care clinics.
Over 25,000 older adults were screened for mental health and alcohol drinking problems in primary care settings. More than
2,200 elderly primary care patients with depression, anxiety, and at-risk drinking participated in this study.
Principal Findings: The PRISM-E study found that many older adults with behavioral health problems could be successfully screened in a primary care setting and treated in either integrated care or enhanced referral care. Screening in primary care revealed a spectrum of behavioral health problems from mild to severe depressive and alcohol disorders, with significant amounts of comorbidty.
For those older adults with minor depression or at risk drinking problems, integrated care was equivalent in its clinical and cost outcomes to enhanced referral care. For those with major depression and more severe alcohol problems, enhanced referral was shown to be more effective.
The PRISM-E study demonstrated that access barriers to referral care can be reduced through relatively modest enhancements, though some older adults may still be reluctant to access any behavioral health services if they are only offered in specialty settings.
Conclusions: The PRISM-E findings have three key implications for federal behavioral health policy. First, given the clinical cost equivalence of integrated and enhanced referral care for many older adults, they strongly support the
New Freedom Commissions recommendations to allow patients to choose the care system that best meets their needs and preferences. Thus, for the majority of patients, individual preference may serve as the most effective guide to choosing between integrated and enhanced referral care. However, there are large numbers of older adults with major depression and severe alcohol disorders who could in many cases be best served through enhanced referral systems. Very little attention has been devoted at the federal level or among foundations in enhancing referral systems and their linkages to primary care settings. Enhancing the referral system is particularly important because of the number of primary care facilities serving older adults that cannot be integrated because of their small size. Finally, Medicare and other payers need to reform their policies so that there is a level playing field between integrated and referral care models. This requires a reimbursement policy that does not favor either model. The
Medicare reimbursement rules currently do not allow midlevel providers to be reimbursed in a primary care setting without onsite supervision by a psychiatrist.
Implications for Policy, Delivery, or Practice: Findings from
PRISM-E will be discussed from the perspectives of different federal agencies. The study calls for a more evidence-based and patient-centered care system for older adults with behavioral health problems.
Primary Funding Source: SAMHSA
● Stakeholder Perspectives on Integrated Mental Health
Services in Primary Care Settings
Cynthia Zubritsky, Ph.D., Hongtu Chen, Ph.D., Joseph Gallo,
Ph.D., James Maxwell, Ph.D., Karen Cheal, MPH, Jack
McIntyre, M.D.
Presented By: Cynthia Zubritsky, Ph.D., Director, Managed
Care Initiatives, University of Pennsylvania, 600 Market Street,
7th Floor, Philadelphia, PA 19104; Tel: (215)662-2886; Email: cdz@mail.med.upenn.edu
Research Objective: To examine patient satisfaction and provider perspectives on integrated care versus enhanced referral models.
Study Design: Older adults aged 65 and above were screened at primary care clinics from 10 study sites throughout the U.S.
Those who met diagnostic criteria for depression, anxiety, and/or at-risk alcohol consumption were randomly assigned to either IC or ES models. We also collected supplementary data on primary care providers' perspectives on integrated versus enhanced referral care.
Population Studied: Older adult (over 65 years old) users of primary care services at eleven study sites throughout the country. Study settings included community health centers,
VA facilities, and other community primary care clinics.
Over 25,000 older adults were screened for mental health and alcohol drinking problems in primary care settings. More than
2,200 elderly primary care patients with depression, anxiety, and at-risk drinking participated in this study.
Principal Findings: Older primary care patients who participated in PRISM-E reported significantly higher satisfaction with the integrated mental health care than with the enhanced specialty referral care. Patients who attended the treatment service twice or more were more likely to feel
satisfied with the IC than ESR. A reduction in depression symptoms as measured by a decrease in CES-D score was significantly associated with patient satisfaction for the subset of patients with depression but clinical outcome was not predictive of patient satisfaction for those with at-risk drinking.
Conclusions: This finding suggests that overall primary care patients would prefer mental health services delivered in the primary care setting. Primary care physicians, in general, preferred the integrated model across several domains. They reported that integrated care led to better communication between PCPs and MH providers, less stigma, and better coordination of mental health and general health care. Mental health clinicians also have positive views of how an integrated model will improve access to mental health services, although they have more confidence in the quality of treatment provided by the enhanced specialty referral, especially for patients with major depression.
Implications for Policy, Delivery, or Practice: The study shows that patient satisfaction is often more dependent on patient preferences than actual clinical outcomes. This supports policy analysts' move for choice of care.
Primary Funding Source: SAMHSA
Call for Panels
Effects of Hospital Competition & Managed Care
Penetration on Outcomes, Inputs & Alternative Sites of
Care
Chair: R. Tamara Konetzka, University of Chicago
Tuesday, June 28 • 10:30 am – 12:00 pm
● Market Forces, Adoption of Cardiac Care Technology, and Outcomes
Laurence Baker, Ph.D., Chris Afendulis, Ph.D.
Presented By: Laurence Baker, Ph.D., Associate Professor,
Department of Health Research and Policy, Stanford
University, Redwood Building, T253, Stanford, CA 94305; Tel:
(650)723-4098; Fax: (650)723-1919; Email: laurence.baker@stanford.edu
Research Objective: Changes in financial incentives brought about by managed care and other health plan activities could influence the diffusion of medical technologies. Competition between hospitals could also play a role. Shifts in adoption patterns could impact health care costs and patient outcomes.
Some previous work explores relationships between managed care, competition, and technology adoption, but only for some technologies and often without the ability to link adoption to treatment patterns and outcomes. This paper studies relationships between managed care, competition, and adoption of three hospital-based technologies used in the care of heart attack (acute myocardial infarction, or AMI) patients: cardiac catheterization, percutaneous transluminal coronary angioplasty (PTCA), and coronary artery bypass graft (CABG).
Study Design: The first part of the study is a hospital-level longitudinal analysis of factors affecting adoption. Using comprehensive Medicare claims files, we identify adoption from the time at which hospitals began billing for each service.
We link data on HMO activity and hospital competition in the hospital’s metropolitan statistical area (MSA). We use hazard models and more general transition models to study relationships between managed care, competition, and adoption, controlling for a wide range of population demographics and market characteristics. We pay particular attention to the possibility that HMOs are more likely to enter areas where hospitals or populations have less strong preferences for new technologies, using early adoption of related but older technologies to proxy for preferences. The second part of the study is an individual-level cross-sectional analysis of the relationship between technology availability, technology use, and outcomes. We use Medicare claims data for patients with new AMIs, and estimate regressions that relate 1) area managed care and competition measures to the probability that a patient’s index hospital has a given technology, 2) the availability of a technology in the index hospital to actual receipt of the technology, and 3) receipt of these technologies to outcomes. We control for an extensive set of demographics and comorbidities.
Population Studied: In the first part, all 2,877 general acute care hospitals in MSAs in operation in 1985. For the second part, a 20% sample of over-65 FFS Medicare patients with
AMI in 1996-2000 (N=148,177).
Principal Findings: Increases in managed care are associated with reductions in adoption of cardiac technologies, and we do not find evidence that this finding is related to HMO entry into areas with preferences for lower levels of technology.
Work on the effects of competition is ongoing. In individuallevel models, decreases in the availability of technologies are associated with less use of the technologies and with worse outcomes.
Conclusions: Financial pressure limited the adoption of these three cardiac technologies. Though other work suggests that slowing adoption of new technologies can sometimes be beneficial, in this case reduced availability may have led to worse outcomes.
Implications for Policy, Delivery, or Practice: Market forces can have important implications on health care delivery and outcomes via effects on technologies. Technology adoption is an important area for continued monitoring, from the standpoint of both costs and outcomes.
Primary Funding Source: No Funding
● Hospital Competition, HMO Penetration, and Growth of
Free-Standing Ambulatory Surgery Centers
John Bian, Ph.D., Michael A. Morrisey, Ph.D.
Presented By: John Bian, Ph.D., Assistant Professor, Division of Preventive Medicine, University of Alabama at Birmingham,
MT 640, 1530 3rd Avenue S, Birmingham, AL 35294; Tel:
(205)934-7608; Fax: (205)934-7959; Email: jbian@mail.dopm.uab.edu
Research Objective: Free-standing Ambulatory surgery centers (ASCs) have been a growing phenomenon for the last
20 years. They provide a lower-priced substitute for hospitalbased outpatient departments. However, their existence is controversial. Some argue that ASCs achieve cost advantages by performing a narrow set of procedures in an efficient fashion. Others argue that such facilities pose a cost advantage to hospitals because of ASCs drawing profitable surgeries away from hospitals, increase unnecessary surgeries because of financial incentives inherent to physician ownership, and provide inferior quality of care because of their
limited ability to handle complications. Little empirical research has examined market determinants of the growth of
ASCs. In markets with substantial hospital competition, hospitals may compete more aggressively for ambulatory surgeries, reducing the potential profitability of ASCs.
Managed care plans, on the other hand, selectively contract with efficient providers such as ASCs. We hypothesize that
ASCs are less likely to be present in markets with higher hospital competition but are more likely to be present in markets with greater managed care penetration.
Study Design: We constructed a 1990-98 Metropolitan
Statistical Area (MSA) panel dataset from the Medicare Online
Survey Certification and Reporting System, excluding rural areas where fewer ASCs are located. The dataset was merged with information from the American Hospital Association annual survey files, an HMO penetration file, and Area
Resource File. An MSA is a healthcare market. The dependent variable is the number of ASCs per MSA. The key explanatory variables are hospital competition, measured by the
Herfindahl Index, and HMO penetration. We used ordinary least squares (OLS) regression with the log-transformed dependent variable for estimation. In addition, negative binomial regression was used to model the count nature of the dependent variable. Both models used MSA fixed effects to control for MSA-level heterogeneity.
Population Studied: The sample has 291 MSAs from 1990-
98, yielding a total of 2619 MSA-years.
Principal Findings: The average number of ASCs per MSA increased from 3.16 in 1990 to 7.74 in 1998. Hospital competition remained stable from 1990-95 but became less competitive from 1996-98. HMO penetration doubled during the period. Although the two sets of estimates of hospital competition and HMO penetration from both models are consistent, only the estimates from the log-transformed OLS are statistically significant (p < .05). Our regression analysis suggests that the growth of ASCs is slower in markets with higher hospital competition, but faster in markets with lower
HMO penetration. Based on the OLS estimates, a 10percentage-point decrease in hospital competition is associated with an increase of 30% in the number of ASCs; a
10-percentage-point increase in HMO penetration is associated with a decrease of 40% in the number of ASCs.
Conclusions: Our analysis suggests that market forces might play important roles in the growth of ASCs. Additional research on dynamics of outpatient surgery competition and quality of care is planned.
Implications for Policy, Delivery, or Practice: Managed care contracting practices may result in a slower proliferation of
ASCs. More work is needed to understand the effects of hospital competition.
Primary Funding Source: No Funding
● Does Hospital Price Competition Influence Nurse
Staffing and Quality of Care?
Julie Sochalski, Ph.D., Kevin Volpp, M.D., Ph.D., R. Tamara
Konetzka, Ph.D., Jingsan Zhu, M.B.A., Joanne Spetz, Ph.D.
Presented By: Julie Sochalski, Ph.D., Associate Professor,
School of Nursing, University of Pennsylvania, Room 358 NEB,
420 Guardian Drive, Philadelphia, PA 19104; Tel: (215)898-
3147; Email: julieas@nursing.upenn.edu
Research Objective: Price competition between hospitals has been shown to reduce the rate of growth in hospital costs, though little is known about how price competition has affected quality of care. This study examines whether price competition among hospitals induced reductions in nurse staffing and the consequences for patient outcomes.
Study Design: The study focuses on hospitals in California where selective contracting legislation, passed in 1982, created significant price competition among hospitals which resulted in lower hospital cost growth. Annual hospital financial data and patient discharge data for 1983 through 2001 from
California’s Office of Statewide Health Planning and
Development were used to determine whether increasing price competition led to cutbacks in the total hospital workforce and in particular the hospital nursing workforce
(RN, LVN, and nurse aides), and to examine if and how these effects differ across hospital market areas. We assess the impact of these staffing changes on 30-day inpatient mortality.
We include HMO penetration as well as an interaction between HMO penetration and hospital market concentration to examine whether managed care effects are mediated by hospital market area competition. We use hospital fixed effects to control for time-invariant differences between hospitals.
Population Studied: 478 short-term acute hospitals and
657,523 patients with a primary diagnosis of AMI linked with state death certificates.
Principal Findings: RN hours per patient day (RN ratio) rose with increasing HMO penetration from 1988-1997 and then declined significantly through 2001. In contrast, the trends in the most competitive markets showed significantly that competition was associated with lower nurse staffing from
1993-1997 and higher nurse staffing ratios from 1998-2001.
LVN hours per patient day fell consistently with increasing
HMO penetration and this was also true since 1988 for nurse aides. AMI mortality declined significantly as the RN ratio rose, though changes in LVN and nurse aide hours did not influence differential mortality. While mortality rose to a significantly greater degree in more competitive markets, this effect did not appear to be mediated by the changes in the RN ratio.
Conclusions: Price competition has influenced hospital decisions about the use of nursing services among RNs,
LVNs, and nurse aides. Decreases in RN staffing led to higher mortality rates for AMI patients. However, reductions in nurse staffing did not appear to be the mechanism by which increasing price competition negatively affects mortality.
Implications for Policy, Delivery, or Practice: The causal association between mortality and RN staffing could fuel the debate on the advisability of mandatory hospital nurse staffing ratios, recently legislated in California and under consideration nationwide. Increasing staffing levels alone, though, will not
be sufficient to ameliorate the adverse effects of other market forces on quality of care.
Primary Funding Source: Doris Duke Charitable Foundation
● Effects of Financial Stress from Price Competition on
Hospital Quality of Care
Kevin Volpp, M.D., Ph.D., R. Tamara Konetzka, Ph.D., Julie
Sochalski, Ph.D., Jingsan Zhu, M.B.A.
Presented By: Kevin Volpp, M.D., Ph.D., Assistant Professor,
School of Medicine, University of Pennsylvania and
Philadelphia VA Medical Center, 1225 Blockley Hall, 423
Guardian Drive, Philadelphia, PA 19104; Tel: (215)573-9718;
Fax: (215)573-8778; Email: volpp70@mail.med.upenn.edu
Research Objective: Hospital competition in the United
States shifted from a quality/amenity basis to a price basis with the growth of managed care in the 1980s and 1990s. The passage of selective contracting legislation in California in
1982 gave third-party payers new freedom to negotiate discounts with hospitals and to channel patients away from high-cost providers and has been shown to have lowered the rate of increase in hospital costs. Little is known about whether cost containment by hospitals was achieved through reductions in the quality of services provided or through greater efficiency. In this study, we examine whether financial stress in hospitals induced by price competition affected mortality rates for patients with acute myocardial infarction
(AMI).
Study Design: This is an observational study comparing rates of change in 30-day mortality rates for AMI patients in markets of differing degrees of hospital market concentration and increases in managed care penetration. We use hospital fixed effects to control for time-invariant differences between hospitals. An instrumental variables approach is used to adjust for the potential endogeneity of HMO penetration. We test whether more vulnerable populations such as the uninsured were adversely affected to a greater degree than the insured and the robustness of our analyses to market definition and different model specifications.
Population Studied: We examined data on 750,379 patients with primary diagnosis for AMI admitted to California hospitals between 1983 through 2001. These data were linked with state death certificates from 1990-2001.
Principal Findings: Overall 30-day in-hospital mortality from
AMI dropped from 14.9% to 8.3% from 1983 to 2001 in
California. However, the rate of improvement varied significantly across markets. We find that for every 1 percentage-point increase in managed care penetration there is .053 percentage point higher AMI mortality (p<.01). In less competitive markets, managed care penetration appears to lower AMI mortality to a non-significant degree. Using patient flow markets, the adverse effects of price competition are largest among the under 65 insured and effects are not significant among the uninsured. Using HSA-based markets, effects are largest among the uninsured. Adverse effects from price competition were strongest in the late 1980s and early
1990s.
Conclusions: Price competition among hospitals adversely affected AMI mortality and this effect was greatest in the late
1980s and early 1990s. There is some evidence that the uninsured fared worse than the insured under price competition.
Implications for Policy, Delivery, or Practice: Market reforms have swept the country in the past twenty years. The degree to which AMI mortality worsened needs to be considered in the context of reported cost savings from hospital price competition.
Primary Funding Source: Doris Duke Charitable Foundation