General Posters General Posters Poster Session B Monday, June 27 • 6:15 pm – 7:30 pm ●Use of Immunization Registry Data to Evaluate Provider Compliance with Recommendations for Pneumococcal Conjugate Vaccine Conservation During the 2004 Shortage Norma Allred, MSN, Ph.D., Abigail Shefer, M.D., John Stevenson, MA, Maureen Kolasa, MPH, Richard Schieber, M.D., MPH, Kyle Enger, MPH Presented By: Norma Allred, MSN, Ph.D., Epidemiologist, National Immunization Program, Centers for Disease Control and Prevention, 1600 Clifton Road, NE, MS-E52, Atlanta, GA 30076; Tel: (404)639-8722; Fax: (404)639-8614; Email: NAllred@cdc.gov Research Objective: To evaluate the usefulness of immunization registry data to monitor provider compliance with CDC recommendations for conservation of pneumococcal conjugate vaccine (PCV) during the 2004 vaccine shortage. Study Design: We used registry data from the Michigan Child Immunization Registry (MCIR) to evaluate changes in third and fourth dose PCV coverage in Michigan. Data for individual immunization records were obtained for a cohort of children enrolled in MCIR as of February 29, 2004. Vaccination data were collected through July 2004. The proportion of children who received DTaP3 and PCV3 at seven months of age was calculated for each birth month. The proportion who received MMR1 and PCV4 at sixteen months of age was also calculated. DTaP3 and MMR1 vaccine coverage was used for comparison as these vaccines were not in short supply during this time period and they are given at the same ages as PCV3 and PCV4, respectively. Coverage rates were compared between public and private providers and by practice specialty. Population Studied: All children born between January 1, 2000 and February 29, 2004 were eligible for inclusion in the study. Over 600,000 children had at least one shot record and were included in the analysis. All public providers and 74% of private providers in Michigan supply vaccination data to the registry. Principal Findings: As of July 2004, 597,000 (94%) of 600,826 children in the registry had at least one shot record. Coverage for DTaP3 and MMR1 remained steady while PCV coverage declined at each time point when recommendations were announced. The sharpest decline was observed for PCV 3 at seven months of age when coverage dropped from 28% to 10% when recommendations to withhold the third dose were announced in March 2004. PCV coverage trends were similar for public and private providers as well as for pediatricians and family practitioners. Conclusions: Michigan immunization providers appear to be complying with CDC recommendations for PCV administration during the vaccine shortage. However, it is unknown if the third and fourth doses of PCV were not given by providers due to a shortage of vaccine or due to adherence to the guidelines. Additional information on variability of vaccine supply at the individual provider level would be needed to answer this question. Implications for Policy, Delivery, or Practice: Immunization registries can be useful in evaluating vaccination coverage trends during a vaccine shortage. These data may also be used to determine the extent to which children who had doses withheld during a shortage received those doses at a later date. Primary Funding Source: CDC ●Extending a Broken Chain? The Continuum of Care Implications of Rural Hospital Safety Problems Joseph Angelelli, Ph.D., Mary Fennell, Ph.D., Vincent Mor, Ph.D., Dequan Jiao Presented By: Joseph Angelelli, Ph.D., Assistant Professor, Health Policy & Administration, Penn State, 116 Henderson, University Park, PA 16802; Tel: (814)865-5177; Email: jangel@psu.edu Research Objective: To examine the relationship between the pressure ulcer (PU) rates of rural hospitals and the PU rates of nursing homes that admit patients discharged from those hospitals. Study Design: The four most discharged to nursing homes for each of 440 rural hospitals were identified via matched Medicare claims from the year 2000. The AHRQ pressure ulcer prevalence was computed for each hospital based on discharge claims. Two independent measures of PU prevalence for each nursing home were obtained via the CMS Nursing Home Compare site (admission PU prevalence) and the Online Survey Certification and Reporting (OSCAR) survey (PU prevalence that includes long-stay residents). Logistic regression models were used to examine the effect of being in the worst performing quartile of rural hospitals on the odds of having a referral relationship with a nursing home in the worst quartile of PU performance. Population Studied: Medicare beneficiaries over age 65 discharged from a nationally representative sample of 440 rural hospitals in the year 2000. Principal Findings: For every 1% increase in a hospital's pressure ulcer rate, the odds of that hospital having a low quality NH partner as measured by admission PU prevalence increase 9%. For every 1% increase in a hospital's pressure ulcer rate, the odds of that hospital having a low quality NH partner as measured by OSCAR PU prevalence increase 13%. Conclusions: Patient safety problems in rural hospitals have "downstream" implications for quality of care problems commonly observed in settings further along the continuum of care. Implications for Policy, Delivery, or Practice: These findings highlight the need for hospitals to devote more attention and resources to inter-organizational relationships as they attempt to "extend the chain" and improve the transition from acute to post-acute care. Primary Funding Source: NIA ●A Social Indicator Methodology for Estimating Need for Substance Abuse Treatment Services Hermik Babakhanlou-Chase, MPH, William E. McAuliffe, Ph.D., Teresa E. Anderson, Ph.D. Presented By: Hermik Babakhanlou-Chase, MPH, Epidemiologist, Department of Public Health, Executive Office of Health and Human Services, 250 Washington Street, Boston, MA 02108; Tel: (617) 624-5170; Email: Hermik.Babakhanlou-Chase@state.ma.us Research Objective: The Social Indicator Study developed indexes of the level of drug, alcohol, and substance abuse treatment need in 1997-2001 among Massachusetts’s community network areas (CHNAs). The analysis used the indexes to assess existing gaps in the utilization of treatment services relative to needs. Study Design: The study focused on indicators of serious substance-use disorders rather than on indicators of casual use of alcohol or controlled drugs. Using data obtained from state and federal agencies, the need indicators were mean rates of drug and alcohol related mortality, hospital discharges, and arrests. The dependent variable was treatment admission rates. Population Studied: The population is the entire state population from 1997-2001. Principal Findings: The psychometric analysis of the indicators found that overall they possessed good reliability and validity. Cronbach’s alpha reliability estimates were .99 for treatment admission rates, between .86 and .91 for the mortality rates, and between .95 and .98 for the hospital discharge rates. The arrest rates obtained from the Federal Bureau of Investigation (FBI) had missing data, especially small towns but also in some larger cities. The study imputed missing data, but as a result could not assess the reliability of the arrest indicators. The drug indicators had clear evidence of convergent and discriminate validity. There was good evidence of convergent validity for the alcohol mortality and hospital discharge rates, but such evidence was limited for the alcohol (driving under the influence) arrest rates. Despite the limitations of the alcohol arrest rates, the alcohol need index performed better with the arrest variable included. The Drug Need Index (DNI), the Alcohol Need Index (ANI), and the Substance Abuse Need Index (SNI) were unweighted linear composites of standardized versions of the mortality, hospital discharge, and arrest rates. The Cronbach’s alpha estimates of the reliability for the DNI was .86, for the ANI was .58, and for the SNI was .81. The study developed evidence of construct validity for the DNI and to a lesser extent for the ANI. Conclusions: The study found that there was a close match between relative need and existing services delivered in the CHNAs. The DNI correlated .89 with drug treatment admission rates, the ANI correlated .79 with alcohol treatment admission rates, and the SNI correlated .90 with substance abuse treatment admission rates. Thus, 80% of the variance in the average treatment admission rates for 1997-2001 was attributable to variations in the substance abuse treatment needs of the CHNA residents. Implications for Policy, Delivery, or Practice: Analysis of CHNA substance abuse treatment admission rates between 1997-2001 found that there was a six-fold difference between the highest and lowest rates. The range was greater for drug treatment admissions rates than for alcohol treatment admission rates. A primary goal of the study was to determine the extent to which these wide differences in rates reflected differences in treatment needs or some other factor such as differences in access to treatment. Primary Funding Source: Massachusetts Department of Public Health ●Deny, Delay, or Pay: Consequences of Barriers to Coverage for Immigrant Parents Seeking Healthcare for their Children. Tamar Bauer, JD, Linda Weiss, Ph.D., Cayce Hill Presented By: Tamar Bauer, JD, Director, New York Forum for Child Health, The New York Academy of Medicine, 1216 Fifth Avenue, New York, NY 10029; Tel: (212) 822-7200; Fax: (212) 822-7369; Email: tbauer@nyam.org Research Objective: Medical debt and health-related financial problems affect many Americans. For immigrant families, these issues may be compounded by immigration status, language proficiency and lack of familiarity with our health care system. This study examines the causes and consequences of health-related financial problems among immigrant parents seeking healthcare for their children. Study Design: We conducted 75 semi-structured interviews of immigrant parents living in New York City, from August 2003 December 2003. Respondents were recruited using purposive sampling through collaboration with community-based organizations. Respondents had to be an immigrant parent caring for a child (under age 19). Population Studied: The population studied was Haitian, Latino, and Russian immigrant parents and children living in New York City. We limited our sample to children who were uninsured at the time of the parent's interview, or publicly insured with a period of uninsurance in the past two years. Principal Findings: Difficulties enrolling, recertifying and navigating public insurance, and inadequate coverage, create financial difficulties for immigrant families. Payment issues affected access in 67% of the families. Providers turned children away when their parents could not provide a working insurance card or cash in advance. Some participants delayed bringing children to doctors or purchasing medicine during gaps in insurance. Well-child care was often skipped because of uninsurance. Parents cared for sick children as well as they could, using over-the-counter medications and home remedies. Pharmacy costs were managed by splitting prescriptions or using leftover medications. Rather than delay care, 49% of the parents paid cash for care at the point of service, despite financial hardship. In some cases, parents thought their children were insured, only to learn they were under- or uninsured. When billed by providers, parents felt compelled to pay, unsure whether they would be reimbursed. These out-of-pocket funds were often meant for other basic necessities like rent and food, and ranged from $50 for a doctor’s visit to $10,000 in bills for a surgery not covered by insurance. 39% of participants currently, or recently, had unpaid medical bills. Haitians in particular reported struggling with debt. Immigrants with strong personal networks turned to them for financial help; others paid by credit card, or were struggling to pay off large bills in small monthly payments. Some providers helped uninsured but eligible patients seek reimbursement or accepted payment over time. Conclusions: Barriers to coverage, and inadequate coverage, often lead to denied or delayed health care for the children of immigrants, or to out-of-pocket payment for their children’s health care, threatening both the health of the children and their parents’ financial stability. Parents frequently pay for care with cash they can ill afford, or incur burdensome debt to providers, on credit cards or by borrowing from family. This debt negatively influences immigrants’ health-seeking behavior and ability to pay for basic necessities, further reducing their ability to care for their children's health. Implications for Policy, Delivery, or Practice: Hardships reported by participants underscore the difficulty of navigating public insurance systems, and the essential role of charity care funding streams in New York City hospitals. Such funds should be more systemically available to patients in need. Primary Funding Source: Foundation for Child Development ●Gender and Race Disparities in In-Hospital Coronary Artery Bypass Surgery Mortality Rates: 1998-2002 - Have Things Improved? Edmund Becker, Ph.D. Presented By: Edmund Becker, Ph.D., Professor, Health Policy and Management, Emory School of Public Health, 1518 Clifton Road, NE, Atlanta, GA 30322; Tel: (404) 727-9969; Fax: (404) 727-9198; Email: ebeck01@sph.emory.edu Research Objective: The growing scrutiny of disparities in coronary artery bypass surgery in (CABS) in-hospital mortality rates for the nation’s over 350,000 hospital admissions raises questions about whether the trends persist over time after controlling for relevant patient, clinical, hospital, and area characteristics. We control for 41 covariates and evaluate CABS patient in-hospital mortality rate trends by gender and race for the 1998-2002 period. Study Design: With in-hospital mortality as the dependent variable, the logistic regression model controls for: patient characteristics (gender, six age categories, patient=s race Caucasian, African American, and Asian, and insurance coverage - Medicare, Medicaid, private, self-pay, no charge, and other); risk factors and co-morbid conditions (smoker, history of tobacco use, COPD, S/P CABG, S/P PTCA, insulindependent diabetes, non-insulin dependent diabetes, conduction disorders, intracranial hemorrhage, chronic renal failure, unspecified renal failure, cardiogenic shock, hypertension, acute MI, old MI, cardiomyopathy, congestive heart failure, peripheral vascular disease, unstable angina, acute liver necrosis, endocarditis, mitral valve disease, aortic valve disease, mitral and aortic valve disease, and chronic hepatitis); procedure characteristics (number of vessels bypassed, diagnostic cath, intra-aortic balloon pump, hypothermia use, and heart pump); medications (GP2B3A and thrombolytics); year dummies; and site characteristics (dummy variables for each of the sites performing CABS surgery with one site excluded). Procedures, complications, and co-morbid conditions were identified using ICD-9 codes. Population Studied: A total of 1,194,159 patients whose primary procedure was a CABS graft during the years 1998 through 2002 in the Healthcare Cost and Utilization Project (HCUP) national databases were analyzed. Principal Findings: For the 5-year period, for all CABS patients, unadjusted in-hospital mortality rates averaged 2.7% but had fallen 17.1% - from 2.98% in 1998 to 2.47% in 2002. Female in-hospital CABS mortality rates average more than 1.5% higher than male mortality rates (3.79 vs. 2.26; p<0.001). With the exception of African-Americans, who experienced no change in in-hospital mortality, over the 5-year period, white in-hospital mortality declined 17.5%, Hispanics declined 30.4%, and Asian’s declined 11.9%. Multivariate results reveal that over each successive year of the study, in-hospital mortality rates have declined significantly for all races although they remain higher for females and African American patients. Conclusions: The analysis provides evidence that U.S. gender and race disparities in CABS in-hospital patient in-hospital mortality appear to be declining and some of the underlying gender and race disparities are ameliorated when other patient comorbidities and risk factors are included. For CABS patients, our findings suggest that gender differences both within and across the races may be greater than the differences among racial groups. Implications for Policy, Delivery, or Practice: It is essential that analyses of disparities account for the influence of other patient characteristics and be sensitive to their interactions. Primary Funding Source: No Funding Source ●What's Up Doc? Mixing Web and Mail Methods in a Survey of Physicians Timothy Beebe, Ph.D., G. Richard Locke, M.D., Sunni Barnes, Ph.D. Presented By: Timothy Beebe, Ph.D., Associate Professor of Health Services Research, Department of Health Sciences Research, Mayo Clinic, 200 First Street SW, Rochester, MN 55905; Tel: (507) 538-4606; Fax: (507) 284-1180; Email: beebe.timothy@mayo.edu Research Objective: To increase response rates, household surveys often turn to mixed-mode designs whereby instruments are designed to be administered via mail, web, telephone, and/or in-person and respondents are allowed to respond to the form most appropriate for them. The application of mixed-mode designs to physician surveys seems natural given that low response rates to single-mode physician surveys are common. Recent evidence suggests that one particular mode combination, mail and web, might prove useful in extending the coverage of the survey to a broader mix of physicians because the profile of providers responding online was somewhat different from those responding to a mail survey. Study Design: The paper will present the results of an experiment testing two different mixed-mode designs representing two combinations of mail and web surveys. The survey focused on physician evaluations of the Mayo Electronic Medical Record (EMR). Physicians were randomly assigned to receive either an initial mail survey with a web survey follow-up to nonrespondents or its converse – an initial web survey followed by a mailed survey to nonrespondents. In both conditions, all physicians received a reminder one week after the initial mailing via the medium corresponding to the initial contact (web v. mail). Population Studied: 500 Mayo Clinic Department of Medicine physicians representing several specialties such as allergic diseases, cardiovascular diseases, endocrinology, gastroenterology, hematology, infectious diseases, nephrology, pulmonary and critical care medicine, and rheumatology. Principal Findings: The data collection is on-going at the time of the writing of this abstract. Key research questions to be addressed in the analysis include: Can mixing mail and web methods increase response rates among physicians? Which of the two combinations tested proved most effective? How does mixing modes impact the participation of different types of physicians? What effect do the different combinations have on the resulting evaluations of the EMR? Conclusions: Studies that use only a single data collection methodology may face the increased risk of skewing the composition of the responding sample and potentially biasing the survey findings on certain topics. For physician surveys where a web-based method is an option, it may be best to couple it with a mailed version to extend sample coverage. Implications for Policy, Delivery, or Practice: Investigating the attitudes, beliefs, behaviors, and concerns of physicians via survey is vitally important given their role in shaping the rapidly changing health care system. In order to ensure that the perspectives of physicians in one’s responding physician sample represent those of all physicians in the frame, it is incumbent on health policy survey researchers to employ the best methods for achieving that goal. If the results of the present investigation prove positive, the use of mixed-mode web and mail surveys may be among those “best methods.” Primary Funding Source: Mayo Department of Medicine ●Quality Measurement & Reporting System Used by Academic Medical Centers Raj Behal, M.D., MPH, Richard Bankowitz, M.D., MBA, Jodi Neikirk, MA Presented By: Raj Behal, M.D., MPH, Medical Director, Clinical Informatics, University HealthSystem Consortium, 2001 Spring Road Suite 700, Oak Brook, IL 60523; Tel: (630)954-4892; Email: rbehal@uhc.edu Research Objective: Develop a quality measurement and reporting system (QMRS) for national quality and patient safety indicators for use by academic medical centers in the U.S. Study Design: Quality and safety indicators developed by the AHRQ, the Leapfrog Group, the JCAHO and the CMS were organized in a scorecard. Individual performance of 120 U.S. academic medical centers and their affiliates belonging to the University HealthSystem Consortium databases was shown with percantile rankings for each indicator. The QMRS was distributed to senior leadership at each of the medical centers. Population Studied: U.S. academic medical centers. Principal Findings: The QMRS is positively received by the medical center leaders and quality staff as an effective screening tool for identifying potential opportunities for improving quality and safety. Conclusions: Use of comparative data and relative ranking among peer academic medical centers in a quality measurement and reporting system is an effective modality for stimulating improvement initiatives. Implications for Policy, Delivery, or Practice: Hospital performance on quality and safety indicators does not have to be publicly reported to stimulate quality improvement. Distribution of report cards among peer hospitals showing relative rank may be similarly effective. Primary Funding Source: No Funding Source ●Are We Overstating the General Health Disparity Between Latinos and Whites? Jeanne Black, MBA, Ron D. Hays, Ph.D., Ninez Ponce, Ph.D. Presented By: Jeanne Black, MBA, Doctoral Candidate, Health services, UCLA, 4396 Keystone Avenue, Culver City, CA 90232; Tel: (310)836-8876; Email: jtblack@ucla.edu Research Objective: Self-reported health status is widely used as an indicator of population health, based on the question, “In general, would you say your health is excellent, very good, good, fair, or poor?” Using this measure, Latinos report significantly worse health than non-Latino Whites. The objective of this research is to compare the performance of the single-item measure with the Physical Component Summary (PCS) and Mental Component Summary (MCS) scores derived from the SF-12 to evaluate whether the general health question has cross-cultural validity as a measure of health status. Study Design: This cross-sectional study used data from the random-digit-dial California Health Interview Survey 2001. We used weighted logistic regression to examine the impact of demographic characteristics, socioeconomic status (SES), acculturation-related factors, and health conditions on the odds of reporting fair or poor health, and weighted OLS regression to evaluate the effect of these same variables on PCS and MCS scores. Acculturation-related factors were English proficiency (LEP), language(s) spoken at home, years in the US, citizenship status, and immigrant generation. Scoring weights for the PCS and MCS allowed the two summary scores to be correlated. The internal consistency reliability of the PCS and MCS was examined using Cronbach’s alpha, with results for Latinos stratified by English proficiency. Population Studied: 10,942 Latino and 25,679 non-Latino White non-elderly adults residing in California. Principal Findings: The unadjusted odds of reporting fair/poor health were 3.3 for Latinos compared with nonLatino Whites (p<0.0001). Controlling for demographics, SES, acculturation-related factors, BMI and chronic conditions reduced these odds to 1.5. However, there was no LatinoWhite difference in either the PCS or MCS scores after controlling for age, gender, and poverty level. Cronbach’s alpha for the PCS was similar for non-Latino Whites and both Latino sub-groups. However, the PCS item-rest correlation for the general health question was 0.63 for whites and 0.47 for Latinos with LEP. Smaller group differences were observed in the MCS. Conclusions: Prior US studies of the cross-cultural validity of measures from the Medical Outcomes Study were not based on representative population samples that include the uninsured, recent immigrants, or those who do not speak English. The fact that in a representative sample, the singleitem general health measure shows significant disparities between Latinos and non-Latino Whites, while the scale measures of physical and mental health show none, calls into question whether the general health question should be used alone to measure health status differences in a culturally and economically diverse population. The difference between Whites and Latinos in the correlation between the general health question and the other SF-12 items suggests that the general question may not represent the same construct in the two groups. Implications for Policy, Delivery, or Practice: Further psychometric analysis of the SF-12 should be conducted in diverse population samples. Analysis of health disparities should be based on specific clinical outcomes or use multiple measures of self-reported health. Using the general health question to adjust for health status in studies of utilization or satisfaction with care should be done with caution in culturally diverse populations. Primary Funding Source: No Funding Source ●Hospital Characteristics Associated with Pediatric Splenic Injury Management Steve Bowman, MHA, PhC, Fred Zimmerman, Ph.D., Dimitri Christakis, M.D., Sam Sharar, M.D., Diane Martin, Ph.D. Presented By: Steve Bowman, MHA, PhC, Epidemiologist, Washington State Department of Health, Office of EMS and Trauma System, PO Box 47853, Olympia, WA 98504-7853; Tel: (360) 236-2873; Fax: (360) 236-2829; Email: steve.bowman@doh.wa.gov Research Objective: We examined hospital characteristics associated with pediatric splenic injury management and tested two hypotheses: 1) children are more likely to receive non-operative, conservative management in children's hospitals than in general hospitals, and 2) children are less likely to receive non-operative management in for-profit hospitals than in not-for-profit hospitals. Study Design: Retrospective cohort study of data from the Kid's Inpatient Database (KID) for year 2000. Multivariate logistic regression was used to control for patient and hospital characteristics. The main outcome measure was splenectomy performed within 2 days of hospital arrival. Population Studied: All children aged 0-16 years who were hospitalized with a traumatic (non-iatrogenic) spleen injury in non-federal short-stay hospitals in any of the 27 states participating in KID (N=2851). Principal Findings: A total of 370 children (13%) received splenectomies within 2 days. In logistic regression analyses adjusting for patient characteristics, injury severity and hospital characteristics including trauma designation, splenectomy was found to be less likely among children treated at children's hospitals (OR, 0.24; 95% CI, 0.11-0.55) than among children treated in general hospitals. Splenectomies were found to be more likely among children treated at for-profit hospitals (OR, 2.44; 95% CI, 1.32-4.50) than in not-for-profit hospitals. Conclusions: Variation in the management of pediatric splenic injuries remains. Children's hospitals have adopted spleen-conserving management practices. For-profit hospitals perform more unnecessary splenectomies than not-for-profit hospitals. Additional pediatric education and training for providers in general hospitals and especially in for-profit hospitals may be needed to increase the use of spleenconserving management practices. Implications for Policy, Delivery, or Practice: By identifying factors associated with pediatric spleen management variation, state and/or local governments can strengthen organized trauma systems. By demonstrating the level of variability in current medical practice, decision-makers will receive information on the potential for improved efficiency through the adoption of a decision rule – either as a requirement (administrative rule) or as a guideline. Primary Funding Source: No Funding Source ●Changes in Prescribing Patterns, Medication Management, and Hospital and Emergency Room Utilization: The Impact of an Asthma Population Disease Management Model and Individual Asthma Care Management Joan Branin, Ph.D. Presented By: Joan Branin, Ph.D., Associate Professor, Health Services Managment, University of La Verne, 1950 3rd Street, La Verne, CA 91750; Tel: (909) 593-3511 x4247; Fax: (909) 3922702; Email: braninj@ulv.edu Research Objective: The purpose of this research was to investigate the impact of an asthma population disease management model coordinated with an individual asthma care management program for adult asthma patients on prescribing patterns, medication management, and hospital and emergency room utilization. Study Design: An asthma population disease management model coordinated with an individual asthma care management program to monitor and facilitate treatment of adult asthma patients was initiated at a health maintenance organization in Southern California. Patients were enrolled in an asthma care management program that emphasized the use of preventative asthma control measures. These measures included increased use of anti-inflammatory medications and non-pharmaceutical methods such as dust control and peak flow meter usage and better recognition of asthma triggers. One-on-one asthma self care management training, patient education materials, and an individualized follow up program involving an interdisciplinary asthma team approach were instituted. A retrospective review of medical records and data sources was conducted before and after the implementation of the individual asthma care management program for adult asthma patients. Population Studied: A total of 2,423 asthmatic patients between the ages of 18 and 65 years of age identified as having moderate and severe asthma and a history of hospitalizations and emergency room visits for asthma were continuously enrolled in the asthma population disease management program during 2003 and 2004. Principal Findings: There was a statistically significant reduction in uncontrolled acute asthma episodes, hospitalizations, and emergency room visits after the implementation of the asthma population disease management model coordinated with the individual asthma care management program. There was a significant reduction in the number of oral corticosteriod courses, the number of prescribers, and the ratio of inhaled anti-inflammatory medications to beta-agonist usage (p <.05). The number of asthma hospitalizations decreased by 18%. Similarly, the number of emergency room visits decreased by 21%. There was a positive correlation between the number and type of prescribers and the rate of emergency room visits (r = .63) and hospitalizations (r = 54). Within one year of implementation of the asthma care management program, the number of asthma patients identified with moderate and severe high-risk characteristics was reduced by 27%. These findings support the findings of an earlier study. Conclusions: An asthma population disease management model and individual asthma care management can reduce the number of prescribers and result in better medication management and fewer emergency room visits and hospitalizations for asthma among adult asthma patients. Implications for Policy, Delivery, or Practice: These findings suggest that an asthma population disease management model coordinated with individual asthma care management can lead to (1) better self-care medication and symptom management, (2) improved quality of care, (3) changes in prescribing patterns, and (4) reduced hospital and emergency room utilization. Primary Funding Source: ULV Faculty Research Grant ●IRT Evaluation of IHI Safety Climate Survey Sandra Bruggeman, Rosa Cabanela, Ph.D., Arthur R. Williams, Ph.D., Erin K. Mc Murtry, James M. Naessens, MPH Presented By: Sandra Bruggeman, Master's Health Services Analyst, Health Care Policy and Research, Mayo Clinic Rochester, 200 1st Street SW, Rochester, MN 55905; Tel: (507) 284-4582; Email: Bruggeman.Sandra@mayo.edu Research Objective: To evaluate reliability and validity of the IHI 19 item Safety Climate Survey using groups of healthcare providers at a large mid-western hospital as respondents. Study Design: Questionnaires containing the 19 IHI Safety Climate Survey items were distributed via intranet to a random sample of 1476 hospital nursing staff including registered nurses, licensed practical nurses, and patient care assistants; 1106 consulting physicians with recent hospital experience; and 641 resident physicians who had participated at least one year in the Residency Program at the hospital. The response rate was 45.4%. Respondents’ answers were analyzed using principal component factor analyses and Item Response Theory (IRT). A scree test of the eigenvalues from the factor analysis was used to examine dimensionality. MULTILOG 7.03 software (Samejima’s model) was used to calculate item parameters and produce Item Characteristic Curves (ICC) and Total Information Curves. Population Studied: 1451 clinical survey respondents consisting of 733 nurses, 485 physicians and 232 residents at a large mid-western hospital. Principal Findings: After principal component factor analysis (varimax rotation) was done, a scree test identified 2 factors. The factors loaded on 13 of IHI questionnaire items, with one loading on both factors. The first factor loaded on 8 items, and the second loaded on 6, including the cross-loading. The maximum variance accounted for on both factors combined was 38.6%. MULTILOG parameter calculations found that each of these factors had 2 items with alphas less than 1.00, indicating a very poor discrimination of traits. For items on each of the factors, MULTILOG produced four beta values, almost all were negative. As a consequence, no Total Information was provided on the upper part of the trait distributions (> 1 SD) on these factors. Conclusions: It is not clear what exactly is being measured on either of these factors. The Total Information provided on either factor is poor, with information being concentrated at the lower ends of the trait distributions. Many of the 19 items on the IHI questionnaire also appear to be redundant, since 11 alphas are less than 1.00. This suggests that many items could be dropped from the questionnaire. Implications for Policy, Delivery, or Practice: Although measuring “Safety Climate” has become popular, the generalized use of non-validated instruments and unjustified confidence in the results of their use may lead to undesirable consequences including less safety for patients not more. Primary Funding Source: No Funding Source ●Assessing the Validity of the Current Asthma HEDIS Measure to Distinguish Quality Asthma Care Christopher Cantrell, Ph.D., Paul G. Solari, M.D., Deborah L. Fritz, Ph.D. Presented By: Christopher Cantrell, Ph.D., Senior Manager, Applied Outcomes and Analysis, GlaxoSmithKline, 5 Moore Drive, Research Triangle Park, NC 27709-3398; Tel: (919)4838686; Fax: (919)483-0611; Email: christopher.r.cantrell@gsk.com Research Objective: The current asthma Health Employer Data and Information Set (HEDIS) measure is a ratio of the number of persons receiving at least one prescription for a HEDIS-defined asthma controller during the measurement year (numerator) among identified “persistent” asthmatics in the previous year (denominator). Higher HEDIS scores are intended to reflect better asthma care within an organization, however there is some concern around the validity of this measure to distinguish quality care. First, the numerator criterion (one controller prescription) is potentially too conservative. A single controller prescription is insufficient for maintaining control in a “persistent” asthma patient. Second, while HEDIS technical specifications for the measure state that “the best available evidence indicates that inhaled corticosteroids are the preferred primary therapy”, the measure itself does not distinguish between types of controllers. Finally, the medication usage criteria used to identify “persistent” asthma patients may include persons who are not asthmatic and/or who are not chronic in nature (e.g. exercise-induced, rhinitis, etc.). Therefore, these patients may not benefit from future controllers and have very little risk of asthma exacerbation thereby diluting the measurement sample. The purpose of this research was to determine if being numerator compliant is associated with better measurement year outcomes; lower Emergency Department and hospitalization (ED/hospitalization) rates. Study Design: A retrospective longitudinal design replicating the current asthma HEDIS measure using two years, 20022003, of national administrative claims data. Rates per 100 and odds ratio were calculated and chi-square was utilized to test for significance (p<0.05). Population Studied: Integrated Healthcare Information Services’ Managed Care Benchmarks database (Waltham, MA) was utilized to replicate the asthma HEDIS measure and identify 30,687 “persistent” asthma patients. Principal Findings: Of the 30,687 patients identified, 21,124 were numerator compliant yielding a HEDIS score of 68.84% (NCQA reported national average for 2002 was 67.9%). ED/hospitalization rates were 6.9 and 3.7 per 100 for numerator compliant patients and non-numerator compliant patients respectively. The odds ratio for numerator compliant patients = 1.95 (CI = 1.73 - 2.20) with a chi-square value of 112.2 (p<0.0001). This indicates that numerator compliant patients are 95% more likely to have and ED/hospitalization versus their counterparts. Conclusions: The findings of this study suggest that the current asthma HEDIS measure does not validly assess quality asthma care in terms of ED/hospitalizations. In fact, our results indicate that the measure reflects the opposite of what is intended. Numerator compliant patients are significantly associated with higher ED/hospitalization rates and therefore a higher asthma HEDIS score in this case would not reflect better care. Implications for Policy, Delivery, or Practice: The National Committee for Quality Assurance (NCQA), the organization responsible for HEDIS scores, reports that about 90 percent of health plans collect HEDIS data and 60 percent of large employers and public purchasers use HEDIS data to help guide contracting decisions, set performance targets or employee contributions based on plan performance. Our data suggests that the asthma measure may not be useful for the purposes in which it is being employed and that these concerns should be addressed. Primary Funding Source: GlaxoSmithKline ●Racial/Ethnic Health Care Disparities in a Managed Care Medicare Population Joyce Co, MPH, Beatriz Jaramillo, DrPH Presented By: Joyce Co, MPH, Assistant Director, Health Services Analysis, Medical and Quality Informatics, HIP Health Plan of New York, 55 Water Street, 5th Floor, New York, NY 10041; Tel: (646) 447-6372; Fax: (646) 447-3164; Email: jco@hipusa.com Research Objective: Disparities in health care are among the nation’s most serious health care problems. A recent report by the Institute of Medicine found that racial and ethnic minority groups tend to receive a lower quality of health care than nonminorities, even when access-related factors such as income and insurance coverage are controlled. At HIP, the availability of race/ethnicity data for Medicare members makes it possible to determine whether racial and ethnic disparities exist within managed care, and if so where. Study Design: This analysis focuses on selected HEDIS Effectiveness of Care measures for Medicare members. These measures are: breast cancer screening; HbA1c testing and LDL-C screening for members with diabetes; antidepressant medication management: optimal practitioner contact, acute phase treatment, and continuation phase treatment; mental health follow-up after hospitalization: 7 days and 30 days; and osteoporosis. Administrative data was collected for these measures. Using this administrative data, rates for the selected HEDIS 2004 measures were calculated by racial/ethnic group. The race/ethnicity groups existing in HIP’s system are African-American, American Indian/Alaskan Native, Asian, Hispanic, Other, Unknown, and White. Due to the small number of members in American Indian/Alaskan Native, Asian, Other, and Unknown categories for measures such as breast cancer screening and the two diabetes measures, these groups were collapsed into one group, “Other”. In addition, due to the small number of members within the Hispanic ethnic category for the mental health measures and the osteoporosis measure, Hispanic members were also included in the “Other” category for these measures. Rates by racial/ethnic groups were compared for each measure using a chi-square statistic. Statistical significance was established at alpha < 0.05. Population Studied: Medicare members who fit criteria for the following HEDIS measures: breast cancer screening; HbA1c testing and LDL-C screening for members with diabetes; antidepressant medication management: optimal practitioner contact, acute phase treatment, and continuation phase treatment; mental health follow-up after hospitalization: 7 days and 30 days; and osteoporosis. Principal Findings: The following measures showed no significant difference across the races measured: breast cancer screening and osteoporosis. Two sets of measures showed significance: mental health measures and both diabetes measures. For all mental health measures, the AfricanAmerican and Other groups had lower rates than the White group. For the diabetes HbA1c testing measure, the two groups with the lowest rates were the Hispanic and Other groups, and for the diabetes LDL-C screening measure, the two groups with the lowest rates were the African-American and Other groups. Conclusions: This analysis suggests that some racial and ethnic disparities in health care exist at HIP, specifically in the areas of mental health treatment and diabetes management. Therefore, factors associated with certain races and ethnicities, including specific cultural beliefs and attitudes about diet and nutrition, perceptions of mental health, treatment, and the health care system in general warrant increased attention. Implications for Policy, Delivery, or Practice: By gaining a clearer understanding of the race- and ethnic-specific factors that influence treatment and management, specific interventions may be designed to break down the barriers to obtaining quality health care and eliminate the disparities that are currently observed. Primary Funding Source: HIP Health Plan of New York ●Measuring Patient Safety Climate: A Review of Surveys Judith Colla, Sc.D., Ann Bracken, M.D., Ph.D., Linda Kinney, MPA, William Weeks, M.D., MBA Presented By: Judith Colla, Sc.D., Research Associate, Community & Family Medicine, Dartmouth Medical School, Center for Evaluative Clinical Studies, Hanover, NH 03755; Tel: (802)295-9363 x5806; Fax: (802)291-6286; Email: judith.colla@dartmouth.edu Research Objective: Five years ago, the Institute of Medicine recommended improving patient safety by addressing organizational cultural issues. Since then, surveys measuring a patient safety climate considered predictive of outcomes have begun to emerge. The objective of this work is to compare the following aspects of patient safety climate surveys: history, dimensions covered, strength of psychometrics, and use in patient outcomes studies. Study Design: We conducted a systematic review that included Medline search, review of abstracts, published manuscripts, references from papers, relevant websites, and contact with authors. Population Studied: Not applicable. Principal Findings: We found nine surveys that measure an organization’s patient safety climate. Most cover five common dimensions of patient safety climate: leadership, policies and procedures, staffing, communication, and reporting. All surveys used Likert scales. The thoroughness of psychometric testing varied considerably. While all of these surveys have been used to compare units within or between hospitals, only one has been used to explore the association between organizational climate and patient outcomes, such as mortality, reductions in medication errors and length of stay. Conclusions: Although several good surveys of patient safety climate exist, we found considerable variation in the length, dimensions covered, and reported reliability of these surveys. While achievement of a patient safety culture is an admirable goal, more effort should be expended on understanding the relationship between measures of patient safety climate and patient outcomes. Implications for Policy, Delivery, or Practice: Patient safety climate surveys may detect employee concerns and may help foster communications around this topic. However, since there is only limited evidence that patient safety climate scores are associated with patient safety outcomes, administrators and policymakers should be cautious in their use of these surveys. Primary Funding Source: No Funding Source ●Establishing the True Effect of Dental Insurance on Dental Use. Philip Cooper, Ph.D., Richard J. Manski, DDS, MBA, Ph.D. Presented By: Philip Cooper, Ph.D., Health Economist, Center for Financing, Access and Cost Trends (CFACT), Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850; Tel: (301)427-1675; Email: pcooper@ahrq.gov Research Objective: Identify the true insurance component of dental care coverage or pre-paid dental care. Study Design: We attempt to determine the true effect of dental coverage by isolating the health seeking behavioral component of dental care. We use health insurance and self identified measures of attitudes toward risk as proxies for health seeking behavior. Population Studied: Representative sample of U.S. population aged 19 in 2001 from the Medical Expenditure Panel Survey - Household Component and Insurance Component. Principal Findings: While roughly 88 percent of the U.S. population had some type of Medical coverage in 2001, 46 percent had dental coverage. Virtually all persons with dental coverage also had medical coverage. While nineteen percent of those without any coverage had a dental visit, thirty-nine percent with only medical coverage (no dental) had a dental visit and fifty percent of those with dental coverage had a similar visit. Conclusions: Since medical coverage generally does not provide for dental reimbursement, the positive effect of medical insurance on the dental use suggests a indirect health seeking behavior component of having medical insurance on the use of dental care. Implications for Policy, Delivery, or Practice: By isolating and understanding the health seeking behavioral component of dental care coverage, employers and other plan sponsors may be able to more accurately design future dental care plans that better meet the needs of employees and their dependents. Primary Funding Source: AHRQ ●A Comparison of Four States' Strategies for Reducing the Use of Physical Restraints in Nursing Facilities Leslie Cortes, M.D., Jennie Yu Chou, MS, RPh Presented By: Leslie Cortes, M.D., Director, Medical Quality Assurance, Quality Assurance / Quality Improvement, Department of Aging and Disability Services, 701 West 51st Street, W-510, Austin, TX 78751; Tel: (512)438-2567; Fax: (512)438-4415; Email: leslie.cortes@dads.state.tx.us Research Objective: This project compares the impacts of four states’ quality improvement strategies for reducing restraint use in nursing facilities. The strategies include usual regulation, technical assistance provided by Quality Improvement Organizations or QIOs to volunteer facilities, and the technical assistance provided by QIOs to volunteer facilities combined with a state-based technical assistance program in which all facilities are required to participate. The effect of restraint reduction on the prevalence of falls in each state is examined, and the relative impacts of the voluntary and compulsory technical assistance programs in Texas are compared. Study Design: Two-year, statewide quality indicator trends for the four states with the highest prevalence of restraint use during the first quarter of 2002 are examined. The impact of restraint reduction on fall prevalence is examined through time-series analysis. The relative impacts of Texas’ voluntary and compulsory technical assistance strategies are determined through analysis of attributable fractions. Population Studied: The study population consists of nursing home residents in Medicaid-certified nursing facilities in Arkansas, California, Louisiana, and Texas. Principal Findings: The state of California, with no technical assistance program for restraint reduction, achieved a 19.7 percent reduction in restraint use. Arkansas and Louisiana, states providing QIO-based technical assistance for restraint reduction to 10 and 15 percent of their facilities, respectively, achieved 23.5 and 16.8 percent reductions in restraint use. Texas, the state providing voluntary QIO-based technical assistance to 13 percent of facilities and compulsory technical assistance to all facilities achieved a 47.5 percent reduction in restraint use. In Texas, 89.7 percent of the total reduction in restraint use was attributable to the state’s compulsory technical assistance program; the remainder was attributable to the voluntary QIO-based intervention. Cross-correlation analysis shows that the variance in fall prevalence attributable to restraint reduction varies among the states. In California and Texas, there was a moderate to strong tendency for fall prevalence to decrease as restraint use was reduced. In contrast, Arkansas and Louisiana showed a moderate tendency for fall prevalence to increase as restraint use was reduced. Since the first quarter of 2002, only Louisiana has experienced an absolute increase of 1.7 percent in fall prevalence. Conclusions: Both voluntary and compulsory quality improvement initiatives showed a positive impact on reducing restraint use. The combined use of both strategies was associated with the largest decrease in restraint use, with the compulsory quality improvement strategy showing a greater impact than the voluntary strategy. Major reductions in nursing home restraint use can be achieved without increasing the prevalence of falls. Implications for Policy, Delivery, or Practice: Greater statewide quality improvements are achieved through compulsory state-based technical assistance. However, other factors can influence the degree of improvement. These factors include the effectiveness of the technical assistance interventions used, the readiness of nursing facilities to implement recommended changes, the types of technical assistance provided, the frequency of on-site contacts with the facilities, and the training of the quality consultants who provide the technical assistance. Future studies should examine the impact of these factors on the effectiveness of quality improvement programs. Primary Funding Source: Title XIX funding for Medicaid services ●Relation of Member Characteristics and Utilization Patterns to Response to Reminders Sent for Overdue Health Screenings Jun Zhu, Ph.D., James Davis, Ph.D., Deborah A. Taira, Sc.D. Presented By: James Davis, Ph.D., Research Analyst, Care Management, HMSA (BCBS of Hawaii), 818 Keeaumoku Street; PO Box 860, Honolulu, HI 96808-0860; Tel: (808)9485495; Fax: (808)948-6043; Email: james_davis@hmsa.com Research Objective: To understand the relation of member characteristics, physician specialty, and previous medical utilization to the response to screening reminders sent for breast, cervical, colorectal cancer, diabetes and cholesterol screening. Further, to examine the relation of reminders sent for multiple health screenings on the likelihood of members receiving the screenings. Study Design: A longitudinal design examining member characteristics and medical history in the year before reminders were sent on the rates of subsequent health screenings. The study utilized administrative data from a large health plan in Hawaii. Population Studied: Health plan members who were sent reminders for overdue health screenings between 1999 and 2003. The number of people receiving reminders was 44,331 for breast cancer screening, 73,875 for cervical cancer, 130,593 for colon cancers, 84,584 for diabetes, and 54,843 for cholesterol. Principal Findings: Screening rates ranged from 21% for colorectal cancer to 31% for diabetes in response to an initial reminder. Rates consistently decreased for members requiring increased numbers of further reminders. In multivariable models, hazard ratios indicated that the likelihood of receiving screening after reminders were sent declined with age for all the screenings except for cholesterol. The highest hazard ratios occurred for members having an intermediate morbidity level, suggesting that moderate-ill members were most likely to get screened. Screening rates were 17% to 26% higher in June than in November and December. The number of office visits before the reminders were sent was significantly associated with diabetes screening but not other health screenings. Members seeing an OB/GYN in the past year had increased rates of cancer screenings, whereas members seeing a primary care physician or an endocrinologist had significantly increased rates of diabetic and cholesterol screenings. Members sent multiple reminders for overdue screenings were less likely to obtain health screenings. However, of these members, those who received one cancer screening were more likely to get another cancer screening; and those obtaining a non-cancer related screening were more likely to get another non-cancer related screening. Conclusions: Member characteristics and medical history in the year before reminders were sent were associated with receiving health screenings. For members receiving multiple reminders, those who responded to one reminder were more likely to respond to others. This finding was especially strong among cancer related screenings, and between diabetes and cholesterol screening. Implications for Policy, Delivery, or Practice: Barriers to obtaining health screenings may be identified prior to sending health screening reminders. Directed efforts may increase health screening, such as efforts to encourage greater cholesterol screening at earlier ages, or possibly mailing additional reminders in summer when screening rates are greatest. Primary Funding Source: HMSA (BCBS of Hawaii) ●Member and Physician Characteristics Associated with Variations in Receiving Care Recommended by a Physician Quality Service Recognition Program James Davis, Ph.D., Deborah Taira, Sc.D. Presented By: James Davis, Ph.D., Research Analyst, Care Management, HMSA (BCBS of Hawaii), 818 Keeaumoku Street; PO Box 860, Honolulu, HI 96808-0860; Tel: (808) 9485495; Fax: (808) 948-6043; Email: james_davis@hmsa.com Research Objective: To examine member and physician characteristics associated with receiving health screenings and recommended practices for the management of acute and chronic conditions Study Design: The study employed a cross-sectional design examining the associations between member and physician characteristics and 30 indicators of recommended clinical care. The study used measures of clinical outcomes based on administrative data for a one-year interval (April 2002 through March 2003). Population Studied: The eligible population included all members in the fee for service, health maintenance organization, or Medicare managed care plans of the largest secondary insurer in Hawaii. Members were included in the analysis for specific clinical indicators if they met the eligibility criteria for the indicator. Eligibility criteria varied from simple age, gender criteria to more complex clinical algorithms. The number of members eligible for the various indicators ranged from 839 for follow-up after emergency visits for asthma to 70,542 for compliance with anti-hypertensive drugs. Principal Findings: Twenty of the 30 recommended practices varied significantly (p < 0.05) by age. Older members received recommended care more often for 13 practices whereas younger members received the recommended care more often for seven. Thirteen indicators varied significantly by gender; males received the recommended practice more often for 12 of the 13 indicators. The one measure where females had a better result was osteoporosis screening for patients on corticosteroids. Members with a high morbidity index received recommended care more often for 10 practices, including diabetic retinal exams, shoulder x-ray for adhesive capsulitis and beta blocker use in coronary artery disease. Members with a low morbidity index only did better on endoscopy or barium enema after diagnosis of diverticulitis. With physician characteristics, one or more of four specialties examined (primary care, obstetrics and gynecology (ob/gyn), endocrinology) was significantly associated with receiving recommended care for 24 of the 30 practices examined. Seeing an ob/gyn or a primary care physician was more frequently associated with receiving recommended health screenings. All four specialties showed associations with several disease management indicators. For example, seeing an endocrinologist was positively associated with ace inhibitor and lipid lowering drug use in coronary artery disease, whereas seeing a cardiologist was associated with ace inhibitor use in congestive heart failure and beta blocker use in coronary artery disease. Conclusions: The results suggest that recommended practices for health screening and for the management of acute and chronic conditions vary by both member and physician characteristics. The characteristics associated, however, varied substantially by the recommended practice. The results should be interpreted cautiously as they are based on administrative data using a single statistical model chosen to facilitate comparisons across the clinical indicators. Implications for Policy, Delivery, or Practice: The results suggest efforts to improve recommended clinical practice might focus on either member or physician characteristics. Targeted interventions may selectively improve adherence to recommended clinical guidelines. Primary Funding Source: HMSA (BCBS of Hawaii) ●Response to Health Screening Reminders Related to Office Visits with New and Previously Seen Physicians James Davis, Ph.D., Jun Zhu, Ph.D., James Davis, Ph.D., Deborah A. Taira, Sc.D. Presented By: James Davis, Ph.D., Research analyst, Care Management, HMSA (BCBS of Hawaii), 818 Keeaumoku Street; PO Box 860, Honolulu, HI 96808-0860; Tel: (808)9485495; Fax: (808)948-6043; Email: james_davis@hmsa.com Research Objective: To examine the association between the characteristics of physicians seen in the months following the mailing of health-screening reminders and the likelihood of receiving screening for breast, cervical, and colorectal cancers, and for diabetes and cholesterol. Study Design: The study was a longitudinal study focusing on the year after reminders were sent for health screenings. Results were analyzed using a pooled logistic regression procedure examining office visits and other health care by months after the reminders were sent. We specifically investigated whether members seeing new physicians, defined as physicians they had not visited in the past year, were especially likely to obtain health screenings. Population Studied: Members of the largest health insurer in Hawaii who received reminders for overdue health screenings between 1999 and 2003. The population was studied by analyzing their administrative data. The number of people receiving reminders was 44,331 for breast cancer screening, 73,875 for cervical cancer, 130,593 for colon cancers, 84,584 for diabetes, and 54,843 for cholesterol. Principal Findings: Health screenings were significantly increased in the months after reminders were sent, peaking for all five screenings during the second month after the reminders were mailed. Other results were also consistent across the various health screenings. In adjusted, multivariable models, members who had visited new physicians had odds ratios for obtaining health screenings 7-9 times higher than other members. This enhancement in response from seeing a new physician remained among members receiving their second and third reminders, the most recalcitrant members. Odds ratios for these members, however, were about half as great. In general, regardless of the number of reminders, the more new physicians visited the more likely members were to receive health screenings. Other physician characteristics were also associated with obtaining the recommended health screenings. Members having more overall office visits (new and old physicians) were more likely to be screened. Having an office visit with an OB/GYN improved the odds for cancer screenings, whereas having an office visit with a primary care physician enhanced the odds for diabetes and cholesterol screenings. Conclusions: Members seeing physicians they had not visited in the past year had significantly higher screening rates than members only seeing physicians they had previously visited. Other physician characteristics were also associated with obtaining recommended health screenings. Implications for Policy, Delivery, or Practice: If seeing a new physician proves to have a causal association with obtaining recommended health screenings, then better understanding of the management of new versus previously treated patients may offer insights into how to improve the rates of recommended health screenings. Primary Funding Source: HMSA (BCBS of HawaiI) ●Group Medical Visits for Patients with Type 2 Diabetes: Patient and Physician Perspectives Lynn Deitrick, RN Ph.D., Deborah Swavely, RN, MSN, Larry N. Merkle, M.D., W. Glenn Stern, MA, Sam Bub, M.D., Jack Lenhart, M.D. Presented By: Lynn Deitrick, RN Ph.D., Ethnographer, Health Studies Unit, Lehigh Valley Hospital and Health Network, 17th & Chew Streets, PO Box 7017, Allentown, PA 18105; Tel: (610)402-2293; Fax: (610)402-2247; Email: Lynn.Deitrick@lvh.com Research Objective: Recent research suggests that group medical visits for patients with chronic illness are clinically effective as well as time efficient. Group visits for chronically ill patients might also facilitate patient empowerment and activation. This study looks at the value of group medical visits conducted in a primary care practice as a treatment option for patients with type 2 diabetes, from both the patient and physician perspective. Study Design: Type 2 diabetic patients who completed 10 hours of diabetes education were invited to participate in group medical visits conducted by their primary care physician. Group medical visits in both practices were structured in a similar way. Each group visit was scheduled for 90 minutes during a weekday morning. Groups were held in a room in the physician’s office. Group visits included didactic education sessions by the physician or diabetes educator, a question and answer period, socialization, nurse and physician assessment of vital signs, lab data and foot exams. The hospital ethnographer observed several cycles of group medical visits at both primary care practices. Interviews were conducted with patients, the diabetes educator and physicians in both groups to elicit information about patient empowerment, activation and patient and physician perspectives on the value of group medical visits. The Patient Activation Measure was used to quantify activation among the patients. Population Studied: Two groups of type 2 diabetic patients n=20 at two different primary care practices engaged in monthly group medical visits for the first three months, then quarterly group visits for the next year. Group 1 patients n=8, 6 females, and 2 males and Group 2 patients n=12, 9 females and 3 males were homogenous for ethnicity and age. Principal Findings: Interviews with diabetic patients participating in group medical visits suggest that patients are very positive about the value of these visits. Group visit patients report benefits including peer learning, peer support and physician and diabetes educator encouragement as factors that enable them to better care for themselves and manage their diabetes. These patients also reported that participating in group visits allowed them to connect with their peers and provided helpful information and encouragement for them as they learned to live with their diabetes. Interviews with both participating physicians and two certified diabetes educators verify these findings. Physician perspectives of group medical visits indicate that connection with peers and provision of an extended support network, as well as development of a trusting relationship with the physician, are added benefits of group medical visits. Patient Activation Measurements to quantify patient activation in group visit participants from both practices have been collected and are being analyzed. Conclusions: Group medical visits appear to be a useful model for providing continuing education, medical management, and empowering patients to self-manage their type 2 diabetes in the primary care setting. Implications for Policy, Delivery, or Practice: Group medical visits appear to be a useful model for providing continuing education, medical management, and empowering patients to self-manage their type 2 diabetes in the primary care setting. Primary Funding Source: AHRQ ●Influence of Medicare Reimbursement on Prescribing Patterns of TNF-alpha Inhibitors Esi DeWitt, M.D., Henry A. Glick, Ph.D., Daniel A. Albert, M.D., Marshall Joffe, M.D., Ph.D., Frederick Wolfe, M.D. Presented By: Esi DeWitt, M.D., Clinical Research Fellow, Rheumatology, The Children's Hospital of Philadelphia, 3405 Civic Center Boulevard, Room 236, Philadelphia, PA 19104; Tel: (267)251-8071; Fax: (215)590-4750; Email: dewitt@email.chop.edu Research Objective: Anti-TNF alpha biologic therapies for the treatment of rheumatoid arthritis have shown comparable efficacy in randomized clinical trials. Thus, there is no well defined clinical rationale for differential prescribing of these therapies in RA. However, preferential Medicare reimbursement for infusion drugs such as infliximab, INF, in contrast to the injection drug etanercept, ETN, may influence prescription. Our objective is to characterize prescribing patterns for ETN and INF in clinical practice. We hypothesize that physicians, aware of differential reimbursement, prescribe drugs in part to maximize insurance coverage. Controlling for insurance plan may thus account for some of the observed differences between groups prescribed the two drugs. Study Design: Observational cohort study of patients with RA on anti-TNF-alpha biologic drugs. Disease and demographic variables prior to start of the two drugs were compared, without and with stratification by insurance type, private or public. Multivariate logistic regression models were constructed to characterize predictors of biologic prescription choice. Population Studied: 1808 RA patients, referred from the practices of 447 doctors, who were newly started on ETN or INF after enrollment in the National Data Bank for Rheumatic Diseases. Principal Findings: 42.4% of patients were initially prescribed ETN and 57.6% INF. Overall, ETN patients were quite different than INF patients. They had 0.15 (p<0.001) lower HAQ functional disability scores, were 7 years younger (p<0.001), and were 49% (p<0.001) less likely to have public insurance than were INF patients. However, stratification by insurance type narrowed these differences. Between ETN and INF groups the difference in HAQ scores among privately insured patients fell to 0.08, and there was only a 0.04 difference in HAQ scores among the publicly insured. Within the privately insured strata, the difference in mean age between groups was only 1.1 years. Compared to those with private insurance, regression analyses revealed publicly insured patients to be less likely to receive ETN, with an adjusted odds ratio for ETN prescription of 0.55 (CI 0.40-0.69, p<0.001). Predictors of INF prescription were, in addition to having public insurance, advanced age and prior methotrexate use. Whereas higher education and SF-36 physical function component score had a weak effect in favor of ETN prescription. Evaluation of the covariates disability (HAQ), disease duration, sex, and income did not affect the result. Conclusions: This observational cohort study demonstrates differential prescribing of ETN and INF for treatment of RA. In multivariate analyses, public insurance was a strong predictor of INF prescription, a reflection of preferential Medicare reimbursement for infusion drugs. But controlling for insurance type did not always equate patient characteristics in the two groups, suggesting other factors also influence this prescribing. Implications for Policy, Delivery, or Practice: This finding adds to the evidence that insurance reimbursement policies influence physicians’ prescribing behavior. These results raise questions about the current policy of differential governmental reimbursement for drugs with similar efficacy but different routes of administration. Primary Funding Source: Arthritis Research Center Foundation ●Emergency Colorectal Cancer Resection: A Measure of Disparities? Jessica Diggs, BS, Siran Koroukian, Ph.D., Mireya Diaz-Insua, Ph.D., Gregory Cooper, Ph.D. Presented By: Jessica Diggs, BS, M.D./Ph.D. Candidate in HSR, Epidemiology and Biostatistics, Case Western Reserve University School of Medicine, 10900 Euclid Boulevard, LOC 4945, Cleveland, OH 44106; Tel: (412)877-7438; Fax: (216)7911416; Email: jessica.diggs@case.edu Research Objective: Colorectal cancer, presenting as a surgical emergency is associated with increased morbidity and mortality, which is influenced by disease stage, comorbidity, and patient age. An acute presentation of colorectal cancer may reflect deficiencies in previous care, including issues of access, quality, screening policy, and healthcare utilization. It therefore has the potential to be used as an indicator of health disparities. The factors that determine whether a patient will present with a surgical emergency have not been described previously. The objective of this study is to characterize this population and to determine the impact of age, comorbidity, and insurance on the likelihood of presenting with a surgical emergency. Study Design: This is a cross sectional study using data from the 2002 National Inpatient Sample of the Healthcare Cost and Utilization Project. ICD9 codes were used to identify 25,789 colorectal cancer patients undergoing resection procedures. A subset, 2,548 patients, presenting with a diagnosis requiring emergency intervention, including peritonitis, occlusion, or perforation, were identified. Patients were stratified into younger, less than 65, and older, 65 and over, age groups. Bivariate relationships between age, comorbidity, and insurance status were assessed using chi squared analysis. Multivariate logistic regression was used to predict the association between patient characteristics and the likelihood of an emergent presentation. Population Studied: The study population includes all colorectal cancer patients undergoing a colorectal resection procedure. Principal Findings: The majority, 71 percent of this population was over age sixty five and 63 percent were insured by Medicare. In both the younger and older age groups, the privately insured had the lowest percentage of emergent presentations, 8.1 percent, 9.3 percent compared with Medicaid, 15.6 percent, 13.8 percent, and those who self pay 19.6 percent , 11.9 percent. Multivariate analysis showed that among younger people, those without private insurance had an increased risk of emergency presentation compared to privately insured patients, with odds ratios of 1.75 to 3.12. Among older patients, being dually eligible for Medicaid and Medicare significantly increased the risk of emergency presentation, with an odds ratio of 1.47 compared to Medicare alone. There was no consistent association of comorbidity and emergent presentation. For example, among patients less than 45 years of age, individuals with three or more comorbidities were less likely to present emergently and other conditions such as diabetes, that require more frequent interactions with healthcare professionals, were associated with a lower likelihood of emergent presentation. Conclusions: These results show a significant effect of insurance status on the risk of presenting with a surgical emergency, placing those lacking private insurance at increased risk of emergent presentation, and therefore at greater risk of morbidity and mortality. The lack of association between comorbidity and emergent presentation, suggests that contact with the healthcare system may increase the probability of early detection. Implications for Policy, Delivery, or Practice: Emergency presentation may be an indicator of healthcare disparities and may contribute to increased healthcare expenditures. Early identification of patients at risk may reduce disparities, decreasing the burden of these procedures on both the healthcare system and patients. Primary Funding Source: No Funding Source ●Early Effects of the Healthy City Program on Social and Living Environments in Jing-An District of Shanghai Xiaocang Ding, MSMHM, Jay J. Shen, Ph.D. Presented By: Xiaocang Ding, MSMHM, Deputy Director, Jiang-An District Bureau of Health, 422 Jiang-Ning Road, Shanghai, 200041; Tel: 86-21-627733; Fax: 86-21-6279375; Email: dingxc2001@21cn.com Research Objective: Although China’s economy has grown dramatically during the last two decades, it has been concerned that its social development is behind the economic development. In order to address health issues resulting from the rapid urbanization, such as degradation of the ecological and social environments, the government of Jing-An District, Shanghai, China, in 2001, adopting the World Health Organization (WHO)’s concept of “healthy city,” started implementing a planning process and the following goalattaining program to enhance infrastructures of social and health systems that supports the fast growing economic development. This research is to examine whether any progress has been made since the initiative. Study Design: Descriptive analysis on comparing a set of social and health indicator measures before and after the implementation of the plan. Measures being used covered aspects of housing or living space, green coverage, air pollution, health insurance coverage, access to care, unemployment insurance coverage, and social security. Only were some short-term measures examined and more measures will be assessed in the future. Population Studied: There were about 305,000 people residing in Jing-An District in 2004. Principal Findings: Since 2003, with 90 million yuan (unit of Chinese currency) being spent on varied projects, 190,000 square meters of old style subdivisions have been renovated, 80,000 square meters of public space and facilities have been upgraded, and kitchens and toilets of 40,000 households in 7,000 old-style buildings have been remodeled. A two-year project (2004-2005) of installing fire protection devises in old style buildings, which is spending 52.5 million yuan and will affect 50,062 households in 10,148 buildings, is on its halfway. The percentage of green areas has been increased from 12.7% in 2001, 14.1% in 2003, to 15.0% in 2004. The average green space per capita increased to 0.91 square meter in 2004. Passing rates of the automobile emission test were 86, 91, 100, and 96 in 2001, 2002, 2003, and 2004, respectively. By 11/30/2004, all coal-burning stoves were replaced and coalburning stoves no longer were used in the district. Percentages of days that the air quality meeted the national standard were 85%, 77%, and 85% in 2001, 2002, and 2004, respectively. Finally, in 2004, almost 100% of employees had health insurance coverage, social security, and unemployment insurance. Conclusions: Some short-term progress has been observed since 2001 in improving the area’s social and living environments, which, along with the economic progress, have improved local residents’ living standards, quality of life, and general wellbeing. Implications for Policy, Delivery, or Practice: Using the healthy city concept to enhance social and living environments is important in developing countries like China that has placed its major focus on economic development and growth in recent years. Although some short-term improvement has been observed in Jing-An, more measures including health status and longer effects need to be assessed to see weather the progress is broader and sustainable in the long-run. Primary Funding Source: Jing-An District Bureau of Health ●Expanding Organ Donation among African Americans in the United States through Religious Organizations: Results from an Experimental, Educational Intervention Roger Durand, Ph.D., Phillip J. Decker, Ph.D., Lamon Atkins, BS, Kimberly Davis, RN, CPTC, Samuel Holtzman, BS Presented By: Roger Durand, Ph.D., Professor of Healthcare Administration, University of Houston-Clear Lake, 2700 Bay Area Boulevard, Houston, TX 77058; Tel: (281)283-3141; Fax: (281)283-3136; Email: durand4321@aol.com Research Objective: Among African Americans in the United States, the number of organ donors continues to be low relative to the demand for transplantable human organs. Among the oft-cited reasons for this low donation rate is that many African Americans believe organ and tissue donation to be against their religion. In view of this reported belief, an educational intervention aimed at influencing African American attitudes toward, behavior regarding, and intentions about organ donation, was designed for implementation in predominantly African American religious organizations (churches, tabernacles, and mosques). In this research, the results of this educational intervention were studied. Study Design: A true experimental design was utilized. A total of 60 experimental subjects and 41 control subjects were studied over a two-year period. All subjects were African Americans enrolled in formal religious education classes at one of four predominantly African American congregations. A pretest, consisting of a survey designed to ascertain subjects’ attitudes, behavior (especially engaging in family discussion regarding donation) and behavioral intentions was initially administered to study subjects. Experimental subjects were then presented with educational materials on donation, materials that especially emphasized religion and donation. Finally, approximately two years later, a post-test of the same subjects, again consisting of a survey eliciting the same information gathered on the pre-test, was administered. A university IRB approved the study design and measuring instruments. Population Studied: The population studied consisted of African Americans who attended adult, religious education classes at one of four religious organizations in the greater Houston, Texas, area. Principal Findings: In comparison to controls, experimental subjects were found to have substantially increased (as indicated principally by lambda and Kendall’s Tau-C statistics) in their attitude that organ donation is important to the community, in their feeling that donation is a comfortable topic for discussion, in their trust in healthcare professionals regarding the declaration of brain death, and, perhaps most importantly, in their belief that organ donation is consistent with their religion. On the other hand, no differences between experimental and control subjects were found with regard to either family discussion regarding organ donation or in intention to donate upon death. Conclusions: Educational programs offered through religious organizations appear to offer an important means to improve the attitudes of African Americans in the United States regarding organ donation. This is especially so with regard to the feeling reportedly held by many African Americans that donation is against one’s religion. On the other, education alone does not seem likely to bring about appreciable shifts in actual donation behavior (i.e., engaging in family discussion about the subject) or in intention to donate. Implications for Policy, Delivery, or Practice: The findings and conclusions of this research suggest that organ procurement organizations (OPOs) can alter the beliefs of African Americans regarding donation by designing educational programs for religious organizations and by working with those organizations to implement the programs in formal classes. However, increasing the supply of transplantable organs from among African Americans depends upon behavioral changes that seem to be difficult to influence through religious classroom instruction alone. Primary Funding Source: HRSA ●Patient-Physician Colorectal Cancer Discussions in Primary Care Jennifer Elston Lafata, Ph.D., Christina Moon, MA, George Divine, Ph.D., L. Keoki Williams, M.D., MPH Presented By: Jennifer Elston Lafata, Ph.D., Director, Center for Health Services Research, Henry Ford Health System, 1 Ford Place, Suite 3A, Detroit, MI 48202; Tel: (313)874-5454; Fax: (313)874-7137; Email: jlafata1@hfhs.org Research Objective: To describe how primary care physicians and patients discuss colorectal cancer (CRC) screening and how these discussions impact CRC screening use. Study Design: We used a mailed survey and retrospective claims data to compile information on a cohort of insured, primary care patients eligible for CRC screening. The survey collected information on the content of patient-physician CRC screening discussions (including the “5 As:” Assess, Advise, Agree, Assist, and Arrange) as well as patient preferences for shared decision making. Survey responses were linked with 5year claims data on prior CRC screening use. We estimate the proportions of primary care patients receiving recommended CRC screening, discussing CRC screening with their physician and, among those discussing CRC with their physician, reporting different elements of discussion content. Population Studied: 4,966 health plan enrollees aged 50-80 years with a recent visit to a primary care physician. Principal Findings: Among the 2,513 survey respondents (50.6% response rate), 58.7% were female, 68.1% were married, 53.7% were employed, and 34.4% were African American. Just over half (54.0%) received recommended CRC screening and 79.6% reported discussing CRC screening with their physician. The most frequently discussed screening modality was colonoscopy (70.7%), followed by sigmoidoscopy (41.4%) and fecal occult blood testing (40.6%). Approximately two thirds indicated discussing their interest in screening (“assess”), 36.1% reported being offered a choice among different screening modalities (“advise”) and 31.1% were asked about their preferences for different types of tests (“agree”). Over half (55.5%) reported receiving help making an appointment (“assist”) and 60.9% indicated receiving information on how to get test results (“arrange”). Approximately half of respondents (50.3%) indicated that they prefer to share decision-making responsibility with their doctor, 74.7% indicated they were involved in the CRC screening decision-making process as much as they wanted, and 13.9 % indicated there was information they wanted but did not discuss with their physician. Conclusions: The majority of primary care patients report discussing CRC screening with their physicians. Yet, the content of these discussions varies and almost half have not received recommended CRC screening. Although only a third reported being offered a choice among available screening modalities or asked about their preferences, the majority indicated being involved in the CRC decision-making process as much as they wanted to be. Implications for Policy, Delivery, or Practice: Routine screening is known to be effective at reducing CRC morbidity and mortality. Yet, many people (including those receiving routine primary care) fail to receive recommended screening. Given the limited time physicians and patients have to discuss CRC screening it is important that discussions be as productive as possible. Whether the use of a shared decisionmaking process and the “5 As” will lead to improved CRC screening adherence remains an important question. Primary Funding Source: Henry Ford Research Fund ●Brand and Generic Medication Prescribing for Medicare Beneficiaries with Hypertension Alex Federman, M.D., MPH, Ethan A. Halm, M.D., MPH, Tsivia Hochman, MS, Albert L. Siu, M.D., MSPH Presented By: Alex Federman, M.D., MPH, Assistant Professor of Medicine, Mount Sinai School of Medicine, 1470 Madison Avenue, Box 1087, New York, NY 10029; Tel: (212)241-8605; Email: alex.federman@msnyuhealth.org Research Objective: Little is known about the use of generic drugs by elderly patients in different income and insurance strata. We examined the relationship of income and prescription coverage with use of generic and brand antihypertensive medications in Medicare beneficiaries with hypertension. Study Design: We conducted a cross-sectional analysis of 2001 data from the Medicare Current Beneficiary Survey (n = 12,864, 71% response rate). MCBS interviewers record drug names verbatim from pill bottles or pharmacy receipts 3 times per year. We included community-dwelling beneficiaries over 65 with hypertension and excluded those with end-stage renal disease. We further selected respondents who used one or more drugs from 5 classes: angiotensin converting enzyme inhibitors (ACEI, 2/10 agents available as generics), alpha1antagonists (AA, 3/3 generics), beta-blockers (BB, 9/9 generics), calcium channel blockers (CCB, 4/8 generics), and thiazides (3/4 generics). The primary outcome was use of a brand drug in any of the 5 drug classes. Multisource drugs listed by brand name were coded as brand agents. When respondents used both brand and generic drugs we coded the outcome as generic, although coding as brand in such cases produced similar analytic results. We modeled the outcome using generalized estimating equations with a logit link function in SUDAAN to correct standard errors for the MCBS sampling design. Income was coded as percent of poverty: less than 100%, 100-135%, 136-150%, 151-200%, 201-300%, greater than 300%. Prescription coverage included none, Medicaid, managed care, self-purchased, employersponsored, and other. All analyses adjusted for age, sex, race, ethnicity, urban residence, Census region, and number of chronic diseases and number of antihypertension medications used. Population Studied: Community-dwelling Medicare beneficiaries over 65 with hypertension who used one or more antihypertension drugs in 2001 (n = 4,639). Principal Findings: Fifty-five percent of subjects had incomes less than 200% of poverty and 26% lacked prescription coverage. Rates of brand antihypertension medication use in each class were: ACEI, 87%; BB, 25%; CCB, 69%; thiazides, 29%; and AA, 45%. In multivariate analysis income had no significant association with brand use, either in the primary analysis or in subgroup analyses of each drug class. In contrast, lack of prescription coverage and managed care coverage were inversely associated with brand use compared to private coverage (adjusted OR 0.74, 95% CI 0.61-0.89, and AOR 0.80, 0.64-0.99, respectively). The effect was driven by BB use: in the subgroup of BB users, brand BB use was lower for those without prescription coverage (AOR 0.67, 0.49-0.91) and those with managed care (AOR 0.62, 0.43-0.91), whereas there were no differences by coverage type in the 4 other drug class subgroups. In the full cohort, neither income nor prescription coverage were associated with whether a beneficiary used any type of BB. Conclusions: Type of prescription coverage has a limited association with the use of brand versus generic antihypertension medications by Medicare beneficiaries, and income has no association. Implications for Policy, Delivery, or Practice: Physicians are missing opportunities to reduce the out-of-pocket healthcare spending of many low-income and underinsured elderly hypertensives, and reduce future costs for Medicare itself. Current policy and interventions that promote generic prescribing should be reevaluated and enhanced. Primary Funding Source: RWJF ●A Comparison of Dental Costs for Medicaid and S-SCHIP Programs Stephen Flach, M.D., Ph.D., Peter Damiano, DDS, MPH, Elizabeth Momany, Ph.D., Knute Carter, BS Presented By: Stephen Flach, M.D., Ph.D., Assisstant Professor, University of Iowa, Public Policy Center, 217 South Quadrangle, Iowa City, IA 52242; Tel: (319) 384-8266; Email: stephen-flach@uiowa.edu Research Objective: The State Children’s Health Insurance Programs (S-SCHIP) extended dental coverage to children not eligible for Medicaid. Creation of this newly covered population led us to ask whether dental costs would be higher than in traditional Medicaid programs due to pent-up demand derived from untreated dental need. Using administrative data, we evaluated the costs of the first episode of dental care for newly enrolled children in the Iowa Medicaid program and the three different S-SCHIP dental plans. Study Design: We identified all new enrollees (who had not been enrolled in the previous 12 months) in the Iowa Medicaid and S-SCHIP programs from July 1, 2001 to June 30, 2003, who utilized dental care and were enrolled for at least four months after their initial dental visit. We defined the four months following the initial dental visit as the first “episode of care”, based on dental judgment and an analysis that indicated that 94% of the six-month aggregate dental costs after the first visit occurred within the first four months. We used claims data to record all dental care utilization, and categorized services into: preventive/ diagnostic, routine restorative, complex restorative, and all other services using procedure codes. Costs were applied to procedure codes using charges from the 2003 American Dental Association (ADA) survey of dental fees. We categorized subjects as “early utilizers”; first dental visit within three months of enrollment, “middle utilizers”; first seen in 4-12 months, and the “late utilizers”; first seen after more than 12 months. We performed multivariate ordinary least squares (OLS) regression with total costs and the natural logarithm of total costs as dependent variables, and the time to first visit and health plan type as the key explanatory variables, controlling for age, gender, and urbanicity. Population Studied: Enrollees in the three S-SCHIP and the Medicaid programs in Iowa. Principal Findings: The mean total cost for the first episode of dental care was significantly higher in Medicaid ($376 versus $285, $223, and $ 328 for the three S-SCHIP plans). More S-SCHIP subjects were seen within three months of enrolling (37%, 50%, and 46%) than in Medicaid (33%). All SSCHIP dental plans had a greater share of expenses spent on preventive and diagnostic care (42%, 49%, 40%) compared to Medicaid (33%), while children in the Medicaid program had a higher percent of total costs devoted to complex restorative costs (24% verses 19%, 17%, and 18% for the S-SCHIP plans). OLS regressions indicated that costs for “early utilizers” during the initial episode of care was $71 higher than “middle utilizers” and $35 higher than “late utilizers”. Each S-SCHIP plan was less costly than the Medicaid plan, with average differences of $99, $193, and $37 per first episode of dental care. Conclusions: Total costs of the first episode of dental care are greater for those seeking care in the first 3 months after enrolling in the program. Costs were lower for children in the S-SCHIP program, and more focused on prevention. Implications for Policy, Delivery, or Practice: This report provides important information for policymakers concerning the costs of dental coverage in S-SCHIP plans. Primary Funding Source: AHRQ ●Direct Cost of Treatment in Autism: The Burden of Illness Scott Flanders, Ph.D., MS, Julie Whitworth, PharmD, MPA, Mohamed A. Hussein, MSCS, MSPH, Daniel R. Vanderpoel, PharmD, Timothy Sandman, MBA, Luella Engelhart, MS Presented By: Scott Flanders, Ph.D., MS, Associate Director, Outcomes Research, Janssen Medical Affairs, 740 Waterford Drive, Grayslake, IL 60030; Tel: (847)548-8977; Fax: (847)5489008; Email: sflander@janus.jnj.com Research Objective: To compare the direct treatment costs for children newly diagnosed with the pervasive developmental disorder autism, versus children newly diagnosed with diabetes or asthma, in a privately insured population. Study Design: A retrospective database analysis was completed using claims derived from the administrative database of a health benefits organization. Six months of prediagnosis and 12 months of post-diagnosis direct healthcare cost claims were analyzed for children newly diagnosed with autism, diabetes, or asthma. Healthcare costs included inpatient, outpatient, pharmacy, and special costs, which included physical and speech therapy, and emergency room visits. Non-parametric (Dunn Test) and parametric (Generalized Linear Model) comparisons of post-diagnosis costs were completed. Age, gender, insurance plan type, and costs in the pre-diagnosis period were controlled for within the multivariate model. Population Studied: The study population was comprised of children aged 3 – 17 years newly diagnosed with autism (N = 470), diabetes (N = 522), or asthma (N = 550). All subjects were continuously enrolled in an HMO, PPO, or Medicaid plan type throughout the study’s duration. Principal Findings: The autistic cohort exhibited a higher proportion of male subjects (80.6%) and a lower mean age (9.4 years) than the diabetic children (49.8%, 12.6 years) or asthmatic children (52.9%, 9.9 years). Autistic children exhibited significantly greater median total healthcare costs than children with asthma ($2103.58 vs. $850.27, P < 0.05) or diabetes ($2103.58 vs. $1605.20, P < 0.05) in the 12-month post-diagnosis period. After controlling for confounders, autistic children exhibited significantly greater 12-month postdiagnosis total healthcare costs than asthmatic children (P < 0.0001); the difference in total healthcare cost between autistic and diabetic children was non-significant (P = 0.84). Conclusions: Results from this study in a privately insured population suggest that autistic children may incur equivalent or significantly greater healthcare costs than children diagnosed with diabetes or asthma in the 12-month postdiagnosis period of their diseases. Implications for Policy, Delivery, or Practice: This study highlights the costs of care associated with autism as they compare to other growing public health concerns for children, such as asthma and diabetes. Primary Funding Source: Janssen Medical Affairs ●Using the SF-12 to Predict Medical Expenditures John Fleishman, Ph.D., Joel Cohen, Ph.D., Williard Manning, Ph.D., Mark Kosinski, MA Presented By: John Fleishman, Ph.D., Senior Sociologist, AHRQ, 540 Gaither Road, Rockville, MD 20850; Tel: (301)4271674; Fax: (301)427-1276; Email: jfleishm@ahrq.gov Research Objective: Relatively few studies have used selfreported health status in models to predict medical expenditures, and most of these have used the SF-36. The extent to which the shorter SF-12 predicts expenditures is not known. This study examines the extent to which the SF-12 measure of health status prospectively predicts medical expenditures in a nationally representative sample. Study Design: Interviews obtained data on demographics and selected chronic conditions. Respondents completed the SF12 in a questionnaire. Data on expenditures incurred subsequent to the interview were obtained in part from provider records. We examined different regression model specifications and compared different statistical estimation techniques (e.g., 2-part OLS models versus generalized linear models). Population Studied: We used nationally representative data from the 2000-2001 panel of the Medical Expenditure Panel Study. Analyses were based on 5,542 respondents with complete data. Principal Findings: Adding the SF-12 to a generalized linear regression model with a Poisson distribution improved the prediction of subsequent medical expenditures. In a model with only age and gender, adding the SF-12 increased Rsquared from 0.06 to 0.13. The coefficients for the Physical Component Summary (PCS) and the Mental Component Summary (MCS) of the SF-12 for this model were -0.045 (p=<.01) and -0.012 (p< .01), respectively. In a model including demographic characteristics, chronic conditions and prior expenditures, adding the SF-12 increased the R-squared from 0.26 to 0.29. The coefficients for the PCS and the MCS for this model were -0.025 (p<.001) and -0.005 (p=.15), respectively. Models estimated using ordinary least squares had undesirable properties. In terms of R-squared, a generalized linear model (GLM) with a Poisson variance function was consistently superior to a GLM with a gamma variance function. Conclusions: Information on self-reported health status, derived from the SF-12, is useful in predicting medical expenditures. The extent to which the SF-12 adds predictive power over a comprehensive array of diagnostic data remains to be examined. Implications for Policy, Delivery, or Practice: The association between SF-12 scores and medical expenditures illustrates the potential use of self-reported health status in the risk stratification of patients. The SF-12 may be useful for disease management programs and for adjusting capitation rates. Primary Funding Source: AHRQ ●Using the SF-12 Health Status Measure to Predict Medical Expenditures John Fleishman, Ph.D., Joel W. Cohen, Ph.D., Willard G. Manning, Ph.D., Mark Kosinski, MA Presented By: John Fleishman, Ph.D., Senior Social Scientist, Center for Financing Access and Cost Trends, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850; Tel: (301)427-1674; Fax: (301)427-1276; Email: jfleishm@ahrq.gov Research Objective: Relatively few studies have used selfreported health status in models to predict medical expenditures, and most of these have used the SF-36. The extent to which the shorter SF-12 predicts expenditures is not known. This study examines the association between SF-12 summary scores and prospective expenditures, controlling for a variety of demographic and clinical variables. Study Design: Interviews obtained data on demographics and selected chronic conditions. Respondents completed the SF12 in a questionnaire. Data on expenditures incurred subsequent to completing the questionnaire the were obtained in part from provider records. We examined different regression model specifications (e.g., including versus excluding information on selected chronic conditions) and compared different statistical estimation techniques (e.g., 2-part models versus generalized linear models). Population Studied: We used data from the 2000-2001 panel of the Medical Expenditure Panel Study, which comprises a nationally representative sample. Analyses were based on 5,542 respondents with complete information and who completed the SF-12 themselves. Principal Findings: Adding the SF-12 to a generalized linear regression model with a Poisson distribution improved the prediction of subsequent medical expenditures. In a model with only age and gender, adding the SF-12 increased Rsquared from 0.06 to 0.13. The coefficients for the Physical Component Summary (PCS) and the Mental Component Summary (MCS) of the SF-12 for this model were -0.045 (p=<.01) and -0.012 (p< .01), respectively. In a model including demographic characteristics, chronic conditions and prior expenditures, adding the SF-12 increased the R-squared from 0.26 to 0.29. The coefficients for the PCS and the MCS for this model were -0.025 (p<.001) and -0.005 (p=.15), respectively. Models estimated using ordinary least squares had undesirable properties. In terms of R-squared, a generalized linear model (GLM) with a Poisson variance function was consistently superior to a GLM with a gamma variance function. Conclusions: The link between SF-12 scores and medical expenditures illustrates the potential use of self-reported health status in the risk stratification of patients. Information on self-reported health status, derived from the SF-12, is useful in predicting medical expenditures. The extent to which the SF-12 adds predictive power over a comprehensive array of diagnostic data remains to be examined. Implications for Policy, Delivery, or Practice: The SF-12 may be useful for disease management programs and for adjusting capitation rates. Primary Funding Source: AHRQ ●Variations in Educational Debt Among USMGs Completing Graduate Medical Training in New York State, 2000-2003 Gaetano J. Forte, BA, David P. Armstrong, BA Presented By: Gaetano J. Forte, BA, Director of Information Management, Center for Health Workforce Studies, University at Albany, School of Public Health, 1 University Place, B334, Rensselaer, NY 12144; Tel: (518)402-0250; Fax: (518)402-0252; Email: gjf01@health.state.ny.us Research Objective: To describe variations in educational debt among new physicians. To explore the impact of variations in educational debt on career decisions upon completion of graduate medical training. Study Design: The data analyzed for this report were collected via an annual survey of all physicians completing a residency or fellowship training program in New York State. The survey was conducted in collaboration with the teaching hospitals in the state from 2000 to 2003. The questionnaire, completed by each individual resident, collected extensive information on their demographic and educational backgrounds, their posttraining plans and their perceptions of the job market for new physicians. Between 2000 and 2003 almost 12,000 physicians completing training in New York State responded to the survey (approximate response rate of 65%). A general linear model was used to estimate levels of educational debt among graduates of medical schools in the US (USMGs) controlling for a number of important variables, including age, gender, race/ethnicity, type of medical education, and specialty training. A multivariate logistic regression model was employed to explore the effects of educational debt on decisions to continue training in a subspecialty among USMGs completing General Pediatrics and General Internal Medicine training programs. Population Studied: USMG residents and fellows completing a residency training program in New York State between 2000 and 2003. Principal Findings: Average levels of debt increased in each year of the survey. Levels of educational debt varied significantly by specialty, type of medical education (allopathic and osteopathic), and race/ethnicity. Educational debt varied by as much as two to one across the specialties. Higher levels of education debt were associated with increased likelihood of going into patient care practice rather than going on to subspecialty training. Conclusions: Educational debt continues to increase for new physicians. There are a number of important factors that affect the educational debt of new physicians. The findings of this study suggest that specialty, type of medical education, and race-ethnicity are among the most influential factors affecting levels of educational debt of new physicians. Level of debt is associated with specific career choices among new physicians. Implications for Policy, Delivery, or Practice: With a renewed discussion of bolstering the national physician supply through the expansion of medical education capacity in the US, determining how to finance this endeavor will have to consider the role educational debt plays in the career decisions of physicians. Because levels of educational debt vary significantly, certain groups are saddled with high and increasing debt upon completion of their graduate medical training. Understanding the relationship between educational debt and career choices is critical in developing policies that are most likely to ensure that appropriate care is available to those who need it. Primary Funding Source: HRSA ●Use of Indirect Measures of Race/Ethnicity to Examine Disparities in Managed Care Allen Fremont, M.D., Ph.D., Arlene Bierman, M.D., MS, Mona Shah, MS, Chloe Bird, Ph.D., Steve Wickstrom, MS Presented By: Allen Fremont, M.D., Ph.D., Natural Scientist, Health, RAND, 1776 Main Street, Santa Monica, CA 90407; Tel: (310) 663-7916; Email: fremont@rand.org Research Objective: Although managed care plans routinely track quality measures, few examine whether their performance differs by enrollee race/ethnicity, in part because plans do not collect that information. We sought to assess the validity of using indirect measures of race/ethnicity based on geocoding to examine disparities in care. Study Design: Validation study comparing estimates of managed care enrollee’s racial characteristics and associated disparities in quality of care for diabetes and cardiovascular disease. Enrollee’s address was used to determine the Census Block Group (e.g. a small neighborhood) where they lived. Those living in a predominantly black neighborhood (i.e., with > 66% of the residents black) were identified as likely to be black. Medicare+Choice enrollee race data was also obtained from the CMS Denominator file. Individual-level measures of race were not available for commercial enrollees. Quality was measured by performance (i.e., percentage of eligible patients who received indicated care) on six NCQA HEDIS 2000 process of care measures including: in diabetics, an annual check of glycosolated hemoglobin (HgbA1c), low density lipoprotein (LDL), and urine protein levels, as well as a dilated eye exam; beta-blocker prescription for myocardial infarction (MI) patients; and LDL check in patients after a cardiac event. Population Studied: Managed care enrollees from nine Medicare+Choice and 10 commercial plans serving approximately 195,116 Medicare and 2,151,050 commercially insured enrollees across four regions. The analytic sample included the 16,262 Medicare+Choice enrollees and 37,404 commercial enrollees who were met NCQA eligibility criteria for the HEDIS diabetic or cardiovascular measures. Principal Findings: 89% of Medicare+Choice enrollees identified as black in CMS files were also identified as likely to be black based on geocoding. Of the 11% of black patients misclassified, 9% of those were black enrollees living outside of predominantly black neighborhoods. In the analyses using geocoded measures, racial disparities were present on 5 of 6 measures, with the largest disparities between racial groups on Beta-blocker after MI and LDL checks, which both exceeded 20 percentage points (P<.01). Estimated disparities were nearly identical in the analyses using the CMS measure of race with all estimated disparities within 1 percentage point of the estimates using the geocoded measure of race. A similar pattern of racial disparities was observed among commercial enrollees when race was based on a geocoded measure. However, no individual-level measure of race was available for commercial enrollees to assess disparities or compare results to those based on the geocoded measure. Conclusions: Using geocoded measures of race, we found disparities within both Medicare+Choice and commercial plans on HEDIS measures of diabetes and cardiovascular care effects. Estimated disparities in Medicare+Choice were similar regardless of whether a geocoded measure of race or one based on CMS data was used. Implications for Policy, Delivery, or Practice: Our study suggests that health plans can use indirect measures of race based on geocoding to begin examining racial and SES disparities among their enrollees. However, additional validation of the use of such indirect measures in commercial plans is still needed. Primary Funding Source: AHRQ ●The Effect of Depression and Functional Impairment on Healthcare Services Use of Functionally Impaired Elderly Primary Care Patients Bruce Friedman, Ph.D., MPH, Yeates Conwell, M.D., Rachel L. Delavan, MS Presented By: Bruce Friedman, Ph.D., MPH, Assistant Professor, Community and Preventive Medicine, University of Rochester, 601 Elmwood Avenue, Box 644, Rochester, NY 14642; Tel: (585)273-2618; Fax: (585)461-4532; Email: Bruce_Friedman@urmc.rochester.edu Research Objective: For functionally impaired, elderly primary care patients living in the community (not in nursing homes), to describe how use of 24 healthcare services compares (1) among those with major depression, clinically significant depressive symptoms but not major depression, and no depression, and (2) among those with mild, moderate, and severe functional impairment. Study Design: A longitudinal observational design examining health services use data from the control group (N=343) of the Medicare Primary and Consumer-Directed Care Demonstration. Patients and/or their caregivers completed a daily healthcare services utilization diary over 24 months. Data on hospital, nursing home, and assisted living use were obtained from facilities. Baseline data were collected on depression and functional impairment. Mild impairment was defined as dependence in 3-6 instrumental activities of daily living (IADL) but no activities of daily living (ADL), moderate impairment as dependence in 1-3 ADLs, and severe impairment as dependence in 4-6 ADLs. Chi Square tests and tests for trend were used to compare probability of use, and amount of use conditional on some use. Population Studied: Elderly Medicare primary care patients living in the community in 8 New York, 6 West Virginia, and 5 Ohio counties. Individuals had to be functionally impaired (2+ ADLs or 3+ IADLs) with recent significant healthcare services use. Principal Findings: Comparing major depression with no major depression at the time of study entry, probability of use differed significantly for only one of 24 services over the next 2 years. Patients with major depression were more likely to use adult day care. Among users, amount of use differed significantly for only one service. Patients with major depression had fewer psychiatrist office visits. Among the 283 cognitively intact patients, probability of use was lowest for those with no depression, intermediate for clinically significant depression but not major depression, and highest for major depression for specialist physician office visits, hospital admissions, emergency room visits, therapist office visits, and taxi/van trips. The opposite pattern occurred for physician home care visits. Among users, depressed patients had significantly more podiatrist office visits and taxi/van trips while patients with no depression had the most audiologist office visits. Among the 301 patients who had at least 3 IADLs at baseline (42 other patients were no longer impaired by that time), probability of use was lowest for those with mild functional impairment, intermediate for moderate impairment, and highest for severe impairment for physician assistant/nurse practitioner, nurse, therapist, home health aide/personal care aide, and companion home visits and for adult day care. The opposite pattern occurred for primary care and dentist office visits. Among users, hospital admissions and days and ambulance trips were highest for the severely functionally impaired. Conclusions: Depression appears to affect different categories of healthcare services than does functional impairment, both for probability of use and for amount of use conditional on some use. Implications for Policy, Delivery, or Practice: More attention should be paid to the differential impacts of depression and functional impairments on health services use of elderly Medicare primary care patients. Primary Funding Source: NIMH ●Geographic Factors and Hospital Characteristics that Prevent Admission to High-volume Surgical Hospitals in Japan Kiyohide Fushimi, M.D., Ph.D., Sayuri Shimizu, Noritoshi Yoshii, M.D., Toshiro Kamata, M.D., Hidenori Imai, Kaname Yamamoto, M.D. Presented By: Kiyohide Fushimi, M.D., Ph.D., Associate Professor, Department of Health Care Informatics, Tokyo Medical and Dental University Graduate School of Medicine, 1-5-45 Yushima, Bunkyo-ku, Tokyo, 1138519; Tel: 81(3)52835788; Fax: 81(3)5283-5788; Email: kfushimi.hci@tmd.ac.jp Research Objective: It has been shown that the quality of surgical care of hospitals may be associated with the volume of surgical procedures. Measures have recently been taken to increase the volume of surgical procedures in regional operation centers by Japanese government; however, the effectiveness of the measures has not been confirmed. It has been pointed out that geographic factors may prevent, especially in the rural areas, the establishment of regional operation centers to effectively provide high-quality surgical cares. The purpose of this study was to elucidate geographic, hospital, and patient factors that determine the admission to high-volume surgical hospitals or low-volume alternative ones. Study Design: High-volume surgical hospitals for cardiac surgery, lung surgery, brain surgery, percutaneous coronary intervention (PCI) were determined as the number of the corresponding operations being 50 or greater per hospital per year according to the government guidelines. Low-volume surgical hospitals were determined as other hospitals where at least one corresponding operation was performed per year. Hospital factors were obtained from the Survey of Medical Care Institutions and patient factors were obtained from the Patient Survey operated by the Ministry of Health, Labour and Welfare of Japan in 1996, 1999, and 2002. The distances between patient residences and hospitals were determined from the location of the hospitals and the representative points of patient residence areas. Multivariable logistic regression models were used with dichotomous dependent variable for the admission to high-volume hospitals or lowvolume hospitals. Independent variables were bed volume, teaching status, ownership, facilities, work force, transfer status, and distances from patient residence of admitted hospitals and of control alternative hospitals. The control alternative hospitals were determined as the nearest lowvolume hospitals for the patients admitted to high-volume hospitals and as the nearest high-volume ones for the patients admitted to low-volume ones. Covariates included gender, age, outcome, and the length of stay of patients. Statistical analyses were performed by Stata/SE 8.0. Population Studied: Records for 1387 patients with cardiac surgery, 1848 with lung surgery, 2759 with brain surgery, 12252 with PCI admitted for acute care from a nationally representative sample of 2,047,289 discharged patients from the Patient Survey were analyzed. Principal Findings: Factors in favor of admission to highvolume hospitals were transfer from other hospitals (OR=2.58, CI: 1.82-3.67), referral by physicians (OR=2.59, CI=1.82-3.67), teaching status of high-volume hospitals (OR=2.57, CI: 1.624.04), and private ownership (OR=1.73, CI: 1.05-2.88) of lowvolume hospitals. Factors against to the admission to highvolume hospitals were distances to high-volume hospitals (OR=0.997 per km, CI: 0.995-.999). The distances to lowvolume hospitals were not significant (OR=1.00). Conclusions: We found that physicians referral and teaching status contributes to induce patients to high-volume surgical hospitals and that geographic factors are statistically significant but substantial only when the high-volume hospital is very far. Implications for Policy, Delivery, or Practice: Health policy measures to promote referral to high-volume surgery hospitals and to provide information of ‘good’ hospitals to patients who might choose low-volume surgery hospitals will improve the quality and efficiency of surgical care. Primary Funding Source: Other Govt, Health and Labour Sciences Research Grants for Research on Applied use of Statistics and Information, for Comprehensive Research on aging and Welfare, and for Special Researches from the Ministry of Health, Labour and Welfare of Japan ●Analysis of Obesity in Hawaii's Children Krista Gronley, MPH, MBA, Deborah A. Taira, Sc.D. Presented By: Krista Gronley, MPH, MBA, Project Manager, Research, Care Management, Blue Cross Blue Shield of Hawaii, 818 Keeaumoku Street, Honolulu, HI 96814; Tel: (808)948-6979; Fax: (808) 948-6043; Email: krista_gronley@hmsa.com Research Objective: To evaluate the rate of obesity of Hawaii’s children age 2 through 14. Study Design: This study used data from the 2004 member satisfaction survey conducted by a large health plan to investigate the rate of obesity by age, island, and ethnicity grouping in children age 2 through 14 (n=4,096). We converted self-reported height and weight to a body mass index (BMI) and used the Clinical Growth Charts (body massfor-age) from the National Center for Health Statistics to categorize the BMI’s into underweight, normal, at risk and overweight. Members who did not report height and weight were excluded from the data. Zip-codes were used to identify isle of residence and ethnicity was self-reported. Members who indicated more than one ethnicity were place in a mixed category, except for Hawaiian. Members who marked Hawaiian along with any other ethnicity were categorized as Hawaiian. Population Studied: The study sample included health plan members ages 2 through 14 who completed, or whose parents completed, the 2004 member satisfaction survey and reported their height and weight (n=2,280). Each of the children participating in the survey had seen a physician at least once in the prior 12 months. Principal Findings: We found that 21.3% of the population are above the 95th percentile of the Clinical Growth Charts and considered to be overweight and 11.8% are in the risk category, between the 85th and 95th percentile. Those in the overweight category were more likely to be female, age 2 through 5, live on Maui or Molokai and be of Hawaiian, Samoan or Filipino ethnicity. The rate of over-weight or atrisk-weight declines as age increases. Conclusions: A full third of the study population are above the 85th percentile on the Clinical Growth Charts (body mass index-for age). There are significant differences by age and by ethnicity but not by island of residence, gender or health plan product. Implications for Policy, Delivery, or Practice: Targeted interventions that are age and ethnicity specific may provide members with the needed information and tools to achieve and maintain a healthy body-mass-index. Primary Funding Source: No Funding Source ●Child Health Beliefs and Behaviors of Child Care Directors: Head Start vs. Non Head Start Centers Ruchi Gupta, M.D. MPH, Paula Duncan, M.D., John M Pascoe, M.D., MPH, Linda Southward, Ph.D., Peter Gorski, M.D., MPA, Robert Greenberg, M.D. Presented By: Ruchi Gupta, M.D., MPH, Pediatrics, Institute for Health Services Research and Policy and Children's Memorial Hospital, 339 East Chicago Avenue., Room 712, Chicago, IL 60611-3071; Tel: (312)503-0429; Fax: (312)503-2936; Email: r-gupta@northwestern.edu Research Objective: Approximately 75% of preschool children (23 million) are in some form of child care. Head Start is the only national child care program developed to provide comprehensive education, health, and nutrition services to low-income children. Our objective was to compare directors child health beliefs and practices in both Head Start and nonHead Start child care centers. Study Design: A random telephone survey was conducted with directors of licensed child care centers in four states. Chisquared and T-tests were used to evaluate differences between Head Start and non-Head Start programs. Population Studied: Child care directors in four states including Mississippi, Vermont, Florida and New Mexico were randomly selected for the study. Principal Findings: 2201 surveys were completed (response rate of 89%), of which 191(9.1%) were Head Start centers. Forprofit and non-profit centers were analyzed together after preliminary analysis showed concordance. Fewer non-Head Start centers reported screening for health problems compared to Head Start centers (63% vs. 93%, P<0.01). 97% of Head Start programs felt screening was moderately/extremely useful compared to 81% of non-Head Start programs (p<0.01). 98% of Head Start programs provide parents with information about their child's health needs compared to 91% of non-Head Start programs (p<0.01). Head Start centers used more consultation by health professionals including physicians, nurses, dentists, dental hygienists, dieticians and mental health professionals (p<.01). Barriers including, lack of time, someone to provide information, funds and English comprehension were not significantly different between the Head Start and non-Head Start centers. Conclusions: Head Start continues its strong commitment to child health by providing universal screening and 2-3 times more consultation by health professionals than non-Head Start child care centers. Implications for Policy, Delivery, or Practice: Improving the health of preschool children may involve using Head Start as a model and advocating for similar health services in all child care centers. Primary Funding Source: SSRC, Mississippi State University ●Reduction in Emergency Room Visits, Hospital Admissions, and Clinical Outcomes After Enrollment in a Mediation Access Program Gary Harmon, MPH, Robert Federici, MSPH, Wendy Roy, BS, Keith Ashby, M.D., Danny Jackson, John Lefante Jr., Ph.D. Presented By: Gary Harmon, MPH, Epidemiologist, Biostatistics, Tulane University School of Public School and Tropical Medicine, 1440 Canal Street, New Orleans, LA 70112; Tel: (504)988-2891; Fax: (504)988-1706; Email: gharmon@tulane.edu Research Objective: To determine change in emergency room visits, hospital admissions, and clinical outcomes for participants in the Cenla Medication Access Program in rural central Louisiana. Study Design: Pre-post longitudinal cohort where outcomes for participants are assessed before and after enrollment to look for change. Participants are enrolled in the CMAP and begin receiving chronic care prescription medications at greatly reduced costs ($3 per prescription). Participants are consented to allow the evaluators access to their hospital records. Emergency room visits and hospital admissions in the 6 months prior to enrolling in CMAP are compared to usage in the 6 months after enrollment to assess change. Chart abstractions are performed to look for changes in blood pressure, blood glucose, and blood lipids from the period before enrollment into CMAP compared to the period after enrollment. Population Studied: Over 7000 residents of central Louisiana who access the public hospital in Pineville, Louisiana. Patients are eligible if they are receiving medications for a chronic illness, meet income guidelines, and do not have any type of insurance that pays for medications. The top five diagnoses for this population are hypertension, diabetes mellitus, hyperlipidemia, arthritis, and gastroesophageal reflux disease. Patients are seen in the hospital clinic and are enrolled through Social Services after meeting eligibility requirements. Principal Findings: Through the first 2 years of the project, there has been an observed reduction of 33.6% in ER visits and a 72.2% reduction in hospital admissions when comparing 6 months prior to enrollment to 6 months post enrollment. Reductions of 4% and 5% for systolic and diastolic blood pressure, respectively, have been observed after adjusting for numerous covariates (age, race, gender, primary diagnosis, number of medications patient receives). There has been an observed reduction of 17% in average blood glucose, and an 11% reduction in average hemoglobin A1c. Reductions in total cholesterol and LDL cholesterol of 20% and 25%, respectively, have been observed. Conclusions: Preliminary analyses indicate that patients in the Cenla Medication Access Program have reductions in the use of the emergency room and fewer hospital admissions. Participants have also experienced reductions in blood pressure, blood glucose, and blood lipids. Implications for Policy, Delivery, or Practice: Results indicate that providing medications at greatly reduced costs to individuals who otherwise cannot afford them results in reduced usage of emergency care and hospital admissions. Also demonstrated is that patients will experience reductions in clinical measures targeted by the medications. The reduction in emergency room admissions and hospital admissions can lead to a lowering of the financial burden incurred by state hospitals, and the clinical improvements seen in the patients can lead to better quality of life. Primary Funding Source: The Rapides Foundation ●Exploring the Link between Asthma Hospital Services and Air Quality in New Hampshire Aichatou Hassane, MA, Ross Gittell, Ph.D., Robert Woodward, Ph.D., Cameron Wake, Ph.D. Presented By: Aichatou Hassane, MA, Ph.D. Student, Whittemore School of Business and Economics, University of New Hampshire, 15 College Road, McConnell Hall, Durham, NH 03824; Tel: (603) 862-7071; Fax: (603) 862-3461; Email: ahassane@cisunix.unh.edu Research Objective: This study used the spatial, temporal, and age-specific patterns of asthma hospital services to identify potential asthma triggers. Study Design: We merged hospital respiratory services data with data on air pollutants for Ozone (O3), Sulfur Dioxide (SO2), and Carbon Monoxide (CO) obtained from local Environmental Protection Agency (EPA) and University monitoring sites. To analyze the relationship between asthma admission and air pollution, we first ran multiple OLS regression of asthma daily services on daily air pollutants and weather. Second, we compared cumulative daily asthma services relative to cumulative daily concentrations of air pollutants for various age groups, locations, and years. Pollutants were measured in parts per billion (ppb) with data on daily 1-hour maximum for O3 and CO and daily average for SO2. Population Studied: We collected detailed admissions and outpatient contacts (services) from three New Hampshire Seacoast area hospitals. Principal Findings: We found a strong seasonal variation with the lowest services in the summer and a strong peak in the fall, but at different times in different years. We also observed that individuals age 0-4 and 18-24 had the highest rates of asthma services and the largest seasonal variation. The multiple OLS regression of daily asthma on pollutants (CO, SO2, and O3), controlling for daily outside temperature and relative humidity, but without seasonal adjustment, revealed only one significant correlation, a negative relationship between asthma services and O3. An increase of 1 ppb in the daily 1-hour maximum O3 led to a 0.6% decrease in the number of daily asthma services (P=0.08). The O3 regression coefficient became insignificant when a dummy for August was added. Our second and more detailed method of analyzing the cumulative daily asthma services relative to the cumulative daily concentration of air pollutants showed no evidence of a link between asthma services and the levels of air pollutants. The occasional changes in the slope of the cumulative daily asthma graphs occurred at substantially different times than the changes in the slope of the accumulative concentration levels of CO, SO2, or O3. Conclusions: In contrast to the substantial literature establishing a link between air quality and asthma, our preliminary analyses of more detailed data found no confirmation of causality. We note that our finding that high levels of ozone were associated with lower asthma services is similar to that reported by Neisdell (2004). He argued that California smog alerts in period of high ozone led to avoidance behavior. The EPA also issued smog alerts in New England during this period. Implications for Policy, Delivery, or Practice: An understanding of the exact causes of asthma hospital services is important both for hospital management teams and employee benefit managers. Our study confirmed the findings of Bates et al. (1998) and Silverman (2003), which identified low summer and high fall asthma services and greater fluctuations in the very young. But our results conflict with Garty, et al. (1998) and Schwartz et al. (1993) who linked asthma services to poor air quality. Additional studies of other pollutants, pollen, and weather conditions are necessary to determine the environmental triggers of asthma services. Primary Funding Source: National Oceanic and Atmospheric Administration ●Health Care Experiences among Working-Age Adults with Physical Disabilities Enrolled in a Medicaid Managed Care Program Pei-Shu Ho, Ph.D., Susan E. Palsbo, Ph.D., Phillip W. Beatty, MA Presented By: Pei-Shu Ho, Ph.D., Senior Research Associate, Center for Health & Disability Research, National Rehabilitation Hospital, 102 Irving Street NW, Room 1067, Washington, DC 20010; Tel: (202) 877-1787; Fax: (202) 7267521; Email: pei-shu.ho@medstar.net Research Objective: We compared the health care experiences of fee-for-service versus a Medicaid managed care program among working-age adults with physical disabilities living in Minneapolis/St. Paul metropolitan area. We hypothesized that the need for care, quality of care and access to care would increase in the Medicaid managed care program that used care coordinators and disability-sensitive provider panels. Study Design: We used retrospective, one-group pretestposttest study design in which adults age 18-64 with physical disabilities were asked about their health care experiences before and after they enrolled in a Medicaid managed care program. Study participants responded to the CAHPS®-like survey over the phone at enrollment and one year later. Data, collected from the longitudinal surveys and intake assessment, were used for analyses. We used paried McNemar test and Marginal Homogeneity test for comparative analyses. Population Studied: Study participants were 124 adults enrolled in the Minnesota Disability Health Options (MnDHO) program, a demonstration Medicaid managed care program for working age adults with physical disabilities. Principal Findings: The need for care of our study participants was greater in the year after they enrolled in the MnDHO program except for rehabilitation therapy and alternative care. In particular, the need for primary care doctor (p=0.005), personal care attendants (p=0.001), and transportation assistance (p<0.001) was significantly increased after enrolled in the MnDHO program. The percentage of participants who reported having received needed care also increased in the year following MnDHO enrollment, especially in accessing their primary care doctor (p<0.001) and getting needed medical equipment (p=0.008). Significant improvement was also found in areas such as health care coordination, quality of interaction with health care providers, and promptness of care. Conclusions: Our findings show that the use of communitybased health care services has significantly increased after the MnDHO enrollment. The respondents are also more likely to receive needed care and are satisfied with the care they have received after they enrolled in the MnDHO program. Implications for Policy, Delivery, or Practice: Our results underscore the positive impact of disability literate care coordinators and providers on meeting the health care needs and providing quality care to people with physical disabilities. Primary Funding Source: RWJF ●Gender Disparities in Percutaneous Coronary Interventions for Acute Myocardial Infarction in Pennsylvania Christopher Hollenbeak, Ph.D., Carol Weisman, Ph.D., Michael Rossi, M.D., Steven Ettinger, M.D. Presented By: Christopher Hollenbeak, Ph.D., Assistant Professor, Surgery and Health Evaluation Sciences, Penn State College of Medicine, 500 University Drive, H113, Hershey, PA 17033; Tel: (717)531-5890; Fax: (717)531-4185; Email: chollenbeak@psu.edu Research Objective: Guidelines for the management of patients with acute myocardial infarction (AMI) recommend percutaneous coronary interventions (PCI) as an alternative to thrombolytic therapy when the procedure can be performed in a timely manner by experienced personnel. The purpose of this research was to determine whether there were gender disparities in the use of PCI in the treatment of AMI in Pennsylvania and, if so, whether outcomes were affected. Study Design: We performed a retrospective analysis of a statewide database. Multivariate analyses were performed using binary logistic regression, controlling for relevant covariates. Retrospective matching on propensity scores was performed using a “greedy” matching algorithm. Population Studied: Data were provided by the Pennsylvania Health Care Cost Containment Council (PHC4) and included all 31,351 AMI patients at all acute care hospitals in the state of Pennsylvania during the year 2000. The population was stratified by gender to identify disparities in treatment and outcomes. Principal Findings: During the year 2000, 10,170 (32.4%) AMI patients were treated with PCI and 21,181 (67.6%) patients were medically managed in Pennsylvania hospitals. Across all patients, women were significantly less likely to undergo PCI than men (23.9% vs. 39.9%, p<0.0001) and were also more likely to die in the hospital during the index admission (12.7% vs. 9.7%, p<0.0001). Restricting the sample to patients treated at hospitals where PCI was available, women were still significantly less likely than men to undergo the intervention (34.4% vs. 65.4%, p<0.0001). They were also more likely to die at hospitals offering PCI than men (10.1% vs. 7.3%, p<0.0001). These disparities in treatment and outcomes were confirmed in multivariate analyses. After controlling for age, race/ethnicity, severity at admission, location of infarct, and source of admission, women still had 24% lower odds than men of receiving PCI (p<0.0001) regardless of the availability of PCI. Finally, we used propensity score methods to match 3,023 women who received PCI to 3,023 women who did not. Results showed that women who received PCI were significantly less likely to die than similar women who did not (2.3% vs. 10.4%, p<0.0001). We also matched 3,329 women who received PCI to 3,329 similar men who received PCI. Among these patients, the difference in mortality was not statistically significant (1.59% vs. 1.92%, p=0.39). Conclusions: In the Commonwealth of Pennsylvania, women appear to be less likely to receive primary PCI for treatment of AMI. Furthermore, women who received this form of therapy experienced better outcomes. Implications for Policy, Delivery, or Practice: These results suggest that the morbidity and mortality associated with AMI in women could be reduced by increased utilization of primary PCI. The reasons for the gender disparity in treatment are not understood and require further investigation in order to improve treatment outcomes for women. Primary Funding Source: No Funding Source ●Performance Evaluation of the Multipurpose Australian Comorbidity Scoring System (MACSS) C D'Arcy J Holman, MBBS MPH Ph.D., David B. Preen, Ph.D., James B Semmens, Ph.D., Judith C Finn, Ph.D. Presented By: C D'Arcy J Holman, MBBS, MPH, Ph.D., Professor of Public Health, School of Population Health, The University of Western Australia, 35 Stirling Highway, Nedlands, Western Australia, 6009; Tel: 61-8-6488-1251; Fax: 61-8-6488-1188; Email: darcy@sph.uwa.edu.au Research Objective: To characterise the MACSS, a new 102item comorbidity scoring system, in terms of its performance in risk adjustment, validation of item-recording, and optimisation of look-back. Study Design: A series of large population-based cohort studies using the Western Australian Data Linkage System; and hospital chart and primary medical care record validation reviews in a random sample. Population Studied: Hospital separations of adult medical (n=326,456), surgical (n=349,686) and pyschiatric (n=16,895)patients; and a cross-section validation study based on 2,037 patients. Principal Findings: MACSS substantially outperformed the Charlson index. Administrative health data identify only 45% of comorbid conditions. Optimal look back varies with clinical group and health outcome. Conclusions: MACSS is a high-performance comorbidity scoring an adjustment system, but relies on good ascertainment of comorbid conditions and use of the appropriate look back period. Implications for Policy, Delivery, or Practice: Health services researchers may consider using the MACSS in studies requiring risk adjustment for comorbidity due to its high performance across a wide range of patient groups. Primary Funding Source: Australian National Health and Medical Research Council ●Cope with National Health Insurance: Strategic Behaviors of Taiwan's Hospitals Kuo-Cherh Huang, Dr.Ph, MBA, Ning Lu, Ph.D., MPH, Yi-Hua Chen, Ph.D., Hui-Chih Chang, MHA Presented By: Kuo-Cherh Huang, Dr.Ph, MBA, Associate Professor, Graduate Institute of Health Care Administration, Taipei Medical University, 250 Wu-hsing Street, Taipei, 110; Tel: +886-2-23785339; Fax: +886-2-23789788; Email: kchuang@tmu.edu.tw Research Objective: The primary goals of this study were to profile the strategic behaviors of Taiwan hospitals under the National Health Insurance Program, identify the predictive factors for such behaviors, and assess the influences of hospitals' strategic behaviors on performance. Study Design: Self-administered questionnaire survey. Population Studied: Our study population included 127 district teaching hospitals or above in Taiwan. Upper level managers or executives of the hospitals such as superintendent, vice-superintendent, or other upper-level administrators who were knowledgeable of their hospital practice were explicitly asked to respond to the survey, to ensure the accuracy and reliability of the data collected. Principal Findings: 1. The findings indicated that the most popular strategic behavior adopted by hospitals was strategic alliance, while the least likely embraced strategy was merger and acquisition. 2. Multinomial logistic regression analysis results demonstrated that hospital ownership and level were significant predictors of hospitals' strategic behaviors. Private and non-profit private-sector hospitals were significantly more likely to embrace the low-cost strategy, as opposed to the unrelated diversification strategy, compared to public hospitals. Furthermore, compared to medical centers and regional hospitals, district teaching hospitals were more likely to implement various strategies such as low cost, as opposed to the strategic behavior of unrelated diversification. 3. Multiple regression analysis results showed that the strategies of differentiation and focus exerted significantly positive influences on hospitals' performance. 4. Finally, we observed that the degree of local market competition had no impact on hospitals' choices of strategies and performance. Conclusions: This investigation shows that Taiwan's hospitals conduct a variety of strategies when they encounter emerging opportunities and threats created by important forces, in this case, the creation of a universal healthcare system. The findings also suggest the correlates of such behaviors. Moreover, we demonstrate that different strategic behaviors exert different degrees of impact on hospitals' performance. Implications for Policy, Delivery, or Practice: This empirical research embarks on the issue of the strategic behaviors of hospitals when they are facing a significant environmental change. From a management perspective, it is imperative for hospital administrators to embrace a more suitable strategy in order to survive in today's highly competitive healthcare market. Taken together, the study's results shed light on some interesting issues regarding hospitals' strategic behaviors in response to dramatic changes in the external environment, and provide some suggestions for hospital administrators. Primary Funding Source: National Science Council, Taiwan. ●An Examination of Risk Factors For Stroke Among Two Medicare Elderly Cohorts Baqar Husaini, Ph.D., Barbara Kilbourne, Ph.D., Pamela C. Hull, Ph.D., Darren Sherkat, Ph.D., Robert Levine, M.D., Majaz Moonis, M.D. Implications for Policy, Delivery, or Practice: Our findings add to the evidence that controlling risk factors is crucial for prevention and reduction of high cost of treatment for stroke patients. Primary Funding Source: AHRQ Presented By: Baqar Husaini, Ph.D., Professor and Director, Center for Health Research, Tennessee State University, 3500 John A. Merritt Boulevard, Nashville, TN 38209; Tel: (615)3203005; Fax: (615)320-3071; Email: bhusaini@tnstate.edu Research Objective: In this paper we estimate the direct and indirect effects of various risk factors by race and gender on the likelihood of developing stroke for two sub-cohorts of Tennessee Medicare beneficiaries. Medicare data provide a unique opportunity to analyze physician diagnoses of cardiovascular risk factors for stroke in order to assess differential morbidity patterns across two age cohorts. Study Design: Using ICD-9 codes we examined seven (7) risk factors in 1996: (i) Transient ischemic activity, (ii) high cholesterol, (iii) hypertension, (iv) diabetes, (v) myocardial infarction, (vi) congestive heart failure and (vii) atrial fibrillation. Outcome variables included whether or not an individual was diagnosed with a stroke during 1997- 2000. Further, we excluded individuals from the analysis individuals with a diagnosis of stroke in 1996 as well as those who died during 1997-2000. We then performed a set of block recursive logistic regressions to estimate the direct and indirect effect of risk factors on stroke. Population Studied: Combined data from the Medicare enrollment database (EDB 1996-2000) and physician billing records was used for this paper. We used a closed cohort of Tennessee Medicare beneficiaries, aged 65 or older in 1996 (n=698,697). We selected only African Americans and whites. We used two cohorts of elderly in 1966: individuals 65-69 years old (n= 21,898) and another sub-cohort of individuals 70-74 year olds (n=25,723 ) as of January 1, 1996. Principal Findings: Our analyses show that among the 65 – 69 year-old cohort, 160 individuals were diagnosed with stroke during 1997-2000. Race did not appear to directly affect the likelihood of the diagnosis. Over this period, a diagnosis of hypertension increased the likelihood of stroke 52% (p<0.001). African Americans were significantly more likely to receive diagnoses of diabetes and hypertension than their white peers. Thus, among this cohort race disparities occur primarily in the conditions that predict stroke rather their incidence. Among the 70-74 year-old cohort, 263 individuals were diagnosed with stroke. Race did not directly predict the incidence of stroke. However, African Americans were more than twice as likely (p<0.001) than Whites to be diagnosed with diabetes or hypertension, and 69% more likely (p<0.001) to be diagnosed with congestive heart failure; all three conditions increase the likelihood of stroke. Conclusions: The incidence of stroke increases markedly between the younger and older cohort. Among the younger cohort of beneficiaries race does not appear directly associated with a diagnosis of stroke. However significant indirect associations between race and stroke appear through diabetes among African American elders. In the older cohort (70-74 years old), while a direct association between race and stroke is non significant, Race indirectly associated with stroke through diabetes, congestive heart failure and hypertension. ●Socio-Economic Determinants Associated with Length of Psychiatric Hospitalization hirohisa imai, M.D., Ph.D. Presented By: hirohisa imai, M.D., Ph.D., associate professor, health science, asahikawa medical college, midorigaoka, E2-11-1, asahikawa, 078-8510; Tel: +81-166-68-2400; Fax: +81-16668-2409; Email: hiroimai@asahikawa-med.ac.jp Research Objective: The present study examined socioeconomic determinants associated with length of stay in regional psychiatric hospitals in Japan, in order to identify variables that might help to reduce length of stay. Study Design: This cross-sectional study analysed the complete database of one district in Japan. Analyses were conducted using aggregate data from 24 hospitals. Multiple regression analysis was performed to determine socioeconomic factors influencing length of stay in psychiatric hospitals. The information contained the patient profiles, the basic demographic data, the economic factors, the admission and a diagnostic assessment according to ICD-10 as the patient characteristics and the number of medical stuffs, the number of outpatient and inpatient. Population Studied: All 24 psychiatrics hospitals and a total of 6854 patients staying with them in Miyazaki prefecture during 1 April 1998 to 1 March 1999 formed the study population. The patients comprised 46.2% males and 53.8% females. As their age distribution 4 groups were divided under 20, from 21 to 40, from 41 to 65, over 65 years old. Principal Findings: Significant positive associations were identified between length of stay and proportion of involuntary admissions as social factors. Our finding of an association between involuntary admission and longer length of stay supports the results of previous investigations. Unlike many other developed countries, time in seclusion is not limited in Japan. A system of periodic reviews or audits for psychiatric hospitals is thus needed to safeguard against unnecessary confinement and excessive length of stay. Multiple linear regression analysis revealed that proportion of patients receiving public aid was independently associated with shorter length of stay. Shorter length of stay for patients receiving public aid may reflect the fact that the stigma associated with mental illness remains high in Japanese families. Historically, hospitalisation has represented the most socially acceptable solution for many families of psychiatric patients in Japan, and a researcher indicated that many families do not want mentally ill family members to return home. In our experience, patients who have no family and receive no other financial support sometimes receive public aid. These patients may be discharged more rapidly because no pressure is exerted to delay the process. Conclusions: Our study clarified socio-economic determinants associated with length of stay in regional psychiatric hospitals. To date, policy makers have not implemented evidence-based health policies for mental care. The factors identified in this study offer a basis for effective psychiatric health policies to reduce length of stay and should contribute to health policies promoting de-institutionalisation. Implications for Policy, Delivery, or Practice: Average length of stay for inpatients in psychiatric hospitals has consistently been longer in Japan than in any other developed nation, at 330 days in 1996. If government authorities or policy makers focus on data regarding socio-economics determinants affecting length of stay at individual psychiatric hospitals when devising and implementing psychiatric care policies, and if policies can in fact alter practices at individual psychiatric hospitals, hospital length of stay for individual psychiatric patients may be effectively shortened. Primary Funding Source: No Funding Source ●Analysis of Pediatric Emergency Care in Massachusetts How Can It be Improved? Rachel Immekus, MS/MPH Presented By: Rachel Immekus, MS/MPH, Health Policy Analyst, Health Systems Measurement and Improvement Group, Division of Health Care Finance and Policy, 2 Boylston Street, Boston, MA 02116; Tel: (617) 988-3273; Email: rachel.immekus@hcf.state.ma.us Research Objective: To provide a comprehensive picture of emergency medical services utilized by children in the state of Massachusetts in order to better direct pediatric care. Study Design: Analysis was conducted on all emergency visits, including those that resulted in a hospital admission or observation stay. Detailed patient-level data from the Division of Health Care Finance and Policy’s Emergency Department Database, Outpatient Observation Database, and the Inpatient Discharge Database were used to obtain total volume, regional areas of high visit concentration, common diagnoses and procedures, mode of transportation, and admission patterns. ArcView software was used to produce maps illustrating zip code areas by concentration of patient utilization in relation to emergency department locations. Population Studied: Children between the ages of 0 and 14 who visited a Massachusetts emergency department in the hospital fiscal year 2003 Principal Findings: In FY 2003, 427,132 ED visits were made by children. These visits accounted for 15.4% of all ED visits in Massachusetts. Approximately 11% of outpatient visits were registered in the early morning hours between midnight and 8:00, 34% were admitted between 8:00AM and 4:00 PM, while the majority of visits, 55%, were registered between 4:00 PM and midnight. Five diagnoses, based on ICD-9 codes, accounted for 20% outpatient visits and included 1) Unspecified otitis media, 2) Fever. 3) Acute upper respiratory infection 4) Unspecified viral infection, and 5) Acute pharyngitis. The majority of pediatric visits resulted in an outpatient discharge with only 7% being admitted to inpatient or observation care. Those that were admitted however carried the more emergent diagnoses common to inpatient admissions: 1). Volume depletion, 2) Asthma with (acute) exacerbation 3) Pneumonia 4) Single live born (where the mother was admitted through the ED for the birth of a child) and 5) Acute appendicitis. Private insurers and Medicaid covered 92% of the visits (64% and 28% respectively) that did not result in admission. Conclusions: The majority of pediatric ED visits occur in the day and evening hours when most physicians’ offices should or could be open. Also, the top outpatient diagnoses suggest treatment at a clinic or physician’s office to be feasible. The vast majority of the kids who presented for ED care were insured (92%) theoretically facilitating their primary and secondary care in a non-ED setting. Implications for Policy, Delivery, or Practice: Given the high volume of pediatric visits to the ED during day and evening hours, especially for non emergent conditions, policy makers and insurers might want to consider incentives to large pediatric providers and CHCs to expand hours into the early evening for sick calls and/or visits. Primary Funding Source: No Funding Source ●Employment of Mid-level Providers in Primary Care Programs and Control of Diabetes George Jackson, Ph.D., MHA, Shoou-Yih Daniel Lee, Ph.D., David Edelman, M.D., MHS, Morris Weinberger, Ph.D., Elizabeth Yano, Ph.D., MSPH Presented By: George Jackson, Ph.D., MHA, Postdoctoral Fellow, HSR&D Service (152), Durham Veterans Affairs Medical Center, 508 Fulton Street, Durham, NC 27705; Tel: (919)286-0411 x7103; Fax: (919)416-5836; Email: george_jackson@alumni.unc.edu Research Objective: Considering the different training orientations of nurse practitioners (NPs) [biopsychosocial] and physician assistants (PA’s) [biomedical], we examined the association between employing mid-level providers in Veterans Affairs primary care practices on diabetes control, as measured by hemoglobin A1c (HbA1c). Study Design: A cohort of 224,221 diabetes patients (alive October 1, 1999) was identified using the Veterans Health Administration (VHA) Diabetes Registry and Dataset and VHA Medical SAS Datasets. Data on the organization of programs at large VA facilities (>= 4000 patients and >= 20,000 outpatient visits in 1998) was obtained via the 1999 VHA Survey of Primary Care Practices (n = 219). To simultaneously control for the impact of individual and organizational characteristics on the individual outcome of HbA1c, two-level hierarchical modeling was used to assess the association between having NPs and PA’s in primary care programs and the last HbA1c value recorded during the follow-up period (fiscal year 2000-2001). Organizational-level covariates that represent integration of providers into programs included: staffing sufficiency, on-site training, and practicing at multiple VA locations for both nurses and NPs/PAs. Also included were nursing full-time equivalents per 1000 general medicine patients, state of transition to service-line structure, needed primary care staff hired year before survey, total facility general medicine patients, geographic region, academic affiliation, and location in a hospital. At the individual level, results were risk adjusted based on comorbidities, gender, body mass index, race/ethnicity, service-connected disability rating, VA meanstest category (partial proxy for socioeconomic status), use of glucose control medication, and primary care utilization. Population Studied: The analysis included 82,692 VA primary care patients with diabetes at 190 separate facilities. Principal Findings: Presence of NPs in primary care programs was associated with patients’ HbA1c results being lower by 0.34%. Having PAs in the program was not associated with HbA1c. Also significant at the p=0.05 level were: (1) having NPs/PAs practice at multiple VA sites (HbA1c higher by 0.20%), (2) sites reporting NP staffing was barely or not at all sufficient compared to being more sufficient (HbA1c higher by 0.20%), (3) having a partially as opposed to fully implemented service-line structure (HbA1c higher by 0.18%) and (4) having needed new staff acquired the year before the survey (HbA1c lower by 0.13%). Conclusions: Inclusion of NPs in primary care practices was associated with better control of diabetes. Further, programs reporting that staffing of mid-level providers is sufficient, having the mid-level providers practice only at one facility, having a fully implemented service-line structure, and actively acquiring needed staff are also associated with better diabetes control. Implications for Policy, Delivery, or Practice: The associations reported here point to the possibility that including NPs in primary care programs may lead to better diabetes outcomes. Primary Funding Source: No Funding Source ●Institutional Factors Associated with ADL Decline in Pennsylvania Nursing Homes Amy Jessop, Ph.D., MPH, Jocelyn Andrel, MSPH, Naomi Hauser, RN, MPA, David Wenner, DO Presented By: Amy Jessop, Ph.D., MPH, Senior Epidemiologist, Quality Insights of Pennsylvania, 585 East Swedesford Road, Wayne, PA 19087; Tel: (877)346-6180; Fax: (610)688-5276; Email: ajessop@wvmi.org Research Objective: In 2002, the Centers for Medicare & Medicaid Services (CMS) implemented the National Nursing Home Quality Initiative, designed to improve quality of care and provide public access to MDS-based quality measures. Pennsylvania nursing homes consistently lead the nation in reported decline in activities of daily living (ADL) (PA 24%, National 15%), while other quality measures approximate the national averages. No reason is readily apparent for the comparatively high level of ADL decline. The research objective of this project was to explore and identify nursing home characteristics associated with ADL decline in Pennsylvania nursing homes. Study Design: This was a cross-sectional study. CMSreported quality measures for 2002 were merged with facility characteristic data from the 2002 Pennsylvania Department of Health’s Long-term Care Facility Questionnaire. Characteristics explored included type of ownership (size, profit status, hospital ownership, reimbursement rates, occupancy, agreements, and special units), utilization characteristics (length of stay, admission and discharge profiles, and payor profile), resident characteristics (age, race, gender), and staffing characteristics (medical, nursing, and therapy staff hours). Univariate and multivariate analyses were employed to identify factors associated with ADL decline. Population Studied: The study population consisted of 622 nursing homes in Pennsylvania licensed by the Pennsylvania Department of Health, who completed the 2002 PA LongTerm Care Facility Questionnaire, and had ADL decline measures reported for at least one quarter in 2002. Federal facilities and intermediate care facilities were excluded. Principal Findings: Significant and non-significant associations will be presented. Factors positively associated (p less than 0.05) with ADL decline included hospital governance, occupancy level, Medicare covered days, days paid by private insurance, percentage of residents admitted from hospital, higher percentage of residents 65-74, and more therapy staffing hours. Factors negatively associated with ADL decline were a higher percentage of residents admitted from other nursing homes, length of stay, and a higher percentage of residents 85-94, and more nursing hours. Data was analyzed using SAS version 8.2. Conclusions: This study was the first step toward understanding ADL decline within Pennsylvania nursing homes, as well as the first time that these existing data sources were linked to explore ADL decline. The analysis showed that there was no one obvious explanatory variable, nor did the multivariate model explain much of the variance in ADL decline. Future examination of ADL decline will include collecting primary data such as nursing home practices and staff experience. In addition, we plan to examine characteristics that distinguish states from one another, such as regulatory policy, and plan to make comparisons between states and state characteristics. Implications for Policy, Delivery, or Practice: Identification of facility-level factors provides a basis for development of facility and statewide initiatives designed to address ADL decline in nursing homes. Primary Funding Source: CMS ●Regional Trends in Racial Disparities in Major Procedure use among the Elderly Ashish Jha, M.D., MPH, Elliott S. Fisher, M.D., MPH, Zhonghe Li, MA, E. John Orav, Ph.D., Arnold M. Epstein, M.D. Presented By: Ashish Jha, M.D., MPH, Assistant Professor, Health Policy & Management, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115; Tel: (617) 432-5551; Fax: (617) 648-5000; Email: ajha@hsph.harvard.edu Research Objective: Racial disparities in major procedure rates are well known and data suggest that national gaps in care have remained stable. However, given substantial regional variation in disparities, we sought to identify local regions that have successfully eliminated gaps in care. Study Design: We used 100% data file from the Medicare Part A program from 1992 through 2001 to calculate ageadjusted procedure rates for black and white enrollees 65 years of age or older. We calculated the white minus black gap for men and women separately for three procedures (coronary artery bypass graft surgery, carotid endarterectomy, and total hip replacement). Population Studied: Medicare enrollees who received procedures in hospital referral regions large enough to provide stable estimates for both blacks and whites. Principal Findings: During 1992-1994, compared with white enrollees, black enrollees received lower rates of all three procedures in all 158 HRRs examined (p<0.05 in 152 HRRs). During the decade of monitoring, the white minus black gap widened in 89 HRRs (p-value for trend <0.05 in 42 HRRs) and narrowed in 69 HRRs (p-value for trend <0.05 in 22 HRRs). At the end of the study period, in 1999-2001, blacks still received less of each procedure in all 158 HRRs examined (p<0.05 in 154 HRRs). Conclusions: We found that while the white minus black gap narrowed in a few regions, the large gaps present in the early 1990s persisted at the end of the decade. Implications for Policy, Delivery, or Practice: Despite the national attention on racial disparities, we found no evidence that any local regions have eliminated the racial gap in care for the three procedures examined. New efforts are needed to reduce the gaps in care between white and black Americans. Primary Funding Source: RWJF ●Follow Up Rate and Related Factors of Medical Utilization after an Abnormal Result in Korean National Cancer Screening Program Heui Sug Jo, M.D., MPH, Ph D, MS, Ji Sook Choi, M.D., MPH, Ph D, Joo Hon Sung, M.D., MS, Yong Jun Choi, M.D., MPH, DrPh D, Sang Soo Bae Presented By: Heui Sug Jo, M.D., MPH, Ph D, Assistant Professor, Dept. of Preventive Medicine, Kangwon National University, 192-1 Hyoja- Dong, Chunchon city, 200-701; Tel: 82-33-250-8872; Fax: 82-33-250-8875; Email: choice@kangwon.ac.kr Research Objective: Cancer is a primary cause of death and its prevalence and mortality rate has increased in Korea. These truths have caused economic loss because of loss of human resources and productivity not only on the country level but also on the individual level. In order to minimize the losses from cancers, secondary preventive activities, which can raise the survival rate through early detection and treatment as well as primary preventive activities which reduce the risk factors, should be performed progressively. The Korean government has provided organized cancer screening programs such as a gastro-scope, a Colono-scope, a mammogram, and a cervical pap-smear test to the people enrolled in the Korean national health insurance system since 1999 through National Health Insurance Corporation (NHIC) cancer screening program. There is not, however, a management program for people with abnormal results in screening tests but only an announcement of the results to an individual exists. The purposes of this study are to identify the follow up rate and to analyze related factors of medical utilization among the people with an abnormal result in cancer screening tests through the Korean National Cancer Screening Program. Study Design: We studied cross sectional study. We used three kinds of data; the Korean normative cohort data, the data from cancer screening tests of NHIC, and medical insurance claim data from hospitals. Also we set up an individual serial data file through merging the data according to their Korean identification number. Population Studied: The Korean medical insurance cohort includes 1,205,470 people who were selected representatively according to the same rate of age, sex, and region registered in the Korean national health insurance system from 1996 to 2001. Finally, 8,284 people in the Korean cohort took a screening test from January 1, 2001 to December 31, 2001. There were 2,647 people of these who received abnormal results from cancer screening tests. Medical utilization for diagnosis was defined as medical care according to the problems with organs related to those examined through that cancer screening test as well as the International Classification of Disease code within one year after the announcement of abnormal results. A chi-square test was performed to identify the differences in follow-up rates by kinds of cancer tests, gender, age, economic state, and amount of time before follow-up. A logistic regression analysis was also performed to examine variables that account for the differences. Principal Findings: 1. Approximately 54.6% of people with an abnormal result did not utilize medical care after notification of an abnormal result. Concerning each cancer form, 44.4% of the people with an abnormal result for stomach cancer, and 52.1%, 51.8%, and only 41.8% results for colorectal cancer, breast cancer, and cervical cancer respectively used medical care after abnormal results. 2. Furthermore, medical utilization of people with an abnormal result within 1 month after the notification was very low; 38.4% for stomach cancer, and 32.0%, 37.9%, and 28.2% of the people with an abnormal result for colorectal cancer, breast cancer, and cervical cancer, respectively. 3. If all other factors are controlled, female was more likely to use follow up care after an abnormal result (odds ratio=1.47, 95% confidence interval: 1.22-1.78). Older people over the age of 60 were also more likely to use followup care (odds ratio= 1.61, 95% confidence interval: 1.05-2.58). Conclusions: In conclusion, the follow-up rate after an abnormal result was very low. It can be one of the factors that may cause the low effectiveness of the National Cancer Screening Program. Due to the high coverage rate of medical insurance, owing to the national health insurance system, barriers to medical care are not a serious problem to people with a low incom Implications for Policy, Delivery, or Practice: To Improve the effectiveness of National Cancer Program, younger people should be targeted as groups needing assistance in adhering to follow-up recommendations. Primary Funding Source: This work was supported by Health technology Planning & Evaluation Project Grant, No. 01PJPG1-01CH10-0007. ●Quality Control of Disease Coding Error on Health Insurance Claim Data - The Korean Burden of Disease ProjectHeui Sug Jo, M.D., MPH, Ph.D., Joo Hon Sung, M.D., MPH, Ph.D., MS, Ji Sook Choi, MS, Moon Sun Hwang, Heon Jae Jeong, M.D. Presented By: Heui Sug Jo, M.D., MPH, Ph.D., Assistant Professor, Dept. of Preventive Medicine, Kangwon National University, 192-1 Hyoja- Dong, Chunchon city, 200-701; Tel: 82-33-250-8872; Fax: 82-33-250-8875; Email: choice@kangwon.ac.kr Research Objective: We attempted to verify and to improve the quality of disease coding on health insurance claim data used in assessing of Korean Burden of Disease. Also, we understood health care organizational characteristics and patient’s characteristics affecting to coding error. Study Design: 1. We organized reviewer group consisting of clinical specialists and researchers of health care management. The target diseases for validation are epilepsy, coronary heart disease, cerebro-vascular disease, rheumatoid arthritis, asthma, lumbar hernrated disc disease, frequently utilized disease . 2. Considering the amount of the medical service use, the hospitals to be examined were selected by stratified sampling; onsequently, 30 specialized general hospitals, 52 general hospitals, and 49 hospitals were chosen(a total of 131 medical facilities). 3. The validation of data was performed by turning out the agreement rates among the diagnostic code of insurance claim, diagnosis based upon medical records, and judgement by medical recorders. 4. The medical records and electric database of total 5,842 cases within our Korean disease cohort -700 of coronary heart diseases, 892 of stroke, 977 of rheumatoid arthritis, 399 of systemic lupus erythematous disease, 787 of bronchial asthma, 834 of epilepsy, and 1253 of the frequent diseases were checked from May 2003 to December 2003. Principal Findings: 1. The accuracy rates were 79.7% in epilepsy, 61.4% in coronary heart disease, 75.8% in cerebrovascular disease, 84.1% in rheumatoid arthritis, 52.6% in asthma, 81.0% in lumbar herniated disc disease, 84.1% in frequently utilized diseases. 2. Agegroup, utilization type, type of operation, Admission length,and medicare expenditure are related factor. 3. We also developed adjustment Model. Conclusions: Our results show that the validity of data is various among diseases,and also suggest that there are many related factors related validity. As a results, our health insurance cohort through validation and adjustment can be used in broad researches of health care management area such as, assessment of health status and health related quality, choice of priorities in health sector, allocation of resources, and international comparative study. Primary Funding Source: This work was supported by Ministry of Health and welfare in Republic of Korea. Grant No. 01-PJPG1-01CH10-0007 ●Doctors’ Attitude and Satisfaction about the Review System of National Health Insurance Claims in Korea Heui Sug Jo, M.D., MPH, Ph.D., Heon Jae Jeong Presented By: Heui Sug Jo, M.D., MPH, Ph.D., Assistant Professor, Dept. of Preventive Medicine, Kangwon National University, 192-1 Hyoja- Dong, Chunchon city, 200-701; Tel: 82-33-250-8872; Fax: 82-33-250-8875; Email: choice@kangwon.ac.kr Research Objective: The purposes of this study are to understand the doctors’ attitude and satisfaction about the review system of national health insurance claim in Korea and to suggest the way to improve this system. Study Design: The national health insurance system in Korea has adopted fee-for-service payment system and has performed retrospective review system about claimed service as the method of quality control and evaluation of appropriate services. However, as review system has been carried out, it has been engaged in a policy for the restriction of medical service expenditure, and consequently in a policy of the curtailment of medical service claim; therefore, considering doctors’ perspective, this review system is thought to be very distorted from the quality control policy originally intended. Moreover, the fact that the clear criteria and cases of the criteria have not been offered, and the complex and long-term needed procedure for interposing objections have been suggested as the obstacles for doctors to better understanding of the review system. Population Studied: This study conducted a survey of the doctors registered in the medical association in Seoul city. The survey was performed as a form of self-administered questionnaire form January 2004 to February 2004. The contents of questionnaire dealt with doctors’ attitude and satisfaction about the review system of medical service claim. Principal Findings: 536 doctors opening their clinics replied. 1. About the influences of the review system on improvement of medical service quality, 88.6% of repliers showed a negative attitude. 2. 97.2% of repliers have had experiences that they have given distorted-insufficient-medical services in order to evade the curtailment of service claim. Also, 91.4% of repliers stated that they have had experiences of intentional modification or alteration of diagnostic code to shun the curtailment. 3. Moreover, 50.5% could not raise an objection owing to the complex administrative procedure and loss of time; 23.2% did not even check out the causes of their curtailment. 4. 32.7% of doctors said that they could not gain any information about the criteria of curtailment, and 64.4% were in state of being partially aware of the curtailment criteria. 5. For 47.6%, it was from the peer group of doctors that doctors had obtained information about the criteria of curtailment; however, merely 7.7% of doctors have acquired information from formal education materials. Conclusions: Most of the doctors showed negative attitude to the curtailment procedure and the review system of service claim originally intended to be one of the quality control methods of medical service in Korea; also, the development of both scientific and reasonable parameters and criteria for claim is needed. Implications for Policy, Delivery, or Practice: 1. Through the improvement of review system for appropriate medical service, there is a need of a way to increase the satisfaction of medical service providers, and to encourage the motivation for quality control. 2. Education is strongly needed to provide doctors with sufficient information about review criteria and curtailment cases. Primary Funding Source: No Funding Source ●Factors Associated with Screening Behavior of Stomach Cancer Based on Socio-Ecological Model: Korean National Study Heui Sug Jo, M.D., MPH, Ph.D., Dong Hyun Kim, M.D., Ph.D., Sang Soo Bae, M.D., Ph.D., Hun Jae Lee, Ph.D., Hyung Won Cho, M.D.. MS, Yong Jun Choi Presented By: Heui Sug Jo, M.D., MPH, Ph.D., Assistant Professor, Dept. of Preventive Medicine, Kangwon National University, 192-1 Hyoja- Dong, Chunchon city, 200-701; Tel: 82-33-250-8872; Fax: 82-33-250-8875; Email: choice@kangwon.ac.kr Research Objective: Stomach cancer is one of the most frequent cancers in Korea. In order to reduce the mortality and occurrence rate of stomach cancer, early detection through cancer screening, proved to be effective especially in stomach cancer, as well as primary preventive activities reduce the risk factors, should be recommended intensively. The purposes of this study were to investigate screening rates of stomach cancer and to analyze factors associated with screening behavior based on socio-ecological model. Study Design: A socio-ecological model was used as the theoretical model for variables, which included the intrapersonal level factors such as anxiety for cancer, individual value on efficiency of test, economic burden for test, difficulty of accessibility, interpersonal level such as familial recommendation, familial tendency for regular check-up, doctors recommendation, and societal level such as recognition and knowledge for national cancer screening, experience of health education, social norm about cancer screening. Population Studied: Screened (n=600) and non-screened (n=800) people were surveyed by telephone interview. The sample was selected by random digit dialing method according to the same rate of age, sex, and region in Korean population. Principal Findings: For intra-personal factors, anxiety for cancer (odds ratio=1.62, 95% confidence interval: 1.07-2.45) was associated with having been screened. For interpersonal factors, people with familial member who regularly have a check-up (odds ratio=2.10, 95% confidence interval: 1.47-2.99) were more likely to have been screened. In addition, many experiences of health education about importance of cancer screening test were associated with having been screened on the societal level (odds ratio=4.45, 95% confidence interval: 1.30-15.22). Conclusions: These findings suggest that aggressive education and promotion should be performed to improve the cancer screening rates. Also multi-disciplinary approaches should be favored for improving rates based on socioecological model. Implications for Policy, Delivery, or Practice: The socioecologic framework is a good explanatory model for stomach cancer screening behavior. These results can guide intervention amied at ncreasing cancer screening rates. Primary Funding Source: Supported by Korean National Cancer Institute Grant. no 0320460-1 ●Practice-Based Electronic Billing Systems and their Impact on Immunization Registries Maureen Kolasa, RN, MPH, Janet Cherry, MA, Andrew Chilkatowsky, BA, David Reyes, MS, MPH, James Lutz, MPA Presented By: Maureen S Kolasa, RN, MPH, Lead Epidemiologist, National Immunization Program, Centers for Disease Control and Prevention, 1600 Clifton Road, E52, Atlanta, GA 30333; Tel: (404)639-8759; Fax: (404) 639-8615; Email: mkolasa@cdc.gov Research Objective: Many providers of children’s immunizations rely on electronic billing systems to report information to computerized immunization registries. If billing systems fail to capture some administered immunizations, the registry will not reflect a child’s true immunization status. This study assessed differences between immunizations administered and immunizations reported to a registry via provider’s electronic billing systems in Philadelphia, Pennsylvania. Study Design: In Philadelphia, 45 private and public immunization practices served >50 children age 7-35 months and submitted immunization data to the registry via electronic billing systems in 2002-2003. The Philadelphia Department of Public Health (PDPH) conducted chart audits for all children who had received their last known immunization at one of these 45 practices and for whom the registry identified as missing at least one recommended immunization. Chart records were compared to registry records to identify immunizations administered in these practices but not reported to the registry by electronic billing systems. Lost revenue was calculated conservatively by considering only potential reimbursement of administrative fees. Population Studied: All 45 immunization practices serving > 50 children age 7-35 months and using electronic billing systems to report data to Philadelphia’s immunization registry were evaluated. A total of 20,611 children, about 43% of all children age 7-35 months in Philadelphia, had at least one immunization recorded in the immunization registry and had received their last known immunization at one of these 45 practices. Principal Findings: The registry identified 12,321 (60%) of the 20,611 children served by the 45 study practices that were missing recommended immunizations. Of the 256,969 immunizations administered to the 20,611 children in the study practices, 62,213 (24%) were not in the registry. The electronic billing systems submitted all administered immunizations for 69% of immunization visits, some but not all for 11% of visits, and none for 20% of visits. Pneumococcal conjugate vaccine was the vaccine most frequently administered but not reported to the immunization registry. Immunizations administered but not billed cost these providers up to $980,477 in lost revenue from administrative fees alone. Conclusions: When providers used electronic billing systems to enter data into Philadelphia’s immunization registry, the electronic billing systems did not report almost 25% of immunizations administered, resulting in incomplete registry data. Improvement of billing data quality would result in more complete registries, higher reported immunization coverage rates, and recovered revenue for immunization providers. Implications for Policy, Delivery, or Practice: A national Healthy People 2010 objective is to increase the proportion of children <5 years fully participating in an immunization registry to 95%. In order to meet this objective, policies and interventions are needed to improve the completeness of data reported to immunization registries via electronic billing systems. Primary Funding Source: CDC ●Evaluation of Physician Compliance with Pneumococcal Conjugate Vaccine Shortage Recommendations within a Managed Care Organization in 2004 Maureen S Kolasa, RN, MPH, Stephen M. Tannenbaum, M.D. Presented By: Maureen S Kolasa, RN, MPH, Lead Epidemiologist, National Immunization Program, Centers for Disease Control and Prevention, 1600 Clifton Road, E52, Atlanta, GA 30333; Tel: (404)639-8759; Fax: (404) 639-8615; Email: mkolasa@cdc.gov Research Objective: Shortages of pneumococcal conjugate vaccine (PCV) were experienced in the first half of 2004. To maintain adequate vaccine for high risk children, the Advisory Committee for Immunization Practices recommended delay of administration of the 3rd and 4th doses of PCV (PCV3 and PCV4, respectively) to healthy children, but no delay in administration of the 1st and 2nd doses of PCV (PCV1 and PCV2, respectively). This study assessed the pattern of PCV administration during periods of vaccine shortage and changing recommendations in a managed care setting. Specific research objectives were to determine physician compliance with PCV shortage recommendations, describe the overall pattern of vaccination of children with high risk diagnoses and whether this was in accordance with recommendations, and demonstrate the usefulness of Kaiser registry data in assessing immunization administration patterns related to vaccine shortages. Study Design: Using their immunization registry, Southern California Kaiser Permanente evaluated PCV doses administered to all enrolled children at age 3 months, 5 months, and 7 months in 2004. PCV coverage prior to shortage-related recommendations was compared to coverage after recommendations were made to discontinue PCV dose 3 and 4 to healthy children. PCV coverage of children with diagnoses placing them at high risk for pneumococcal disease will be compared to PCV coverage for children without high risk diagnoses. Population Studied: This study included all Kaiser Permanente providers in Southern California that immunize children < 17 months of age. All children < 17 months of age, enrolled with Southern California Kaiser Permanente in 2004, and receiving at least one immunization from Kaiser Permanente providers in Southern California were included in the analysis. Principal Findings: Immunization coverage levels with PCV1 for children age 3 months remained similar (range 80-86%) between January and July 2004. The same pattern was seen with PCV2 for children age 5 months. PCV3 immunization coverage for children age 7 months dropped significantly (p< .0001) from 73% coverage in February to 17% coverage in April, one month after ACIP recommendations to suspend this dose to healthy children. PCV3 coverage among 7 month olds remained <20% through July 2004. Conclusions: Southern California Kaiser Permanente immunization providers decreased delivery of PCV3 to enrolled children in accordance with ACIP recommendations. These providers demonstrated rapid adoption of recommendations and maintained compliance with the recommendations through July 2004. Implications for Policy, Delivery, or Practice: The methods used by the Advisory Committee for Immunization Practices and by Kaiser Permanente Advisory in Southern California were effective in communicating vaccine-related recommendations to Kaiser Permanente providers. Use of these same methods therefore may promote compliance with future vaccine-related recommendations. On a programmatic level, this type of study can assist managed care organizations in identifying and intervening with providers not following recommendations. Primary Funding Source: No Funding Source ●Children Victims of Homicide are Unlikely to use Injury Related Health Services Prior to the Fatal Event Angelique Stubblefield, BSN, MPH, Siran Koroukian, Ph.D. Presented By: Siran Koroukian, Ph.D., Assistant Professor, Epidemiology and Biostatistics, Case Western Reserve University, School of Medicine, 10900 Euclid Avenue, Cleveland, OH 44106-4925; Tel: (216)368-3197; Fax: (216)3683970; Email: skoroukian@case.edu Research Objective: Numerous studies have reported increased incidence of violence induced fatalities or VIF among socioeconomically disadvantaged subgroups of the population, especially among children. However, the extent to which victims of violence could be identified through their pattern of utilization of health services is unknown. This study aims at describing patterns of use of injury related services by Ohio Medicaid children victims of homicide, in the period preceding the fatal event. Study Design: This was a retrospective study using linked Ohio Medicaid and death certificate files, 1992 to 1996. All Medicaid children under age 10 and victims of homicide as identified through the cause of death documented in the death certificate files, were included in the analysis, with a final sample of 138 children. Bivariate associations were tested using chi square tests to assess statistical significance in the difference between proportions. Odds ratios were obtained using multivariable logistic regression analysis, adjusting for demographics and months of enrollment in Medicaid prior to the fatal event. Population Studied: All Medicaid children under age 10 and victims of homicide as identified through the cause of death documented in the death certificate files, were included in the analysis. Principal Findings: The yearly average of violence induced fatalities during the study period was 27.6 deaths. The majority,70 percent of decedents had no prior contact for injury related services and 86 percent were 0 to 4 years old. Fatal child abuse, fatal assault, and death by firearms were listed as the cause of death on the death certificate respectively for 46 percent, 37 percent, and 17 percent of the victims. Children age 0 to 4 were more likely than their older counterparts not to have had previous contact with the health care system for injury related services after controlling for demographics and length of enrollment in Medicaid prior to the fatal event. The adjusted odds ratio is 4.11 with a 95 percent Confidence Interval of 1.03 to 16.5. Those residing in Metropolitan, Appalachian, and Suburban counties were 14.8 times more likely not to have received injury related services when compared to those living in rural counties, with a Confidence Interval of 3.2 to 69.1. Conclusions: We estimated that a rather small proportion, thirty percent, of children may have been in contact with the health care system prior to the fatal event for injury related services, a statistic that translates into a limited window of opportunity for health care providers to prevent such events from happening. Implications for Policy, Delivery, or Practice: The analysis of patterns of use of injury related services in the period preceding the fatal event may be used to develop a risk assessment tool for health practitioners to identify children at risk for becoming victims of homicide, and hopefully prevent VIFs. Although our study indicated that a rather small proportion of children have any contact with the health care system for injury related services, this finding should by no means impede further research efforts in this direction, as all such deaths are preventable. Primary Funding Source: Ohio Department of Public Safety ●The Impact of Interstate Migration Patterns on Nursing Workforce Planning Efforts Linda M. Lacey, MA Presented By: Linda M. Lacey, MA, Associate Director: Research, North Carolina Center for Nursing, 222 North Person Street, Raleigh, NC 27601; Tel: (919) 715-3523; Fax: (919) 715-3528; Email: lmlacey@northcarolina.edu Research Objective: Policy decisions are currently being made to enlarge nursing education programs in order to attenuate the expected shortage that will develop in the next 10 years. Given the expensive nature of these programs it is important for states to be able to accurately estimate the need for expansion. The objective of this research was to understand the relative impact of interstate migration versus production of new professionals in the growth of the nursing workforce at the state level. Study Design: Repeated cross-sectional analyses of annual RN licensure files were conducted to identify newly licensed RNs and to determine whether they were new graduates and where they had received their entry-to-practice education. Eleven years of data (from 1991 - 2001) were examined. Population Studied: This study focused on the population of newly licensed registered nurses acquiring an initial license to practice in North Carolina during the period from 1991 to 2001. Principal Findings: During the 1990s, approximately 6,000 7,000 RNs each year obtained an active license to practice in North Carolina for the first time. Throughout the decade, more than half of those newly licensed nurses were found to have received their entry-into-practice education in a state other than North Carolina. Conclusions: As states institute solutions to the growing nursing shortage, it becomes important to understand the dynamics operating within that labor force. Patterns of interstate in- and out-migration should be taken into account when planning for educational expansion. The impact could be either positive or negative. Implications for Policy, Delivery, or Practice: Entry-level nursing education programs are expensive both in terms of faculty manpower and clinical resources. Knowing how both in- and out-migration patterns across state lines are affecting the nursing labor force in a particular state will allow for more accurate planning and budgeting. Primary Funding Source: The Duke Endowment and the North Carolina Center for Nursing ●Estimates of Potential Health Gains from Reducing Multiple Risk Factors of Stroke in Korea Heeyoung Lee, M.D., MPH, Seok-Jun Yoon, M.D., Ph.D., Hyeong-sik Ahn, M.D., Ph.D., Ok Ryun Moon, M.D., Ph.D. Presented By: Heeyoung Lee, M.D., MPH, Principle Researcher, Department of Preventive Medicine, College of Medicine, Korea University, 126-1, 5ka Anamdong, Seongbukku, Seoul, 136-705; Tel: 82-2-920-6346; Fax: 82-2-927-7220; Email: imstone@korea.ac.kr Research Objective: We estimated the burden of diseases caused by stroke using SMPH (Summary Measure of Population Health), and compared the attributable burden of risk factors with the avoidable burden of risk factors. Study Design: Firstly, the disease burden of stroke due to premature death and disability was estimated by using morbidity and mortality data in 2001. Secondly, we selected the risk factors and exposure variables of stroke and decided the prevalence and relative risk of risk factors by systematic reviews. Thirdly, we calculated the attributable burden of stroke at the present prevalence of risk factors and the avoidable burden of stroke at the counterfactual prevalence of risk factors were calculated. Population Studied: Korean(South) people Principal Findings: The burden of stroke for a 1,000,000 population was 3322.62 person-years for males and 2532.20 person-years for females. The burden of stroke at present prevalence for males per 100,000 population was attributed to smoking (1940.41 person-years), alcohol (864.31 personyears), and hypertension (667.28 person-years). The burden of stroke at present prevalence for females per 100,000 population was attributed to alcohol (462.76 person-years), physical inactivity (455.65 person-years), and smoking (407.68 person-years). The joint population attributable fraction (PAF) by risk factors was 80.2% percent for males and 52.4% percent for females. Conclusions: For reducing the burden of stroke, a management policy for risk factors is required essentially. Modifying the risk factors reduced the burden of stroke could be more effective intervention to consider an initial consideration of the differences of risk factors, gender, and age could be more effective intervention. Implications for Policy, Delivery, or Practice: Primary Funding Source: Korea Health 21 R&D Project, Ministry of Health & Welfare, Republic of Korea ●Estimation of the Burden of Disease in Korea Heeyoung Lee, M.D., MPH, Seok-Jun Yoon, M.D., Ph.D., Sang-Cheol Bae, M.D., Ph.D., Heuisug Jo, M.D., Ph.D., Jin Yong Lee, M.D., MPH, Hycjung Chang, M.D., Ph.D., Jaehyun Park, M.D., MPH Presented By: Heeyoung Lee, M.D., MPH, Principle Researcher, Department of Preventive Medicine, College of Medicine, Korea University, 126-1, 5ka Anamdong, Seongbukku, Seoul, 136-705; Tel: 82-2-920-6346; Fax: 82-2-927-7220; Email: imstone@korea.ac.kr Research Objective: We estimated the burden of diseases in Korea using disability adjusted life year(DALY), a composite measure of premature mortality and disability that equates to years of healthy life lost. Study Design: We calculated DALYs of 123 diseases for the year 2002. The burden of disease due to premature death was estimated by using years life lost due to premature death(YLLs) measurement developed by the global burden of disease study group. For the calculation of the years lived with disability(YLD), the following parameters were estimated in the formula; incidence rate, mortality, prevalence rate and disability weight of disease. Population Studied: Korean(South) People Principal Findings: By disease category, DALYs per 100,000 population were attributable mainly to cancer(1,525 personyear), cardiovascular disease(1,492 person-year), digestive disease(1,140 person-year), diabetic mellitus(970 person-year), respiratory disease(951 person-year). The leading five diseases of DALYs for males per 100,000 population in Korea were diabetic mellitus(1020 person-year), cerebrovascular accident(937 person-year), cirrhosis of the liver(671 personyear), asthma(663 person-year), ischemic heart disease(601 person-year). For females, the leading five diseases of DALYs per 100,000 population were diabetic mellitus(919 personyear), cerebrovascular accident(900 person-year), peptic ulcer disease(794 person-year), asthma(755 person-year), rheumatoid arthritis(531 person-year). Conclusions: These results presented a substantially different ranking of disease burden than did mortality rates alone. Also, we found differences in the rank order by gender and age group. Implications for Policy, Delivery, or Practice: We found the DALY method emplyoed was appropriate to quantify the burden of disease. Thereby, it would provide a rational bases to plan a national health policy regardiing the burden of diasese in Korea. Primary Funding Source: Korea Health 21 R&D Project, Ministry of Health & Welfare, Republic of Korea ●Predictors of Prescription Drug Skimping among Enrollees in SeniorCare Musetta Leung, MS, Donald S. Shepard, Ph.D., William Stason, M.D, MS, Grant Ritter, Ph.D. Presented By: Musetta Leung, M.S., The Heller School for Social Policy and Management, Brandeis University, MS 035, Waltham, MA 02454-9110; Tel: (617) 699-5239; Email: mleung@brandeis.edu Research Objective: Increasing concerns over prescription drug affordability has prompted states to adopt measures that ensure access among low-income populations. Previous research has shown that people who could not afford prescriptions often did not fill their prescribed medicines, “split pills,” skipped doses, or went without other lifenecessities in order to pay for drugs. This study evaluated the predictors of prescription drug skimping, defined as skipping doses or not filling prescriptions for financial reasons, among elders enrolled in two states’ pharmacy assistance programs (PAPs). Study Design: In mid-2002, Illinois and Wisconsin implemented “SeniorCare,” the states’ Pharmacy Assistance Waivers (Pharmacy Plus) program that provided low-income persons aged 65+ with Medicaid-funded prescription drug assistance. A random sample of SeniorCare enrollees participated in a survey, which asked pre- and post-SeniorCare questions on topics including drug utilization, affordability, and insurance. Questions on health encompassed selfperceived health status, as well as number of conditions that were later defined as perceived life threats (e.g. heart disease, cancer) or not (e.g., osteoporosis, ulcer). Multivariate logistic regression was used to assess predictors of skimping. Population Studied: A random sample of communitydwelling SeniorCare enrollees was interviewed via telephone by an academic survey research organization in Spring 2004. Proxy respondents (7%), such as a family member, were accepted, and a response rate of 61% was obtained. All enrollees with non-missing data were included in the analysis. Principal Findings: A total of 1,824 respondents (54% Wisconsin and 46% Illinois) were included in the analysis. The average age was 77 years, and the majority of respondents was white (84%), female (73%) and had annual household income at or below 160% of federal poverty level (66%). On a scale from 1 to 7 (very poor to excellent), the mean health status was 3.7, and respondents reported having an average of 3 medical conditions. Fifty-seven percent of enrollees had coverage through private health insurance plan prior to PAP, and 19% of the entire sample reported using private insurance to pay for medications. Logistic regressions (odds ratios with 95% confidence intervals) showed that having more comorbid conditions was associated with a higher probability of skimping (OR=1.3, CI=1.18-1.44), but having life-threatening conditions was associated with less skimping (OR=0.83, CI=0.71-0.98). Being less poor, older, having better selfperceived health, and having private drug insurance were all associated with less medication skimping (OR=0.49, CI=0.350.70; OR=0.94, CI= 0.93-0.96; OR=0.80, CI=0.70-0.90; OR=0.69, CI=0.50-0.95, respectively). Gender, marital status, and race did not have a statistically significant effect on preSeniorCare medication skimping. Conclusions: Prescription drug skimping was more prevalent among those who were poor, uninsured, and sicker (i.e., those with worse self-perceived health and more comorbid conditions). Having life-threatening conditions, however, was associated with better medication adherence. Implications for Policy, Delivery, or Practice: Understanding who is at risk of medication non-adherence may help policy makers to better design programs and target them to persons with the greatest need of assistance. Primary Funding Source: CMS ●“Do the Conclusions Look as Good as They Seem?” A Review of Quality Improvement Intervention Studies Linda Li, B.Sc.(PT), Ph.D., Lorenzo Moja, M.D., MSc, Alberto Romero, M.D., MSc, Jeremy Grimshaw, M.D., Ph.D. Presented By: Linda Li, B.Sc.(PT), Ph.D., Post-Doctoral Fellow, Clinical Epidemiology Program, Ottawa Health Research Institute, 1053 Carling Avenue, Administration Building, Room 2-010, Ottawa, Ontario, K2B 7T4; Tel: (613) 798-5555x19749; Fax: (613) 761-5402; Email: lli@ohri.ca Research Objective: To assess the appropriateness of conclusions reported in recently published quality improvement (QI) intervention studies in relation to the study design. Study Design: This study consisted of a systematic review and an expert rating panel. We hand-searched 11 major medical journals or health services research journals for randomized and non-randomized evaluations of QI interventions (RCTs, non-RCTs) published between January 2002 and December 2003. Eligible studies were those evaluating interventions that aimed to change health professional behaviours based on research evidence. Two independent reviewers extracted data for each trial, including study characteristics, methodology, and all statements addressing the causal effect between the intervention and outcomes in the abstract and the main text. Population Studied: A 38-member expert panel was assembled to assess the strength of causal inference suggested by the causal statements. This panel consisted of clinical epidemiologists, clinical trial methodologists, health services researchers, and clinical researchers. Half of the panelists were English first-language speaking. They rated the strength of causality of each quote on a Likert scale (range 1-7, higher = stronger causal relationship), assuming that all quotes were from well designed RCTs. Each panelist rated 60 to 70 randomly assigned quotes and each quote was rated by 10 to 12 panelists. Student’s t-tests were used to compare the ratings from RCTs and non-RCTs. Separate analyses were done for statements recorded from the abstract and those recorded from the main text. Subgroup analyses were conducted for studies that reported statistically significant results and those reported no effect or mixed results. Principal Findings: Of the 4543 titles hand-searched, 73 articles were included (RCTs = 38; non-RCTs = 35) and 207 causality quotes were extracted (abstract = 68; main text = 139). 7 studies had no abstract quote and 5 had no main text quote. In studies where more than one quote was extracted, only the highest score was used since the strongest causal statement provided the most definitive message to readers. Hence, ratings of 66 abstract quotes (RCTs = 34; non-RCTs = 32) and 68 main text quotes (RCTs = 34; non-RCTs = 34) were included in the analysis. Preliminary analysis was conducted with the 29 responses that we have received thus far. Among the abstract quotes, the mean causality rating was significantly higher in the non-RCTs (5.09 ± 1.12, versus 4.02 ±1.62 in RCTs; p=0.03). 26.5% of RCTs versus 56.3% of non-RCTs reported positive results in the abstract (p=0.01). A similar trend was found in the main text quotes (RCTs = 4.60 ± 1.62; non-RCTs = 5.22 ± 1.29; p=0.09). In the subgroup analysis, a statistically significant difference was only found in studies reporting no effects or mixed results in the abstracts (RCTs = 3.38 ± 1.34; non-RCT = 4.50 ± 1.26; p=0.015). Conclusions: RCTs are the gold standards for examining causal relationships between interventions and outcomes. Conclusions from non-RCTs are expected to be more conservative due to intrinsic weaknesses of the study design. However, our preliminary results suggest that non-RCTs evaluating QI interventions might have overstated the strength of causality in their abstracts and main text. Implications for Policy, Delivery, or Practice: QI studies often employ quasi-experimental or observational designs. Conclusions of these studies can be misleading to policy makers and clinicians if authors are overzealous in stating the causal relationship. Findings of this study will be the first step in improving the appropriateness of conclusions stated in QI intervention studies. Primary Funding Source: No Funding Source ●Predictors of Adherence to Colorectal Cancer Screening Recommendations Su-Ying Liang, Ph.D., Kathryn A Phillips, Ph.D., Jennifer Haas, M.D., Mika Nagamine, Ph.D. Presented By: Su-Ying Liang, Ph.D., Assistant Research Professor, Clinical Pharmacy, University of California, San Francisco, 3333 California Street, Suite 420, San Francisco, CA 94143; Tel: (415)514-0457; Fax: (415)502-8271; Email: sliang@itsa.ucsf.edu Research Objective: The objective of this study was to examine the associations of health plan and individual characteristics with adherence to colorectal cancer (CRC) screening recommendations and whether the associations differ between specific types of CRC tests. Study Design: Data were from the 2000 National Health Interview Survey. The main outcome measure was selfreported adherence with CRC screening within the recommended timeframe (i.e., , having home fecal occult blood test (FOBT) in the prior year or sigmoidoscopy in the last 5 years or colonoscopy in the last 10 years). The primary independent variable was the referral requirements of a health plan – the one specific health plan characteristic that may be relevant to CRC tests. Other individual level variables included were individual socioeconomic status (age, gender, household income, education), types of health plans (private vs public, Medicare, Medicaid, private managed care vs private nonmanaged care, public managed care vs public non-managed care), having a usual source of care, and clinical risk factors (family history of colorectal and non-colorectal cancer, personal history of non-colorectal cancer). Sub-analyses were conducted for each of the three CRC tests. All analyses were conducted using sampling weights to reflect the U.S. civilian, non-institutionalized population and standard error adjustment to account for complex survey design. Population Studied: The population contained a nationally representative sample of adults ages 50 and over without a history of CRC. Our sample included 11,574 individuals who meet our selection criteria. Principal Findings: Thirty-seven percent of individuals had received CRC screening consistent with current recommendations. Among those ever screened, 69% was adherent to the screening recommendations. Specifically, 14% was adherent using combinations of CRC tests while 17%, 11%, and 27% were adherent to FOBT only, sigmoidoscopy only, and colonoscopy only, respectively. We found that enrollment in plans with referral requirements and enrollment in managed care plans were not significantly associated with the overall adherence measure but their associations with adherence to specific types of CRC tests were different. Individuals enrolled in plans with referral requirements were less likely to be adherent to colonoscopy guidelines (p=0.002) than those without referral requirements, while individuals enrolled in plans with referral requirements had a trend to be adherent to FOBT (p=0.11) and sigmoidoscopy (p=0.22) guidelines. Individuals enrolled in public managed care plans were more likely to be adherent to FOBT (p=0.08) and sigmoidoscopy (p=0.02) guidelines than those enrolled in public non-managed care plans, while the trend was reversed for colonoscopy (p=0.28). As for other individual level variables, we found in our bivariate analyses that individuals were more likely to be currently adherent if they were male (p=0.07), race/ethnicity being White (p=0.09), non-Medicaid beneficiaries (p=0.09), having a personal history of non-CRC cancer (p=0.02), having a family history of colorectal cancer (p=0.08), and having a usual source of care (p=0.04). Conclusions: The overall adherence rate was low (37%), however, the adherence rate of those ever screened was high (69%). Most people who get screened are up-to-date on screening so the adherence rate of all eligible adults including never screened gives a less optimistic picture than may be warranted. In addition, we found one health plan characteristic - referral requirements - was negatively associated with adherence to colonoscopy guidelines while this specific health plan characteristic did not have a negative impact on adherence to FOBT and sigmoidoscopy guidelines. Other individual level predictors of overall adherence were similar and consistent to the literature. Implications for Policy, Delivery, or Practice: Understanding the predictors of utilization of and adherence to CRC recommendations can help identify and reduce barriers to health care. Furthermore, it is important to understand whether specific health plan structures have differential impacts on various types of CRC tests. Insurance benefit designs that may wish to promote specific screening tests should take into account the potential differential impacts of health plan organizational structures on these tests. Primary Funding Source: NCI ● Implementing a Voluntary Prescribing Error Reporting System in Outpatient Primary Care Practices Benjamin Littenberg, M.D., Amanda Kennedy, PharmMD Presented By: Benjamin Littenberg, M.D., Professor of Medicine, University of Vermont, 371 Pearl Street, Burlington, VT 05401; Tel: (802) 847-8268; Fax: (802) 847-0319; Email: benjamin.littenberg@vtmednet.org Research Objective: To design a voluntary outpatient prescribing error reporting system for primary care practices, implement the system in a sample of practices, analyze the reports, and provide feedback to prescribers about reported errors. Study Design: We asked triage nurses and office staff to submit copies of all phone or fax communications with community pharmacists about prescribing errors for inclusion in our database. Reports could be telephoned, mailed, or collected weekly by a courier. We did not specify a standard reporting form because we wished to minimize office intrusion and encourage participation. Although the reporting system was designed to be a job function of the triage nurses and office staff, prescribers were encouraged to report their own errors. Population Studied: Seven outpatient primary care practices in Northwestern Vermont. Principal Findings: All 7 practices have contributed reports and the majority of practices have submitted reports via courier. To date, 143 prescription error reports have been submitted to the database. Only 6 reports were contributed directly by prescribers. Preliminary analysis of the reports reveals 43 omission errors, 52 commission errors, 17 clarifications due to illegibility, 10 pharmacy errors, and 14 pharmacist clarifications that were not errors. 7 reports could not be categorized due to missing data. Reports most frequently concerned antidepressants (33/143) and antihypertensives (22/143). 43 reports involved problems with strength, including 11 prescriptions written for strengths not commercially available. At least 10 reports described errors with the potential to harm patients if not detected or corrected by pharmacists prior to dispensing. Conclusions: Implementing a voluntary outpatient prescribing error reporting system using triage nurses and office staff is feasible. Alternative methods for submitting reports need to be explored, as courier services are expensive. Striking a balance between data collection and office burden requires continued assessment. Implications for Policy, Delivery, or Practice: Many of the errors described in the reports can be prevented using inexpensive, available, and accessible tools. Continued surveillance, with deeper analysis of specific reports, and the development of appropriate mechanisms for feedback to prescribers require continued research. Primary Funding Source: AHRQ ●Care Fragmentation and Psychiatric Illness is Associated with Increased Emergency Department Use Among Complex Diabetic Patients Connie Liu, M.D./Ph.D. Candidate, Andrew O'Connor, DO, MPH, Doug Einstadter, M.D., MPH, Randall Cebul, M.D. Presented By: Connie Liu, M.D./Ph.D. Candidate, Department Epidemiology and Biostats, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, OH 44106; Tel: (216)650-2745; Email: cwl6@case.edu Research Objective: The care of complex diabetic patients often is fragmented among multiple specialists. The objective of the current investigation is to determine if the presence of psychiatric (Psych) comorbidity among diabetic patients aggravates their already fragmented care and results in excessive use of the Emergency Department (ED). Study Design: This is a retrospective analysis of ED use in diabetic patients with chronic kidney disease (CKD). Diabetes, CKD, Psych and non-Psych comorbidities, visit adherence, a1c levels and ED visits were identified from visit records, recorded lab values and ICD-9-CM codes from the system´s electronic medical record; Psych comorbidities were classified as mood/anxiety, psychotic, or “other” disorders. “Fragmentation” was defined as the number of separate nonpsych specialty clinics visited. A negative binomial model was used to estimate the effect of Psych disorders and Fragmentation on the number of ED visits after adjusting for age, sex, ethnicity, number of non-Psych comorbidities, diabetes control (average a1c level over 2 years), adherence (proportion of “kept” primary care appointments), and insurance (commercial, Medicare, Medicaid, uninsured) status. Population Studied: Diabetic continuity patients with chronic kidney disease (CKD, estimated GFR 20-60) seen at 10 primary care practices in a large urban health care system over a 2-year period. "Continuity" was defined as having at least one primary care visit per year within the system during the study period. Principal Findings: Of 3873 patients with diabetes seen during 2002-2003, 623 (16.1%) had CKD, of whom 241 (38.7%) had one or more Psych co-morbidities. Psych patients were younger (66.7 yrs vs. 69.7 yrs, p < 0.0005), had more non-Psych comorbidities (2.92 vs. 2.56, p<0.0003), poorer adherence (84.7% vs. 87.8% kept primary care visits, p<0.004), and a greater level of Fragmented care (1.77 vs. 1.47 specialty clinics visited, p<0.0009) than non-Psych patients. On bivariate analysis, the 141 (22.6%) diabetic CKD patients with mood and anxiety disorders were more likely than those without to have at least one ED visit (63.1% vs. 44.6%; OR 1.42, 95% CI, 1.21 to 1.66). After adjusting for age, sex, ethnicity, non-Psych comorbidities, diabetes control, adherence, and insurance, the presence of a mood/anxiety disorder increased the predicted number of ED visits by a factor of 1.50 (95% CI: 1.14 to 1.98), while greater Fragmentation increased the number of ED visits by a factor of 1.32 (95% CI: 1.19 to 1.47) for each additional clinic visited. Psych diagnoses other than mood or anxiety disorders were not associated with increased number of ED visits. Conclusions: Mood and anxiety disorders are prevalent among patients with complicated diabetes, and are associated with increased fragmentation of care and increased use of the ED. Implications for Policy, Delivery, or Practice: We believe that fragmentation of non-psychiatric care further exacerbates the discontinuous nature of psychiatric care, and may contribute to poorer self-management of diabetes and its complications. Primary Funding Source: No Funding Source create bad medicines. Some propose that US Medicare reduce the rates so that doctors are reimbursed at prices much closer to what is actually paid. The NHI in Taiwan has long recognized potential conflicts of interest in the system and determined to gradually decrease drug profits through large scale reimbursement rate reduction initiatives. As little empirical evidence is available concerning effects of such regulatory initiatives on cost control and prescribing patterns, results obtained from this study should help health service researchers and health policy officials design similar policies in the US or elsewhere. Primary Funding Source: National Science Council, Taiwan ●Do Drug Reimbursement Rate Reductions Work?Evidence from Taiwan’s Outpatient Hypertension Treatments in the Elderly Shuen-Zen Liu, Ph.D, James Romeis, Ph.D., Hsuan-Lien Chu, Ph.D. ●Variations in Depression Treatment Quality: Medicaid versus Commercial Claims Stacey Long, MS, Rebecca Robinson, MS, Stella Chang, MPH, Stephen Able, Ph.D., Onur Baser, Ph.D., Ralph Swindle, Ph.D. Presented By: Shuen-Zen Liu, Ph.D, Professor, Accounting, National Taiwan University, Room 1009, Building II, College of Management, Taipei, 106; Tel: 011 886 2 2736-7817; Fax: 011 886 2 2738 5108; Email: sliu@mba.ntu.edu.tw Research Objective: This study investigates the initial effects of Taiwan’s prescription drug reimbursement rate reduction policy (a supply-side cost control mechanism) in the elderly with hypertension based on outpatient treatment data. Study Design: We use regression analysis to examine whether average prescription drug cost decreases as a result of the nation’s drug reimbursement rate reduction program. The probit model is invoked to investigate changes in prescribing behavior for drugs with and without reimbursement rate reductions. In addition, we further investigate whether or not the drug rate reduction policy results in potential adverse effects on health status. Population Studied: About 137,000 patients aged 65 and older with hypertension were drawn from 21 hospitals in the Taipei area for the study. Principal Findings: We found that average drug cost per prescription increased slightly despite the implementation of the rate reduction policy after controlling for relevant variables in regression analyses. About 8,900 items of drugs (roughly 45% of the total) experienced rate reductions in the National Health Insurance [NHI] formulary. Those drugs, however, appeared only in about 3% of prescriptions. Thus, the policy had limited impact on total outpatient drug expenditures. As expected, in the probit model, we found evidence that physicians substituted drugs experiencing rate reductions with drug experiencing no rate reductions. Conclusions: Physicians appeared to be reluctant in reducing the use of essential drugs even when facing rate reductions, possibly attempting to avoid an adverse health outcome. The result may suggest less concern over the quality of health service because of the policy. Overall, the National Health Insurance reimbursement rate reduction policy offers a mechanism to change prescription behavior of physicians by reducing excessive profits on certain drug items based upon reliable database of reported drug transaction prices. Implications for Policy, Delivery, or Practice: In the US, chemotherapy concessions, referring to drug profits oncologists earn in treating US Medicare patients, have recently sparked heated criticisms as conflicts of interest that Presented By: Stacey Long, MS, Director, Outcomes Research & Econometrics, Medstat, Inc., 39 Sophie Lane, Hampden, ME 04444; Tel: 207-862-6360; Fax: 207-862-6360; Email: stacey.long@thomson.com Research Objective: Achieving universal access to quality health care is challenging. We compared antidepressant treatment patterns between patients with Medicaid and private insurance claims to assess differences in treatment quality. Study Design: MarketScan Commercial Claims and Encounter (CCAE) and Medicaid claims data from the same three states (2001-Q12003)were used to assess depression treatment quality based on antidepressant dose, duration, and "appropriate use" as defined by National Committee for Quality Assurance (NCQA) guidelines. Population Studied: Adults with employer-sponsored commercial insurance or Medicaid coverage residing in the same three states with depression claims and initiating on antidepressants were followed for 12 months. Claimants with prior antidepressant use or diagnoses of schizophrenia, bipolar, and psychoses were excluded. Principal Findings: Of the 10,383 CCAE and 20,170 Medicaid claimants, Medicaid patients were younger, sicker, and more likely to have capitated insurance (all p<0.001). 13.6% of commercial and 1.3% of Medicaid patients were compliant across all 3 NCQA measures. After controlling for demographics and antidepressant type, Medicaid patients were more likely than commercial patients to initiate below recommended doses (OR 1.55) and less likely to have continuous therapy (OR 0.54), or switch/augment therapies (OR: 0.27-0.67) (all p<0.001). Mental health specialty care was also associated with less early discontinuation (OR 0.76) and more continuous care (OR 1.18) (all p<0.001). Conclusions: Compliance with NCQA guidelines was less than optimal in both datasets but significantly worse in Medicaid. Greater equity of care and patient follow-up could result in less illness burden. Implications for Policy, Delivery, or Practice: 1. Participants will learn about variations across Medicaid and Commercial data sources in antidepressant treatment patterns and associated care in terms of appropriate use according to NCQA guidelines and dose recommendations over time. 2. Participants will recognize demographic, clinical characteristics including medical comorbidities, and antidepressant classes variations associated with utilization patterns and guideline compliance. 3. Participants will be able to identify trends in drug utilization and associated care across various antidepressants in both public and private settings. Primary Funding Source: Eli Lilly and Company ●Evaluating the Patient Safety Indicators (PSIs): How Well Do They Perform on VA Data? Susan Loveland, MAT, Peter E Rivard, MHSA, Shibei Zhao, MPH, Anne Elixhauser, Ph.D., Patrick Romano, M.D., Ph.D., Cindy L Christiansen Ph.D., Amy K Rosen, Ph.D. Presented By: Susan Loveland, MAT, Research Systems Manager, Bedford VAMC, Center for Health Quality, Outcomes and Economic Research, 200 Springs Road (152), Bedford, MA 01730; Tel: (781)687-2961; Fax: (781)687-3106; Email: slvland@bu.edu Research Objective: Although the VA has established a National Center for Patient Safety (NCPS) and taken initiatives to improve patient safety, accurate information on the epidemiology of patient safety events in the VA remains unavailable. The Agency for Healthcare Research and Quality (AHRQ) has developed the Patient Safety Indicators (PSIs), which are increasingly being used to screen for potential inpatient safety problems. Our objectives were to: 1) apply the AHRQ PSI software to VA administrative data to identify potential instances of compromised patient safety; 2) determine occurrence rates of PSI events in the VA; and 3) examine the construct validity of the PSIs. Study Design: Because VA administrative databases differ from standard databases in both structure and data definitions, we developed algorithms to create an “acute-only” hospital database from the VA 2001 inpatient administrative data, adding cost information to each hospitalization. We applied the PSI software to calculate VA unadjusted and riskadjusted safety event rates, based on HCUP case-mix, and compared them to non-VA rates – AHRQ’s 2000 HCUP National Inpatient Sample (NIS) and Medicare (2000 MedPAR) data. We also examined construct validity of the PSIs by investigating the difference in outcomes between hospitalizations with and without a PSI event. Population Studied: Veterans discharged from VA hospitals during the time period 10/1/00 through 9/30/01. Principal Findings: We identified 11,411 PSI events in the VA. Observed PSI rates per 1,000 discharges ranged from 0.007 for “transfusion reaction” to 155.5 for “failure to rescue.” Risk adjustment decreased the observed rates of seven PSIs and slightly increased the rates of seven others. VA riskadjusted rates were significantly lower than both HCUP-NIS and Medicare event rates for four indicators: “decubitus ulcer,” “infection due to medical care,” “postoperative respiratory failure,” and “postoperative sepsis.” Conversely, VA PSI event rates were significantly higher than both HCUPNIS and Medicare event rates for two indicators: “postoperative physiologic and metabolic derangements” and “technical difficulty with procedure.” Among the remaining relevant indicators, VA PSI event rates were significantly higher than HCUP-NIS event rates, but were significantly lower than Medicare event rates, for “death in low mortality DRGs,” “failure to rescue,” “postoperative hip fracture,” and “postoperative pulmonary embolism/DVT.” Hospitalizations with PSI events usually had longer lengths of stay, higher mortality, and higher costs than those without events. Conclusions: Our results suggest that the PSIs may be useful in the VA as a tool to screen for potential patient safety events that warrant further investigation. Our PSI rates were consistent with the national incidence of low rates; however, significant differences between VA and non-VA rates indicate that inadequate case-mix adjustment may be contributing to these findings. Implications for Policy, Delivery, or Practice: The PSIs can serve as case-finding tools, providing an initial assessment of the potential scope of patient safety events in the VA. Primary Funding Source: VA ●The Role of Poster Presentations in Knowledge Transfer and Exchange in Quality Improvement or QI Collaboratives Anu MacIntosh-Murray, Ph.D. Presented By: Anu MacIntosh-Murray, Ph.D., Postdoctoral Fellow, Knowledge Translation Program, University of Toronto, 122 Langley Avenue, Toronto, M4K1B5; Tel: (416)466-4725; Email: anu.macintosh@utoronto.ca Research Objective: Many organizations have multidisciplinary teams collaborating with both clinical experts and improvement experts to improve patient care processes and outcomes;e.g., IHI's Breakthrough Series, Vermont Oxford Network. Collaboratives often invest considerable time in poster presentations to share lessons learned beyond teams who initially test changes. Little is known about how this actually works and what information exchange takes place. Study aim: to learn how poster presentations facilitate dissemination and exchange of improvement knowledge between teams in an improvement collaborative, and between the teams and others in the organization. Study Design: Case study of the final poster presentation session of an organization's two improvement collaboratives, one with ten teams, the other with twenty. Data are based on: observation of the poster presentation activities; two sets of semi-structured interviews with the collaborative coordinator and five presenters from four improvement teams, two weeks and two months after the presentation session; document review of posters, templates, guidelines used; and a survey of 100 staff attending the event. Qualitative methods, interpretive and discourse analysis, are used for the interviews and observations, which are the focus here. Population Studied: Team members and staff particpating in and attending a QI collaborative presentation session in a large, muliti-site, urban academic health sciences centre. Principal Findings: 1. Most posters were based on a template provided by the organizers and were professionally prepared; large format, colour-printed, and laminated. They reflected a summative approach, chronicling the background, aim, changes, and results of the projects, not unlike formal research posters. Of the 19 posters displayed for the 75 min. session, six never had a team member there to answer questions, and several others were left unattended at various times. The level and duration of interactions between presenters and viewers varied markedly, depending in part on the efforts of presenters to engage viewers. No presenters had handouts of the posters available at the session. 2. Interview participants said they did not prepare an oral component - the project "story" - to present at the poster, nor were they instructed to do so. They reported that the posters did not prompt any discussion about explicit information about improvement methods, such as rapid cycle, PDSA, change ramps, during or after the session. Most questions were requests for copies of the more portable "tools" that tangibly embodied change ideas, such as education posters, or cards and pens with pain scales. 3. Interview participants spoke more about the conceptual, or awareness raising, and symbolic uses of the poster and the presentation session; instrumental use, to gain ideas for making improvements, was secondary. Conclusions: 1. The purpose and communicative intent of the presentations vary depending on the stage of the collaborative, so formats need to be adapted for each stage. Posters for communicating the work in progress during ongoing learning sessions can be flexible and less elaborate; e.g., pages of Powerpoint slides. These are often referred to as storyboards. Posters for the final presentations may resemble research posters in format and style, and may be a more formal end-product used for multiple purposes; e.g., produced as a single large panel. Problems arose when the purposes were unclear. 2. The poster presentation session had both practical and symbolic functions. It provided opportunities to network with staff from other units about practice issues unrelated to the projects. However, the dominant themes were the more symbolic aspects of the presentation session, including the rituals of celebration and recognition of the collaboratives as team and organizational accomplishments. Implications for Policy, Delivery, or Practice: Persons designing collaboratives and staff who are participating need to understand the purposes and formats of poster presentations. In addition to guidance about the structure and design of posters, poster presenters could benefit from coaching about how to prepare an oral component to support their presentation. Organizers should also consider how to prepare viewers to get the most out of the poster sessions. Primary Funding Source: Canadian Health Services Research Foundation Postdoctoral Fellowship ●Building Relationships Between a Health Plan and an Academic Medical Center Richard Mathis, BA, MA, Ph.D., Ken Patric, M.D., Steve Martin, RN, BS, MS.Ed, John Barnes, MBA, MA Presented By: Richard Mathis, BA, MA, Ph.D., Manager, Medical Policy Research and Development, BlueCross BlueShield of Tennessee, 801 Pine Street, Chattanooga, TN 37402; Tel: (423) 755-5801; Fax: (423) 785-8149; Email: richard_mathis@bcbst.com Research Objective: To improve the quality of health care delivery and to foster relationships between a health plan and an academic medical center. Study Design: Qualitative study. Population Studied: Academic medical center staff; Payor staff; patients. Principal Findings: Collaborations between BlueCross BlueShield of Tennessee (BCBST) and the University of Tennessee College of Medicine Internal Medicine Department have the potential of improving health care delivery to the patients served by the Internal Medicine Department. BCBST has worked with the Internal Medicine Department to provide deidentified data that assists in addressing conditions treated by Internal Medicine at Erlanger hospital. Among these are potential improvements in treating patients presenting with such diagnoses as intoxication. The Internal Medicine Department and BCBST also gain in their knowledge of one another's approach to such issues as fostering and implementing evidence based medicine. Conclusions: Health Plans and academic medical centers should work more closely together to foster quality and evidence based medicine. Implications for Policy, Delivery, or Practice: New relationships need to be forged between provider, payors, and researchers. This is an example of one effective approach. Primary Funding Source: No Funding Source ●Financial Incentives in Outlier Utilization Early in the Medicare Home Health Prospective Payment System Ann Meadow, Sc.D., Philip Cotterill, Ph.D. Presented By: Ann Meadow, Sc.D., Social Science Research Analyst, Division of Payment Research, Centers for Medicare & Medicaid Services, 7500 Security Bouleveard, Mail Stop C319-26, Baltimore, MD 21244; Tel: (410)786 6602; Fax: (410)786 5534; Email: ameadow@cms.hhs.gov Research Objective: Since October 2000, Medicare has paid for home health services under a case-mix adjusted prospective payment system. The unit of payment is a 60-day period of care. Outlier payments are available for care periods with high numbers of visits. Medicare’s outlier payment per period of care is 80% of the difference between total standard per-visit costs and the sum of the PPS payment plus a large deductible. Unlike outlier payments in the hospital DRG system, the provider’s specific costs do not affect the outlier payment. As a result, providers may face varying incentives to serve outlier cases depending on how their cost structure compares with the standard per-visit costs in the payment formula. They may also seek to limit losses on outliers by adjusting delivery practices for outlier cases to reduce costs. The objective of this research is to develop evidence on financial factors potentially affecting the operation of the outlier policy. Study Design: This is an observational study using administrative data from the Medicare program. We compare outlier cases from a 10% random sample of claims with nonoutliers from a 2% random sample. Data from claims detail clinician time in the home during each visit, payments, and the case mix group. We linked to each record the standardized clinical assessments used for case mix assignment, known as the Outcome and Assessment Information Set (OASIS), for purposes of measuring clinical and social characteristics of the patient. We also linked Medicare administrative files describing agency characteristics and, for agencies with usable cost reports, we linked cost report data. Analytic methods include bivariate tables and multivariate models to test several hypotheses, such as: (1) The probability that a case is an outlier is negatively related to the agency’s cost per visit for the services associated with outlier cases, skilled nursing and home health aide services. (2) Outlier cases have shorter average visit lengths for skilled nursing care, after controlling for clinical and social characteristics of the patient and agency practices. Population Studied: The study population consists of Medicare 60-day service periods for Medicare beneficiaries nationally during 2002. The main sample covers more than 5,000 home health agencies representative of Medicarecertified agencies throughout the nation. Usable cost reports came from approximately 1,200 agencies broadly representative of the nation. Principal Findings: Outlier cases average 88 visits, compared to 19 visits for nonoutlier cases. Outliers predominantly use skilled nursing (68 visits on average) and aide services (14 visits on average). After controlling for case mix group, disability program coverage, other demographic variables, urban residence, caregiver and living arrangement variables, cognition, receipt of home health aide services, and measures of clinical nursing need (such as diagnoses for wound conditions and diabetes care, intravenous or infusion treatments, and need for assistance in managing injectable medications), the probability that a case is an outlier varies inversely with the agency’s cost per visit for both skilled nursing care and home health aide visits: Odds ratios for skilled nursing cost (relative to the highest cost quartile) ranged from 4.3 for the lowest quartile to 1.61 for the secondhighest quartile (p <.0001). Odds ratio estimates for home health aide services were about 1.25 for the lowest and secondlowest cost quartiles (p <.05). Plots of average length in minutes per skilled nursing visit exhibited a marked reduction for outlier cases throughout the range of total visits per 60-day period. After adjusting for selected clinical and social patient characteristics, case mix group, type of agency ownership and affiliation, total visits, and the agency’s ownership, affiliation, and state location, we estimated that outlier cases’ average nursing visit length was shorter by about 6 to 8 minutes (in models with and without random intercepts for agencies, p <.0001), a reduction of approximately 15%. The average length of a home health aide visit is about one hour, but in outlier cases using home health aide services (about 48% of outlier cases), it was 3 to 5 minutes longer (p <.05 in random intercept model). Conclusions: The results suggest that the incidence of outlier cases is influenced by home health agency production costs, as well as clinical and social characteristics of patients and service delivery practices in large urban areas. The evidence of shorter nursing visits for outliers is consistent with a strategy of either targeting outlier admissions towards patients with needs (unmeasurable here) that lend themselves to shortened visits, or adjusting visit lengths for patients with large numbers of visits likely to lead to outlier payments. Such responses may serve to reduce agency losses incurred from serving especially costly patients. Home health aide services appear less liable to be length-adjusted in outlier cases, and a slightly longer visit length for outliers could be due to unmeasured patient service needs, labor market factors, or an interaction with nursing services which remains to be explored. Implications for Policy, Delivery, or Practice: Outlier payment policies are intended to supplement the case mix adjustment of payments in situations where poorly understood factors influence the cost of care and put providers at high financial risk. The home health outlier payment formula warrants continued study of its potential for introducing financial incentives that may affect patient access and service delivery. Primary Funding Source: No Funding Source ●A Comparative Study of Canadian and U.S. Academiclinked Health Policy Centers Michele Mekel, JD, MHA, MBA, Samuel E.D. Shortt, MPA, M.D., Ph.D. Presented By: Michele Mekel, JD, MHA, MBA, Visiting Fulbright Scholar, Centre for Health Services and Policy Research, Queen's University, Abramsky Hall, 3rd Floor, Kingston, ON K7L 3N6; Tel: (613) 533-6387; Fax: (613) 533-6353; Email: michelemekel@yahoo.com Research Objective: Identify and compare structural models and operational success strategies of university-linked health policy centers in Canada and the United States. Study Design: This largely qualitative, descriptive study was conducted using semi-structured telephonic interviews with Canadian and U.S. health policy center directors. The interview tool, containing approximately 50 questions, covered five overarching areas of center operations: 1. history; 2. target audience; 3. staffing and structure; 4. funding; and 5. performance evaluation. The U.S. component concluded in 2003, and the Canadian component was completed in 2004. Population Studied: Selected centers were university-linked, devoted to health policy research, and located in either Canada or the United States. For inclusion in the Canadian study, centers had to have a: 1. primary focus on health services and or health policy research; 2. designation in name as center, unit, institute, group, or an equivalent; 3. Canadian situs; and either 4. university affiliation; or 5. inclusion in the Canadian Health Services Research Foundation’s Network of Applied Health Services Research Centre Directors. The U.S. study’s inclusion criteria were similar, except centers had to have a: 1. designation in name as a health policy center or an equivalent; 2. university affiliation; and 3. United States situs. Thirteen, 87 percent, of the identified Canadian centers and 12, 48 percent, of the identified U.S. centers participated. Principal Findings: While the health systems in which they operate differ greatly, health policy centers on both sides of the border share numerous similarities in terms of function and structure. These commonalities included: dependence on governmental funding; governmental policy makers as key audiences; significant funding challenges; utilization of affiliated investigators and collaborative partners in research initiatives; and a dearth of systematic, center-specific performance measurement tools. Nevertheless, some significant differences emerged, including: Canadian centers’ greater focus on provincial health issues than that devoted by U.S. centers to state health concerns; Canadian centers’ discomfort with engaging legislative policy makers; Canadian center differentiation based on health care data access, i.e., whether the center served as a delegated repository for provincial health data, versus U.S. center differentiation based on the type of research undertaken; and a more collaborative environment among Canadian centers than exhibited among their U.S. counterparts. Conclusions: Similarities dominated as Canadian and U.S. health policy centers face virtually identical challenges and utilize analogous operational strategies. Where Canadian and U.S. centers differ, these divergences are primarily attributable to the health systems and governmental structures under which the centers function. Implications for Policy, Delivery, or Practice: While sample sizes were small, results from this study are important because they allow health policy centers to identify, explore, and adopt success strategies that are particularly suited to their operational model. By doing so, centers can enhance their contributions to health policy debates and formation, ultimately benefiting the end users of health services. Moreover, identification of challenges shared by both Canadian and U.S. centers, such as the lack of systematic, center-specific performance metrics, presents a prime opportunity for international collaboration. Primary Funding Source: Fulbright Grant for Research Abroad ●Primary Care Physician Characteristics and Involvement in a Program to Improve Outpatient Clinic Operations David Mohr, Ph.D., Carol VanDeusen Lukas, EdD, Mark Meterko, Ph.D., Marjorie Nealon Seibert, MBA Presented By: David Mohr, Ph.D., Health Services Researcher, VA Boston Healthcare System (152M), Department of Veterans Affairs, 150 South Huntington Avenue, Boston, MA 02116; Tel: (617)232-9500 x5679; Email: david.mohr2@med.va.gov Research Objective: The study sought to 1.) identify factors that predict primary care physician involvement in an innovative program designed to improve outpatient access to care and coordination and 2.) determine if extent of involvement predicts patient satisfaction and diffusion of the innovation. Study Design: A total of 454 primary care physicians from 78 medical facilities in the Veterans Health Administration system responded to a survey about their level of participation in Advanced Clinic Access (ACA), a program designed to improve clinic operations and access to care, and their perceptions of the extent to which ACA had been implemented and diffused at their facilities. The program consisted of ten specific practice activities. Respondents indicated their extent of participation in each activity. An overall extent of involvement score was computed from these data for each physician and regressed at the individual-level on two control variables (clinic size and number of exam rooms per clinician) on four individual characteristics (fulltime status, tenure, supervisory role, and belief in the effectiveness of ACA) and degree of management support for the innovation. In a separate, facility-level analysis, extent of physician involvement was used to predict patient satisfaction with several dimensions of their VA healthcare as measured by an independent survey and employee judgments regarding the extent of program diffusion. Population Studied: Primary care physicians and outpatients in a cross-national hospital system. Principal Findings: Factors significantly associated with individual extent of involvement were tenure of more than one year in the clinic, having a supervisory role (e.g. team leader), belief of program effectiveness and management support (adj r-sq = .18). After including the variables from the individualmodel, extent of involvement significantly improved the percent of variance accounted for patient satisfaction with overall coordination (change in r2 = .05, p <.05, overall adj r-sq = .19), but not with access to care or overall quality. Additionally, extent of involvement significantly improved the percent of variance accounted for program diffusion (change in r2 = .06, p = .01, overall adj r-sq = .47). Conclusions: Physicians who have been with the clinic for a year or more, hold a supervisory role, and see the benefits of the innovation program will be more likely to actively participate in the program. Management support also positively influenced individual’s participation rates in the program. After controlling for these and other factors, higher levels of participation also impact patient satisfaction with overall coordination of care and spread of the program practices to other clinic areas within the medical facility. Implications for Policy, Delivery, or Practice: Individual physician participation in the change program may be facilitated with management support and a greater level of individual belief in the program benefits. Greater participation by physicians can produce noticeable benefits to patients and with the spread of the program to other clinic areas. Primary Funding Source: VA ●Perspectives of Persons with Opioid Dependence: An Analysis of Online Discourse Robert O. Morgan, Ph.D., Loralee Capistrano, MPH, LeChauncy D. Woodard, M.D., MPH, Paul Haidet, M.D., MPH, Richard L. Street, Ph.D., Robert O. Morgan, Ph.D., Barbara F. Sharf, Ph.D. Presented By: Robert O. Morgan, Ph.D., Associate Professor, Houston Center for Quality of Care and Utilization Studies, Baylor College of Medicine, Michael E. DeBakey VA Medical Center, 2002 Holcombe Boulevard (152), Houston, TX 77030; Tel: (713)794-8601; Email: rmorgan@bcm.tmc.edu Research Objective: Efforts to integrate opiate addiction treatment into medical practice has the potential to significantly expand access to care. However, studies indicate that physicians feel unprepared to treat patients with opiate addiction. We conducted this study to explore issues surrounding patients’ perspectives of opiate addiction. Study Design: We conducted a thematic analysis of postings by an Internet discussion group for people actively using or recovering from opiate addiction. We collected discourse from July 13 to July 20, 2004, which generated 180 pages of text. Four investigators analyzed transcripts to identify key themes. Analyses were guided by a focus on the cause, control, severity, treatment, and meaning of opiate addiction from the patients’ perspective. Population Studied: “The Patient Buprenorphine Web Board” consisted of 857 members, of which 665 posted information. Principal Findings: Four themes emerged. First, participants discussed the meaning of addiction, including their reasons for opiate use and feelings about their addiction. One participant commented, “Addiction fills the void. When you get clean, you’re like a doughnut with an even bigger void. You suck in meaninglessness; drown in a sea of futility. Unless you can replace it with something solid, it seems only a matter of time before you’re back where you started.” Second, participants described the impact of addiction on daily life. One participant stated, “We are not rich by any means ($38k per year gross) and I have been spending $1200 per month on my habit. It has put us in bad debt and I have to stop.” Third, participants conveyed their experiences with recovery, stating that relapse is common. One participant noted, “I have detoxed and relapsed before and I know the tricks, traps, and lies we fall into before we start getting high again. It is an ongoing process, and instead of being anxious, frustrated and impatient, look at the things that are better.” Finally, patients perceived their addiction to be a health problem similar to other chronic illnesses and described the impact of addiction on interactions with their physicians. One participant stated, “Some people need an opiate to function, just as a diabetic needs insulin.” Another commented, “I was at my doctor’s office today for the second time in almost a year. The first time, I never told him I was on meth (methadone). Today, he asked me three times if I was on any type of meds, and I told him no.” Conclusions: Our analysis identified patients’ views regarding the meaning of addiction, the impact of addiction on their daily lives, their experiences with recovery, and their perception that addiction is a health problem similar to other chronic illnesses. These findings provide valuable insight into patients’ views of opiate addiction and reveal a complexity and depth that is often not solicited or understood by physicians Implications for Policy, Delivery, or Practice: Because patients’ perspectives often impact illness behaviors, such as disclosure during the medical interview, further work is needed to promote sharing and understanding of patient and physician perspectives regarding substance abuse. Primary Funding Source: AHRQ mix of drugs used drove trends over time. Regional variation in age-standardized drug utilization was greatest for antidepressants, analgesics, and lipid reducing agents: coefficients of variation = 17.5, 14.8, and 14.2 percent, respectively. Populations in urban areas had lower standardized rates of drug utilization and expenditure than populations in rural areas. Conclusions: The methodology developed in this study allows the analyst to track across region, years, and populations multiple factors related to drug utilization and expenditure. While the findings show variations in prescription use are on a scale lower than previous studies have documented for medical and hospital services, the results draw policy attention to potentially important regional and urban-rural disparities. Implications for Policy, Delivery, or Practice: The challenge for policy makers is to ensure that drug use is both appropriate and cost-effective. Our analysis -- based on methods that can be applied in other settings -- helps policy makers identify therapeutic areas deserving of investigation for potentially inappropriate prescribing and to identify factors driving drug costs that may be amenable to policy intervention. Primary Funding Source: Canadian Institutes of Health Research ●Mapping Drug Utilization Patterns: A Canadian Case Study Steve Morgan, Ph.D., Sherin Rahim-Jamal, MSc, Peter Schaub, BA, Meghan McMahon, BA, Dawn Mooney, BA, Lixiang Yan, MSc ●The Association of Depression Status with Functional Status, Role Function, and Disease Impact among adults with Persistent Asthma David Mosen, Ph.D., MPH, Renee N. Saris-Baglama, Ph.D., Michael DeRosa, MA, Elizabeth Fortin, MSW Presented By: Steve Morgan, Ph.D., Assistant Professor, Centre for Health Services and Policy Research, University of British Columbia, 429 - 2194 Health Sciences Mall, Vancouver, BC, V6T 1Z3; Tel: (604)822-7012; Fax: (604)822-5690; Email: morgan@chspr.ubc.ca Research Objective: Expenditure on prescription drugs in North America has been growing at double-digit rates since the mid-1990's. The objective of this study was to develop and implement methods to examine population trends and regional variations in drug utilization and cost. Study Design: A model was developed to quantify causes of trends and variations in drug spending. The model depicts per capita expenditure as a function of the average individual's needs, the rate at which they receive drug therapy, the type of drugs selected, and the prices of the drugs purchased. The study utilized a dataset containing a record of every prescription filled by residents of British Columbia from 1996 to 2002 to quantify the components of the analytical model. The magnitude and causes of trends and regional variations were illustrated using geographic information systems. Population Studied: The study cohort included all of the 4.1 million residents of BC except for registered First Nations, veterans, and the Royal Canadian Mounted Police -approximately 4 percent of the provincial population. Analyses were conducted for five age categories in sixteen geographic regions. Principal Findings: Per capita expenditures on pharmaceuticals increased by 87 percent over the period 1996 to 2002, with cost growth being most rapid among residents aged 45 to 64. Increased volume therapy and changes in the Presented By: David Mosen, Ph.D., MPH, Program Evaluation Consultant, Care Management Institute, Kaiser Permanente, 500 NE Multnomah, Portland, OR 97236; Tel: (503)813-3827; Fax: (503)813-2428 Research Objective: Depression may impact the ability of persistent asthmatics to manage their asthma symptoms, possibly leading to poorer functional health status, decreased role function, and decreased productivity. However, despite these possible health effects, little is known regarding the association of depression on health outcomes for adults with persistent asthma. The objective of this study is to twofold: 1) identify the prevalence of likely depression within the persistent asthma adult population and 2) examine the association of depression status with physical function status, role function and disease-specific asthma impact. Study Design: Survey and administrative data for 428 adult persistent asthmatics enrolled in a large staff model HMO were examined. Persistent asthmatics were identified during calendar year 2002 using HEDIS inclusion criteria. In Fall 2003, the same patients were surveyed regarding functional health status using the SF-8™ Health Survey, which yields a physical component summary score (PCS), a mental component summary score (MCS), and two role functioning scale scores (role-physical and role-emotional). Based on previously published research, patients with an MCS score of 42 are likely to have depression. Additionally, asthma impact was measured using the Asthma Impact Survey™ (AIS-6), a valid and reliable six item scale that measures the extent to which asthma interferes with usual activities such as work, school, or social activities (78 = highest disease impact, 36 = lowest disease impact). Ordinary Least Squares Regression was used to examine the independent association of depression status on physical functional status, role function, and disease-specific asthma impact, adjusting for age, gender, race/ethnicity, educational attainment, and geographic location. Population Studied: Adults (predominantly white women with a mean age of 46.9) with persistent asthma enrolled in large staff model HMO in the United States. Principal Findings: Overall, 28% of those studied had MCS scores below the cut-off of 42, indicative of a high likelihood of depression. Those with a high likelihood of depression reported poorer physical functional status (Standardized Beta Coefficient = -0.27, t = -5.31, p< .0001), poorer role functioning due to emotional problems (Standardized Beta Coefficient = 0.73, t = -19.5, p< .0001), poorer role functioning due to physical problems (Standardized Beta Coefficient = -0.44, t = 8.97, p< .0001), and greater disease-specific asthma impact (Standardized Beta Coefficient = 0.40, t = 7.94, p< .0001); even after adjusting for age, gender, race/ethnicity, educational attainment, and geographic location. Conclusions: Results of this study suggest that a large proportion of the adult asthma population is likely to experience depression. Compared to asthmatics with MCS scores above the cut-off of 42, asthmatics who were likely to be depressed reported significantly lower physical functional status, role function, and disease-specific asthma impact. Implications for Policy, Delivery, or Practice: Because depression is associated with poorer health outcomes for adults with persistent asthma as compared to persistent asthmatics without depression, further efforts are needed to better recognize and treat depression in this population. Future research should evaluate the impact of interventions designed to recognize and treat depression on health-related outcomes. Primary Funding Source: Kaiser Permanente ●Initial Stroke Severity and Health of Informal Caregivers of Stroke Patients; the Minnesota Stroke Survey Melissa Nelson, BS, Maureen A. Smith, M.D., MPH, Ph.D., Brian Martinson, Ph.D., Russell Luepker, M.D. Presented By: Melissa Nelson, BS, Graduate School Fellow, Health Services Research and Policy, School of Public Health, University of Minnesota - Twin Cities, Mayo Mail Code 729, 420 Delaware Street SE, Minneapolis, MN 55455; Tel: (612) 331-5627; Email: nels2404@umn.edu Research Objective: In 1999, 1.1 million Americans had difficulties with everyday activities resulting from strokes. The large proportion of care provided informally to stroke survivors suggests that stroke may significantly impact caregiver health. Studies have examined the impact of the stroke caregiving experience on caregiver burden, but few have examined the impact on caregiver health and none have examined whether initial stroke severity affects subsequent caregiver health. The objective of this study is to examine the relationship between initial stroke severity and subsequent caregiver health. Study Design: Preliminary data were obtained from an ongoing longitudinal study of stroke patients and their caregivers in the Minneapolis-St. Paul metro area. Medical records for index stroke hospitalizations were abstracted and patients or their proxies and caregivers were interviewed 2 months and 12 months after the patient was discharged from the hospital. The patient/proxy response rate was 66% and caregivers were interviewed for 98% of patients or proxies who reported the presence of a caregiver. Caregiver health was measured using self-reported health on a scale of excellent to poor, with higher values indicating worse health. The patient’s stroke severity was measured using the Canadian Neurological Scale, with scores reversed so that higher values indicated worse health. Other patient characteristics included sociodemographics, previous stroke, functioning prior to stroke and upon discharge, and comorbidities. Caregiver characteristics included age, gender, and education. Environmental characteristics included coresidence of caregiver and patient after discharge and the use of paid caregiving services. Population Studied: Caregivers and their corresponding patients or proxies who responded to the 2 month interview (N=211). Principal Findings: 53% of caregivers reported excellent or very good health, 34% reported good health, and 13% reported fair or poor health at the 2 month interview. The average stroke severity was 2.3 (standard deviation = 2.1) on a scale from 0-11.5. After adjustment for patient, caregiver, and environmental characteristics using ordinal logistic regression, the odds of poorer caregiver health at 2 months were 1.21 times greater for each unit increase in stroke severity, Odds Ratio (OR) = 1.21, 95% Confidence Interval (CI) = 1.03-1.42. Other factors associated with poorer caregiver health were greater caregiver age, OR = 1.04, 95% CI = 1.02-1.07, and the patient having less than a high school education, OR = 3.77, 95% CI = 1.42-10.03, while use of paid care prior to stroke was associated with better caregiver health, OR = 0.31, 95% CI = 0.10-0.92. Conclusions: These data suggest that increasing stroke severity was associated with poorer caregiver health 2 months after hospital discharge. Subsequent research will incorporate additional 2 month data and measures of caregiver health at 12 months after discharge. Implications for Policy, Delivery, or Practice: The potential negative impact of stroke severity on caregiver health emphasizes the need for improved understanding of factors that place caregivers at risk. This could assist in developing interventions targeted at caregivers who are at risk for deteriorating health. Primary Funding Source: National Institute of Neurological Disorders and Stroke ●Use of Prescription Medications in the Treatment of Pervasive Developmental Disorders Donald Oswald, Ph.D., Neil Sonenklar, M.D. Presented By: Donald Oswald, Ph.D., Professor, Psychiatry, Virginia Commonwealth University, Box 980489, Richmond, VA 23298; Tel: (804)828-9900; Fax: (804)828-2645; Email: doswald@vcu.edu Research Objective: The study examines the use of prescription medicine among individuals with a diagnosis of an autism spectrum disorder. We characterize patterns in the use of psychoactive medications and examine trends with regard to demographic variables. Study Design: The study consisted of a descriptive analysis of prescribed medications in a large sample of individuals with an autism spectrum disorder. Comparisons of type and number of medications were made across groups defined by age and gender. Pooled administrative data of the 2002 MarketScan® Research Database from MedStat, a healthcare information company, were used to investigate medications prescribed for enrollees with a diagnosis of an autism spectrum disorder. Diagnosis data for inpatient admissions and outpatient services were used to identify subjects with an autism spectrum disorder and to capture comorbid diagnoses. Outpatient pharmaceutical data were used to identify medications prescribed. Population Studied: The sample consisted of all persons with a diagnosis of an autism spectrum disorder who had at least one prescription drug claim from an outpatient provider in the year 2002 (n= 2,206). The mean age for persons in the sample was 14.7 years (SD=10.95; range 0 – 64) Principal Findings: The number of drug claims during the year varied considerably across individuals (mean = 17; SD = 23.49; range 1 – 770). The number of different drugs prescribed to an individual over the course of the year also varied substantially. The median number of drugs was 5 (mean=6.05; SD = 4.93; range 1 – 41). Prescribed drugs came from 143 different therapeutic classes. The five most frequently prescribed classes of drugs were: “Pyschotherapeutic, Antidepressant” “Antibiotic, Penicillins” “Stimulant, Amphetamine Type” “Psychotherapeutic, Tranquilizers / Antipsychotics” and “Antihistamines and Comb NEC”. Prescribed medications were also examined in relation to comorbid diagnoses to characterize the number, type, and probable target of psychoactive drugs employed in the treatment of individuals with autism spectrum disorder. Frequency of use of psychoactive medications was examined by gender and age. Conclusions: Individuals with autism are treated with a wide variety of psychoactive medications over the course of their lives. Prescription patterns reflect extensive use of antidepressants, stimulants, and antipsychotics. Use of these drugs varies across subgroups of the population. Implications for Policy, Delivery, or Practice: The study offers the first large-scale investigation of prescribing practices based on drug claims for the population of individuals with an autism spectrum disorder. Characterization of patterns for the prescription of psychoactive drugs in the population provides valuable baseline data. Age and gender differences reflect different targets of treatment across these groups. The findings from this study will serve as a point of comparison as the number and type of medications targeting common symptoms in the population increase. Primary Funding Source: CDC ●Parent's and Youth's Perception of Mental Health Services MariaEva Pangilinan, Ph.D., Bruce Copley, MA, Jaime Lopez, MA, Teea Gilbert, MA Presented By: MariaEva Pangilinan, Ph.D., Health Program Specialist, Mental Health Department, Santa Clara County, 828 South Bascom Avenue, San Jose, CA 95128; Tel: (408)8855768; Email: eva_pangilinan@yahoo.com Research Objective: The Youth Satisfaction Survey for Families (YSSF) and the Youth Satisfaction Survey (YSS) were developed to help monitor performance in mental health services. Its developer had reported observed differences in domain average scores between parents and their adolescent children, except for the General Satisfaction domain on these surveys. Given that these instruments were used statewide for the first time in California in November 2003 no baseline data and studies on the association between parents and youths perceptions of services as measured by these instruments were available at the state and county levels. This study examined baseline information on the association between perceptions of parents and youths receiving mental health services at the Santa Clara County. Study Design: Following the statewide protocol, all Santa Clara County family and children (F & C) consumers receiving face-to-face mental health services, case-management, day treatment, and medication services from county and noncounty operated organizations were recruited from November 3 to 17, 2003, regardless of the funding source. Excluded were consumers served in acute hospitals, Psychiatric Health Facility (PHF), crisis (stabilization, residential and intervention), jail and jail hospital settings, long-term care residential placements, such as: State hospitals, Institute for Mental Disease (IMD), and individual / group contract managed-care network providers. The YSSF and YSS surveys were to be completed by the consumers on their own, as much as possible. HIPAA Guidelines were followed and confidentiality was observed. To assess the parents’ perceptions in relation to their adolescent children, correlation analysis for matched ParentYouth data was performed between the parent YSSF domain average scores and the YSS Youth domain average scores. Paired sample t-test was also performed by domain and by item. Population Studied: Data from clients who have at least one question answered on the Youth Satisfaction Survey for Families (YSSF) and Youth Satisfaction Survey (YSS) were included for analysis. This sample included a total of 660 YSSF for Parents and 470 YSS for Youths. From which, a total of 188 matched Parent-Youth records across domains was found. Principal Findings: Results showed low correlations between Parent and Youth ratings for Services Access (r = 0.20 at p < .015, n = 155) and Service Outcomes (r = 0.21 at p < .010, n = 154). No significant correlations were found for the rest of the domains. The Parent-Youth domain score differences were found to be significant for Involvement in Treatment Planning Good Services Access, General Satisfaction and Cultural Sensitivity domains but not for Good Service Outcomes. The Parent-Youth item score differences were also found to be significant except for all of the Good Service Outcomes items as well as for 2 of 4 items for Cultural Sensitivity (at p<.01). Except for Outcomes measures, the Youths tended to have lower ratings of services than their Parents. Conclusions: This finding provided additional evidence regarding the observed difference in perceptions of services between Parents and Youths on the YSSF and the YSS. The study suggests independent perceptions about Involvement in Treatment Planning, General Satisfaction and Cultural Sensitivity and a trend towards agreement regarding Services Outcomes and Good Service Access and sensitivity to religious beliefs and cultural background.This study acknowledges that in addition to the simple correlational analyses performed in this study, analyses of simultaneous effects that may well be related to age, ethnicity and gender in relation to perceptions of services could still be conducted. Implications for Policy, Delivery, or Practice: Findings certainly lent support to maturational differences. From the health services perspective however, findings suggest the need to assess the age-appropriateness and sensitivity of survey questions to assess perceptions vis-à-vis ageappropriateness of conditions of services that could benefit quality improvement efforts to yield a more similar experience of services by all consumers. Caution is recommended in application of findings. Primary Funding Source: Work related ●Nursing Care Job Design and Care Quality in Long-Term Care Facilities Victoria Parker, DBA, Gary H. Brandeis, M.D., Amy Rosen, Ph.D., Fei Wang, Ph.D., Elaine Hickey, MS, Denise A. Tyler, MA Presented By: Victoria Parker, DBA, Assistant Professor, Health Services, Boston University School of Public Health, 200 Springs Road (152), Bedford, MA 01730; Tel: (781)2757500 x6054; Fax: (781)687-3106; Email: vaparker@bu.edu Research Objective: To determine the relationships between nursing care job design characteristics and care quality in long-term care facilities. Study Design: Observational study of nursing care workers in long-term care facilities, employing primary data collection through surveys and secondary analysis of minimum data set (MDS) records to assess resident outcomes. Employeeperceived job design characteristics (skill variety, task identification, task significance, autonomy, intrinsic feedback) were assessed via the Job Diagnostic Survey (JDS). MDS data were used to assess 3 resident outcomes (pressure ulcer incidence, decline in late-loss activities of daily living (ADLs) and increase in problem behavior) over two consecutive quarters. Employee data were linked with resident-level MDS data via a facility identifier. Logistic regressions were developed for risk-adjustment models, while linear models were used to examine the relationships between risk-adjusted resident outcomes and individual job design characteristics. Population Studied: 1146 nurses and nursing assistants employed in 20 facilities were surveyed. MDS records for 1,976 residents residing in one of those 20 facilities for at least 45 consecutive days during the study time frame were analyzed to ascertain changes in condition between two assessments. Facilities were recruited on a convenience basis to fill pre-specified categories (size, profit status, ownership, location). Subsequent analysis of these 20 with reference to national and state comparison groups revealed that they were at or near average in terms of quality indicators and survey deficiencies. Within each facility, study personnel recruited employees to complete the JDS on all three shifts and weekends until a 60-70% response rate was achieved within each facility. Principal Findings: We found significant inter-facility variation in characteristics of nursing care job design. Three aspects of job design (task identity, intrinsic feedback and autonomy) were positively linked (p<.01) with lower rates of decline in late-loss ADLs. There were mixed findings (at p<.05) for increase in problem behavior (positive association with intrinsic feedback; negative for skill variety and task significance). No characteristics of job design were significantly linked to rates of pressure ulcer development. Conclusions: These mixed findings are consistent with prior attempts to link employee variables to resident outcomes. The more significant positive links between job design and ADL decline suggest that nursing employees who identify a piece of work as their own, receive feedback provided by the work itself, and have control over the timing and pacing of their work are better able to work with residents in ways that support ADL maintenance. The mixed findings with respect to problem behavior may reflect the bi-directional nature of this relationship in that resident problem behavior may reduce the employees’ perceptions of more positive job design attributes. The lack of any significant association with pressure ulcer incidence rates is likely due to the very low incidence rate of pressure ulcers within the study population. Implications for Policy, Delivery, or Practice: Concern about the adequacy of the long-term care workforce to meet the projected increase in demand for this type of health care service implies that knowledge of employee perceptions of job design must be improved in order to help employers optimize job design for both care quality and employee retention. Primary Funding Source: AHRQ ●Parents’ Quest for Knowledge: The Search, Utilization and Meaning of Child Psychiatric Diagnosis Vaishali Patel, MPH, Ph.D. Candidate, Susan DosReis, Ph.D. Presented By: Vaishali Patel, MPH, Ph.D. Candidate, Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, 6 Char Street, Edison, NJ 08820; Tel: (732)2618323; Email: vapatel@jhsph.edu Research Objective: Understanding parental help-seeking processes is vital as parents’ recognition of their children’s mental problem and the need for professional intervention serves as an important link to access, use, and continuity of care. This qualitative study sought to explore how parents obtain mental health services on behalf of their children by identifying strategies parents employ and the resources parents rely upon, especially clinicians, during this process. Study Design: We applied a grounded theory approach that consisted of successive iterations of data collection, inductive analysis of interview and focus group transcripts and field notes, and development of theory. First, we conducted two semi-structured interviews with parents of children who had received mental health services (n=7). Second, we conducted semi-structured interviews with child mental health providers (n=7). Third, we conducted a parent focus group (n=4) to elaborate on emergent themes from the in-depth interviews. Population Studied: Two mental health consumer advocacy organizations recruited parents and clinicians were recruited through professional contacts of the research team. All the parents were middle-aged; a majority of participants were female, the biological parent and married. The clinicians consisted of three psychiatrists, a mental health hospital administrator (and former clinician), a psychologist, and two school-based clinicians (clinical social worker; licensed clinical professional counselor). Principal Findings: As parents became aware of their children’s mental health problems, they began a “quest for knowledge,” a search for information related to their children’s psychiatric diagnoses and services available. Parents knowledge of their child’s psychiatric diagnosis was cited by parents as enabling them to better advocate on behalf of their children to obtain care and deal with stigma. Clinicians’ views regarding the value and utility of diagnosis differed from parents: clinicians were more wary of diagnosis’ meaning and reported that diagnosis did not necessarily inform treatment. Clinicians instead emphasized providing parents with skills to manage their child’s illness. Parents attributed clinicians’ reluctance to disclose diagnosis due to stigma and fears of labeling a child with a severe psychiatric disorder. Parents in this study were unaware of providers’ concerns regarding the ambiguous nature of diagnosis, and the complexity involved with making a proper diagnosis in children. Parents use of alternative resources, such as the Internet, books, consumer advocacy organizations and other parents of children with mental health problems was in part fueled by parents’ inability to obtain information from traditional mental health providers. Conclusions: Parents’ “quest for knowledge” emerged as a key mechanism that enabled parents to circumvent barriers to care and cope with their children’s mental health problems. Differences between parents and providers regarding the value of diagnosis potentially limits their communication, contributing to parents’ use of alternative resources to search for information regarding their child’s mental health and services available. Implications for Policy, Delivery, or Practice: These findings highlight parents’ struggles to obtain and make sense of information that potentially impacts their ability to make informed treatment decisions for their children. To address parents’ need for information, clinicians can potentially play an important role in providing comprehensive psychoeducation or referring parents to programs provided by consumer advocacy organizations that educate parents about mental illness. Primary Funding Source: NIMH ●The Role of Sociocultural and Discrimination Factors in predicting Support for Health Policy among US Latinos Debra Joy Pérez, MA, MPA, Ph.D. Presented By: Debra Joy Pérez, MA, MPA, Ph.D. Program Officer, WKKELLOGG Fellow in Health Policy, Research, Multicultural Mental Health Research Fellow, NIMH mentee, Research & Evaluation Unit, Interfaculty Program in Health Policy at Harvard, The Robert Wood Johnson Foundation, Harvard University, PO Box 3311, Princeton, NJ 08543; Tel: (917)586-5955; Email: djperez@fas.harvard.edu Research Objective: Latino politics influence policy agendas and candidate support. There is little research on the specific importance of health policy to the Latino electorate. This study explores how Latinos differ in the following domains: support for healthcare policy as a priority for government action and healthcare as a determinant of voting. This study uses national data to model how Latinos who consider healthcare as an important factor for government action might differ from those Latinos who do not. It explores the relationship between support for health policy issues and multiple demographic, socioeconomic, acculturation and discrimination measures on this important question. This study compares Latinos to whites with regard to support for health policy issues. This study examines which SES, sociodemographic and political factors are correlates for supporting a health policy agenda. Study Design: This study estimates a model using logistic regression to predict which Latinos are more likely to support health policy issues and to vote on health issues. This study tested the following four hypotheses: 1. Latinos with low acculturation are less likely to identify health policy as an important issue as compared to Latinos with high acculturation. 2. Latinos who experience barriers to care are more likely to identify health policy as an important issue than those who do not report barriers. 3. Latinos with low acculturation are more likely to support larger government and increase taxes as compared to Latinos with high acculturation. 4. Latinos reporting greater barriers to care are more likely to say that healthcare is a factor in vote decision as compared to those reporting fewer barriers. 5. Latinos who report higher levels of perceived discrimination are less likely to support health policy as priority issue as compared to those Latinos reporting little or no discrimination. DATA AND METHODOLOGY: The data was collected by the Kaiser Family Foundation/The Pew Hispanic Center in their 2002 National Survey of Latinos from April 4 to June 11, 2002. A nationally representative sample of 4,214 adults, 18 years and older were selected at random for this phone survey. The Random Digit Dial sample includes over-samples of Puerto Ricans, Salvadorans, Dominicans, Colombians and Cubans. Of the total sample, 2,929 were self-identified Latino or Hispanic. The sample was weighted to reflect the distribution of Latino adults in the US by country of origin, age, gender and region. There are 1008 non-Hispanic whites in the sample and 171 African Americans. The margin of sampling error varies for each of the different subgroups. Population Studied: Latino voters and non-voters, Puerto Ricans, Cubans, Mexicans Principal Findings: Results showed that Latinos with higher levels of acculturation were morely to report health policy issues as important determinant of their presidential vote. More than 40% of US born Latinos reported using health policy as a voting determinant compared to 32% of foreing born voter (P = 0.0001.) Latinos who have been in the US for more generations were more likely to support health policy issues compared to Latinos who have been in the US for fewer generations. Thirty two percent of first generation reported that health policy was a determinant of their presidential vote compared to 48% of third and fourth generation Latinos (P = 0.0000.)Language was a strong predictor of support for health policy issues. Only 31% of Spanish dominant Latinos reported health policy issue as important to their vote compared to support 40% of English speaking Latinos (P = 0.0000.) Finally, Latinos who identified themselves as having an American identity were more likely to report health policy as an important determinant of their vote compared to only 30% of Latinos with a national origin Latino identity. Conclusions: Latino voting behavior differed based on the degree of acculturation (nativity, generational difference, and language dominance) and identity differences (Puerto Rican, Cuban, Mexican, and other Latinos). Latinos are heterogenous in their knowledge and support for health policy issues. Important changes in demographics among Latinos will influence how central platforms for political parties incorporate health policy issues. Implications for Policy, Delivery, or Practice: Few studies attempt to explain the significance of health policy to Latino voters and how increases in number of Latino voters and non-voters might affect the health policy agenda in the US. This study is motivated by the need to better understand Latino support for health policy and what if any factors might galvanize Latinos to become part of a national call for healthcare reform. This study helps estimate how Latinos will impact health policy if projected changes to the socioeconomic variables such as increases in income or insurance status occur. Policymakers need to consider acculturation in framing health policy issues among Latinos. Primary Funding Source: WKK, NIMH ●Comparison of LDL Screening and Control Rates for Various Reporting Entities and Populations Donna Pillittere, MS, Philip Renner, MBA, Sarah Hudson Scholle, MPH, DrPH Presented By: Donna Pillittere, MS, Senior Health Care Analyst, Quality Measurement, National Committee for Quality Assurance (NCQA), 2000 L Street, NW, Suite 500, Washington, DC 20910; Tel: (202)955-1736; Fax: (202)9553599; Email: pillittere@ncqa.org Research Objective: Low density lipoprotein (LDL) has been identified as the primary target for cholesterol-lowering therapy; elevated levels have been shown to be a powerful risk factor for coronary heart disease (CHD). Guidelines recommend the LDL cholesterol goal for patients with CHD and CHD risk equivalents (such as diabetes) to be LDL<100 mg/dL. As a consequence of these guidelines, there has been much discussion in the clinical quality improvement world around achievement of the suggested goal. The purpose of this study is to examine how well different reporting entities and populations are doing so. Study Design: A comparison of LDL screening and control (LDL<130mg/dL; LDL<100mg/dL)rates in two Health Plan Employer Data Information Set (HEDIS) populations (cardiovascular event and diabetes) covering services provided by managed care, collected from administrative data supplemented by medical record review and in two physician level recognition program populations (iscemic vascular disease (IVD) and diabetes) covering services provided by physicians, collected from physician level medical record data. All data are reported based on a standardized set of performance measures using detailed specifications. Recognition data probably represent top performers, as they are physicians who voluntarily come forward to be measured and if achieved, awarded recognition for quality of care in heart/stroke and in diabetes. Population Studied: 2000-2003 HEDIS commercial managed care data, 1997-2003 physician level medical record data from the Diabetes Physician Recogntion Program (DPRP) and 2003 physician level medical record field test data from the Heart/Stroke Physician Recognition Program (HSRP). Principal Findings: HEDIS commercial managed care data showed a steady increase in performance for the diabetes and CHD population for both the LDL screen and LDL<130 indicators. For example, from 2000-2003, mean LDL screen rates for the diabetic population increased from 76.5 to 88.4 and mean LDL<130 rates increased from 44.3 to 60.4. From 2000-2003, mean LDL screen rates for the CHD population increased from 74.2 to 80.3 and mean LDL<130 rates increased from 53.4 to 65.1. For the first time with HEDIS 2004, NCQA collected rates at LDL<100 with the diabetic population achieving a mean rate of 34.7 and the CHD population achieving a mean rate of 47.6. In comparison, 1997-2003 physician level data from the DPRP showed an increase in mean screen rates from 60 to 90, mean LDL<130 rates from 37-75, and mean LDL<100 rates from 17-45. Fieldtest data from the HSRP showed mean rates of 79.2 for LDL screen, 68.7 for LDL<130 and 48.9 for LDL<100. Conclusions: On average, physician level data showed better performance on all three inidicators than healt plan data. HEDIS health plan data showed LDL<130 and LDL<100 rates were consistently higher in the CHD population than in the diabetic population. On average, LDL<100 rates were 20-30 percentage points lower than LDL<130 rates in both health plan and physician level data. Implications for Policy, Delivery, or Practice: Recognized physicians tend to be top performers, showing what is possible to achieve. Lower LDL control rates for the diabetes population may indicate more difficulty in getting this population's cholesterol under control. Just as initial measurement on the LDL<130 rates for different populations and reporting entities showed improvement over time, as the industry becomes more aware of the LDL<100 goal, rates here should also improve with time. Primary Funding Source: NCQA ●Health Care Needs of Women in Shelter Stacey Plichta, Sc.D., Terri Babineau, Sc.D., Elizabeth Vogel, MS Presented By: Stacey Plichta, Sc.D., Associate Professor, School of Community and Environmental Health, Old Dominion University, 105 Spong Hall, Norfolk, VA 23529; Tel: (757) 683-4989; Fax: (757) 683-4410; Email: splichta@odu.edu Research Objective: This study examines the acute and chronic health care needs of women seeking refuge in a domestic violence shelter. The battered women's shelter movement in the U.S. began as a feminist grassroots effort in the 1970's and grew to over 1800 places sheltering 200,000+ women per year. Abused women have worse physical and mental health status than other women but few studies have examined the health care needs of women in shelter. Study Design: 65 women sought medical care from a family medicine MD and residents who work in an urban shelter one day each month. Their medical records were abstracted with no identifiers and entered into an SPSS dataset. Women were predominately African-American (71%) with an average age of 35.2 years (80% are under 44). 60% were known to have children and 18% were pregnant or postpartum. This study was approved by the Medical School IRB. Population Studied: Battered women seeking medical care at a monthly program offered as a partnership between the local medical school and a community-based domestic violence shelter. Principal Findings: Women presented with many conditions, including injuries & pain (32%), mental health (19%), upper respiratory infections (19%), gynecological concerns (16%), hypertension (15%) and dermatological issues (8%). Half (51% )had one or more preexisting diseases including: hypertension (32%), asthma (22%), arthritis (19%), diabetes (9%) and seizure disorders (5%). Conclusions: Shelters provide an invaluable resource to women who are harmed by interpersonal violence. However, due to a lack of resources, few can help women address their health concerns. These results clearly indicate a need for further work towards models that bring medical care to abused women in shelter. Implications for Policy, Delivery, or Practice: The model of partnering a medical school with a domestic violence shelter needs to be better studied. Different mechanisms for delivering care and ensuring continuity of care should be explored. Primary Funding Source: No Funding Source ●Diabetes Healthcare Quality Report Cards: How Accurate Are The Grades? Leonard Pogach, M.D.,MBA, Minge Xie, Ph.D., Yue Shentue, BS, Chin-Lin Tseng, DrPH, Stephen Crystal, Ph.D., Monika Safford, MD Presented By: Leonard Pogach, M.D.,MBA, Director, Center for Healthcare Knowledge Management, Medicine, VA New Jersey Healthcare System, 385 Tremont Avenue, East Orange, NJ 07018; Tel: (973)676-1000 x1693; Fax: (973)395-7111; Email: len.pogach@verizon.net Research Objective: To evaluate the accuracy and precision of single random samples compared to bootstrapping methodology in identifying facility outliers in adherence to performance measures. Study Design: Cross-sectional analysis of diabetes performance measures obtained from administrative databases from 79 Veterans Health Administration (VA) facilities serving 250,317 diabetes patients (1,015-13,598/ facility) between October 1, 1999 and September 30, 2000.Patients with diabetes were identified using HEDIS algorithms and had at least one face-to-face encounter in both years as a proxy for continuous enrollment. To minimize bias potentially attributable to non-random errors in measurement or data collection, we eliminated facilities using nonstandardized laboratory methods and those greater than 4 standard deviations from the mean. Measures were poor hemoglobin A1c (test not done or above threshold), good lowdensity lipoprotein cholesterol (LDL-C test performed and below threshold), and good blood pressure (BP) control. Facility adherence to each measure was calculated using 150 separate single random sample sets of 411 subjects (comparable to HEDIS criteria) with and without age adjustment. The accuracy and precision of the random sample was determined by comparison to bootstrapping of 1000 samples as the criterion standard in defining outlier status defined as greater than two standard deviations or 5% difference from the mean, or classification within the 10th or 90th percentile. Population Studied: The population was largely male (97.5%), with 53.7% of patients > 65 years of age. For glycemic control, 22.8% had last A1c>9.5% and 33.1% had A1c>8.5%; for lipids, 53.1 % had last LDL-C<130 mg/dL and 29.8% had LDL-C<100 mg/dL; for BP, 44.7% had last BP>140/90 mmHg and 67.3% had BP>130/80 mmHg. Principal Findings: There was substantial variation among the facilities, with 2 to nearly 3-fold differences between best and worse performers. For A1c, the worst performing facility had 34.7% of its diabetes patients with A1c >9.5% and the best performing facility had just 13.2%. For LDL-C, the worst performing facility had 32.3% of its diabetes patients with LDLC <130 mg/dl, and the best performing had 68.2%. For BP, the worst performing facility had 43.2% of its diabetes patients with BP <140/90 mmHg, and the best performing had 67.8%. Variations were similarly large across lower thresholds for the performance measures (A1c >8.5%, LDL <100 mg/dl, BP >130/80 mmHg). Evaluation of positive predictive value (PPV) and sensitivity (Sn) suggested that all of the random sampling approaches were both inaccurate and imprecise in comparison to bootstrapping: PPV for A1c >9.5% mostly near .65 (range .18 to 1); Sn mostly near .75 (range .17 to 1).The results were similar for all performance measures for each of two thresholds. Stratified random sampling on age did not improve precision, and different outlier detection approaches did not change the results. Conclusions: The use of a single random sample to identify VA facility outliers using individual diabetes performance measures results in considerable misclassification compared to the bootstrapping method applied to the eligible facility population. Implications for Policy, Delivery, or Practice: Currently used methods to determine outliers on performance measures lack sufficient reliability for accountability. Performance measures that reflect the entire eligible population may greatly alleviate this shortcoming. Primary Funding Source: VA ●A Comparison of Populations and Orthopaedic Outcome among an Elderly Patient Population Served by two Distinct Urban Hospitals Wendy Preisman, DrPH, Charles Cornell, M.D., Margaret Peterson, Ph.D., Stephen Lyman, Ph.D., Laura Robbins, DSW, Carol Mancuso, M.D. Presented By: Wendy Preisman, DrPH, Research Analyst, Hospital for Special Surgery, 535 East 70th Street, 8 West, New York, NY 10021; Tel: (212)606-1201; Fax: (212)794-1309; Email: preismanw@hss.edu Research Objective: To assess postoperative patient satisfaction outcome among elderly patients served by two distinct New York City based institutions--a specialty orthopaedic hospital and a general community hospital. Study Design: From April to December 2004, patients from both study institutions who had undergone total hip or total knee arthroplasty were invited to participate in this survey based study. Patients completed a self-administered measure of patient satisfacion, "Patient Judgements of Hospital Quality", during their 6 week postoperative visit. Population Studied: 33 patients completed the satisfaction measure; 21 from the specialty hospital and 12 from the community hospital. The average age was 70 years for patients treated at the specialty hospital, and 66.7 years for those treated at the community hospital. 68% of the study population was female at the specialty hospital, and 55% at the community hospital. Principal Findings: Those treated at the specialty orthopaedic facility rated the following variables significanty higher (p=.05 or less) than those treated at the community institution: medical facilities, overall efficiency, consideration of their needs, ease of getting information, nursing skill and competence, and information given by nurses. Moreover, the condition of the hospital room and efficiency of discharge were also rated significantly higher at the specialty institution (p=.05 or less). Significant differences did not emerge in regards to satisfaction with surgical care. Interestingly, a nonsignificant trend emerged among those treated at the community hospital indicating that patients were more likely to return for care or recommend that hospital to others. Moroeover, this cohort reported nonsignificant, yet higher overall satisfaction ratings than those treated at the specialty institution. Those treated at community hospitals also reported that the location of the institution (p=.002) and financial considerations (p=.003) were important factors that highly influenced their choice of hospital. Conclusions: Despite significantly higher ratings on questions that inquire about satisfaction within specific care domains (ie: nursing care, discharge), patients treated at community hospitals reported that they are generally satisfied with their medical care and hospital stay. Moreover, they stated they are likely to return to the same institution and recommend it to others. The apparent importance of choosing a local, affordable neighborhood institution may override certain clinical and service factors in elderly patients' assessment of their overall satisfaction with their hospital stay. Implications for Policy, Delivery, or Practice: This study highlights the importance that accessing local and affordable medical care holds for elderly patient populations. Future research should further investigate which care factors are most valued by elderly patients. Primary Funding Source: United Hospital Fund ●Psychometric Properties of Scales Measuring the Quality of Cancer Care at an Urban Safety Net Hospital Laurel Radwin, Ph.D., RN, Gail Wilkes, MS, RN, AOCN, Joanne Garvey, MS, RN, Linda Curtin, MS, RN, John Whitehouse, BS, Lisa Tracey, BS, RN Presented By: Laurel Radwin, Ph.D., RN, Assistant Professor, Nursing, University of Massachusetts Boston, 100 Morrissey Boulevard, Boston, MA 02125; Tel: (617)287-7572; Email: laurel.radwin@umb.edu Research Objective: To assess the preliminary psychometric properties of scales measuring the quality of cancer nursing care in a diverse sample of hospitalized patients. Study Design: Twenty-two cancer patients participated in a qualitative study of the characteristics and outcomes of excellent nursing care. Characteristics of excellent nursing care, as described by those participants, were conceptualized as patient-centered nursing interventions. Patient-centered nursing interventions were operationalized by the Oncology Patients’ Perceptions of the Quality of Nursing Care Scale (OPPQNCS) and comprise the following: responsiveness, individualization, coordination and proficiency. The OPPQNCS demonstrated acceptable psychometric properties in a sample of 436 patients with cancer in active treatment. These patients were mainly female (66%), mainly White (93%), and well educated (81% attended some college). Desired outcomes of excellent nursing care, also described by the qualitative study participants, were operationalized by the Trust in Nurses Scale, Cancer Patient Optimism Scale, Fortitude Scale, and Authentic Self-Representation Scale. These scales demonstrated acceptable properties in a pilot sample of 66 community-based cancer support group attendees. These patients were mainly female (77%), mainly White (97%) and well educated (80% attended some college). The OPPQNCS and the outcome scales are being used to measure the quality of cancer nursing care in an ongoing large scale study of the relationships among nursing interventions, patient and system characteristics, and outcomes. The study will include psychometric analyses of the OPPQNCS and desired outcome scales. Properties of the OPPQNCS and outcome scales for the sample as of May 1, 2005 will be described by computing the ranges, means, standard deviations, quartiles, and skewness statistics of the items and scales. Preliminary psychometric properties will be assessed, including the scales’ internal consistency reliability measured as Cronbach’s alpha; item-scale correlations; intra-scale correlations; and item-level missing data. Population Studied: Participants are being recruited from a 29-bed inpatient hematology-oncology unit at an urban medical center that is the largest safety net hospital in New England. January 1, 2005 analyses of patient (n=106) characteristics revealed the following indicants of sample diversity: 53% male; 59% White, 31% African American/Black; 1% Asian, 3% American Indian or Native Alaskan; and 12% Hispanic. Most participants (63%) have a high school education or less. Principal Findings: The diversity of the current study sample provides an opportunity to analyze the psychometric properties of the OPPQNCS and the outcome scales in a racially, ethnically, and educationally diverse sample. Analyses as of January 1, 2005 indicate that psychometric properties for the OPPQNCS in the current more diverse sample are similar to those found in the earlier sample. However, properties for the desired health outcome scales appear thus far to be different in the current sample. Conclusions: The results of this study will provide estimates of the psychometric properties of scales measuring the quality of care for hospitalized cancer patients at an urban safety net hospital. Implications for Policy, Delivery, or Practice: Reliable and valid scales are needed to evaluate and improve the quality of care for patients traditionally at risk for unequal treatment. Primary Funding Source: AHRQ ●Clinical Practice Guideline for Acute Gastroenteritis in Children Virginia Rahm, MSN Presented By: Virginia Rahm, MSN, Pediatric Nurse Practitioner, Pediatric Care Center, Children's Mercy Hospital and Clinics, 2401 Gillham Road, Kansas City, MO 64108; Tel: (816)234-3898; Fax: (816)234-3081; Email: grahm@cmh.edu Research Objective: Acute Gastroenteritis in the pediatric population has traditionally been evaluated by laboratory blood work and subsequently treated with intravenous fluids. The research based literature has shown that there are more cost efficient and less intrusive ways to treat Acute Gastroenteritis. The objective was to translate cutting edge research into the clinical environment by developing and implementing a Clinical Practice Guideline (CPG.) Study Design: A multidisciplinary team collaborated to create an algorithm which reflects the care of a child with Acute Gastroenteritis as well as review the current research literature and assess the Evidence Based findings related to Acute Gastroenteritis. Population Studied: There are two populations effected by the Acute Gastroenteritis CPG. These populations are the patients, as practice changes, and the medical staff (physicians, pediatric residents, nurse practitioners and staff nurses) in three environments: inpatient, outpatient and the community. Principal Findings: The CPG went into effect December 2004. The literature findings proved that oral rehydration is the gold standard for rehydrating mild and moderately dehydrated patients rather than intravenous (IV) rehydration. Also the evidence showed that the physical assessment of the patient performed by the medical provider was superior to obtaining blood work for determining the extent of the dehydration. A major component of the CPG was educating the medical staff on the rationale for changing their practice. Conclusions: The main impetus for developing the CPG was to determine best practice based on current evidence. There is an identified lag time in translating research into practice. In order to assist the medical staff, the CPG Team researched and synthesized the literature, provided a working algorithm as a visual, and created standardized order sets found in the Physician Order Management computer system. Implications for Policy, Delivery, or Practice: The use of this CPG will benefit our families by educating them on how to comfortably perform oral rehydration therapy in their own environment, the home. It will also have an economic affect on our organization evidenced by a decrease of laboratory work and IV rehydration performed on these children. By adopting these standards the organization should experience a decrease in Emergency Department utilization for the patient with Acute Gastroenteritis. Primary Funding Source: No Funding Source ●3D: A Tool for Improving Patient Satisfaction and Medication Safety at Discharge Ahmed Rahman, BS, Danica Myhre, BS, Dennis M. Manning, MD, FACP, FACC, John O’Meara, Pharm.D, Karyl Tammel, Lisa Carter, RN, MSN, Arthur R. Williams, Ph.D. Presented By: Ahmed Rahman, BS, Senior Analyst/Programmer, Health Care Policy & Research, Mayo Clinic, 200 First Street SW, Rochester, MN 55905; Tel: (507)538-1543; Fax: (507)284-9542; Email: rahman.ahmed@mayo.edu Research Objective: To determine whether a newly designed form, Durable Display at Discharge (3D) compared to a standard form, Medication Discharge Worksheet (MDW) improves patient satisfaction, understanding and safety of hospital discharge medications. Study Design: This is a randomized, prospective, clinical trial. Over 200 adult subjects, discharged from medical services at Mayo Clinic Rochester are being randomized to either 3D or MDW. The 3D includes several new elements: a) number of tablets/pills, b) indications, c) cautions, d) reconciliation (medications you should no longer take…), e) space for display of the tablets and f) suggested time of day to take each medication. A secure custom webbased system was designed to generate 3D forms and accommodate data collection. In order to assure accurate and complete data collection, the system provides automatic data validations and pre-loads patient data from enterprise databases when possible. A study recruiter who has not been involved in patient care enrolls and randomizes the patients into the study using the web-based system. Before leaving the hospital, patients in the 3D arm are given a personalized 3D form generated by the web-based system. The MDW is generated by the patient’s nurse in the usual fashion. A nurse then educates the patient about their discharge medications and the form they are given. A Study Investigator (blinded to the hypothesis) calls the patient 7-14 days after the patient is discharged and conducts a follow-up survey using the webbased system. Population Studied: Adults 21 years of age and older being discharged from a teaching hospital to home self care or to the care of a relative. Patients are receiving 4 or more regularly scheduled medications. Principal Findings: At this point, subjects eligible for analysis have exceeded 150. Qualitative analysis of our primary endpoint (satisfaction) shows a 10% but non-significant difference favoring 3D (85% vs. 75%, t-test = 1.47). We are “under-powered” at present. At the study conclusion, we will have final data on satisfaction, and secondary (though important) endpoints related to medication errors and medication knowledge – which have not yet been preliminarily analyzed. Conclusions: Little is known about the relative merits of types of Medication Discharge Education tools. At the conclusion of our 200+ subject study, our data will add several new elements of knowledge to this seldom studied topic. The strengths of the study are: a) focus on patient safety and satisfaction, b) randomized design, c) web-based tool application for data maintenance, d) endpoints of selfreported errors, and e) an objective test (quiz) of personalized medication knowledge and understanding. Implications for Policy, Delivery, or Practice: Optimizing medication discharge education is a daunting challenge. Our study will add to knowledge concerning efficacious methods for medication regimens. Improving the design of a discharge medication education tool may enhance patient knowledge and lead to improvements in both patient safety and understanding. Primary Funding Source: Mayo Foundation for Medical Education and Research ●Enhancing Patient-Centered Communication Behaviors and other Outcomes: Do Physician and Patient Interventions make a Difference? Jaya Rao, M.D., MHS, Lynda Anderson, Ph.D. Presented By: Jaya Rao, M.D., MHS, Medical Epidemiologist, Health Care and Aging Studies Branch, Centers for Disease Control and Prevention, 4770 Buford Highway NE, MS K-51, Atlanta, GA 30341; Tel: (770)488-5091; Fax: (770)488-5486; Email: jrao@cdc.gov Research Objective: Effective physician-patient communication, a critical component of health care quality and patient safety, is associated with improved patientcentered outcomes. We conducted a systematic review of interventions designed to enhance the communication behaviors of physicians and patients in outpatient clinical settings. Study Design: Nine online databases were searched between 1966 and 2004 using search terms focusing on communication partners and behaviors, relationship-building behaviors, and intervention type(s), and the references of potentially relevant articles were hand searched. Eligible studies: 1) tested an intervention designed to improve the communication skills of physicians or patients; 2) were randomized controlled trials (RCTs); 3) objectively assessed verbal communication behaviors as the primary outcome; and 4) were published in English. Two investigators abstracted information on the study sample, intervention, and communication and non-communication outcomes. Interventions were characterized by type (e.g., information, modeling, feedback, practice), delivery strategy, and overall intensity. Communication outcomes were categorized into interpersonal or information-exchanging behaviors within the context of patient-centered care principles. The intervention’s effects on communication and non-communication outcomes were assessed. Population Studied: Physicians and patients in outpatient clinical settings Principal Findings: Thirty-two RCTs were reviewed: 14 involved physicians; 15, patients; and 3, both physicians and patients. Nearly all physician interventions were rated as moderately or highly intense, 78% included 3 or 4 types, and most were delivered over time. Of all patient interventions, half were low intensity, 41% involved only 1 type, and most were delivered in the clinic waiting room. Compared to controls, physicians receiving these interventions were more likely to exhibit a patient-centered communication style and elicit patients’ perspectives. Similarly, intervention patients obtained more information per question and exhibited greater involvement in the visit than controls. Most studies found no post-intervention effects on visit length or satisfaction. Some studies, especially those involving patient interventions, found improved patient adherence, recall and health status. Conclusions: Interventions to enhance communication were associated with improved communication behaviors among physicians and patients. The intervention intensity appeared to be clearly related to improvements in physicians’ communication behaviors, but this relationship was less pronounced among interventions involving patients. Communication interventions directed toward patients resulted in improved health status, adherence and recall. Implications for Policy, Delivery, or Practice: The studies included in this review demonstrate that the communication behaviors of patients and physicians are modifiable. However, most of the interventions examined are not feasible for practice because they require additional personnel and multiple sessions. The future challenge in this field is to design effective communication interventions that can be integrated into routine practice. Primary Funding Source: No Funding Source ●Quantity and Cost of Informal Care Before Death Among Community Dwelling Elderly YongJoo Rhee, Dr.PH. Presented By: YongJoo Rhee, Dr.PH, Norfolk, VA 23513; Tel: (757)853-5230; Email: yjrst@yahoo.com Research Objective: This study estimates informal care cost for the dying elderly in communities. Study Design: The replacement cost for informal care was estimated using a two-part model owing to skewed data distribution. A probit model is used to estimate the probability of informal care use. For cases with positive hours, ordinary lease squares regression is used to estimate the natural logarithm of hours. The derivative of expected hours is estimated for each observation and, then averaged over full sample. The derivates of expected hours are summed to get the effect of the independent variables on total hours of care Population Studied: This study uses data from the 1993 Asset and Health Dynamics Study, a nationally representative survey of people age 70 or older (N = 7,443). Proxy informants for people who died (n=727) before 1995 follow-up were interviewed Principal Findings: These adjusted hours are multiplied by 1998 national average home aide wage ($8.20 per hour). . Sensitivity tests are performed using the 10th percentile wage rate ($5.90) and 90th percentile wage rate ($10.80). All analyses are conducted using sample weights to generate estimates of the economic costs of informal caregiving before death at a national level. Conclusions: Family burden is still significant to take care of the dying elderly. The reasonable financial estimates should be prepared when this informal care is replaced with formal care givers. Implications for Policy, Delivery, or Practice: This study results could use this estimation in planning for individual or family, community for the end-of-life. These estimates should be compared with institutionalized care, nursing home and home care. It should take into consideration to provide federal funded programs or support for families who deal with the dying elderly at their homes. Substantial financial support should be designed for family caregivers in communities Primary Funding Source: Other ●Controlling for Patient Case Mix at the End of Life: Issues in Identifying Cause of Death Samuel Richardson, BA, Wei Yu, Ph.D. Presented By: Samuel Richardson, BA, Research Health Science Specialist, Health Economics Resource Center, VA Palo Alto Health Care System, 795 Willow Road (152 MPD), Menlo Park, CA 94025; Tel: (650) 493-5000 x22816; Fax: (650) 617-2639; Email: Samuel.Richardson3@med.va.gov Research Objective: Researchers investigating health care at the end of life often control for patient case mix using the death certificate underlying cause of death. It is unclear whether the reported underlying cause of death is consistent with the reason to control for case mix. The purpose of this study was to examine whether the reported underlying cause of death was consistent with the medical condition that required the most resource use or the last important principal diagnosis during the final year of life. Study Design: We obtained underlying cause of death as reported on the death certificate for 12,808 VA patients who died in California between October 1999 and September 2001. Using administrative data from both VA and Medicare, we compared the reported underlying cause of death with the major principal diagnosis closest to death and with the diagnosis responsible for the plurality of health care costs in the final year of life. We measured agreement between methods at three different levels of identification: the top two causes of death, the top four causes of death, and the top 10 causes of death, using CDC cause of death categories. Population Studied: Our study sample consisted of veteran decedents in California, but the results should be generalizable to a broader population of decedents. Principal Findings: All three methods classified similar numbers of patients into each cause of death category. However, neither of the methods using administrative data was consistent with the death certificate underlying cause of death at the patient level. For the top 10 causes of death, the patient-level consistency was slightly above 50%. As a whole, these methods were more consistent in identifying cancer deaths than deaths due to other causes. Conclusions: The three methods generate considerable discrepancies in patient classification. The principal diagnosis and cost-based methods are relatively consistent with one another, but not with death certificate data. Implications for Policy, Delivery, or Practice: Our results suggest that methods using health utilization data should not be used as a proxy for death certificate data. However, for many research purposes, methods based on costs in the final year of life or principal diagnoses near death may in fact be more appropriate than death certificate data. Primary Funding Source: VA ●Exploring Racial Differences in Infectious Disease Mortality Risk in the United States Kakoli Roy, Ph.D. Presented By: Kakoli Roy, Ph.D., Economist, Office of Workforce and Career Development, Centers for Disease Control and Prevention, 1600 Clifton Road NE, MS E90, Atlanta, GA 30333; Tel: (404)498-6298; Fax: (404)498-6145; Email: kjr3@cdc.gov Research Objective: To measure the extent to which blacks and whites differ in their risk of infectious disease mortality and whether these differences persist on controlling for socioeconomic factors. Study Design: Several multivariate logistic regressions were used to measure the association between mortality risk and race and examine whether racial differences in mortality risk persist upon controlling for socioeconomic status (SES) and other demographic factors. Population Studied: The nationally respresentative National Health Interview Surveys, 1990-1994, were linked to the National Death Index, 1990-1997. The final sample includes five cohorts with 335,918 persons aged 18 years or more, 5.13% of whom died during the 4-8 years of follow-up. Principal Findings: Mortality risk from infectious diseases is higher among blacks compared to whites; controlling for SES narrows the black-white gap with odds ratios declining from 1.88 to 1.60. Controlling for the same factors also narrows the black-white gap in mortality risk for both all-cause and noncommunicable diseases as odds ratios decline from 1.34 to 1.13 and 1.28 to 1.09, respectively. However, the race effect on injury-related deaths, on adjusting for SES, was found to be statistically insignificant. Conclusions: Even after controlling for SES and other demographic factors, we determined that blacks have an increased risk of death from infectious diseases compared with whites. However, because infectious diseases are diverse with respect to race predilections, future research is needed to examine whether race-based effects persist for specific infectious diseases. Implications for Policy, Delivery, or Practice: Such information could also contribute toward more targeted policy formulation by identifying individual diseases for which the effect of race is most pronounced. Primary Funding Source: No Funding Source ●Educating Culturally Competent Nurse Practitioners in Virginia Leorey Saligan, MS, APRN, BC-FNP, CCRN, Laurel Garzon, DNSc, CPNP, Carolyn Rutledge, Ph.D., CFNP, Richardean Benjamin, Ph.D., MPH, Stacey Plichta, Sc.D. Presented By: Leorey Saligan, MS, APRN, BC-FNP, CCRN, Lecturer, School of Nursing, Old Dominion University, Hampton Boulevard, 2nd floor, Hughes Building, Norfolk, VA 23529; Tel: (757)683-5010; Fax: (757)683-5194; Email: LSALIGAN@odu.edu Research Objective: 1. To equip nurse practitioner students with the knowledge, skills, awareness, encounters, and desire to work with underserved or disadvantaged populations. 2. To create a Cultural Competency Infrastructure that will facilitate the integration of cultural competency programs and recruitment of diverse students and faculty into the family, women’s health, and pediatric nurse practitioner programs. Study Design: Utilized both qualitative and quantitative methodologies. Qualitative data gathered from focus groups using people representing various ethnic, economic, religious, gender and sexual preference situations, as well as special health needs such as mental illness, HIV and AIDS. Data gathered from these focus groups are utilized by standardized cases using professional actors as patients, where nursing students practice their cultural sensitivity skills. These actors represent patients from various socioeconomic backgrounds and they provide feedback to nursing students to better understand patient care. This role-playing is part of a program with the Eastern Virginia Medical School’s Thomas Center in Norfolk, Virginia, but they also travel to several Old Dominion University’s TELETECHNET sites to provide similar experiences to the distance learning nursing students. The entire didactic sessions would include utilizing lectures, guest speakers, and panels representing various cultures; clinical activities; training and feedback utilizing standardized patients; and participation in community multicultural experiences. Pre, middle, and post cultural competency measurements are obtained from the participants utilizing Campinah-Bacote's cultural competency model. Population Studied: Nurse Practitioner students enrolled in the Family, Women's health, and pediatric nurse practitioner programs of Old Dominion University, School of Nursing. Principal Findings: Ongoing study Conclusions: Ongoing study Implications for Policy, Delivery, or Practice: The goals of this project are consistent with the national goals of improving access to quality health care through appropriate preparation, composition, and distribution of the health profession’s workforce; and to improve access to a diverse and culturally competent and sensitive health profession’s workforce. Primary Funding Source: HRSA ●Factors Associated with Depression Screening by Certified Nurse Midwives/Certified Midwives Lorraine Sanders, DNSc, CNM Presented By: Lorraine Sanders, DNSc, CNM, Assistant Professor of Nursing, Adelphi University, 3087 Shore Drive, Merrick, NY 11566; Tel: (516)867-7840; Email: sanders@adelphi.edu Research Objective: To describe the depression screening practice of CNMs/CMs and to examine relationships, using the Theory of Planned Behavior, among attitude, perceived ability, knowledge, and significant demographics on depression screening. Study Design: Self-report survey. Population Studied: Certified nurse midwives/certified midwives (n=387) attending the 2004 American College of Nurse Midwives annual meeting. Principal Findings: Only 25% of respondents (n=95) reported that they always screened women for depression and only 12.4 % of respondents (n=47) knew of the depression position statement published by ACNM. Using multiple regression, a relationship was established between attitude (pr =.27, p = .000), perceived ability (pr =.25, p = .000), knowledge ( pr =.16, p = .004), education level (pr =.12, p = .036) and depression screening with attitude and perceived ability having the strongest relationship. Conclusions: Attitude, perceived ability, knowledge, and education account for 20% of the variance of depression screening among CNMs/CMs. Study findings suggest that depression screening is not fully integrated into the practice of many CNMs/CMs. Implications for Policy, Delivery, or Practice: Educational intervention may positively influence attitude, ability, and knowledge, which in turn, may increase integration of depression care into midwifery practice. The provision of universal depression screening may improve health outcomes to women and their families. Further research is needed to assess screening methods, interventions for the treatment of depression, and evaluation of institutional barriers to depression screening. Primary Funding Source: No Funding Source ●Emerging Uses of Historical New Jersey Health Statistics Information in Statewide Reports from 1877 to 2000 Marcia M. Sass, Sc.D., Mark C. Fulcomer, Ph.D., Brian Little, DO, Donna DiBruno, DO, Michael W. Holton, Ann Jones, Presented By: Marcia M. Sass, Sc.D., Assistant Professor, Health Systems and Policy, UMDNJ-SPH, 32 Rebecca Court, West Trenton, NJ 08628; Tel: (609) 771-9086; Fax: (609) 7719086; Email: MarciaSass@aol.com Research Objective: Research Objectives: (1) to describe the background and methods employed in developing a major resource of historical health information; and (2) to describe some of the primary uses of the resulting electronic reports, particularly in student projects. Study Design: After first describing the development of electronic versions of the historical New Jersey health statistics reports from 1877 to 2000, this presentation highlights emerging uses of the unique collection that resulted. Work on assembling this complete collection of public documents began in 1999 as part of the preparation of two major bibliographies that have been described earlier (Fulcomer et al, 2002; Fulcomer et al., 2003; and Fulcomer et al., 2004). Containing over 1,000 distinct entries, the first bibliography was started in 1969 and covers the literature related to discriminant functions and classification up until 1978. Based on some research on the application of somewhat similar prediction methods to sudden infant death syndrome (Fulcomer et al., 1981), a second bibliography on infant mortality (IM) and adverse reproductive outcomes (AROs) was organized and continues to expand, now in excess of 300 entries. Population Studied: Population of New Jersey from 1877 to 2000. Principal Findings: Initially a few of New Jersey health statistics reports, particularly those from decennial years, were viewed as a small subset of the ARO collection and incorporated into a 1999 epidemiology course. However, as that set of reports joined some other contributions from early peer-reviewed sources (e.g., Holt, 1913), in subsequent courses, it soon became apparent that a complete set of the reports in conveniently accessible electronic formats would provide a valuable resource for courses. Because the reports are one of the country’s earliest series of uniformly collected, population-based health statistics, the information they contain can provide valuable insights into the evolution of the public health system, not only within the state and the nation but also in some instances with international perspectives. After completing the systematic copying of the over 45,000 pages found in the 131 reports, a process that began in the fall of 2002, the scanning of the page images and converting them into PDF versions (for use with Acrobat 5.0 or later) was finished by August, 2004. First used in their entirety in the fall 2004 epidemiology courses, the entire collection is now available for distribution at minimal cost on three CDs or a single DVD. Conclusions: In addition to dissemination to libraries throughout New Jersey and the integration of some of the information into lectures, including bioterrorism preparedness and health disparities, this presentation focuses primarily on the use of the reports in student projects. Among some of the reports that will be summarized are: (1) preventing infant mortality through vaccination; (2) the impact of milk safety on infant mortality; (3) an overview of segregation and economic status on infant mortality; (4) the influence of maternal age on fetal mortality; (5) a critical review of risk factors for infant mortality from an international and historical perspective; and (6) a history of the reporting of fetal deaths. Implications for Policy, Delivery, or Practice: Clearly, despite the large number of years they span, the reports provide historical perspectives that enrich the understanding of many current-day health issues. Primary Funding Source: No Funding Source ●Smoking Status, Health Status and Health Care Utilization among Chinese Americans Donna Shelley, M.D., MPH, Marianne Fahs, Ph.D., Rajeev Yerneni, Ph.D. candidate Presented By: Donna Shelley, M.D., MPH, Assistant Professor, Sociomedical Sciences, Columbia University Mailman School of Public Health, 722 West 168th Street, New York, NY 10032; Tel: (212)305 0068; Fax: (212)305 0506; Email: drs26@columbia.edu Research Objective: Chinese Americans constitute the largest segment of Asian Americans in the U.S. and are at particularly high-risk for excess tobacco-related morbidity and mortality. Yet, health care access, health status and utilization of health care services among Chinese Americans remain largely undocumented, and there are no studies examining the association of smoking status with health care access and utilization in this population. This National Cancer Institute funded study describes and compares several indicators of health care access and utilization among smokers and non smokers in a representative sample of Chinese Americans living in NYC. Study Design: In-person interviews were conducted with 2357 representative adults ages 18-74 using a Chinese language instrument based on validated questions from several national surveys. Population Studied: Chinese Americans Principal Findings: Overall 67% of respondents were employed, 70% had some type of health insurance, 67% reported excellent health, 72% saw a health professional in the past 12 months. Ninety seven percent were foreign born and only 22% were acculturated based a composite of two categorical variables regarding language and media. When compared to ever smokers, never smokers were significantly more likely to have health insurance (72% vs 64%, p<.001), report health status as excellent (69% vs. 63% p<.05), and to have seen a health professional in the past 12 months (75% vs. 65%, p<.001). Results comparing current and former smokers and the influence of acculturation and other demographic characteristics will be presented. Conclusions: This population based study demonstrates gaps in access to care among Chinese American immigrants. Results suggest that current and former smokers, both at higher risk for tobacco related morbidity and mortality, are less likely to receive preventive and treatment services. Implications for Policy, Delivery, or Practice: Our findings have important policy implications for immigrant health. Results provide new information about health care access among a high risk population and will inform future research to reduce tobacco-related health disparities among Chinese Americans. Primary Funding Source: NCI, Centers for Disease Control and Prevention ●Evaluation of a Maternity-Related Post-Discharge Program in a Managed Care Organization Alpa Sheth, MPH, Beatriz Jaramillo, Dr.PH, Joyce Co, MPH Presented By: Alpa Sheth, MPH, Senior Health Services Analyst, Medical and Quality Informatics, HIP Health Plan of New York, 55 Water Street, 5th Floor, New York, NY 10041; Tel: (646) 447-6371; Fax: (646) 447-3164; Email: asheth@hipusa.com Research Objective: In a given year, there are approximately 3.9 million births in the United States. Though most efforts for pregnancy care are focused on the pre-birth period, post delivery is also a very critical time for the health of the mother and her baby. Problems, if not treated promptly and effectively in the postpartum period, can lead to ill health and even death for both the child and mother. It is recommended that mothers receive postpartum visits and newborns receive initial follow-up visits in a timely manner. For women enrolled in HIP Health Plan of New York who delivered a baby in 2001, only 43.23% had a postpartum visit in a timely manner. For babies born to mothers enrolled in HIP during 2001, only 6% received a follow-up visit during the first 14 days of their life. Consequently, a post-discharge program was developed at HIP in order to improve performance of these measures and provide better care. The post-discharge program consists of a survey administered telephonically within 48 hours of discharge by HIP staff to all females with a maternity-related discharge in order to followup on the adequacy of the hospital discharge plan and facilitate member’s compliance with postpartum visit and baby’s first pediatric visit. The program was initiated in 2002. The objective of this study is to evaluate the effectiveness of the post-discharge program in increasing compliance with postpartum visits and newborn follow-up visits. Study Design: The program was evaluated by: A) comparing the proportion of female members having a postpartum visit 0 to 60 days after delivery between those members who completed the survey and those who did not during 2002, utilizing a Z-statistic B) comparing postpartum and newborn follow-up visit rates over a period of three years (2001 through 2003) utilizing Bartholomew’s test. In both cases, significance was set at alpha < 0.05. Population Studied: All women with a maternity discharge during 2001, 2002 and 2003 who were continuously enrolled for the 60 days post delivery. All newborns continuously enrolled in the plan for the 30 days after birth during the same 3 years. Principal Findings: The proportion of female members with a postpartum visit was statistically significantly higher among the surveyed (39.93%) members compared to the nonsurveyed (30.07%) ones (z =2.16, p<0.05). In addition, there were statistically significant upward trends in the proportion of women complying with postpartum visits (p<0.05) and the newborn follow-up visit rate (p<0.005) over the three-year period from 2001 to 2003. Conclusions: The results indicate that the post-discharge survey program has been effective in increasing the proportion of members with a maternity-related discharge who receive a postpartum visit and the proportion of newborns with a follow-up visit within 14 days after birth. Implications for Policy, Delivery, or Practice: The positive impact of the program shows that telephonic contact is effective and may offer a useful avenue for identification and timely treatment of postpartum depression. Since this program has delivered significant results, it will continue at HIP. Primary Funding Source: HIP Health Plan of New York ●Perceptions of Racial/Ethnic Discrimination and SelfAssessed Mental and Physical Health among African American, Hispanic, and Asian American Veterans Linda Sohn, M.D., MPH, Nancy D. Harada, PT, Ph.D. Presented By: Linda Sohn, M.D., MPH, Physician, Geriatrics, VA GLAHS, 16111 Plummer Street, Sepulveda, CA 91343; Tel: (818)891-7711 x4028; Fax: (818)895-5817; Email: Linda.Sohn @med.va.gov Research Objective: The objective of this research is to examine the association between perceptions of racial/ethnic discrimination and self-assessed mental and physical health among African American, Hispanic, and Asian American veterans. Study Design: Secondary data analyses of the 2001 Veteran Identity Program (VIP) Survey, which collected healthcare utilization data from 3,227 veterans residing in the Veterans Integrated Service Network 22 (VISN 22) covering Southern California and Southern Nevada. The dependent variables, mental and physical functioning, were measured by the SF-12 Health Survey. Independent variables included perceived racial/ethnic discrimination (during past military service and/or with a healthcare provider during an ambulatory care visit in the past year) and sociodemographic variables (age, race/ethnicity, income, education, employment status, marital status, social support, racial/ethnic residential composition, health insurance, regular source of medical care and smoking history). Univariate and bivariate statistics were run to examine distributions, frequencies, and associations between pairs of variables. Ordinary least squares (OLS) regression was conducted to model the mental and physical functioning scores as a function of the two discrimination variables and the sociodemographic variables. Population Studied: 1,737 African American, Hispanic and Asian American veterans were included in the analytical sample. The average age of the sample was 57 years, about 50% of the sample was married, and 97% reported having a regular source of care. Fifty-five percent of the sample used the Veteran’s Administration (VA) exclusively for health care, 32% used the VA in addition to another source for health care, and 13% were non-VA users. Principal Findings: Most sociodemographic characteristics were significantly associated with SF-mental and SF-physical measures of health status. Preliminary results of the OLS regression find that both discrimination variables significantly influence mental and physical functioning, controlling for sociodemographic characteristics. In both models, perceived racial/ethnic discrimination is associated with poorer mental and physical health status. Conclusions: In non-veterans, racial/ethnic discrimination has been demonstrated to adversely affect the health of minority populations. Discrimination may be a source of stress, thus contributing towards a decreased sense of well being. Our study demonstrates that this may also be true for veteran populations: Racial/ethnic discrimination (occurring during past military service and/or during the health care visit) may adversely influence the health of veterans, although the mechanism by which this occurs remains to be studied. Implications for Policy, Delivery, or Practice: Our findings point to the need for the Department of Veterans Affairs to begin to develop policies and programs to minimize perceptions of discrimination during the healthcare encounter. This may be achieved by empowering minority veterans with social resources to reduce perceptions of discrimination during a healthcare interaction, or by educating healthcare providers on how to provide culturally sensitive healthcare for specific racial/ethnic groups. Primary Funding Source: No Funding Source ●What Care Management Processes do Medical Groups have for Chronic Disease Care? Leif I. Solberg, M.D., Greg Pawlson, M.D., MPH, Stephen Asche, MA, Sara Hudson Scholle, DrPH, Sarah Shih, MPH, Merry Jo Thoele, MPH, RDH Presented By: Leif I. Solberg, M.D., Director for Care Improvement Research, HealthPartners Research Foundation, PO Box 1524, MS 21111R, Minneapolis, MN 55440; Tel: (952) 967-5017; Fax: (952)967-5022; Email: leif.i.solberg@healthpartners.com Research Objective: To describe the presence and functionality of various care management processes important to quality care for patients with chronic diseases in a varied sample of medical groups in Minnesota. Study Design: We recruited private medical groups for pretesting and assessment of a new survey instrument developed by the National Committee for Quality Assurance (NCQA) for documenting the presence and function of eight care management processes in ambulatory care practices (PSAS or Practice System Assessment Survey). This survey, based on the chronic care model, also included questions about the characteristics and culture of these medical groups. Survey reliability was assessed by both test-retest and inter-rater comparisons and validity was assessed by comparison of survey responses to on-site audits of the processes. The findings were analyzed for presence and function by medical group and in relation to each group’s organizational features. Population Studied: Twenty medical groups that are members of a quality improvement collaborative in Minnesota were identified as representing varied locations, sizes, and quality sophistication, and 17 agreed to participate in this study. The 11 groups that completed the final study (3 others participated in the pretest) had from 1-33 clinic sites apiece. Six medical groups were from rural and five from urban areas, and five of the groups were relatively new to quality improvement. Complete survey responses were obtained from 328 people at 42 clinics in the 11 medical groups out of the 452 mailed out to those with selected roles and professions in each group (73% response rate). Principal Findings: All data have been collected and analysis is underway. We will report on the frequency and perceived function of each of the following processes: 1. Continuity of care 2. Clinical information system 3. Registry 4. Systematic monitoring 5. Clinician reminders 6. Performance tracking 7. Clinical quality evaluation and improvement 8. Care management We also gathered information about each organization’s characteristics as well as the collegiality, quality emphasis, and autonomy as reported by the physicians, nurses, and other selected personnel in each group. This will allow us to suggest possible relationships between the processes, cultural features, and organizational characteristics. Conclusions: The data available from this survey provides information about practice systems important for quality care of people with chronic conditions Implications for Policy, Delivery, or Practice: Although these results are from a relatively small sample of medical groups, they illustrate the type of information on systems and processes now readily available from the PSAS instrument, as well as the extent to which these care management processes are present in these medical groups. This information will be helpful for those interested in measurement of the systems needed for quality care of patients with chronic conditions, whether for internal process improvement, public accountability or performance payment. Primary Funding Source: RWJF ●"It's Broke. Can We Fix It?": A Community Forum Theatre Dialogue on Disparities in Access to Health Care John Sullivan, MA, Mathew Stanford, BA Presented By: John Sullivan, MA, Co-Director: Public Forum & Toxics Assistance Division, Preventive Medicine / Community Health, NIEHS / Sealy Center for Environmental Health & Medicine, 301 University Boulevard, Galveston, TX 77555-1071; Tel: (409)747-1246; Fax: (409)772-1790; Email: josulliv@utmb.edu Research Objective: Purposes of the Forum included: 1) using images and performative tools of TO to analyze social, economic and cultural implications of health care disparities, 2) framing issue experientially to illustrate impacts of service denial, and encourage audience empathy, 3) using interactive facets of TO to encourage participatory dialogue and problemsolving among audience members and performers. (Dialogue was expressed as actions by audience volunteers who entered scene(s), replaced the oppressed protagonist or allied with protagonist, and proposed solutions through stage actions.) Study Design: A workshop convened for training in Theatre of the Oppressed dramaturgy. Participants learned: 1) how image tableaus may represent affective and conceptual elements of the issue, 2) uses of dramatic structure to sequence and focus personal stories, 3) development of scenarios through improvisation, 4) uses of special structures to deconstruct significance of scenic actions, and 5) using audience interaction to invite dialogue, novel actions and processing of spectator interventions. The show was performed for community members in an informal space, using an issue-specific sociometric exercise as a warm-up. Audience selected one scene for the Forum process, and spectators initiated interventions. Interventions were processed, and closure scene was performed by original protagonist. Population Studied: 15 Workshop participants varied by SES / Race / Ethnicity but all had direct experience with denial of access to health care. Audience (64) varied across categories, but medical students and health care professionals composed >50%. Principal Findings: The following perceptions, biases, attitudes and differentiating characteristics emerged: 1) <12% of audience had direct experience with denial of care, 2) audience interventions express judgement that UTMB’s Demand & Access Management Program (DAMP) for rationing care poses a problem for medical ethics, 3) health care access disparities compromise the social fabric of communities, 4) strong ameliorative action is necessary to “transform the system,” 5) none of the proposed interventions gave the scenes satisfying closure. Conclusions: This Theatre of the Oppressed Forum provided a dynamic structure for dialogue and predisposed audiences toward direct/authentic engagement with real life aspects of a difficult and frustrating issue. Though no clear solutions emerged, the audience demonstrated active empathy for the uninsured, and political will to engage public policy-makers. All participants, including primary care physicians, voiced an urgent need for more detailed understanding of DAMP. Audience members generally agreed that DAMP should be simplified and infused with social justice. Implications for Policy, Delivery, or Practice: Care consumers and providers need basic information on mechanics of DAMP – and similar systems – as well as the fiscal parameters behind health care resource allocation. The health care community also needs to engage in further dialogue to develop ethical guidelines for navigating the health care crisis that transcend free market pragmatism, the designation of health care as solely a commodity, and moves the dialogue into the realm of disparities in social justice (eg. John Rawls’s difference principle, per audience member suggestion). Primary Funding Source: Institute for the Medical Humanities @ UTMB / Health Care for All Texas (HCFAT) ●Evaluation of a Diabetes Care Connection Program Using Episode Treatment Groups Deborah Taira, Sc.D., MPA, Kenneth Buchanan, Deborah A. Taira, Sc.D.; MPA, Richard S. Chung, M.D., Brent Hamar, Bill Gandy, James Pope, M.D. Presented By: Deborah Taira, Sc.D., MPA, Health economist and research manager, Care Management, HMSA (BCBS of Hawaii), 818 Keeaumoku Street PO Box 860, Honolulu, HI 96808-0860; Tel: (808)948-5337; Fax: (808)948-6043; Email: deb_taira@hmsa.com Research Objective: To examine the impact of a diabetes disease management program on cost and quality. Study Design: In mid-2000, a comprehensive diabetes care management program was implemented at a large health plan in Hawaii. For this retrospective, observational study, administrative data (1999-2002) was grouped into Episode Treatment Groups (ETGs). Medical directors reviewed the types of episodes and categorized them as being: 1) diabetes (4 ETGs); 2) complication of diabetes (e.g. acute myocardial infarction); 3) comorbidity of diabetes (e.g. obesity); 4) unrelated (e.g. cancer). We estimated pre-program costs for each category, adjusted them using a trend factor based on premium increases, and compared them to post-program costs of care. Quality metrics were also examined pre- and post-program implementation. Analyses were performed separately according to insulin use and age. Population Studied: The study sample included health plan members with diabetes who incurred claims that were able to be grouped into Episode Treatment Groups (n=44,334). Principal Findings: Costs in the post-period were $18.6m less than expected for non-insulin users and $13.6m less for insulin users. For non-insulin users, the cost of complications were $14.1m less than expected, while diabetes-related costs were $2.3m more than expected. Complication reductions were found for renal failure, coronary artery disease, hypertension, and diabetic retinopathy. Rates of A1C, micro albumin, and cholesterol screening all improved significantly after program implementation. Conclusions: Although direct diabetes costs were higher than expected, cost of complications were lower than expected, resulting in estimated overall cost avoidance of more than $32.2m over 2.5 years. Findings were sensitive to trend factor selection. Implications for Policy, Delivery, or Practice: Further research is needed to determine which aspects of the program were most effective. Primary Funding Source: HMSA (BCBS of Hawaii) ●Health-related Quality of Life Among Asian Pacific Americans in Hawaii Compared to Whites Deborah Taira, Sc.D., Kimberly Nakao, Richard Chung, M.D. Presented By: Deborah Taira, Sc.D., Research Manager, Care Management, HMSA (BCBS of Hawaii), 818 Keeaumoku Street; PO Box 860, Honolulu, HI 96808-0860; Tel: (808)9485337; Fax: (808)948-6043; Email: deb_taira@hmsa.com Research Objective: To examine ethnic disparities in healthrelated quality of life for Asian Pacific Americans compared to whites, after adjustment for age, gender, educational attainment, and area income level. Study Design: The Center for Disease Control’s HealthRelated Quality of Life instrument was added to a member satisfaction survey study. The survey had a 40% response rate. Race, ethnicity, and education level were self-reported on the survey. Area income was merged from census files using zip code. We examined both mean number of unhealthy days for each ethnic group and the likelihood of having no unhealthy days. OLS regression and logistic regression models were estimated to determine whether the level of unhealthy days for each Asian Pacific group differed from whites. Population Studied: Adult members of a large health plan in Hawaii who responded to health-related quality of life items on a member satisfaction survey in June of 2004. Principal Findings: After adjustment for age, gender, and socioeconomic status, Japanese, Filipinos, and Chinese had significantly fewer unhealthy days than whites, for both physical and mental health. Hawaiians and patients of mixed race were similar to whites in both areas. The largest disparities were found for days in physical pain, with the odds ratio of not having any pain days being OR=1.9, 95%CI[1.7,2.0] for Japanese, OR=1.6, 95%CI[1.4,1.8] for Filipinos, and OR=1.4, 95%CI[1.2,1.5] for Chinese. The number of unhealthy days increased with age, but decreased with education level. Conclusions: Health-related quality of life differs by race/ethnicity, among health plan members of a large insurer in Hawaii, with Japanese, Filipino, and Chinese members having fewer unhealthy days than whites. Implications for Policy, Delivery, or Practice: Further research is needed to determine whether there are ethnic differences in response bias (i.e. possibly, an understatement of unhealthy days by Asians) and to better understand the implication of these findings on health care cost, utilization, and outcomes. Primary Funding Source: HMSA (BCBS of Hawaii) ●Do Comfortable Bed Positions have a Healing Effect on Patients? Kazue Takayanagi, M.D., Ph.D, Yuri Fujiwara, M.D., Kyoko Sugiura, ME, Nahoko Kakudate, MP, Kazuhiko Okamoto, Dr.Eng, Kazuhiko Nishide, Dr.Eng Presented By: Kazue Takayanagi, M.D.,Ph.D, Associate Professor, Health Services Administration, Nippon Medical School, 1-1-5, Sendagi Bunkyo-ku, Tokyo, 113-8602; Tel: (03)3822-2131; Fax: (03)3822-8144; Email: taka-y@nms.ac.jp Research Objective: We investigated the position of the bed in relation to the comfortableness and a healing effect on human being. Study Design: Apparatus: Four beds are placed parallel to the wall in a narrow cubicle room. The ceiling has the sharp paralleled grid pattern. White partitions between beds were used for the privacy. And then, these beds are placed in slant position to the wall. Procedure: Subjects lie down for 15minutes in each paralleled placed bed to wall. After 30min. rest at outside of the room, subjects lie down for 15minutes in each slant positioned bed to the wall. STAI, A Profile of Mood test (POMS) were administered, and pulse, blood pressure, and catecholamine, ACTH, cortisol, NK cell activity, blood count, CD4/CD8 in the blood were measured, both before and after bed ridden. Population Studied: Fifty-two students (average age 23.1 years, 34 male, 18 female) in Nippon Medical School and The University of Tokyo participated in the study. Informed consent was obtained from all the participants. Principal Findings: The subjects with low and stable anxiety by STAI proved slower to adapt to the new lay down circumstances either on parallel or on slant position of the bed by POMS. On the other hand, the subjects with high and unstable anxiety showed quicker to adapt to new lay down circumstances. In comparison between the parallel and slant bed, the blood pressure increased from av. 102/63 to 103/65 (p<0.05) and ACTH (Stress hormone) increased 5.5% on subjects both in a slant positioned bed. This showed increased stress. CD4 decreased in 5.5% (p<0.01) on subjects in a slant positioned bed, which showed lower immunity. On the contrary, NK cell activity increased on subjects in a slant positioned bed. According to the questionnaire survey, subjects lying down on the slant bed tend to feel a sense of discomfort and some subject were nervous about slant grid on ceiling and partitions. Conclusions: Even healthy subjects showed stress and decreased immunity in 15 minutes bed ridden. Increased physiological indicator, stress hormone showed more stress in the slant positioned bed as well as psychological change. In general, patients may needs more comfortable circumstances. Even subjects with low and stable anxiety by STAI proved slower to adapt to the new lay down circumstances either on parallel or on slant position of the bed by POMS. On the other hand, the subjects with high and unstable anxiety showed quicker to adapt to new lay down circumstances. It is confirmed by the questionnaire survey. The subjects with slant positioned bed may feel occluded space or opened space according to the head position in relation of the wall. Most subjects are used to lie down at a bed situated parallel to the room at home. Position of the bed towards patient's room may have more effect to the patient’s psychological and physiological status in short time bed ridden. Although the healthy subjects can accommodate new environment, it may take time to be acclimatized to new environment for the patient. Implications for Policy, Delivery, or Practice: Bed arrangement for the patient’s room should be chosen as a parallel position to the room or non-grid ceiling should be chosen. Primary Funding Source: No Funding Source ●Health Care Delivery System in Rural Taiwan— What Have we Learned from the Telemedicine Program in Taiwan? Hsiu-Fen Tan, Ph.D., Hung-Fu Tseng, Ph.D. Presented By: Hsiu-Fen Tan, Ph.D., Assistant Professor, Health Care Administration, Chang Jung Christian University, 396 Chang Jung Road, Section 1. Kway Jen, Tainan, 711; Tel: 011886-6-2785110; Fax: 011885-6-2785831; Email: ttan@mail.cju.edu.tw Research Objective: In 1995, the Taiwan Department of Health (DOH) initiated a Telemedicine Pilot Program (TPP) to improve the health services in rural areas, focusing in specialized care, emergency services, and diagnosis consultation. During the past 10 years, the DOH has provided over 200 million NT dollars to the system and the number of service areas has increased from 5 to 24. However, the average number of visits in each location has dropped 80%. The purpose of this study was to find out what the major difficulties have been and what we have learned from the experience of running this pilot program. Study Design: This cross-sectional survey is the first comprehensive evaluation on the TPP in Taiwan. The questionnaire included subjects covering the quality, accessibility, and acceptability of telemedicine care perceived by patients and healthcare providers. Population Studied: The face-to-face interview was conducted among all healthcare providers using the telemedicine system (n=144) and samples of residents living in the areas where telemedicine were available (n=1129). Principal Findings: About 8-20% of the residents were aware of their local telemedicine services, less than 10% of the residents have ever used the system. About 29% of the residents did not think the telemedicine could provide better care, and 25% of the residents would not increase their willingness to seek care in town. About one third of the residents did not agree that telemedicine improve accessibility to specialized care and 10% of the residents expressed their strong opinions against telemedicine. Over 82% of the former telemedicine users expressed that they would have chosen transferring or referral services instead of telemedicine had they have choices. The top three reasons for choosing telemedicine service were saving time and expenses in commuting, needing opinions from specialists, and lack of specialized care in local environment. The top three reasons for not choosing telemedicine were critical emergency, absence of physical contact with doctors, and inconvenient schedule. The top three reasons for local health care providers to initiate telemedicine contact with medical center were consulting for second opinion, lack of specialty, and confirmation of diagnosis. The top three reasons for not initiating the system were self-confidence on diagnosis, inconvenient schedule, and equipment malfunction. Conclusions: Like other countries, part of the difficulty for telemedicine in Taiwan is resistance from healthcare providers and the low acceptability by the public. In small geographic country like Taiwan, the value of the telemedicine is relatively insignificant. Given the better transportation condition, wider distribution of health resources, and freedom of choosing services, the telemedicine inevitably became the last option for rural residents in Taiwan. Implications for Policy, Delivery, or Practice: In smaller country, the value of the telemedicine may not be in providing routine health care services to rural areas, rather, its value is in assisting rural doctors in diagnosis to determine if transferring or referral is necessary. Such value can improve the quality of diagnosis so that the better quality of care. Primary Funding Source: Taiwan Deparement of Health ●Fully Paid Employer Provided Health Insurance: Trends in Benefits over Time Amy Taylor, Ph.D., Alice Zawacki, Ph.D. Presented By: Amy Taylor, Ph.D., Senior Economist, CFACT, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850; Tel: (301)427-1660; Fax: (301)4271276; Email: ataylor@ahrq.gov Research Objective: Despite concerns about increasing health insurance premiums, there continue to be establishments which pay the full cost of health insurance for their employees. As medical care costs and health insurance premiums inevitably rise in the future, it will be important for policymakers to understand the characteristics of these fully paid plans and the benefits they provide. The goal of this analysis is to examine the benefits of fully paid plans, and see whether health care and health insurance are becoming more expensive to employees in terms of higher co-payments, coinsurance, and deductibles when employers pay 100 percent of premium costs. Study Design: We analyze data from 1997 through 2002 from the Medical Expenditure Panel Survey-Insurance Component, which collects establishment level data on the health insurance plans offered to employees. The survey provides information on the plans’ premiums, contributions by employers and employees, and covered benefits. We focus on the plans offered by establishments that are fully paid by the employer and present information on the benefits of these plans, including premium levels, provider type arrangements and out-of-pocket expenses. The benefits of fully paid plans are compared with those of plans not fully paid by employers and changes are examined over time. Population Studied: The nationally representative sample of private establishments found in the MEPS-IC from 1997-2002. Principal Findings: Plans that were fully paid were more likely to require referrals to see specialists, less likely to cover preexisting conditions or outpatient prescriptions, and had higher out-of-pocket expense limits than plans that were not paid 100 percent by employers. These plans also were more likely than plans not fully paid to have had the highest premiums, and were less likely to be self-insured. The percentage of all plans offered by employers who paid 100 percent for single coverage decreased from 45 to 39 percent between 1997 and 2002 and for family coverage the percentage decreased from 25 percent to 21 percent. There was also a decrease in any provider type arrangements between 1997 and 2002. Conclusions: The characteristics and benefits of fully paid health insurance plans, including premiums, out-of-pocket limits, provider type arrangements and outpatient drug coverage, differ in many ways from plans that are not fully paid by employers and tended to be less generous in several ways. The percent of plans that were fully paid for by employers decreased during the time period 1997-2002. Implications for Policy, Delivery, or Practice: As health care costs continue to rise, it is important to look at employers’ responses to increasing health insurance premiums. While some employers will continue to pay 100 percent of premium costs for their employees, they may pass increasing costs on by changing the benefit structure of these plans. Examining the characteristics of fully paid health plans over time will shed light on the financial burden borne by employees in the form of rising deductibles, co-payments, and coinsurance and the effect of these costs on the demand for health care services. This has particular policy relevance in that these plans may affect the attractiveness and growth of Health Savings Accounts. Primary Funding Source: AHRQ ●The Economic Impact of Unrecognized Bipolar Disorder in the California Medicaid (MEDI-CAL) Program: a Six Year Analysis Thomas Tencer, BSc, Jeffrey S. McCombs, Ph.D., Jeonghoon Ahn, Ph.D. Presented By: Thomas Tencer, BSc, Graduate Student, Pharmaceutical Economics and Policy, University of Southern California, 1540 E Alcazar CHP-140, Los Angeles, CA 90033; Tel: (323)442-1460; Fax: (323)442-1462; Email: tencer@usc.edu Research Objective: Compare the healthcare costs of patients with unrecognized and recognized bipolar disorders with depressed-only patients over 6 years. Study Design: Retrospective cohort study. Population Studied: Data from the Medi-Cal program were used to identify 14,809 new episodes of antidepressant therapy with 6 years of post-treatment data. Patients were classified as recognized-BP (RBP) based on a bipolar diagnosis and/or a mood stabilizer (MS) prescription claim on or before initiation of AD therapy. Unrecognized-BP (UPB) patients received a bipolar diagnosis or MS therapy after their initial AD treatment episode. Patients were defined as having switched into mania if they (1) terminated use of their initial AD after one prescription; or (2) used an antipsychotic, anticonvulsant, or MS within 30 days; or received a diagnosis of mania within 180 days of starting AD therapy. Principal Findings: Only 14.5% of new episodes of antidepressant therapy were initiated by RBP patients, while 28.2% of antidepressant users ‘switched’ to a bipolar diagnosis over a 6-year period. Depressed-only patients had the highest health care costs during the first year ($5,465) followed by UBP patients ($4,802) and RBP patients ($4,667). Conversely, UBP patients exhibited the largest increase in total cost per year after 6 years (+92.7%), followed by RBP patients (+78.8%), then depressed-only patients (+57.8%). However, RBP and UBP patients who show evidence of having switched into mania after their initial antidepressant treatment episode exhibit lower 6-year increases in total costs (49.3% and 60.6%, respectively) compared to patients with no evidence of a switch to mania (83.6% and 98.3%, respectively). Conclusions: UBP patients seeking treatment with antidepressants are common and their unrecognized status leads to an inflated trajectory of future cost increases Implications for Policy, Delivery, or Practice: Failure to recognize bipolar disorders in a timely manner may increase healthcare costs. Primary Funding Source: Eli Lilly and Company parent proxies as a source of information about teens’ health care experiences. For example, the National Committee for Quality Assurance excludes 13-17 year olds from the sample of patients surveyed as part of the accreditation processes while the CAHPS Research Consortium allows parents to report for their children up to age 17. Understanding parental involvement in care and how it is related to the assessment of care can be used to make evidenced-based decision about whether adolescents should be included in samples for research and evaluation and whether their information is best provided by parents Primary Funding Source: Minnesota Department of Human Services ●Should Parents be Proxies? Parents’ Involvement in and Assessment of Routine Care for Adolescents. Annette Totten, MPA, Ph.D., Timothy J. Beebe, Ph.D., Kathleen Thiede Call, Ph.D. ●The Association of Distance to Care with Severity of Disease, Length of Stay and Charges for Persons Hospitalized with HIV/AIDS Roberto Vargas, M.D., MPH, Paul Robinson, Ph.D., Susan Ettner, Ph.D., William Cunningham, MD, Ph.D. Presented By: Annette Totten, MPA, Ph.D., Research Associate, Univ. of MN School of Public Health, SHADAC, 2221 University Avenue, Suite 345, Minneapolis, MN 55414; Tel: (612)702-2616; Fax: (612)624-1493; Email: tott0007@umn.edu Research Objective: To determine how the evaluations of quality of care made by parents serving as proxy respondents for their adolescent children are influenced by the level of involvement in their teen’s health care. Study Design: 4,953 (54% response rate) enrollees in Minnesota’s Medicaid program were selected and surveyed between April and July 2003 as part of an effort to assess disparities in and barriers to the use of preventative and other health services. In cases where the selected Medicaid enrollee was a child (under 17), data were collected from a parent. The analyses focus on over 450 assessments provided by parents of the health services received by adolescents 13-17. Parents’ ratings of the health care provided; of the health care provider; of how well the provider listens; and of how well the provider explains things were compared based on whether parents accompany the teen to their appointments and whether the parents go into the examination room. Population Studied: Adolescents (13-17 years old) enrolled in Minnesota’s Medicaid program and their parents who responded to the survey on their behalf. Principal Findings: Parental involvement in health care is less likely with older teens, with both parental accompaniment to the clinic and the examination room falling precipitously with age. Parents who are more involved with their adolescent’s care rate the care significantly better than parents who are less involved. Ratings are more affected by whether the parent goes into the exam room than whether the parent accompanies the teen to the clinic. Conclusions: Declining direct parental involvement in routine health care and declines in parents’ ratings of the care with the increasing age of the teens raise questions about the ability of parents to serve as proxy respondents. Parents may have limited opportunities to directly observe the care provided to older teens and, as a result, their ratings must be based on other sources of information. Implications for Policy, Delivery, or Practice: Currently there is no consistent approach to either the inclusion of adolescents in evaluations of health services or the use of Presented By: Roberto Vargas, M.D., MPH, Assistant Professor, General Internal Medicine and Health Services Research, UCLA School of Medicine, 911 Broxton Avenue, Los Angeles, CA 90024; Tel: (310)794 3703; Fax: (310)794-0372; Email: RBVargas@mednet.ucla.edu Research Objective: Our study examines the association of patient residential distance to the nearest HIV/AIDS medical provider and nearest ancillary services site with severity of disease, length of stay and hospital charges for patients with HIV/AIDS admitted to acute care hospitals Study Design: We combined data from the California Office of Statewide Health Planning and Development 2000 discharge files, the AIDS Project Los Angeles list of HIV services, and the 2000 Census. Our main predictors were distances from the patient’s population-based zip code centroid to the nearest HIV/AIDS medical provider site and nearest ancillary care service site. Our outcomes were dichotomized severity of disease indexes for conditions present at the time of hospitalization: 1) Deyo Charlson 2) Dartmouth Manitoba (DM) Charlson scores and 3) the Severity Classification for AIDS Hospitalizations (SCAH); length of stay and hospital charges were additional outcomes. We used multiple logistic and linear regression to control for differences in patient age, gender, race, and insurance status; and levels of poverty, English language proficiency, racial and ethnic segregation, education level and number of hospital beds in the patient’s neighborhood of residence. Generalized Estimating Equations were used to adjust the variance for clustering by zip-code Population Studied: We identified 3543 hospitalizations for persons with HIV/AIDS in Los Angeles County in 2000. Patients were 85% male, 56% white, 29% black, and 37% Hispanic irrespective of race. Insurance coverage was 26% Medicare, and 47% Medicaid and other indigent care programs. Patients came from communities where, on average, 35% of people had less than a high school education and the ratio of households making less than $20,000 to those making over $60,000 was 2:1 Principal Findings: Mean patient’s distance from care was 0.68 miles (range 0.03-13.3; s.d.0.82) from ancillary care services and 1.81 miles (range 0.05-15.8; s.d.1.5) from medical care services. Multiple logistic regressions suggested that a greater distance to ancillary services was associated with a significantly lower disease severity index among patients admitted to acute care hospitals for both Charlson indexes, Deyo: Odds Ratio (OR) 0.84, 95% Wald Confidence Interval (CI) (0.73, 0.97) and DM: OR 0.86 CI (0.75,0.99) but not for the SCAH index: OR 1.04 CI (0.93,1.16). Multiple linear regressions suggested that greater distance from ancillary care services was associated with a significantly greater length of stay (estimate 0.078; p=0.002) but not with hospital charges (estimate 0.044; p=0.120). Regression models did not show a significant association between distance to medical care provider and length of stay, charges or any severity indexes Conclusions: Our results suggest that greater distance to ancillary, but not medical, HIV services is associated with a lower disease severity at admission and a longer length of hospital stay during admission, implying a lower clinical threshold for admission and a greater threshold for discharge for patients from communities that are further away from ancillary HIV care services Implications for Policy, Delivery, or Practice: Improving geographic access to ancillary services for persons with HIV/AIDS may reduce hospitalizations and length of stay for these patients Primary Funding Source: National Center on Minority Health and Health Disparities ●Methadone Treatment Sites Conduct their intake HIV Testing Onsite James Walkup, Ph.D., Rich Shank, MA, Guldeniz Yucelen, Psy.M, Kevin Moore, Psy.D. Presented By: James Walkup, Ph.D., Associate Professor, IHHCPAR, Rutgers University, 30 College Avenue, New Brunswick, NJ 08901; Tel: (212)724-8362; Email: jaywalks@aol.com Research Objective: To determine which methadone maintenance treatment sites have opted to rely exclusively on on-site HIV testing at intake. Centers for Disease Control estimates suggest injection drug use plays an important role in the spread of HIV, with more than one third of U.S. AIDS cases traceable to it. Opiate addiction treatment settings offer an opportunity for early detection and referral for treatment initiation for clients with HIV. Studies of primary care and MMT indicate on-site integration of services performs better than referrals. Study Design: A mailed survey regarding available treatment policies and practices was sent to MMT sites statewide in a state with a high rate of HIV infection. These data were supplemented with data from the 200 Census regarding the community setting of individual clinics. Descriptive quantitative data were interpreted in the context of prior qualitative work, based on a focus group with a sample of administrators, interview and site visits with a stratified sample of sites statewide Population Studied: Statewide MMT treatment setting in a state with a high rate of HIV among injection drug users. Principal Findings: Response rate was 86% (31/36). Eleven sites rely exclusively on on-site testing. None of these are for profit. They also more commonly provide other, related medical services (e.g., on-site general medical examination, psychotropic prescribing), other HIV-related services (e.g. condoms, specialized case management for clients with HIV), and select mental health services (e.g. individual counseling for those with mental illness). Approximately one fifth (6/31) of sites do not offer HIV testing at intake. Conclusions: Adoption of optimal integration of intake HIV testing is limited, although not rare. Sites with this strategy seem not to be located in more highly disadvantaged areas. One limitation is that our findings provide no information regarding the success of various intake HIV testing strategies at these sites. Implications for Policy, Delivery, or Practice: Optimal HIV testing practices are needed across a range of MMT settings. Research is need to determine (a) which strategies are in fact most successful, under which circumstances; (b) which instances of non-adoption may represent reasonable variation vs. non-adoption that represents neglect or suboptimal service design; (c) how institutional characteristics interaction with service packages. Primary Funding Source: NIMH ●Comparing the Costs and Outcomes of Hospitalists and Private Practice Physician Groups in a Multi-hospital Medical Center. Thomas Wasser, Ph.D., Michael Pistoria, DO, Robert Murphy, M.D., Zubina Mawji, M.D., John Fitzgibbons, M.D. Presented By: Thomas Wasser, Ph.D., Biostatistician, Health Studies, Lehigh Valley Hospital, 5754 Loyola Street, Macungie, PA 18062; Tel: (610)402-2497; Fax: (610)402-2247; Email: thomas.wasser@lvh.com Research Objective: Hospitalists, whose ranks have swelled to more than 8,000 physicians in their 15 year history, are the frequent object of studies examining patient outcomes and costs. Few studies break out cost by hospital service and track this data over time. The objective of this research was twofold and included (1) comparisons of aggregate mean values for hospitalist and non-hospitalist medical groups, and (2) examinations of trends in cost and Length of Stay (LOS) over time. Study Design: Retrospective review of existing cost data. Cost data examined included Laboratory, Radiology, Pharmacy variable cost as well as Total variable cost and LOS. Analysis methods included Analysis of Variance (ANOVA) with Scheffe Post-hoc tests. And regression analysis examining trends in costs and length of stay over time. A p-value of 0.05 was used to determine significance. Population Studied: All patients (n=9,353) admitted and followed at two hospitals whose care was provided by five medical groups (two hospitalist groups, and three private practice groups). Data were collected over a 16 month period from the implementation of the hospitalist program in June 2003 to September 2004. The beginning of the study period coincided with the implementation of the hospitalist programs at both hospitals. Principal Findings: ANOVA results indicated that one of the hospitalist groups had lower costs for Laboratory, Radiology, and Total costs than any other group (all p<0.001). This hospitalist group also had the lowest LOS than all groups 3.37 ± 3.1 (mean LOS and standard deviation) as well. This was significantly lower than all non hospitalist groups (all p<0.001), but not different from the second hospitalist group 3.62 ± 4.93 (p=0.534). The second hospitalist group had generally lower costs and LOS than the non-hospitalist groups but was not statistically significant except for Radiology which was lower than two non-hospitalist groups (both p<0.001). Interestingly there were few significant trends in costs and LOS over time. It was the expectation that hospitalist programs would show consistently lower trends over time as they developed experience. This was not the case as the slopes for Pharmacy cost and LOS decreased significantly for the second hospitalist group (p=0.016 and p=0.024 respectively), however another non-hospitalist group had similar results on the same variables (both p<0.001). Conclusions: These results, while only reflective of the first 16 months of the hospitalist program indicate that costs Laboratory, Radiology, costs are both lower with hospitalist groups as is LOS with Hospitalist care provided. It is important to note that there were no observed differences in Pharmacy costs between Hospitalist and Private Practice groups. Implications for Policy, Delivery, or Practice: This study does not examine the changes in utilization of the hospitalist with regard to severity of illness or diagnostic groupings. That work is currently ongoing. However, these results indicate that Radiology and Lab costs, when drilled down beyond total cost are equally or more affected by the hospitalist. There was no difference in Pharmacy costs. Our study found one hospitalist group to be less expensive than one non-hospitalist group for Pharmacy costs. This may be because Pharmacy costs are highly variable from patient to patient and the resulting large standard deviations require a longer trial to show effectiveness. Primary Funding Source: No Funding Source ●Examination of Practice Patterns and Case Mix in Hospitalists and Private Practice Physician Groups Over Time. Thomas Wasser, Ph.D., Michael Pistoria, DO, Robert Murphy, M.D., Zubina Mawji, M.D., John Fitzgibbons, M.D. Presented By: Thomas Wasser, Ph.D., Biostatistician, Health Studies, Lehigh Valley Hospital, 5754 Loyola Street, Macungiep, PA 18062; Tel: (610)402-2497; Fax: (610)4022247; Email: thomas.wasser@lvh.com Research Objective: One criticism of studies that show Hospitalist programs to be effective is that the new programs could have less severely ill patients which would affect cost and patient outcomes. To examine this, the admission diagnosis, severity of illness, risk of mortality and outcomes (death and discharge location) in patients admitted to a hospitalist service are compared to private practice physician groups. This study examines the change in this patient mix overtime from the implementation of the hospitalist program. Study Design: Retrospective review over a 25 month period (June 2002 through June 2004) of data abstracted from our hospitals care management system. Severity of Illness (SI) and Risk of Mortality (RM) were determined by using APRDRG codes. Discharge (DC) information was coded as Against Medical Advice (AMA), DC to long term facility, Expired, Home, Hospice, or Skilled Nursing Facility (SNF). Analysis methods included chi-square for the four point severity of illness and risk of mortality against pre and post hospitalist implementation. Regression analysis was used to measure the proportion of adverse outcomes over time. Any p-value less than 0.05 was considered significant. Population Studied: Inpatient population (n=10,149) at two hospitals within the same health network. Data included was from a hospitalist program at each hospital and three private practice groups at both hospitals. One hospitalist program began at month 13 within the 25 month follow-up period. Principal Findings: Upon implementation of the hospitalist service there was not a significant difference in the admission diagnosis between Hospitalists and Private Practice groups. However SI was significantly lower in 10 of the 25 months studied for Hospitalists (p<0.05), there was no difference in 12 of the 25 months (p>0.05) and private practice groups had significantly lower SI in three of the months studied (p<0.05). The same relationship was observed in RM as Hospitalists had lower RM for 13 of the 25 months studied (p<0.05), there was no difference in nine of the 25 months studied (p>0.05), and private practice groups had significantly lower SI in three of the months studied (p<0.05). There was no difference in mortality rate between Hospitalists and Private Practice groups, however there were only 293/10149 (2.89%). There were no other significant differences in DC information between any of the five groups studied. Conclusions: This research examines the patient characteristics in Hospitalist based care compared to private practice physician groups in our network. There was no time effect within the data indicating that Hospitalists saw less severe patients immediately after program inception. However, Hospitalist patients do not have the same SI or RM as did private practice patients. There were no observed differences in mortality within the groups. Implications for Policy, Delivery, or Practice: There were no patterns in the data suggesting that SI or RM rates occurred in the initial stages of the Hospitalist program. These data suggest that Hospitalist patients are in general not as severely ill or at the same risk of mortality as private practice groups however neither group lower in the beginning, or after the implementation of the Hospitalist program. No difference in observed mortality rate probably reflects the low mortality rate in general within this patient population. Primary Funding Source: No Funding Source ●Impact of a Multifaceted Quality Improvement Initiative to Implement JCAHO Core Measures for AMI and CHF. Thomas Wasser, Ph.D., Michael Rossi, M.D., Zubina Mawji, M.D., Patricia Parker, RN, Joshua Skibba, M.D., Katrina Fritz, RN, Tamara Masiado, MS Presented By: Thomas Wasser, Ph.D., Biostatistician, Health Studies, Lehigh Valley Hospital, 5754 Loyola Street, Macungie, PA 18062; Tel: (610)402-2497; Fax: (610)402-2247; Email: thomas.wasser@lvh.com Research Objective: Over one million patients will suffer an acute myocardial infarction (AMI) this year. Nearly half will be a recurrent event. Associated congestive heart failure (CHF) will consume more Medicare dollars than any other diagnosis. In response to this disparity in clinical practice, the Joint Commission on Accreditation of Healthcare Organizations identified key quality indicators for the treatment of AMI and CHF. The purpose of this research was to assess the impact of a multifaceted quality improvement initiative to implement Core Measures for AMI and CHF. Study Design: Two hospitals within the same network are participating. Three distinct observational phases including Baseline (prior to 7/03), Phase one (7/03–6/04) data after network campaign to promote the importance of secondary cardiac disease prevention by issuing standardized admission order sets. Phase two (7/04-Present) implemented a Cardiac Core Measure Progress Note to serve as a memory tool and cue physician compliance. Data were obtained by retrospective chart review using a multi-system hospital database. Chi-square was used to contrast Baseline versus Phase one and two. Any p-value less than 0.05 were considered significant for this analysis. Population Studied: Patients who had a principle diagnosis of AMI or CHF at discharge. Some patients were excluded based on published JCAHO guidelines. Quality indicators for this initiative were identified as the percentage of eligible patients who met the Core Measures for AMI and CHF. Data was analyzed at three intervals, baseline (7/02-6/03), phase one, and first quarter of phase two. Combined quality indicator data from both institutions is described below. Principal Findings: For AMI: Ace inhibitor use for left ventricular dysfunction (LVD) increased from 69% to 76% (p=0.044) to 90% (p=0.002). Smoking cessation counseling increased from 89% to 93% (p=0.028) to 97% (p=0.006). Beta blocker use at discharge fluctuated from 86% to 95% (p<0.001) to 93% (p=0.004). Inpatient mortality decreased from 7% to 6% (p=0.465) to 5% (p=0.214). There was no difference in the use of aspirin at admission or discharge (95% and 97% respectively p>0.05). Beta blocker use at admission declined minimally from 93% to 93% (p=0.897) to 91% (p=0.455) all comparisons are from Baseline to Phase one and Baseline to Phase two respectively. For CHF: Discharge instructions increased from 39% to 45% (p=0.013) to 59% (p<0.001). Ace inhibitor use for LVD increased from 65% to 66% (p=0.496) to 72% (0.099). Smoking cessation counseling increased from 43% to 44% (p=0.831) to 78% (p=0.001). The assessment of left ventricular function remained essentially unchanged at 93% to 92% (p=0.880) to 93% (p=0.610). Conclusions: Clearly the interventions have progressively increased compliance with the core measures defined above and are showing progress on outcomes including mortality. While some of the Phase two data have not increased to significance levels the program data is only reflective of two months and is currently being tracked. Implications for Policy, Delivery, or Practice: This quality improvement initiative is proving to be an effective implementation of the JCAHO Core Measures for AMI and CHF. The increased performance has resulted in decreased inpatient mortality. It is likely that this model could be replicated at other institutions. Primary Funding Source: The Dorothy Rider Pool Health Care Trust ●Insights into Provider Attitudes about Pay-for-Quality Incentives in Health Insurance Markets Bert White, MBA, D.Min., Mark Meterko, Ph.D., Gary Young, Ph.D., JD, Matthew Guldin, MPH, Dan Berlowitz, M.D., James Burgess, Ph.D. Presented By: Bert White, MBA, D.Min., Project Director, Health Services Department, Boston University School of Public Health, 715 Albany Street, Boston, MA 02118; Tel: (617) 232-9500 x4380; Fax: (617)232-6140; Email: bertw@bu.edu Research Objective: This research evaluates several demonstration sites that are offering financial incentives to providers. Programs involving monetary rewards to clinicians who achieve certain quality goals are becoming increasingly popular as a mechanism for improving health care and controlling costs. The objective of this research is to make a contribution to those who are involved in designing and implementing pay-for-quality programs. Study Design: Multiple data collection approaches are being used including telephone interviews, written surveys, and site visits in accordance with a conceptual frame work that includes three primary domains: Awareness, Credibility and Control. Practice executives with responsibility for physician contracts in the health insurance market have been identified in each of three demonstration projects. These practice executives provide medical group demographic information and access to physicians exposed to a demonstration’s incentive plan. Practices executives have been interviewed, physicians have been surveyed and lessons have been learned from site visits. Population Studied: A sample of contracting entities in each demonstration has been used to identify and study the practice executives and physicians. Principal Findings: Initial conclusions are that physician attitudes may differ from what is often presented in the media, providers appear confused about administration of programs, and providers, while comfortable with clinical relevance of quality targets, are not impressed with the associated dollars, reporting time lags and impact of the quality targets. Conclusions: Some of the initial conclusions are that physician attitudes may differ from public commentary, providers appear confused about administration of programs, and providers are comfortable with clinical relevance of quality targets but are not impressed with the associated dollars. Implications for Policy, Delivery, or Practice: The study provides business case information to CMS and health insurance markets that is required for a more robust understanding of incentives and their application to providers. With these findings additional research and demonstration projects can be more strategically focused for more efficient and effective rewarding results. Primary Funding Source: AHRQ ●Improving Antipsychotic Adherence for Patients with Cooccurring Schizophrenia and Substance Use Disorders: Assessing Patterns and Effectiveness of Clinical Interventions Josh Wilk, Ph.D., Joyce West, Ph.D., MPP, Steven Marcus, Ph.D., Lisa Countis, Rebecca Hall, BS, Mark Olfson, M.D., MPH Presented By: Josh Wilk, Ph.D., Research Scientist, Practice Research Network, American Psychiatric Institute for Research and Education, 1000 Wilson Boulevard, Suite 1825, Arlington, VA 22209; Tel: (703)90-8618; Fax: (703)907-1087; Email: jwilk@psych.org Research Objective: 1) Characterize interventions used in routine psychiatric practice in the U.S. to manage medication non-adherence among patients with schizophrenia and cooccurring substance use disorders; and 2) Assess the reported effectiveness of interventions for medication non-adherence among patients with co-occurring schizophrenia and substance use disorders. Study Design: A national survey was conducted among a random sample of psychiatrists treating schizophrenia from the AMA Masterfile of Physicians. Each psychiatrist reported on one adult outpatient with schizophrenia who had been in treatment for at least one year and had been non-adherent with oral medications at some point in the last year. 69.3% of eligible psychiatrists responded, resulting in a sample of 295 patients. Patients with schizophrenia only were compared to patients with schizophrenia and a history of a co-occurring substance use disorder. Population Studied: 295 adult outpatients with schizophrenia who had been in treatment for at least one year and had been non-adherent with oral medications at some point in the last year. Principal Findings: Psychiatrists were less likely to use psychological interventions to manage medication nonadherence, such as discussing risks of non-adherence (?2 = 3.73, df=1, p=.05), linking adherence to personal goals (?2 = 7.44, df=1, p=.01), or exploring the meaning of taking antipsychotic medications (?2 = 6.20, df=1, p=.01) among their patients with co-occurring substance use disorders. They were two times more likely to add another antipsychotic to patients with substance use disorders (?2 = 6.93, df=1, p=.01). There were generally few differences between the patient groups in psychiatrists’ perceived effectiveness of psychopharmacological, psychological, behavioral and family interventions; however, depot antipsychotics (t=1.72, df=44, p=.05) and daily observation of pill taking (t=2.06, df=99, p=.03) were reported to be less effective in patients with substance use disorders. Although simplifying dosing was the most commonly used psychopharmacologic intervention among patients with substance use disorders the intervention reported to be most effective for these patients was adding another antipsychotic medication. Conclusions: The less frequent use of psychological approaches by psychiatrists in treating medication nonadherence may reflect less of an emphasis on providing psychosocial services to patients with co-occurring substance use disorders or pessimism about the perceived effectiveness of these interventions. This finding is troubling given that these approaches are recommended by empirically-based practice guidelines, and when used, psychological approaches were found to be as effective in those with co-occurring substance use disorders as in those without these disorders. Implications for Policy, Delivery, or Practice: The practice of providing fewer psychological interventions for medication non-adherence for patients with co-occurring schizophrenia and substance use disorders is troubling. These practices are not only suggested by evidence-based practice guidelines, but also were rated to be as effective for these patients as for those without co-occurring substance use disorders. This finding suggests a target for intervention in changing psychiatrists' routine practice habits. Primary Funding Source: NIMH ●Differences in the Impact of S-SCHIP for Children Enrolled in the First Year vs the Third Year of the Program Jean Willard, MPH, Peter C. Damiano, DDS, MPH Presented By: Jean Willard, MPH, Senior Research Assistant, Public Policy Center, University of Iowa, 227 South Quad, Iowa City, IA 52242; Tel: (319)335-6815; Fax: (319)335-6801; Email: jean-willard@uiowa.edu Research Objective: The primary objective of this study was to evaluated differences between the first and third years of the Iowa S-SCHIP program. Study Design: A pre-test, post-test panel study design was used to evaluate the program. An 80-item questionnaire was mailed to the parents of one randomly selected S-SCHIP enrollee per household upon enrollment in the program. A similar follow-up instrument was mailed regarding each child around the 1-year enrollment anniversary. Telephone interviews were attempted with nonrespondents. Population Studied: Outcomes for children who initially enrolled in S-SCHIP during approximately the first year of the program- January, 1999 through November, 1999, were compared to children who initially enrolled during approximately the third year of the program- July 2001 through June 2002. Data was only included for respondents for which baseline and follow-up data were available- 463 cases for first year enrollees and 1698 cases for third year enrollees. Tests on the equality of proportions were used to determine statistically significant differences for outcomes in first and third years. Principal Findings: 81 percent of new enrollees in the first year had a personal doctor or nurse upon entry into the program vs. 86 percent in the third year. 89 percent of first year enrollees had a personal doctor or nurse by the end of the first year, while the percent for third year enrollees remained at 86 percent. First year enrollees were more likely to have an ER visit in the year prior to enrolling than third year enrollees- 44 percent first year, 39 percent third year, however this declined to about 36 percent after one year in the program for both groups. Children in the first year were less likely to be rated in excellent health- 37 percent vs. 45 percent, however, they were more likely to be in excellent health after a year in the program- 46 percent excellent, vs. 43 percent. Parents of first year enrollees worried more about paying for health care before joining the program- 95 percent worried, vs. 90 percent, however about 60 percent of both worried after one year in the program. 27 percent of first year enrollees and 17 percent of third year enrollees unmet need for care upon entering the program, however after one year in the program 6 percent of first year and 4 percent of third year enrollees reported having unmet need for health care. Conclusions: Children enrolling during the first year of the SSCHIP program were less healthy and less likely to have a regular source of care than children enrolling in the third year. Unmet need for care was significantly reduced in both years. Implications for Policy, Delivery, or Practice: Early pent up demand and/or changes in employer-sponsored health insurance due to the slowdown in the economy and higher insurance costs may have driven a different cohort of children to be enrolled in the program by the third year. Although children enrolling later may be coming in with fewer needs, maintaining their insurance health coverage remains important for the children’s ability to use services. Primary Funding Source: The Iowa Department of Human Services ●Validating a Structural Equation Model (SEM) of Relationships Affecting Well Siblings and Mothers of Chronically Ill Children. Arthur Williams, Ph.D., Phoebe Williams, PhD, RN, FAAN, Todd Huschka, BS, BA Presented By: Arthur Williams, Ph.D., Chair, Health Care Policy and Research, Health Care Policy and Research, Mayo Clinic, 200 1st Street SW, Rochester, MN 55905; Tel: (507)2843356; Fax: (507)284-1731; Email: williams.arthur@mayo.edu Research Objective: To determine the robustness of a Structural Equation Model (SEM) that was estimated with baseline data collected during a randomized clinical trial (RCT). Study Design: A SEM, parts of which had been developed, tested, and reported by two of the authors in publications over a 20-year period, was fit to endogenous and exogenous variables obtained at baseline during the RCT (Journal of Behavioral Medicine, 2002, 25: 411-424). Exogenous variables consisted of age of the well sibling, treatment group placement (full, partial, and control), and diagnosis of the ill child. Endogenous variables consisted of mother’s mood, a latent socio-economic status variable (derived from mother’s education and family income), family cohesion, and the well sibling’s attitude toward the illness of their chronically ill brother or sister, sibling mood, knowledge of illness, selfesteem, social support, and behavior problems. All variables were measured on instruments with known and published psychometric properties. Using AMOS 5.0 and the RCT data, the SEM was first reproduced to match against the published model. The SEM was then fit with data from each of the subsequent time periods and refit using baseline parameters fixed over all time periods. Bootstrapped analysis was performed over all periods. Population Studied: Data were obtained at a large medical center in the Midwestern USA over a 12-month period when an intervention was provided to siblings and mothers of chronically ill children. The mean age of the 252 study siblings was 11 years. About 2/3 were older than the ill child, 50% were male, and 86% were Caucasian. Most of the children (82.1%) lived in two-parent families with median annual incomes above $49,000. Principal Findings: The SEM was robust over all time periods. The SEM had relative chi-squares (CMIN/DF) <=2.00, Comparative Fit Indices (CFI) > 0.90, and Root Mean Square Error of Approximation (RMSEA) <=0.08 over all time periods, exhibiting a good overall model fit. When each subsequent period model was fit with their parameters fixed at baseline estimates, the results were CMIN/DF <= 2.00, CFI > 0.84, and RMSEA <= 0.06. Conclusions: The SEM developed with data from the RCT appears to be valid and reliable, at least as it can be reproduced in subsequent time periods under different assumptions. This SEM provides a clinically useful picture of critical psychosocial interactions that can affect the health and well-being of siblings in families with chronically ill children. Most notably, improved mother’s mood and family cohesion are associated with improved sibling outcomes, and higher levels of sibling knowledge of the chronic illness is associated with positive outcomes on other sibling variables: reduced behavior problems, attitudes toward the illness, mood, selfesteem, and feelings of social support. Implications for Policy, Delivery, or Practice: Financial or inkind support to families with a chronically ill child can be critical to positive outcomes of the well siblings. Further tests of validity of this model should be performed using an independent sample. Primary Funding Source: NIH ●Temporal and Spatial Pooling of BRFSS Data to Describe Overweight and Obesity by County James Wilson, Ph.D., Denise Kirk, MS, Christopher Mansfield, Ph.D. Presented By: James Wilson, Ph.D., Associate Director, East Carolina University, Center for Health Services Research and Development, Physicians Quadrangle Building N, Greenville, NC 27834; Tel: (252) 744-2785; Email: wilsonja@mail.ecu.edu Research Objective: Interest in Behavioral Risk Factor Surveillance System (BRFSS) data at the sub-state level has been increasing in the health care research community. However, many counties still have low response rates and this contributes significantly to statistical instability across counties. One way to improve stability over time is to pool adjacent years of data into an average. To improve geographic stability, there are various techniques that “borrow strength” in numbers by pooling data from adjacent counties. The operation of both techniques on the data results in the smoothing of trends in time series charts and maps. Using weight category data from the NC BRFSS, our study objectives include 1) determine the steps necessary to create smoothed maps using county level data and 2) demonstrate the analytical utility and limitations of smoothing techniques used in the description of sub-state distributions of overweight and obese populations in Eastern North Carolina (ENC). Study Design: SUDAAN was used to calculate weighted percentages and confidence intervals for each county. Obesity and overweight percentages were then mapped. A countystandard error map provided a relative measure of map instability. The map was then smoothed spatially using standard errors of adjacent counties as inverse weights. Smoothed maps were then generated using ArcGIS (ESRI) for studying trends in the distribution of overweight and obese populations. Population Studied: The total 1999 through 2002 BRFSS study population for ENC weight categories was 4,849. County respondents for this 41-county region numbered from 5 to 687. Regionally, the study was comprised of 67.3% Whites and 23.9% African-Americans with the African- American population under-represented by one-third. Corrections in the survey analysis methodology were applied to obtain a closer approximation to regional proportions. Principal Findings: Regionally, the temporally aggregated percentages for overweight and obesity were 36.5% and 24.6%, respectively. Whites (37.7%) were slightly more overweight than African-Americans (34.4%). In contrast, African-Americans were more obese. The percentage of obese African-Americans was 35.0% compared to Whites at 20.8%. For the observed map, there was variation among counties for the two weight categories. In counties with proportionally larger African-American populations, Whites were less overweight and both groups were similar with respect to obesity. Counties with higher White population proportions had African-Americans with higher percentages of obesity, while Whites were more overweight. Because larger counties have higher proportions of Whites, they tended to be similar to the regional weight category distribution. Spatially smoothing the survey results removed extreme inter-county variation, making neighboring counties more similar. Conclusions: At present, the geographic coverage for BRFSS data is not complete. Maps that have been smoothed did not contradict the observed survey distributions and made trends more intelligible where there were statistically sufficient numbers. Implications for Policy, Delivery, or Practice: The geographical portrayal of survey data can have a great impact on policy formation as well as assist in the locating problem areas and the deployment of programs and resources. As BRFSS data collection proceeds, statistical reliability will improve, which will make assessment of programmatic efforts in treating overweight and obesity possible. Primary Funding Source: No Funding Source ●Health Services Use Among Older Women: The Combined Roles of Self-rated Physical and Mental Health Jacqueline Wiltshire, MPH, Ph.D., Michelle McQuirter, BSN, MS, Orlando Taylor, Ph.D. Presented By: Jacqueline Wiltshire, MPH, Ph.D., PostDoctoral Fellow, Center for Women's Health Reseach, University of Wisconsin Medical School, 202 South Park Street, 6W, Madison, WI 53717; Tel: (608)-267-5566; Fax: (608)-267-5575; Email: wiltshire@wisc.edu Research Objective: Research shows that high use of medical care and low use of mental health care by older women is partly attributable to misinterpreting mental health symptoms as medical problems. While self-rated physical health is commonly used to assess use of health services, self-rated mental health is often overlooked, and little has been done to evaluate how the combination of mental and physical health influence health services use. The purpose of this study was to examine the extent to which the combined roles of self-rated physical health and mental health influences health services utilization patterns among older women. Study Design: Data for the study were drawn from the 200001 Household Component of the Community Tracking Study, a nationally representative survey. The key independent variable, a combination of SF-12 physical and mental health summary scores, was measured as poor physical-poor mental, poor physical-good mental, good physical-poor mental, and good physical-good mental health. SF-12 mean scores of <50 are defined as poor health and scores >50 as good. Health services were measured as yes/no to physician, mental health (MH), medical professional (MP), and emergency room (ER) visits. All regression analyses accounted for complex survey design and were adjusted for race/ethnicity, age, marital status, rural/urban, education, employment, federal poverty level, usual source of care, and managed care. Population Studied: The sample included 7,860 insured women aged 55 and over. Principal Findings: Overall, 18.4% of respondents were in poor physical-poor mental health (both mean scores <50), 43.8 in poor physical-good mental health, 6.4% in good physical-poor mental health, and 31.5 were in good physicalgood mental health. 90.6% of the women had physician visits, 16.4% medical provider visits, 15.8% emergency room visits, and 5.1% mental health visits. All comparisons made to respondents in good physical-good mental health. Poor physical-poor mental health was associated with higher physician visits (OR=3.83, 95% CI: 2.86-5.13), MP visits (OR=2.01, 95% CI: 1.64-2.47), ER visits (OR=3.56, 95% CI: 2.92-4.35), and MH visits (OR=6.28, 95% CI: 4.44-8.99), while good physical-poor mental health was associated with lower physician (OR=1.46, 95% CI: 1.05-2.03), MP (OR=1.18, 95% CI: 0.85-1.65), and ER visits (OR=1.53, 95% CI: 1.10-2.11). However, women in good physical-poor mental health had more MH visits (OR=4.55, 95%CI:2.98-6.95) than those in poor physicalgood mental (OR=1.62, 1.11-2.34). Conclusions: While women in poor physical-poor mental health may confuse mental health symptoms for physical medical symptoms which may contribute to their higher services use, there is less indication that women in good physical-poor mental health misinterpret mental health symptoms for medical symptoms. Implications for Policy, Delivery, or Practice: Assessing the combined effects of self-rated physical and mental health is key to understanding older women’s use of health services. Primary Funding Source: NIA ●The Effect of the Medicare Balanced Budget Act on Racial Disparities -An Examination of the Impact of Hospital Financial Stress on Time to Treatment and Mortality for Patients with Acute Myocardial Infarction Jingsan Zhu, MBA, Tamara Konetzka, Ph.D., Lori Parson, Eric Peterson, Ph.D., Kevin Volpp, M.D., Ph.D. Presented By: Jingsan Zhu, MBA, School of Medicine, University of Pennsylvania, 1225 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104; Tel: (215) 573-9731; Fax: (215) 573-8778; Email: jingsan@mail.med.upenn.edu Research Objective: The Balanced Budget Act of 1997 reduced Medicare reimbursement to hospitals by about $116 billion, forcing hospitals to operate under increased pressure of cost containment. Hospital financial stress could exacerbate existing disparities in care. Study Design: A retrospective study using process and outcome measures of patients admitted with acute myocardial infarction (AMI) to 208 National Registry of Myocardial Infarction (NRMI) hospitals from 1996 to 2001. We simulated the impacts of BBA and its subsequent refinements to predict the exogenous effects of Medicare reimbursement cuts on hospital revenues. 1997-2001 Medicare Cost Report data were used to determine pre-BBA payments and to calculate both baseline and changes in actual hospital margins over time. Changes in major outcome and process measures were estimated for black patients and white patients using multivariate linear or logistic regressions as appropriate. We tested whether the BBA impacts were different for white and black patients in a combined model by testing interactions between race and impact group. Population Studied: 226,773 patients admitted with AMI, who were treated in hospitals that submitted at least 20 AMI cases per year to NRMI registry from 1996 to 2001. Principal Findings: Among black patients, we found time from hospital arrival to thrombolytic therapy ordering worsened to a greater degree in high vs. low impact hospitals in post-BBA years 1999 (.12 hours, p=.03) and 2000 (.10 hours, p=.04). For white patients, this measure also worsened in high vs. low impact hospitals in post-BBA years 1999 (.13 hours, p=.04) and 2000 (.14 hours, p=.01). Time to thrombolytic therapy initiation worsened to a similar degree in 1999 and 2000 for both blacks and whites. For whites and blacks, time from hospital arrival to cardiac catheterization improved in high impact vs low hospital to similar degrees (.52 hours, p=.05 for blacks, -.63 hours, p=.04 for whites). Hospital mortality rate for blacks increased in 2001 in high impact relative to low impact hospitals (OR 1.17, p=.02) but changed at similar rates in low and high impact hospitals for whites. Operating margins decreased to a degree commensurate with the estimated degree of revenue reduction in high vs. low impact hospitals. Conclusions: The Balanced Budget Act of 1997 created a moderate financial strain on hospitals but hospital operating margins seemed to have absorbed the bulk of these reductions. Over the time period from 1997-2001, there was no strong and systematic indication that the process of care for patients with AMI was differentially affected based on race. Implications for Policy, Delivery, or Practice: Continued research is needed to examine whether more vulnerable patient groups are disproportionately affected as the financial health of hospitals comes under further pressure. Primary Funding Source: Doris Duke Charitable Foundation