Coverage & Access Call for Papers The Impact of Premiums & Cost Sharing on Access to Care Chair: Jessica Banthin, Agency for Healthcare Review and Quality Sunday, June 25 • 3:45 pm – 5:15 pm ●The Impact of Cost Sharing on Enrollment, Access, and Utilization Among Medicaid Beneficiaries: Evidence from the OHP Cohort Study Wave 2 Matthew Carlson, Ph.D., Bill Wright, Ph.D., Jeanene Smith, M.D., M.P.H., Tina Edlund, M.S., Heidi Allen, M.S.W. Presented By: Matthew Carlson, Ph.D., assistant professor, Sociology, Portland State University, PO Box 751, Portland, OR 97207; Tel: (503) 725-9554; Fax: (503) 725-3957; Email: carlsonm@pdx.edu Research Objective: In March, 2003, changes were made to the Oregon Health Plan (OHP) including increased premiums and copays, benefit reductions, and a 6-month lockout for non-payment of premiums. This prospective cohort study assesses the impact of OHP changes on enrollees’ insurance status and health care access and utilization over the 18month study period. Study Design: Data comes from a prospective cohort study of a subset of OHP beneficiaries who were affected by OHP changes. A baseline, mail-return survey was conducted in October, 2003, 6 months after initial OHP changes were implemented (n=1378, 34% response rate). Twelve months later, a follow-up survey was conducted by mail and telephone (n=991, 72% response rate). This analysis examines the impact of cost sharing by comparing two groups of enrollees who left the system during the 18-month study period: those who reported leaving because of increased cost, and those who reported leaving for other reasons. Population Studied: Adults age 19 and older who were enrolled in the OHP for at least thirty days prior to February 15, 2003, just prior to the initial wave of program changes. Principal Findings: Eighteen months after OHP changes were implemented, 63% of the OHP standard population reported losing coverage for 1 or more months, and nearly half of those who lost coverage (46%) remained uninsured at the end of the study period. Increased out-of-pocket costs was a major driver of disenrollment, nearly half who lost coverage (45%) indicated increased program costs were the main reason. Moreover, when compared to those who left for other reasons, those who left because of cost were more likely to remain uninsured (39% vs. 53%) at the end of the study period. Cost sharing was also associated with increased unmet need. Compared to those who left for other reasons, lost who left because of cost were more likely to report unmet health care needs (47% vs. 69%) and unmet medication needs (36% vs. 59%), less likely to have a primary care visit (73% vs. 64%) and were more than twice as likely to have an ED visit (14% vs. 36%). Additionally, those who left for other reasons were only half as likely to report owing more than $500 in medical debt compared to those who left because of cost (26% vs. 47%). Conclusions: Increased program costs resulted in involuntary loss of coverage for thousands of low-income Oregonians. Evidence from this study, conducted 18 months after OHP program changes suggested that cost sharing was a major barrier to maintaining insurance coverage, and resulted in high levels of unmet health care needs, increased emergency room use, and increased the likelihood of medical debt. Implications for Policy, Delivery, or Practice: Even nominal cost sharing can impact Medicaid enrollees’ ability to maintain coverage. Moreover, those who reported losing coverage as a result of increased program costs fared far worse than those who left for other reasons. In response to findings from this and similar studies, in 2006, the OHP will no longer charge premiums for the lowest income beneficiaries (0-10% FPL) and will remove the lock-out period for non-payment of premiums. Primary Funding Source: CWF, The Robert Wood Johnson Foundation ●Benefit Design and the Price Elasticity of Demand for Health Insurance Sherry Glied, Ph.D. Presented By: Sherry Glied, Ph.D., Professor and Chair, Health Policy and Management, Mailman SPH, Columbia University, 600 West 168th St., 6th Floor, New York, NY 10032; Tel: 212-305-3924; Fax: 212-305-3405; Email: sag1@columbia.edu Research Objective: There is considerable support for expanding health insurance to cover some or all of the uninsured, but there are substantial differences among proposals in the nature of coverage that would be provided. This study examines how the take-up of coverage -- the price elasticity of demand -- will depend on the nature of the coverage. Study Design: I conduct simulation analyses using the Medical Expenditure Panel Survey, the Survey of Program Dynamics, and the Rand Health Insurance Experiment, to estimate reservation prices of alternative coverage packages for uninsured populations. I estimate the proportion of uninsured people who would take-up coverage under each of the options. I use these proportions to compute a demand curve for health insurance offered at different prices under each coverage option. I then use this demand curve to estimate the price elasticity of demand for coverage. Population Studied: Uninsured adults under 65 with incomes <200% FPL. Principal Findings: Different benefit packages will lead to different rates of participation by the uninsured and will achieve different public policy goals. Low-income uninsured people are likely to prefer health insurance that provides frontend protection, rather than coverage that provides protection in case of a catastrophic illness. For a low-income population, the estimated price elasticity for a high deductible policy is about 1/3 as high as for a 100% coverage policy. Conclusions: Uninsured people have much lower income and far fewer assets than do those who are already purchasing coverage. These differences affect the reservation price they would be willing to pay for different forms of coverage. Estimates of take-up for policies should take into account the nature of coverage offered. Implications for Policy, Delivery, or Practice: Thus, coverage expansions that seek to solve the problems that uninsured people face will only be successful if they incorporate coverage that is at least as comprehensive and generous as the private market average. Primary Funding Source: No Funding ●Benefit Design and the Use of Specialty Drugs Dana Goldman, Ph.D., Geoffrey Joyce, Ph.D., Grant Lawless, M.D., Ph.D., Vincent Willey, PharmD., William Crown, Ph.D. Presented By: Dana Goldman, Ph.D., Corporate Chair and Director, Health Economics, RAND Corporation, 1776 Main Street, PO Box 2138, Santa Monica, CA 90407-2138; Tel: (310)451-7017; Fax: (310) 260-8155; Email: dgoldman@rand.org Research Objective: Specialty pharmaceuticals (biologic agents, injectible medications) are high-dollar chronic drug therapies, often costing more than $1,000 per month.. Coverage of these products varies considerably across plans— even more so than traditional prescription drugs. This analysis estimates the financial risk imposed on patients treated with specialty drugs, and the effects of benefit design on the use of these products. Study Design: Our analysis is a retrospective study of patients enrolled in private health plans. We document the variability in coverage of specialty pharmaceuticals and the consequences for plan spending and patients’ out-of-pocket expenses. We examine how responsive use of specialty products is to changes in benefit design, and contrast this demand curve to traditional prescription drugs. We then simulate how alternative benefit designs affect patient out-ofpocket spending and plan use, including the key policy issue of whether to cover these products through the pharmacy benefit. Population Studied: The data contain up to 4 years of administrative claims (2001 to 2004) from 50 different health plans across multiple employers. The data include information on medical and pharmaceutical claims, and these data have been linked to detailed benefit designs for each plan. The study sample consists of privately-insured patients receiving treatment for cancer, kidney disease, rheumatoid arthritis and multiple sclerosis (sample sizes range from 2,000-20,000 per condition). Principal Findings: Patients with these conditions face substantial financial risk for high out-of-pocket expenses, as coverage is less generous for these conditions than for the prototypical beneficiary. The shift towards covering specialty drugs under the pharmacy benefit rather than the medical benefit reduces out-of-pocket costs, at least in the short run. Preliminary results indicate that the demand for specialty drugs is less price elastic than for oral agents (ranging from – 0.07 to –0.22). This suggests that many specialty drugs are prime candidates for full insurance coverage. Conclusions: Specialty drugs are one of the fastest growing and most expensive categories of prescription drugs. The high costs of these drugs ? combined with the high medical costs of treating these conditions ? often result in significant out-of-pocket expenditures. Covering these products under the prescription drug plan can reduce total plan costs and patient out-of-pocket expenditures. Implications for Policy, Delivery, or Practice: The clinical and financial effects of specialty pharmaceuticals require new thinking about the structure of insurance coverage, which is likely to differ from coverage of traditional oral agents. Primary Funding Source: NIA, Amgen ●The Impact of Premium Increases on SCHIP Enrollment in Arizona Joshua McFeeters, M.P.P., Genevieve Kenney, Ph.D. Presented By: Joshua McFeeters, M.P.P., Research Associate, Health Policy, The Urban Institute, 2100 M Street NW, Washington, DC 20037; Tel: (202) 261-5796; Fax: (202) 2231149; Email: jmcfeete@ui.urban.org Research Objective: Premiums and other forms of cost sharing have become a more prominent feature of state public health insurance programs for children. Between 2002 and 2004, 16 states either implemented new premiums or raised premiums in their SCHIP programs. Moreover, pending federal legislation would permit Medicaid programs to introduce greater cost sharing for children living above the federal poverty line (FPL). While there is a growing body of research on the impacts of premiums, there is very little research on the impacts of premiums for families below 150% of FPL. This study examines the experience of Arizona’s SCHIP program, which implemented premiums for children with incomes less than 150% FPL for the first time in July of 2004 and raised premiums for children with incomes between 150% and 200% of the FPL. Study Design: This study uses administrative SCHIP enrollment data from Arizona from 2001 to 2005. The administrative data include information on age, race, gender, and family income. Economic conditions are measured using data on state unemployment and economic growth. A series of descriptive and multivariate analyses performed include monthly SCHIP enrollment totals, rate of enrollment growth, time-series analysis of enrollment trends, and duration analysis using a Cox proportional hazard model to measure the rate of disenrollment from SCHIP. Population Studied: Children in the Arizona Kids Care Program (Arizona SCHIP). Principal Findings: Preliminary results indicate that enrollment in the Arizona SCHIP program for children with family incomes below 150% of the FPL dropped 15 percent six months after the implementation of a new premium in July 2004 of $10/month for one child or $15/month for two or more children. Enrollment declined from 27,692 in June 2004 to 23,506 in December 2004, and enrollment had not returned to pre-premium levels by September 2005. Duration analysis of the below 150% of the FPL income group indicates that after the new premium children were 16 percent more likely to disenroll from SCHIP than before the implementation of the premium. The premium increase that was implemented for the higher income group did not appear to have an adverse effect enrollment in that income group; in fact, SCHIP enrollment grew at a higher rate after the premium hike than it had in the previous period. Conclusions: The implementation of premium of just $10 or less per month per child in Arizona’s SCHIP program for children with family incomes below 150% FPL was associated with a significant enrollment decline for that income group. Implications for Policy, Delivery, or Practice: This analysis suggests that for near poor families, even modest premiums can lead to reductions in caseloads. While this analysis does not provide direct evidence on the coverage implications of the premium increase, other data suggest that many publicly covered children in this income group do not have access to employer-sponsored coverage. Thus, many of the children losing public coverage due to the premium increase may become uninsured and experience access problems. The results suggest that there may be some scope for raising premium levels for higher-income children covered by SCHIP. Primary Funding Source: Children's Action Alliance ●Annual Re-Enrollment in the State Children's Health Insurance Program (SCHIP): Factors That Influence Parents’ Decisions to Continue, Change, or Terminate SCHIP Coverage Laura Shone, Dr.P.H., M.S.W., Jonathan D. Klein, M.D., M.P.H., Alina Bajorska, MS, Peter G. Szilagyi, M.D., M.P.H. 7%, p<.05); “SCHIP and private insurance are different” (75%, 56%, 43%, p<.01). The mean reported “Amount per month SCHIP is worth” was $60 among those who became uninsured, $104 among SCHIP re-enrollers, and $113 among those who obtained other coverage – either Medicaid or private insurance (p<.01). Conclusions: Most children re-enroll in SCHIP, however, the re-enrollment process contributes to coverage lapses and SCHIP exit. Becoming uninsured after SCHIP is associated with having reported no medical visits during SCHIP and negative experiences with or perceptions of SCHIP during the child’s SCHIP enrollment. Implications for Policy, Delivery, or Practice: Future strategies should focus on uninterrupted retention of children already enrolled in SCHIP, whether through passive recertification or assistance with re-enrollment, as well as efforts to identify and address concerns of enrollees who do not use care or are dissatisfied with SCHIP. Primary Funding Source: AHRQ Call for Papers Presented By: Laura Shone, Dr.P.H., M.S.W., Assistant Professor, Pediatrics, University of Rochester, School of Medicine, 601 Elmwood Ave, Box 777, Rochester, NY, 14642; Tel: 585-273-4084; Email: Laura_Shone@urmc.rochester.edu Research Objective: Rationale: Enrollment retention is a significant challenge in SCHIP. Attrition peaks at times when active re-enrollment/renewal is required, and children may become uninsured rather than re-enroll in SCHIP. Little is known about the factors that influence parents’ decisions to continue, change or terminate SCHIP coverage. Objective: Among SCHIP enrollees throughout one large state, 1) Describe the proportions of children who re-enroll in SCHIP for a second year compared to those who obtain other coverage and those who become uninsured; 2) Identify factors associated with changing coverage or becoming uninsured after holding SCHIP coverage. Study Design: Pre-post parent telephone interview shortly after enrollment and 1 year later, for measures of insurance coverage; access, use, and satisfaction with medical care; perceptions of SCHIP and other insurance; and experiences with annual SCHIP recertification process. Bivariate and multivariate analyses were performed using STATA 8.0 to provide weighted statewide estimates. Population Studied: Random sample of new enrollees in New York State’s SCHIP, stratified by age, race, and NYS region (Time One n=2,644; Time Two n=2,290; 87%). Principal Findings: Seven percent of children lost SCHIP eligibility (e.g. “aged out” or “family income too high”). Of the remainder, 86.3% re-enrolled in SCHIP and 13.6% exited -either to other coverage (8.2% overall; 4.6% private and 3.6% Medicaid) or became uninsured (5.4%). Problems with the reenrollment process led to coverage lapses for 6.2% of reenrollers and to SCHIP termination for 17% of exiters. Children who became uninsured after SCHIP, compared with those who re-enrolled in SCHIP or obtained other coverage, were significantly more likely to report: Having no visits during SCHIP (18 %, 10%, 2 % respectively, p<.001); “Less satisfied with SCHIP physician vs. pre-SCHIP physician (18%, 5%, 7%, p<.001); “Physician does not respect me” (24%, 6%, 4%, p<.001); “My child not treated well due to SCHIP” (23%, 9%, Uninsurance: Measurement, Costs & Outcomes Chair: Lynn Blewett, University of Minnesota Sunday, June 25 • 5:45 pm – 7:15 pm ●Why are Survey Counts of Medicaid Enrollees Lower than Administrative Enrollment Counts? A Match Study Between the Current Population Survey and the Medicaid Statistical Information System Michael Davern, Ph.D., Dean Resnick, David Baugh, Robert Stweart, George Greenberg, Kathleen Thiede Call, Ph.D. Presented By: Michael Davern, Ph.D., Assistant Professor, SHADAC, University of Minnesota, 2221 University Ave. SE Suite 345, Minneapolis, MN 55414; Tel: 612-625-4835; Fax: 612624-1493; Email: daver004@umn.edu Research Objective: Survey counts of those enrolled in Medicaid are significantly lower than administrative data counts. Because surveys are the only source of information on key policy populations like the uninsured, and those who are eligible for Medicaid but not enrolled, this disparity can have significant implications depending on why the two counts are systematically different. We examine four main types of explanations and their implications for policy research. The first explanation could be that Medicaid enrollees answer surveys as though they are uninsured (and as a result the count of uninsured people from surveys is incorrect). Second, Medicaid enrollees are systematically left off survey sampling frames or are more likely to not respond to surveys than those not enrolled in Medicaid. Third, Medicaid enrollees answer surveys by naming some other type of health insurance coverage (e.g., “Medicare”). Fourth, administrative data could “over-count” the Medicaid population. Study Design: In an attempt to determine the extent to which any (or all) of these possible explanations are in operation we match the Center’s for Medicare and Medicaid Services (CMS) Medicaid Statistical Information System (MSIS) and Medicare Enrollment Database (MEDB) to two surveys with health insurance coverage information (the Current Population Survey’s Annual Social and Economic supplement -CPSASEC- and the National Health Interview Survey) as well as the Census Bureau’s Master Address File (MAF). We put the CMS enrollment data through the PRED validation system at the Census Bureau to validate social security numbers (SSNs) and then we use SSNs, age and sex to match the data to the survey data and MAF. Ninety two percent of the MSIS data have a valid SSN and 94% of the CPS-ASEC data have a valid SSN. The CPS-ASEC cases for 2001 and 2002 will be matched to the MSIS from 2000-2003. We expect approximately 30,000 matched cases. Population Studied: Medicaid enrollees on the MSIS enrollment data file and the non-institutionalized US population in 2001 and 2002. Principal Findings: By March 2006 we know the extent to which the uninsurance estimates from the CPS-ASEC are biased due to people who have Medicaid coverage reporting that they are uninsured. We will also determine the rates at which people who have Medicaid coverage report some other type of health insurance coverage (either in combination with Medicaid or not) in the CPS-ASEC. And finally, we will determine the rate at which people who probably do not have Medicaid coverage report having Medicaid coverage in the CPS-ASEC. Conclusions: Using the data analysis we will make recommendations concerning survey instrument improvement for measuring the Medicaid population, the uninsured, and those with more than one type of coverage (e.g., Medicaid and Medicare). We will also have recommendations for policy analysts using the CPS-ASEC data to estimate essential policy subpopulations like the eligible but uninsured population. Implications for Policy, Delivery, or Practice: Our conclusions will help policy simulations that estimate important groups like the eligible but uninsured group in their policy simulations that are used to evaluate proposed and implemented health policy. Primary Funding Source: RWJF, Health and Human Services ●Expanding Medicaid May Cost Less Than You Think: Evidence from Health Insurance Churning in the 19982002 MEPS Danielle Ferry, ABD, Sherry Glied, MA, Ph.D., Kathryn Klein, M.P.H. Presented By: Danielle Ferry, ABD, National Bureau of Economic Research, 92 Atlantic Ave., Apt. 4D, Brooklyn, NY 11201; Tel: 917/822-6674; Email: dferry@nber.org Research Objective: Thirty percent of low-income individuals covered by Medicaid in January of 1998 or 1999 experienced at least one spell of uninsurance in the subsequent two years. Many may find extensive and regular recertification requirements too complicated and taxing to bother with. For similar reasons, others may never enroll in the first place. The gap between eligibility and enrollment and the link between lack of health coverage and unmet health needs has spurred interest in expanding Medicaid enrollment by reducing administrative barriers. A more ambitious proposal would extend Medicaid eligibility to anyone without health coverage. This study explores the costs of expanding Medicaid to the uninsured and of using simpler recertification procedures to increase enrollment among those currently eligible. Study Design: We use month-to-month health insurance transitions, expenditures, and service utilization patterns, to compare the health needs of, first: 1) those who were consistently uninsured, 2) those who were consistently enrolled in Medicaid, and 3) those who transitioned from no coverage to Medicaid; and second: 1) those who were consistently enrolled in Medicaid and 2) those who transitioned off of Medicaid. If the decision to enroll in or maintain Medicaid is correlated with health needs, then we expect the pre-transition expenditures of those who gained Medicaid coverage to exceed those of individuals who were persistently uninsured and the pre-transition spending of those who left Medicaid to fall below that of individuals who were enrolled in Medicaid constantly. Population Studied: We use data on 45,650 adults who participated in the Medical Expenditure Panel Survey (MEPS) for two full years between 1998 and 2002 and for whom health insurance information is available for all 24 months. Participants wer ages 19-62 in their first year on the panel. Principal Findings: Our findings support these hypotheses. In the month prior to enrollment, the average expenditures of the consistently uninsured are less than one-quarter of those of “new” Medicaid enrollees and just 17 percent as high as those of long-term Medicaid enrollees. The monthly spending of those who gave up Medicaid to become uninsured, in the month prior to switching, is 65 percent of that of individuals who maintained their enrollment. Conclusions: These results suggests that the costs associated with expanding Medicaid coverage are likely to be less than what can be extrapolated directly from an examination of either the current Medicaid population or those individuals who currently choose to enroll. Our conclusions are robust to a number of sensitivity analyses, including disaggregating expenditures by type, including the cost of uncompensated care in expenditure figures, restricting the sample to individuals with chronic conditions, substituting health care utilization for expenditures, and extending the comparison period from one month to two or four months. Implications for Policy, Delivery, or Practice: States should consider both expanding Medicaid eligibility and working to improve retention among those beneficiaries who are currently eligible as a viable approach to the problem of the uninsured. Primary Funding Source: CWF ●Do Gaps in Health Insurance Make a Difference? Jane Griffin, M.P.H., Tricia Leddy, MS, Christine Payne, Ph.D., Melinda Thomas, MS Presented By: Jane Griffin, M.P.H., Project Director, Rhode Island Department of Human Services, Medicaid Research and Evaluation Project, RIte Care Program, 600 New London Avenue, Cranston, RI 02920; Tel: 401-462-6330; Fax: 401-4626353; Email: jgriffin@dhs.ri.gov Research Objective: To determine the effect of disruption of health coverage on access utilization and unmet health needs for a cohort of children continuously for at least one year in Medicaid managed care program. Study Design: This is a follow-up cohort study. Rhode Island’s Medicaid Management Information System (MMIS) was used to create a retrospective cohort of children <=18 who were continuously enrolled in RIte Care for one year. This group was stratified into gap groups according to number of months uninsured in the subsequent year. A follow-up CATI phone survey was conducted with parents that measured health coverage, utilization, barriers to care, health status, and unmet need. Population Studied: Phone interviews were conducted with 660 parents. In the follow-up year two hundred and ninety (290) of the children had intermittent health coverage (uninsured average of 5.4 months in year) and 370 children had continuous coverage. Principal Findings: Hispanic children, especially if their parents do not speak English, were significantly more likely to have intermittent coverage. Children with intermittent coverage were significantly more likely to live in families that experienced other household disruptions including moving (25.5%), parents changed jobs (34.1%), family without phone (18.6%), parents without health insurance (58.6%) and parents unable to pay mortgage (25.5%). Both children with gaps and children with continuous coverage had a usual source of care (96.7% and 99.2% respectively). However, children with intermittent coverage faced greater difficulty getting medical care (28.3% compared to 7.4%). Children with disruptions in coverage were significantly more likely to have unmet needs for dental care (37.6%), prescription medication (20.8%), specialty care (29.1%), and mental health care (28.0%. Despite problems with access to care, children with intermittent health insurance received the same number of preventive,acute and specialty visits as children with continuous coverage. In addition children with gaps in coverage had the same health status at one year as children with continuous coverage as measured by ability to participate in usual activity and number of chronic conditions. Conclusions: Children on Medicaid managed care with intermittent health insurance coverage for one year following twelve months of continuous public coverage face more barriers and reported more unmet need than children with continuous health coverage. Their families experience more life disruptions including parents changing jobs, moving, and other economic hardships. Children with gaps in coverage were more likely to be minority and speak Spanish. Although having gaps in insurance created difficulty in obtaining care, children with gaps in coverage had the same number of visits as children with continuous coverage and there was no evidence of erosion in health status Implications for Policy, Delivery, or Practice: Disruptions in health care coverage creates hardships in the lives of lowincome families. Children who have gaps in coverage face barriers to obtaining needed care. However, children who have a medical home for at least one year continue to have a usual place of care and similar utlization even if they experience gaps in health coverage. Primary Funding Source: RWJF ●The Effects of Being Uninsured on Mortality and Morbidity Richard Kronick, Ph.D. Presented By: Richard Kronick, Ph.D., AcademyHealth/NCHS Fellow, Division of Health Care Statistics, National Center for Health Statistics, 3311 Toledo Road, Hyattsville, MD 20782; Tel: 301-458-4577; Email: rkronick@cdc.gov Research Objective: The purpose of this research is to improve our estimates of the effects of being uninsured on mortality and morbidity. In a much cited 2002 report Care Without Coverage, Too Little, Too Late, the Institute of Medicine estimated that lack of insurance led to 18,000 excess deaths annually among 25-64 year olds. In a subsequent IoM report, Elizabeth Vigdor estimated that lack of insurance in the United States leads to an annual loss of between $65-$130 billion in ‘health capital’. However, as both the 2002 IoM report and Vigdor acknowledged, estimates of the magnitude of the effect of lack of insurance on morbidity and mortality are quite uncertain. The estimate of 18,000 excess deaths is largely based on two population-based sample surveys comparing subsequent mortality of respondents who were insured at baseline with those who were uninsured. One study was based on approximately 4,700 respondents who were interviewed in the early 1970s. The other uses a larger sample of 150,000 respondents from the Current Population Survey interviewed in the mid-1980s, but is limited to rudimentary controls for baseline health status. And Vigdor’s estimate of the effects of lack of insurance on morbidity estimates an effect of somewhere between $0 and $65 billion loss in health capital annually -- a rather wide range. Study Design: We use data from the National Health Interview Surveys from 1986 through 2000, which have been linked with the 2002 National Death Index. The NHIS is a national multistage probability sample of the civilian noninstitutionalized population. For approximately 500,000 adults in the combined 1986-2000 NHIS samples, we estimated Cox proportional hazard regressions to estimate the probability of death, controlling for a wide range of baseline characteristics. We provide both straightforward estimates of the effects of being uninsured at baseline, as well as Instrumental Variable estimates. In the IV estimates the key instrument is the generosity of the state Medicaid program during the survey year. We take advantage of changes over time in relative generosity (e.g., Tennessee Medicaid generosity increased substantially between 1986 and 2000) to measure whether these largely exogenous changes in Medicaid generosity are related to patterns of morbidity and mortality. We also analyze whether the mortality differential between the insured and uninsured narrows after age 65, on the thought that if poor access to care is the major factor contributing to poor outcomes, then nearly universal Medicare coverage at age 65 should narrow the mortality differential. Population Studied: Approximately 500,000 respondents from the 1986-2000 NHIS, with up to sixteen years of followup mortality data from the NDI. Principal Findings: Similar to the results from earlier work, preliminary results from the Cox proportional hazard regressions estimate that mortality rates for the uninsured are approximately 15% to 25% higher than for the privately insured. However, at age 65 the mortality differential does not appear to narrow significantly. The Instrumental Variables estimates and estimates of the effects of lack of insurance on morbidity have not yet been completed. Conclusions: Not yet known. Implications for Policy, Delivery, or Practice: It is expected that policy makers will benefit from more precise estimates of the effects of lack of insurance on morbidity and mortality. It is also hoped that greater precision in the estimates of the effects of insurance on health will provide the information that estimating agencies (such as CBO, OMB, their counterparts in state legislatures, and private firms such as Lewin) need to be comfortably in expanding the scope of their estimates of health care reform proposals, to include not simply estimates of the number of people covered, and the public and private sector costs, but also of the health benefits of expanded coverage. Primary Funding Source: National Center for Health Statistics ●How Did SCHIP Affect the Insurance Coverage of Immigrant Children? Anthony Lo Sasso, Ph.D., Thomas Buchmueller, Ph.D., Kathleen Wong Presented By: Anthony Lo Sasso, Ph.D., Associate Professor, Health Policy and Administration, University of Illinois at Chicago, 1603 W. Taylor, Chicago, IL 60612; Tel: (312) 4131312; Fax: (312) 996-5356; Email: losasso@uic.edu Research Objective: In the past two decades there have been substantial initiatives at the state and federal levels aimed at increasing insurance coverage among children. However, it is estimated that over 6 million children who are eligible for public insurance remain uninsured. A crucial first step for addressing this problem is to determine how the take-up of public coverage varies within the population. Children of immigrants are an especially important group to consider as foreign-born adults are nearly three times as likely to be uninsured as native-born Americans and their children are also more likely to be uninsured than children whose parents were born in the US. We test whether the effect of the SCHIP expansion was different for children of foreign-born and USborn parents. Study Design: The analysis is based on repeat cross-section data from the Current Population Survey. We use an instrumental variables approach in which the effect of SCHIP eligibility is identified by cross-state differences in the timing and extent of changes in the income eligibility limit over the period from 1996 to 2003. We test for the effect of SCHIP on insurance coverage from any source as well as on the probability of having public insurance (take-up) and on the probability of having private coverage (crowd-out), and how this effect differs between the children of native and foreignborn parents. Population Studied: A nationally representative sample of children aged 0-18. Principal Findings: Our results suggest that take-up among the children of the foreign born was at least as high as natives and in some specifications considerably higher. In our preferred specification, take-up rates for public insurance were roughly twice as large for children of foreign-born parents than for children of natives: approximately 20% versus 9%, respectively. In addition, private health insurance crowd-out was significantly lower among children of immigrants: crowd- out for children of natives was nearly 50% while for children of immigrants it was roughly 25%. Conclusions: Our crowd-out finding contrasts strikingly with recent research on the impact of welfare reform found that reductions in public coverage among immigrants were completely offset by increases in private coverage, implying 100% substitution. However, because immigrants are so much less likely to have private insurance than natives it is perhaps not surprising that crowd-out is less of an issue. Similarly, our results showing stronger take-up among the children of immigrants are at odds with prior research. Yet, given the high rates of uninsurance among immigrants relative to natives, it is again perhaps not surprising that public expansions would find greater success among immigrants. Implications for Policy, Delivery, or Practice: Because SCHIP was enacted just after the 1996 Federal welfare reform legislation, which singled out recent immigrants for harsh treatment, there is reason to expect that SCHIP’s effect on coverage might have been smaller for children of immigrants. However, if outreach efforts to market SCHIP in languages other than English were effective, nativity-related differences in take-up may be less in the case of SCHIP as compared to earlier expansions. Our results have important implications for policy makers looking to further refine efforts to encourage enrollment in public insurance programs by vulnerable populations. Primary Funding Source: RWJF Call for Papers Older & Sicker: Access & Coverage Issues Chair: Deborah Chollet, Mathematica Policy Research, Inc. Monday, June 26 • 8:30 am – 10:00 am ●A Theory of Health Disparities and Medical Technology Geoffrey Joyce, Ph.D., Dana Goldman, Ph.D., Pinar KaracaMandic, Ph.D. Presented By: Geoffrey Joyce, Ph.D., Senior Economist, RAND Corporation, 1776 Main Street, PO Box 2138, Santa Monica, CA 90407-2138; Tel: (310)393-0411 ext 6779; Fax: (310) 260-8155; Email: Geoffrey_Joyce@rand.org Research Objective: The standard Medicare drug benefit (Part D) contains sizeable gaps in coverage for people who spend moderate to high amounts on prescription drugs. This gap or “doughnut-hole” creates a non-linear price schedule with price uncertainty since consumers cannot precisely forecast how much they will spend during the year. Changes in pharmaceutical use—especially discontinuation of efficacious therapy—could have important health consequences for elderly beneficiaries with diabetes, asthma and other chronic conditions. Study Design: We examine outpatient medication use among retirees age 65 and older who have employer-provided drug coverage with a spending cap of $2,500 per year. We examine whether the elderly behave in a myopic or rational manner—that is, do they change their behavior as they approach the cap but do not exceed it. We examine continuation of pharmacy use after patients exceed the cap and their use of prescription medications when coverage resumes in January. We also estimate the use of generic medications and mail-order services before and after exceeding the cap. To assess the potential health consequences, we examine use of medical services in the subsequent year (overall and by type of service) among patients who did and did not exceed the cap. Population Studied: The data include two years (2003-2004) of medical and pharmacy claims on more than 40,000 retirees age 65 and older are enrolled in a private health plan with an annual spending cap of $2,500 for outpatient prescription drugs. We obtained pharmacy transaction data for a subsample of beneficiaries to capture their use (cash purchases) of prescription drugs after exceeding the cap. Principal Findings: Preliminary analyses suggest that the majority of beneficiaries who exceed the annual spending cap continue to purchase medications for a chronic illness. They are more like purchase generic medications and have larger gaps between refills than before reaching the cap and compared to those who did not exceed the cap. Future work will examine whether this pattern is consistent across therapeutic classes and how spending caps affect pharmacy and medical care utilization in the subsequent year. Conclusions: Preliminary findings suggest that beneficiaries with moderate to high drug expenses, particularly the chronically ill, face longer gaps in coverage and are more susceptible to interruptions in insurance. Discontinuation of efficacious therapy could have important health consequences for elderly beneficiaries with diabetes, asthma and other chronic conditions. Implications for Policy, Delivery, or Practice: Cycling into and out of coverage may be more disruptive to care plans than a stable benefit with higher coinsurance. Primary Funding Source: NIA ●Impact of State Mandatory Insurance Coverage on Selected Diabetes Care Services Rui Li, Ph.D, Ping Zhang, Ph.D., DeKeely Hartsfield, M.P.H. Presented By: Rui Li, Ph.D, Prevention Effectiveness Fellow, Division of Diabetes Translation, Centers for Disease Control and Prevention, 4770 Buford Hwy, N.E. MS K-10, Atlanta, GA 30341; Tel: (770)488-1070; Fax: (770)488-1148; Email: Rli2@cdc.gov Research Objective: Diabetes is a costly disease in the United States with serious complications such as blindness, amputation and end-stage renal diseases. The development of diabetes complications, however, can be prevented or delayed by good diabetes care. To improve access to care and quality of care for diabetes patients, many states have passed laws that mandate private insurance to cover certain diabetes care services, devices and supplies. This research is to examine the coverage of these laws and their impact on the diabetes care services. Study Design: We assembled all legal documents related to private insurance coverage of diabetes care in all 50 states, and categorized the coverage. We used a hierarchical logistic regression model to examine the independent association between the state laws and the services, including daily selfmonitoring blood sugar level, annual eye exams, annual foot exams, as well as the use of all three services. Disease conditions, persons’ demographic characteristics and time effects were adjusted. Population Studied: Data for the regression analysis was from the 1996 to 2000 Behavioral Risk Factor Surveillance System (BRFSS), an annual, state-based, random telephonesurvey of 150,000-210,000 community-dwelling US adults. Only people with diabetes and covered by employer-provided insurance or self-bought insurance (n=9215) were included in these analyses. Principal Findings: Forty-six states had laws governing diabetes care but the extent of this coverage varied considerably. Forty states covered self-management education; 12 states covered therapeutic footwear; and only 3 states specified the coverage for eye and foot exams. In addition, 5 states covered HbA1c test and 41 states had general coverage of pharmaceuticals, devices, and supplies. Before the state mandatory law, 42% of persons with diabetes performed daily SMBG. Sixty-three percent received annual dilated eye exams, and 57% received annual foot exams. Only 23% of person with diabetes received all three of the services. The impact of legislation varied depending on the components of the laws and the type of services. Mandatory coverage of devices and supplies increased a person’s likelihood of daily SMBG by 2 percentage points (Not significant) in the first year after the laws were passed, and 4 percentage points (p<=0.05) in the following years. Other coverage had no significant effect on any of the relevant services. However, state laws increased the probability of all three services by 7 percentage points (p<=0.001) in the first year, and 5 percentage points (p<=0.01) in the subsequent years. Conclusions: Coverage of state laws on diabetes services, devices and supplies were heterogeneous. Overall, the state mandatory insurance coverage had some effect on the utilization of diabetes care services. Implications for Policy, Delivery, or Practice: Laws and regulations can be an effective policy tool to improve the quality of diabetes care. Reasons why the coverage of state laws has different effects on different services need to be further investigated. Primary Funding Source: No Funding ●Medicare to the Rescue: Intensity of Health Services and Costs of Care for Previously Uninsured Adults with Chronic Conditions J. Michael McWilliams, M.D., Alan M. Zaslavsky, Ph.D., Ellen Meara, Ph.D., John Z. Ayanian, M.D., M.P.P. Presented By: J. Michael McWilliams, M.D., General Medicine Fellow, Department of Health Care Policy, and Division of General Medicine, Harvard Medical School and Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115; Tel: 617-732-6043; Email: mmcwilliams@partners.org Research Objective: To compare the differential impact of gaining Medicare coverage on rates of hospitalization, use of physician visits, and total medical expenditures for previously insured and uninsured adults with chronic disease. Study Design: Using a natural experiment design and longitudinal data from years 1992-2004 of the nationally representative Health and Retirement Study, we examined differences in self-reported hospitalizations, physician visits, and medical expenditures at multiple biannual intervals before and after adults became eligible for Medicare at age 65. We modeled the number of hospitalizations and physician visits and total medical expenditures in each two-year period as a function of prior insurance status, allowing for differences before and after age 65 and adjusting for gender, race, income, education, and self-reported health at baseline. All analyses included sampling weights and multilevel random effects to account for survey design and longitudinally correlated data. Population Studied: A total of 2263 adults aged 54-61 in 1992 and 66-73 in 2004 who reported being diagnosed with hypertension, diabetes, or heart disease by age 64, and who were classified as continuously insured (n=2008) or continuously uninsured (n=255) from ages 60-64. Because expenditure data were available only through 2002, we expanded the uninsured group to include those who were ever uninsured prior to age 65 (total n=625) to increase power to detect differences in spending. Adults with public insurance before age 65 were excluded. Principal Findings: Adjusted increases in rates of hospitalizations, physician visits, and total medical expenditures were significantly greater for previously uninsured than previously insured adults (greater by 0.26 hospitalizations/2 years(2y), P=0.005; 5.6 physician visits/2y, P<0.001; and $6199/2y, P=0.01)). Despite worse self-reported health status, adjusted hospitalization rates tended to be lower for previously uninsured than insured adults prior to age 65 (0.09 hospitalizations/2y lower, P=0.13) but were significantly higher after age 65 (0.25 hospitalizations/2y higher, P=0.046). Similarly, adjusted physician visit rates were significantly lower for previously uninsured adults before age 65 (2.3 visits/2y lower, P=0.02) but significantly higher after age 65 (5.6 visits/2y higher, P=0.02). Adjusted medical expenditures were also significantly lower for previously uninsured adults before age 65 ($2579/2y lower, P=0.009) but significantly higher after age 65 ($5826/2y higher, P=0.045). Conclusions: After gaining Medicare coverage, increases in hospitalizations, physician visits, and expenditures were substantially greater for previously uninsured than previously insured adults. Moreover, after adjusting for several confounding characteristics, the excess utilization for previously uninsured adults after age 65 greatly exceeded their under-utilization prior to age 65. Implications for Policy, Delivery, or Practice: Uninsured near-elderly adults with chronic conditions may experience greater morbidity, use health services more intensively, and require more costly care as Medicare beneficiaries than they would if insured prior to age 65. Furthermore, their increased use of medical services and related expenditures after age 65 may be substantially greater than their lower costs of care before age 65. Therefore, expanding insurance coverage and access to appropriate care for uninsured near-elderly adults with chronic conditions may offset Medicare expenditures considerably after age 65 by addressing the health needs of this vulnerable population at an earlier age. Primary Funding Source: AHRQ ●Veterans Using and Uninsured Veterans not Using VA Health Care Karin Nelson, M.D., M.S.H.S., Gordon A. Starkebaum, M.D., Gayle E. Reiber, Ph.D., M.P.H. Presented By: Karin Nelson, M.D., M.S.H.S., Staff Physician, Assistant Professor, Department of Medicine, University of Washington, VA Puget Sound, 1660 South Columbian Way, S111-GIMC, Seattle, WA 98108; Tel: 206-277-5118; Fax: 206-7642849; Email: Karin.nelson@va.gov Research Objective: (1) To examine veteran reliance on health services provided by the VA, (2) to describe the characteristics of veterans who receive VA care and (3) to report rates of uninsurance among veterans and characteristics of uninsured veterans. Study Design: We analyzed data from the 2000 Behavioral Risk Factor Surveillance System, a cross-sectional national telephone survey. Using bivariate and multivariate analyses, we examined the association of veteran's demographic characteristics, health insurance coverage and use of VA services. Veterans not reporting VA coverage and having no other source of health insurance were considered uninsured. Population Studied: Nationally representative sample of US veterans (n=23,797). Principal Findings: Among US veterans, 6.2% report receiving all of their health care at the VA, 6.9% report receiving some of their health care at the VA and 86.9% did not use VA health care. Poor, less-educated and minority veterans are more likely to receive all of their health care at the VA. Veterans under 65 who utilize the VA for all of their health care also report coverage with either private insurance (42.6%) or Medicare (36.3%). Of the under-65 veterans, 8.6% (an estimated 1.3 million individuals) were uninsured. Uninsured veterans were less likely to be able to afford a doctor or see a doctor within the last year. Conclusions: Veterans who utilize the VA for all of their health care are more likely to be from disadvantaged groups. A large number of veterans who could use VA services were uninsured. Implications for Policy, Delivery, or Practice: We found that the majority of veterans who are eligible for VA health care services do not use the VA. As the US looks for strategies to reduce the existing number of uninsured, expanding existing federal programs is one possible solution. Although our study suggests that the Department of Veterans Affairs provides services to individuals from disadvantaged groups including veterans from poor, less educated and minority populations, a significant number of veterans remained uninsured. The large number of uninsured veterans who report not using VA services should be a target population for VA recruitment and enrollment efforts, especially given increasing evidence on the detrimental clinical effects of not having health insurance. Health care and preventive practices are indicated to minimize future morbidity among our uninsured veterans. Primary Funding Source: VA ●The Decline in Employer-Sponsored Health Insurance Coverage for Retirees and Its Impact on Older Americans Erin Strumpf, B.A. Presented By: Erin Strumpf, B.A., Health Policy, Harvard University, NBER, 1050 Massachusetts Ave., 3rd Fl., Cambridge, MA 02138; Tel: (617) 588-1483; Fax: (617) 8685023; Email: strumpf@fas.harvard.edu Research Objective: Over the last decade, in the face of rapidly increasing health care costs, the aging of the American population, and accounting policy changes, many employers have discontinued health insurance coverage offerings to retirees. While coverage rates for current retirees have remained relatively constant, offer rates among large employers have declined by about 50 percent since 1988. This decline portends a large drop in coverage rates for future retirees and the consequences of this sea change remain unexamined. This paper explores the likely implications of this coverage decline by investigating the effect of employersponsored retiree health insurance (RHI) offer on the retirement decision, health insurance coverage, and the health care spending, utilization and health status of older Americans. Study Design: I follow a panel of older individuals who all have employer-sponsored health insurance coverage in 1992. I examine differences in the subsequent retirement, health insurance coverage, health care utilization, health spending, and health status of those who were and were not offered RHI by their employer. The under age 65 and age 65+ populations are analyzed separately and controls are included for demographics, health and family characteristics. Population Studied: Health and Retirement Survey (HRS) respondents who are aged 47-63 and have employersponsored health insurance in 1992 are followed over ten years (1992-2002). Principal Findings: Preliminary findings suggest that individuals under age 65 who are offered RHI are more likely to retire and more likely to have health insurance coverage in retirement. There is also some evidence of a positive association between RHI offerings and various measures of health care utilization. In the top half of the spending distribution, out-of-pocket medical spending is significantly lower for those with RHI offer than for those without. While the results on health status show much smaller and statistically insignificant effects, the results from a two-stage model using propensity score matching will account for endogeneity which may bias these results toward zero. Conclusions: These results indicate that we can expect declines in RHI offer rates to have sizeable impacts on the retirement behavior of older workers which may have spillover effects on the labor market opportunities available to younger workers. Declines in RHI offer rates may result in increased financial risk and changes in health care utilization patterns for the near-elderly. Future work will explore the impact on health more carefully. Implications for Policy, Delivery, or Practice: The decline in rates of RHI offer by employers may leave many future retirees with fewer options for affordable coverage. Given the effects of this large change in health insurance options for older Americans, policymakers might consider options to strengthen the system of employer-sponsored coverage and/or ways to make other adequate and affordable insurance options available through public or private means. Furthermore, RHI is an important source of prescription drug coverage and the consequences of changes in these benefits will merit close attention as the Medicare prescription drug benefit is implemented. Primary Funding Source: NIA Related Posters Coverage & Access Poster Session A Sunday, June 25 • 2:00 pm – 3:30 pm ●Patient Assistance Programs in Massachusetts Community Health Centers Alexy Arauz Boudreau, M.D., M.P.H. Presented By: Alexy Arauz Boudreau, M.D., M.P.H., Fellow, MGH Center for Child and Adolescent Health Policy, Massachusetts General Hospital for Children, 50 Stanifor Street, Suite 901, Boston, MA 02114; Tel: 617 726-2337; Fax: 617 726-1886; Email: aarauz@partners.org Research Objective: Patient Assistance Programs (PAP) are charitable programs offered by pharmaceutical manufactures. There are 150-200 different PAP, each with its own application procedures making them difficult to operate. Community Health Centers (CHC) are encouraged to use 340 B pharmaceuticals, a federal regulation allowing qualified entities to purchase pharmaceuticals at prices exempt from Medicaid “best price” arrangements saving states health care dollars. Our goal is to assess 1) how and if Massachusetts’ CHC are utilizing PAP, 2) CHC costs to run the programs, and 3) how they may be integrated with 340 B pharmaceuticals. Study Design: We performed guided interviews with Executive Directors of MACHC. Executive directors included pharmacy directors and other personnel at their discretion. Common themes regarding the benefits and drawback of PAP were identified with conceptual analysis. We used costing analysis to obtain the cost per PAP successfully processed and the benefit to cost ratio per site. Population Studied: We identified a representative sample of CHC (14/50) based on their location in the State, number of patients, and availability of pharmacy services. Principal Findings: Eleven centers accepted interviews with 6 submitting costing data. Nine centers operated PAP in a systematic manner. Three organizational models were identified: PAP operated manually, computerized, or through existing pharmacies. PAP are run through on-site 340 B pharmacies and as separate programs. Four staffing models were found: providers, dedicated personal (nurses or program managers), pharmacy staff and pharmacy faculty in conjunction with a training program. Common benefits of PAP included: increasing access to care, improving quality of care, preventing CHC pharmacy financial loses, and increasing patient satisfaction. Common drawback included: operating PAP is complex, time consuming, and costly operations; and large start-up costs, decreased patient satisfaction due to inconsistencies with PAP, and concerns with quality of care and sustainable access. Benefit to cost ratios ranged from $5.41-$213 for dollar invested, (mean: $78.53, SD: $111.25), with higher ratios for those processing more medications. Cost to operate per medication received ranged from $10-$18.92 varying according to how programs were managed. Conclusions: PAP provide health centers with means to help patients’ access pharmaceuticals. They run at a cost to health centers, with most benefits accruing to patients in the form of medications. PAP run best with dedicated managers and when they are streamlined into existing operating systems. The greater the number of PAP processed, the greater the benefit to cost ratio. They are more economical to operate through in-house pharmacies, however CHC that started using PAP prior to obtaining 340 B pharmaceuticals are more likely to incorporate PAP into their systems than those with established 340 B pharmacies. Implications for Policy, Delivery, or Practice: PAP are valuable adjuvant to CHC to care for the underserved. However, CHC require financial resources to run these programs effectively. States should provide financial incentives to encourage the use of PAP. State and federal grant are needed to support PAP expansion costs. Finally, as a nation we need to decide if it is sufficient to rely on multiple voluntary charitable programs run by private-for-profit organizations to serve as our Nation’s safety net health care system. Primary Funding Source: CWF the publicly insured. These differences across insurance groups are due to both differences in the age distribution and the presence of medical conditions conditional on age. In particular, the uninsured have lower expected expenditures both because they are younger and because they report fewer medical conditions. For children, in contrast, health status appears to be similar among the privately and publicly insured while the uninsured children appear to be healthier. Conclusions: Our results are consistent with selection into health insurance based on health status. Those in worse health are more likely to enroll on average than those in poor health. Particular segments of the uninsured population are characterized by worse health, suggesting that barriers may exist in obtaining coverage for these individuals. Among adults, but not children, public insurance appears to cover a relatively less healthy segment of the population. Implications for Policy, Delivery, or Practice: Our findings suggest that researchers should be cautious in making extrapolations from populations currently enrolled in health insurance for estimates of the cost of covering the uninsured because these populations differ based on their underlying health status. The results also point to high priority segments of the uninsured population, from the perspective of their health, in targeting coverage expansions. Primary Funding Source: No Funding ●The Health Status of the Uninsured M. Kate Bundorf, Ph.D., Donna MacIsaac, MA ●The Effects of Chronic Illness and Chronic Uninsurance on Long-term Health Status Nancy Cheak, MA, Timothy McBride, Ph.D. Presented By: M. Kate Bundorf, Ph.D., Assistant Professor, Department of Health Research and Policy, Stanford University School of Medicine, HRP Redwood Building, Rm T108, Stanford, CA 94305-5405; Tel: 650.725.0067; Fax: 650.725-2586; Email: bundorf@stanford.edu Research Objective: The cost of strategies to reduce the number of uninsured Americans depends on the underlying health status of the uninsured. Relatively little evidence exists, however, on the extent to which the uninsured differ from those with either public or private health insurance in their health status. In this paper, we investigate the relationship between health status and insurance status. Study Design: Using data from the 2-year panel component of the Medical Expenditure Panel Survey, we develop estimates of an individual’s expected health expenditures based on the presence of medical conditions. This measure of health status allows us to aggregate multiple dimensions of health into a single index of health status based on the contribution of medical conditions to health care costs. We also use these estimates to develop measures of health status which distinguish between demographic characteristics and medical conditions in their contribution to expected health expenditures. Using these measures, we compare the health status of the privately insured, the uninsured, and the publicly insured. We also examine differences within insurance groups based on socioeconomic characteristics and examine the possibility of underreporting of medical conditions among the uninsured. Population Studied: A nationally representative sample of the under 65 population in U.S. during 1997-2002. Principal Findings: Among adults, we find that expected health expenditures among the privately insured are higher than those among the uninsured and lower than those among Presented By: Nancy Cheak, MA, Graduate Research Assistant, School of Public Health- Health Policy and Management, School of Public Health, 3545 Lafayette Ave., Saint Louis, MO 63104; Tel: 314-977-8128; Fax: 314-977-1674; Email: cheaknc@slu.edu Research Objective: This study uses a quantitative, longitudinal analysis of the determinants of health status, comparing the chronically ill and chronically uninsured to other groups within the population. Study Design: Using Michael Grossman’s Demand for Health Capital Model as my underlying theory, I will empirically test the hypothesis that persons who are chronically ill, and who have long spells of uninsurance, are unable to buy needed health commodities which results in negative health outcomes. This becomes a cyclical problem in which poor health increases the need for medical care which is not adequately accessible resulting in poorer health of the individual. Data from the 2002-03 Medical Expenditure Panel Survey (MEPS), a health services and expenditure survey cosponsored by the Agency for Healthcare Research and Quality and the National Center of Health Statistics, is used in this analysis. The Physical Component Summary score (PCS) of the SF-12 version 2 and a perceived health variable will be used to measure perceived health status. Population Studied: The sample frame is drawn from the National Health Interview Survey representing a sample from the general civilian non-institutionalized public with oversampling of Hispanic, African American, Asians, and family’s expected to have incomes below 200% of the poverty line. The household component of the MEPS is used in this analysis, and 20,661 adult United States residents completed the SF12v2 and perceived health items. Principal Findings: Reliability and validity for the health status measures, SF12v2 and perceived health status, within this database has been conducted and adequate reliability and validity within the 2003 MEPS was observed. This testing was important because this is the first time in which the SF-12v2 has been used within the MEPS. The PCS was shown to have perfect dose-response change when the mean group scores were compared using participants’ chronic condition status. Initial analysis of the research question supports the hypothesis that health insurance status has an effect on the perceived health status of persons with one or more chronic condition. Conclusions: Previous research has shown that for the average person, the presence or absence of health insurance has a negligible affect on their health. Unfortunately, a growing proportion of our society does not have the luxury of good health. These health problems are often caused by serious health conditions. Improperly treated these conditions will increase in severity. Likewise the cost to treat them will increase exponentially. Governmental policy must be implemented to deter this costly process. Implications for Policy, Delivery, or Practice: The physical problems that plague the chronically ill are extremely burdensome, those that are both chronically uninsured and chronically ill have their health problems exacerbated by the lack of insurance and the simultaneous nature of their disease. Few studies analyze characteristics that makeup this group and how these characteristics affect their long-term health. This study will provide a new view of the chronically-ill and uninsured. Primary Funding Source: No Funding ●Barriers to SCHIP Enrollment and Renewal in an InnerCity Population Julia Costich, J.D., Ph.D., Priyanka Srinivasan Presented By: Julia Costich, J.D., Ph.D., Chair, Dept. of Health Services Management, College of Public Health, 121 Washington Ave, Lexington, KY 40536-0003; Tel: 859-257-6712; Fax: 859-257-3909; Email: julia.costich@uky.edu Research Objective: To identify and classify barriers encountered by low income parents residing in the 16-largest US metropolitan area in obtaining or retaining health insurance for their children. Study Design: A brief survey was distributed by front desk personnel to a total of 202 parents of uninsured children at community health centers. Response options included a checklist generated from interviews with key informants and an open text field. Responses were tallied and classified as reflecting system or knowledge barriers; open text responses were recorded verbatim. Geographically sensitive responses were further classified by zip code. Population Studied: Low-income inner city families. Principal Findings: Initial SCHIP or Medicaid enrollment required a face-to-face interview, while coverage renewal paperwork is sent by the responsible state agency and can be returned by mail. Nearly 20% of parents stated that they did not receive renewal information in the mail as required by state policy and procedure. About 10% were unaware that their children had lost coverage until they registered at the health center, and 11% could not provide the income documents necessary for renewal. When asked what could be done to make the process easier, 32% suggested sending the form earlier so that they would have more time to work on it, 34% wanted information about the process beforehand so that they know what to expect, and 41% wanted to make it easier to call a caseworker if they had questions during the process. Conclusions: The return to in-person interviews for initial eligibility determination unmasked problems with the mailed renewal process that had gone undetected because they were so easily remedied when applications could be mailed. Barriers to continuity in SCHIP/Medicaid enrollment in this metropolitan area are mainly systemic: parents have difficulty because they do not receive the necessary information by mail in a timely manner or lack access to assistance with documentation. Implications for Policy, Delivery, or Practice: Systemic barriers must be eliminated or mitigated to assure children’s continuity of coverage. Primary Funding Source: Primary Care Association ●Expresslane Enrollment in California Michael R Cousineau, Dr.P.H., Eriko Wada, M.P.P. Presented By: Michael R Cousineau, Dr.P.H., Director, Keck School of Medicine, Community Health, 1000 Fremont Dr, Alhambra, CA 91803; Tel: 626 457-4010; Fax: 626 457-5858; Email: cousinea@usc.edu Research Objective: To assess the effects of Expresslane Enrollment on Medi-Cal enrollment and utilization of MediCal-suported services. Study Design: Comparative case studies of numbers and percent of eligible children enrolled into full Medi-Cal as a result of Expresslane Eligibility in six counties in California implementing ELE for Medi-Cal enrollment over a three year period 2003/04 to 2005/06 school years. Overall trends were measured as were comparisons between outcomes among the six participting counties. County-specific data (including size, percent uninsured, presence of a Healthy Kids product,) were used to explain differences in outcomes across piolot sites. Measuring changes in school lunch participation in periods before and after ELE, and betwen ELE sites and non ELE sites were used to determine the effects of ELE on the school lunch program. Population Studied: Uninsured children whose parents sign consent and who are express enrolled into Medi-Cal; participating schools and school districts Principal Findings: In the pilot for year 2, over 3000 potentially eligible children were identified (school lunch eligible). Of these 13% had signed consent authorizing schools to forward application to Medi-Cal. Of these nearly half were already enrolled in Medi-Cal or Healthy Families, and 41% were express enrolled into temporary Medi-Cal;16% were ineligible. Upon further screening, about 400 children about one third of those express enrolled were enrolled into ongoing, regular Medi-Cal. Significant differences in the number and percent of eligible children enrolled into Medi-Cal as a percent of eligible children are observed when comparing the various pilot sites. High Medi-Cal participation rates in pilot sites, parents fear of public charge and immigration concerns, and additional Medi-Cal administrative burdens, and a lack of support staff to assist parents were associated with low participation rates.No change in school lunch participation is noted when comparing schools before and after ELE. Conclusions: The ELE program is not an effective as a standalone enrollment strategy especially in school districts with already high Medi-Cal penetration. Still, there appears little if any danger in decreasing school lunch participation as a result of the Express lane Eligibility program. Administrative obstacles in the enrollment and approval process leading to enrollment, parents’ fear of public charge limit the effectiveness of the program in improving enrollment. Implications for Policy, Delivery, or Practice: Linking public insurance program enrollment to existing benefits program is best deployed as one of many enrollment strategies. In order to improve success, administrative obstacles should be removed, and parents fears of public charge alleviated. Moreover, Expresslane helps to expose (but does not reduce) the overall complexity and fragmentation in the health insurance enrollment system for low income children and families. Primary Funding Source: The California Endowment, Blue Shield Foundation ●Implementation of Project ACCESS for the Uninsured in Dallas, Texas was Associated with Overall Improvement of Access to the Emergency Department. Dan Culica, M.D., Ph.D., Barbara Foster, Ph.D., Nora Gimpel, M.D., James Walton, DO, Mark DeHaven, Ph.D. Presented By: Dan Culica, M.D., Ph.D., Assistant Professor, Management, Policy and Community Health, UTSPH, 5323 Harry Hines Blvd, V8.112N, Dallas, TX 75390; Tel: 214-6481070; Fax: 214-648-1081; Email: dan.culica@utsouthwestern.edu Research Objective: One of the main hypotheses of this project was that utilization of emergency department in a large inner-city hospital will be lower after the implementation of the Project ACCESS and the health status and quality of life of the uninsured patients enrolled in this model of care will improve. Study Design: A case–control design was used to enroll patients over 13 months (April 2003-July 2004) at the emergency department of a large metropolitan hospital in Dallas. Overall utilization of the emergency department by the patients enrolled in Project ACCESS and their control group, and the average length of stay for hospital admissions of the same patients were measured during the enrollment period and the 12 month follow-up. Population Studied: Six hundred and twenty three patients participated in the study, of which 270 were enrolled for the Project ACCESS intervention while 353 represented the control group who followed the routine practice. Principal Findings: There was a significant decrease in the number of patients who visited the emergency department in both groups between the enrollment year and the observation period. There was a significant greater average length of stay among the patients enrolled in the Project ACCESS compared with the control patients during the enrollment year. There was a significant lower length of stay among patients enrolled in the Project ACCESS and those in the control group in the observation year that followed the intervention. Conclusions: These results provide evidence of the overall benefits produced by a program of medical care designed and implemented at local level to assist a population sector with poor access to health care in a large metropolitan area. Implications for Policy, Delivery, or Practice: Similar interventions developed at community level to assist inner-city population that faces multiple barriers to medical care due to low socio-economic status create the premises of a complementary care system that not only improves the health status of the targeted population but also alleviates the burden on the entire safety-net system. Primary Funding Source: CDC ●Extending the Measurement of Bias in Uninsured Estimates Gestur Davidson, Ph.D., Gestur Davidson, Ph.D. Presented By: Gestur Davidson, Ph.D., Senior Research Associate, Div. of Health Services Research & Policy, Univ. of Minnesota, 2221 Univ. Ave. S.E. Suite 345, Minneapolis, MN 55414; Tel: 612-625-2339; Email: david064@umn.edu Research Objective: To estimate the possible downward bias in estimated rates of uninsured in statewide surveys. Study Design: We use a random sample of knonwn enrollees of public programs in the state of PA and survey them and their family memebers as if they had been selected in a RDD statewide survey of coverage. We focus on the members of these families--which contain 1+ members on public programs--who are not themselves on any public program. Population Studied: Among these family members not on a public program, we focus on those who self-report being on a public program when they are not enrolled, i.e. the False Positives (FP). And among these FP, we focus only on those who report no other types of coverage,i.e. (FP(N). Principal Findings: We provide various estimated lower bounds for this number of FP(N)s who would be counted as uninsured were they in a general survey of households "but for" falsely reporting public coverage. The reasonable range of these estimates of downward bias in estimated rates of uninsured is larger than the upward bias contributed by public program enrollees who fail to report their coverage and report no other coverage, the so-called False Negatives of the Medicaid Undercount. Conclusions: Considering only the combined impacts of people on public programs and other family members in their households, we conclude it’s likely that there is a net downward bias in the estimated percent of the population without any insurance in the Pennsylvania health insurance survey. Implications for Policy, Delivery, or Practice: This analysis begins to identify and carefully measure alternate sources of bias in surveys of the uninsured, and by so doing it corrects the presumption that the Medicaid undercount provides the only or the largest component of bias in the estimated rate of uninsured. Primary Funding Source: RWJF ●Geographic Disparity in Health Insurance Coverage: Inner Cities versus Outer Metropolitan Areas Derek DeLia, Ph.D., Dina Belloff, MA Presented By: Derek DeLia, Ph.D., Assistant Professor/Senior Research Analyst, , Rutgers Center for State Health Policy, 55 Commercial Avenue, New Brunswick, NJ 08901-1340; Tel: (732)932-4671; Fax: (732)932-4695; Email: ddelia@ifh.rutgers.edu Research Objective: To measure the disparity in health insurance coverage between inner cities and outer metropolitan areas (OMA’s) and document factors affecting the disparity. Study Design: The New Jersey Family Health Survey (NJFHS) is used to measure health insurance coverage and socioeconomic characteristics of NJ residents. Population size and density criteria are used to distinguish inner city and OMA’s. Regression decomposition is used to determine the most important factors behind the geographic coverage disparity (GCD) defined as the uninsured rate in inner cities minus the rate in OMA’s. Population Studied: Children (0-18) and non-elderly adults (19-64) living in NJ in 2001. Principal Findings: The GCD for children was 11.9 percentage points (19.7% minus 7.8%). More than half of this disparity is explained by lower family income among inner city children. Two-fifths is explained by the lower likelihood of inner city children to live with both parents. Although other factors such as immigration and race/ethnicity also play a role, the GCD for children is actually smaller than one would expect after adjusting for geographic differences in socioeconomic variables. The GCD for non-elderly adults was 19.4 percentage points (32.5% minus 13.1%). Two-fifths of this disparity is explained by lower family income in inner cities. One-fifth is explained by a higher concentration of non-citizen adults in inner cities. A significant portion (28%) of the GCD for adults remains unexplained after accounting for geographic differences in socioeconomic variables. All findings are robust to alternative criteria used to distinguish inner cities from OMA’s. Conclusions: There is a large geographic disparity in health insurance coverage for both children and non-elderly adults, which is driven primarily by lower income among inner city residents. The disparity for children may be somewhat surprising given the availability of coverage under the State Children’s Health Insurance Program (SCHIP). However, the findings suggest that the disparity among children is the result of much greater socioeconomic disadvantage in the inner cities, which may work against SCHIP enrollment efforts. While low income and non-citizen status are important, a great deal of the GCD for adults remains unexplained. Future research should investigate whether this part of the disparity is the result of unmeasured characteristics of adults living in inner cities or characteristics of the inner cities themselves (e.g. availability of safety net facilities providing free care). Implications for Policy, Delivery, or Practice: Previous research on geographic disparities in coverage is based on broad rural/metropolitan classifications. This research shows how these broad classifications can be misleading as there is significant coverage variation within metropolitan areas. The predominance of income as a driver of coverage disparities within metropolitan areas suggests that income-based subsidies will be an essential part of expanding coverage in inner cities. Special consideration of family structure may also be important to maximize the enrollment of inner city children into SCHIP. The concentration of immigrants in inner cities, and their importance in driving the GCD, will continue to present challenges for urban safety net providers as recent immigrants remain ineligible for most public coverage. Primary Funding Source: HRSA ●Uninsurance Among Children Whose Parents are Losing Medicaid Coverage Jennifer DeVoe, M.D., D.Phil., Lisa Krois, M.P.H., Tina Edlund, MS, Jeanene Smith, M.D., M.P.H. Presented By: Jennifer DeVoe, M.D., D.Phil., Postdoctoral Research Fellow, Department of Family Medicine, Oregon Health and Science University, 3181 Sam Jackson Park Rd, Portland, OR 97239; Tel: (503) 494-2826; Fax: (503) 494-2746; Email: devoej@ohsu.edu Research Objective: One way to improve children’s access to health insurance is to expand parental coverage. Less is known, however, about how termination of parental coverage affects children. Recently, thousands of adults lost coverage after Oregon’s Medicaid program (the Oregon Health Plan) implemented tighter administrative rules and benefit reductions. This study was conducted shortly after these cutbacks to examine possible impacts on children’s access to health insurance. Study Design: Cross-sectional analysis. Our collaborative team of researchers created a unique survey instrument to collect statewide primary data from low-income families about their current health insurance coverage and their experience accessing healthcare services for their children. Population Studied: To gather information about children eligible for publicly funded health insurance programs, the sample included all 84,087 Oregon families with children enrolled in the food stamp program at the end of January 2005. This report presents descriptive data from 2,681 completed surveys. Using administrative demographic data, we were able to weight all results back to the entire sample population and control for non-response. Principal Findings: Among the children in Oregon’s food stamp population, 10.9% were uninsured, 73% of children had only public insurance coverage, and 16.1% had private coverage. More than one-quarter (26.3%) of the children in the study went without health insurance coverage at some time during the 12-month period immediately prior to the study. Low-income children who were most likely to be uninsured or have coverage gaps were Hispanic; were teenagers over age 14; were in families at the higher end of the income threshold; had an employed parent; or had a parent who was uninsured. Nearly 80% of uninsured children had a parent without health insurance, compared with less than 20% of privately-insured children (p<0.0001). Fifty percent of uninsured children lived in a household with at least one adult who had recently lost coverage under the Oregon Health Plan, compared with only 40% of children with health insurance (p<0.0001). Similarly, over 51% of children with a recent gap in insurance coverage had an adult in the household who lost Oregon Health Plan, compared with only 38% of children without coverage gaps (p<0.0001). Conclusions: Over 10% of children in Oregon eligible for publicly funded health insurance programs were uninsured, and over 25% of these children had gaps in insurance coverage during a 12-month period. Four out of five uninsured children had an uninsured parent, compared with just one out of five insured children. Uninsured children and those with recent coverage gaps were more likely to have parents who lost Oregon Health Plan coverage after recent cutbacks. Implications for Policy, Delivery, or Practice: Although current fiscal constraints prevent many states from expanding public health insurance coverage to more parents, states need to be aware of the impact on children when parents lose coverage. It is critical to develop strategies to keep parents informed regarding continued eligibility and benefits for their children and to reduce administrative barriers to children’s enrollment and retention in public health insurance programs. Primary Funding Source: HRSA ●A Tool based on a Model of Personal and Ecologic Characteristics to Explain Deprivation in US Healthcare access. Martey Dodoo, Ph.D., Robert Philips, Jr,, M.D., M.S.P.H. Presented By: Martey Dodoo, Ph.D., Senior Economist, The Robert Graham Center, 1350 Connecticut Ave, N.W. Ste 201, Washington, DC 20036; Tel: 202-331-3360; Fax: 202-331-3374; Email: mdodoo@aafp.org Research Objective: To develop a tool for identifying small area communities and/or populations at risk of health care access deprivation in the US. Study Design: There have been many attempts to use personal characteristics to explain barriers or delays in accessing healthcare appropriately. There have also been attempts to use ecological measures like poverty as a nearproxy for health access problems. In this study we combined these efforts to develop indices of health access deprivation, or risks of delaying or not getting healthcare in the US. We used the National Health Interview Survey (NHIS) to identify individual level predictors of health care access deprivation, and constructed indices of access deprivation using logistic and principal component analyses methods. We geo-coded the individual level data that included the access deprivation indices, and merged them with 2000 US Census data at block group level. Using multi-level logistic modeling techniques we undertook pair wise substitution of Census for NHIS variables that allowed us to determine the ecologic variables that best predict reported health access problems in the US. We mapped the small area deprivation indices and showed how they can be overlaid with health care provider availability and other statistics. Population Studied: 2000 US Census data at block group level and a 2002 NHIS national probability sample of 31,044 adults. Principal Findings: We identified 4 strong predictors of health care access deprivation and 6 other predictors, out of 23 potential predictors from the literature. We found that someone is more likely to be deprived of care, if they are female, do not own a home, live in a single adult household, have a functional limitation, smoke, have no health insurance or are less than 65 years old. Conclusions: Attempts to use individual level data to model barriers or delays in accessing health care can be combined with the use of ecologic characteristics to determine health care access problems, to identify small areas or populations at risk of health care access deprivation Implications for Policy, Delivery, or Practice: This study has provided a tool that may be used as a basis for local community assessments, health planning, and to identify specific communities and/or populations at risk of having health access deprivation problems or issues. Primary Funding Source: No Funding ●Michigan Medicaid’s Healthy Kids Dental program: Assessment of the first 4 years Tracy Finlayson, Ph.D., Stephen A. Eklund, D.D.S., M.H.S.A., Dr.P.H. Presented By: Tracy Finlayson, Ph.D., AHRQ Postdoctoral Scholar, School of Public Health, UC Berkeley, 140 E Warren MC7360, Berkeley, CA 94720-7360; Tel: 510-642-5652; Fax: 510643-4281; Email: tracyf@berkeley.edu Research Objective: In 2000, Michigan’s Medicaid dental program initiated Healthy Kids Dental (HKD), a demonstration program offering dental coverage to Medicaidenrolled children in 37 of its 83 counties. HKD is administered through the Delta Dental Plan of Michigan (DDPM), using DDPM-affiliated dentists who are reimbursed at usual customary rates. This study evaluates trends in children’s access to dental care, treatment patterns, dentist participation, and cost in the first four years of HKD. Study Design: DDPM dental claims were obtained for children in the 37 HKD counties and enrollment and utilization data were analyzed. HKD was originally implemented in 22 counties on May 1, 2000, and expanded to another 15 counties on October 1, 2000. All trends were examined for 12-month periods for the 22 and 15 county groups separately over time between 2000 and 2004. The proportion of children by age enrolled and utilizing dental services in each 12-month period reflected trends in access to care. Differences in utilization patterns by length of enrollment were also investigated. Children with two or more cleanings or topical fluoride treatments, two common preventive procedures, were examined to explore whether or not they likely had a dental home and went to the same provider for all of these services. The total number of dentists participating in HKD was also counted, and the proportion of in-county providers tabulated. The average travel distance between children and providers was approximated using a GPS methodology to calculate distance between ZIP code centroids for each visit. Total costs and the distribution of payment for different select major procedure groupings (diagnostic/preventative, restorative, endodontics, and surgical) were also calculated. Population Studied: HKD enrolled children aged 0 to 20 in Michigan’s 37 HKD counties were included in this study. Principal Findings: In the first year, children’s backlog of dental needs appeared to be treated. Since the beginning of the program, access to care has continued to improve and more children are receiving dental services each year. The largest increase in utilization was observed between the first and second years of the program. Children have a high probability of having a dental visit if enrolled, and the proportion of continuously enrolled children has been growing. The number of dentists providing care has been increasing and children continue to receive needed services from local providers close to home (average travel distance was consistently 13-15 miles). At least 40% of HKD children likely have a dental home and more children appear to be entering regular recall patterns. Costs have increased each year, attributable to more children receiving care and a customary rate of reimbursement for services. Conclusions: HKD has been successful in increasing access to dental care for the child Medicaid population in the demonstration counties. The program has grown between 2000 and 2004; more children have enrolled each year, and more local dentists are participating in the program and integrating these children into their practices and providing comprehensive treatment. Implications for Policy, Delivery, or Practice: HKD can serve as a template for other states reforming their Medicaid program’s dental component. Primary Funding Source: NIH ●Step by Step: Local Coverage Expansion Initiative 2005 Evaluation Findings Annette Gardner, Ph.D., M.P.H., Patricia Mintz, M.B.A. Presented By: Annette Gardner, Ph.D., M.P.H., Academic Specialist, Institute for Health Policy Studies, University of California, San Francisco, 3333 California Street, Suite 265, San Francisco, CA 94143; Tel: (415)514-1543; Fax: (415)476-0705; Email: annette.gardner@ucsf.edu Research Objective: In 2005, the California HealthCare Foundation (CHCF) funded Year 2 of Step by Step: Local Coverage Expansion Initiative to increase the capacity of local organizations to expand insurance coverage for low-income, uninsured Californians. The goals of the program are to: 1) Foster development and implementation of local health insurance programs; 2) Encourage efforts to increase enrollment in local and state public insurance programs; and, 3) Increase the number of insured Californians. Evaluation research objectives focused on achievement of the broader goals or outcomes resulting from the Step by Step program, specifically: increased grantee and stakeholder knowledge of the scope of the problem and potential solutions; development and maintenance of critical partnerships; the ability to identify and secure new funding; actual and anticipated enrollment gains for children and adults in new and existing programs; and changes in grantee capacity to design and launch an insurance program. Study Design: The study design for the evaluation of the Step by Step Program focused on the assessment of individual achievement of grantee objectives using grantee mid-point and final reports and assessment of overall program achievement of key short-term outcomes using end-point interviews. Additionally, grantees completed a pre/post survey on specific skills in planning and implementing an insurance coverage initiative. Last, grantees provided information on 2005 actual and 2006 enrollment in existing and new insurance programs, as well as 2005 actual and 2006 estimated funding secured that might be due to their Step by Step participation. Population Studied: Thirteen grantees representing local coalitions, agencies, local governments and health plans were awarded grants to plan or implement a coverage expansion in 2005. Of that group, six were awarded one-year planning grants: four for projects focused on expanding coverage for children; two for projects focused on expanding coverage for adults. The remaining seven grantees focused on key implementation activities: six for projects focused on launching a specific insurance program for children; one for a coverage program focused on adults. Grantees were diverse in geographic focus, target population and the lead agency. Principal Findings: The findings on individual and collective grantee achievements point to the continued success of the Step by Step program in expanding coverage for children and adults under diverse (and often adverse) conditions. Planning grantees successfully positioned themselves to move forward with implementation of their designs, e.g., 3 grantees will be funded for implementation during Year 3 of the Step by Step program. Implementation grantees, particularly those focusing on children, have or will shortly launch their respective coverage programs, achieving significant enrollment gains in public and private insurance programs. Adult grantees hold great potential for new models of funding and partnerships. However, it hasn’t been easy. Grantees of all types have had to contend with a variety of operational and external challenges while weighing their choices. Conclusions: The findings from Years 1 and 2 of the Step by Step program point to the importance of providing technical assistance to localities to reduce the barriers to care for the uninsured. Similar to 2004, significant uncertainty persists in terms of local funding and the possibility of statewide coverage. However, the apparent capacity and commitment to undertake local coverage expansions, coupled with partnerships with the state and significant commitment from private funders, bodes well for local coverage expansions in the near future. Implications for Policy, Delivery, or Practice: The implications of these local coverage expansions for stakeholders interested in expanding coverage are two fold: 1) achieving statewide coverage based on these local efforts, a goal sought by many stakeholders and advocates, presents some challenges, such as recognizing differences in target populations, provider capacity and political commitment; and 2) the insurance and funding gains, while significant at the local level, may collectively fall short of the larger demands for coverage. Similarly, the implication for funders and others desiring to further local coverage expansions is to proceed cautiously with a “one size fits all” approach, particularly for adult expansions. Additionally, grantees benefit greatly from a “hands on” grant program that holds them accountable to their respective objectives and provides assistance as the need arises. Primary Funding Source: California HealthCare Foundation ●Expanding Policy Advocacy Capacity: Results from the Clinic Consortia Policy and Advocacy Program Evaluation: 2001-2004 Annette Gardner, Ph.D., M.P.H., Claire Brindis, Dr.P.H., Sara Geierstanger, M.P.H., Coline McConnel, MA, Melissa MartinMollard, M.S.W., Nadine Chabrier, BA Presented By: Annette Gardner, Ph.D., M.P.H., Academic Specialist, Institute for Health Policy Studies, University of Californa, San Francisco, 3333 California Street, Suite 265, San Francisco, CA 94143; Tel: (415)514-1543; Fax: (415)476-0705; Email: annette.gardner@ucsf.edu Research Objective: In 2001, The California Endowment awarded more than $9 million to fifteen California regional community clinic associations and four statewide clinic organizations to strengthen the role and capacity of consortia to support the management, leadership development, policy and systems integration needs of community clinics. The three year grant supported specific activities related to policy advocacy or activities that mobilize resources to support a policy issue or create a shift in public opinion, technical assistance, media advocacy, as well as shared services. Eighteen grantees were refunded for three years in 2004 to undertake or continue a similar set of activities. A unique funding opportunity directed to a relatively unknown grantee population, our research objectives were to assess the effectiveness of key policy advocacy activities, such as educating policymakers and media advocacy and the achievement of short-term outcomes, such as increased funding to California clinics and their target populations and policy outcomes to increase access to care for low-income populations. Study Design: We are using a case-study approach to identify differences and similarities among grantees and longitdudinal approach to assess achievement of program outcomes in the aggregate. Using a logic model framework, we administered a combination of quantitative and qualitative instruments to the 19 grantees to measure grantee processes and outcomes from 2001-2004. The UCSF Team collected longitudinal data from grantees on cross-site indicators, such as type, number and effectiveness of policy advocacy activities and funding to clinics and consortia, and consortia partnerships. Second, we conducted interviews and focus groups with the targets and beneficiaries of grant-funded activities. To assess changes in policymaker awareness of the grantees and clinics as well as their activities, events, and issues, we administered a 15minute Stakeholder Awareness Survey to over 70 policymakers and community leaders in 2003. In 2004, the UCSF Team readministered the Stakeholder Awareness Survey to 43 state and local policymakers and community leader perceived to be less familiar with consortia and/or clinics but were nonetheless considered important stakeholders in California’s health care safety net. To assess consortia capacity in engaging in media advocacy, the UCSF Team interviewed 17 media representatives and 6 media consultants in 2004. Last, to assess the benefits of grant-funded activities to clinics and their target populations, we interviewed up to 3 member clinics per grantee (for a total of 47 clinics) in 2002 and conducted a focus groups with 11 clinic Executive Directors in 2004. All data was aggregated, tabulated and crosscutting themes were identified. Population Studied: Clinic consortia are membership organizations for primary care clinics and related safety net providers. (There are approximately 768 primary care clinics in California that serve over three million low-income and uninsured Californians.) California is unique in that it has 19 community clinic consortia, with the oldest dating back to 1969 and the most recent launched in 2001. Consortia are flexible organizations that help individual clinics meet the needs of their clinic populations, providing a unified voice in the policymaking arena and offering economies of scale for business and program shared services. Principal Findings: The results of our assessment of key Program outcomes, include:1) Increased policy advocacy on behalf of clinics and their target populations. For example, the precent of policy advocacy activities undertaken by grantees increased from 73% in 2002 to 84% in 2004; 2) Increased clinic financial stability. For example, 17 grantees secured $248 million on behalf of clinics and consortia from 2001 to 2003 (90% went to clinics); 3) Increased policymaker awareness and support. For example, respondents from the two survey groups were familiar and supportive of clinic policy issues (72%); 4) Policy “wins” benefit clinics and their target populations. For example, grantees were involved with 58 pieces of legislations at the federal and state level from 2001 to 2003. of these, 15 (30%) federal policies were passed and 26 (60%) state bills were passed; 5) Member clinics appreciate consortia contributions. For example, grantee policy advocacy viewed as increasingly indispensible, increasing ability of clinics to influence policy; 6) Consortia develop new partnerships. For example, grantees developed and nurtured diverse partnerships, such as partnerships among member clinics, partnerships between member clinics and non-clinic organizations, and partnerships among grantees; 7) Strengthened clinic operations. For example, clinics and grantees report cost savings and revenue generation. The findings indicate that clinic consortia are achieving many Program outcomes and they are an increasingly important and credible voice on behalf of clinics and the populations they serve. Additionally, these activities contribute to longer term outcomes of increased access to care and improved client health status. Conclusions: The Clinic Consortia Policy and Advocacy Program represesents a unique funding opportunity, supporting policy advocacy activities that would not be funded otherwise. The results are encouraging at two levels: funding for policy advocacy activities can result in benefits to clinics and their target population and the Program played a critical role in expanding clinic consortia capacity and contributing to the sustainability of these activities. These achievements are remarkable given the many challenges that grantees have faced in California, particularly decreased funding to counties and the state budget shortfall. In all likelihood, the threats to community clinic expansions will persist, including potential cuts to the Medi-Cal program that could trigger significant revenue losses to clinics. Also, some of the program activities (such as quality improvement and media advocacy) may be more difficult to implement fully because clinic needs and resources vary considerably, and these resources may be more constrained in the future. However, there are policy opportunities in the political arena at the federal, state and local levels, as well as opportunities for funding diversification through new partnerships and programmatic expansions. Implications for Policy, Delivery, or Practice: The implications for policymakers and associations (or networks) are significant. Increasingly, funders are supporting policy advocacy activities undertaken by diverse grantees. Policy advocacy activities are proving useful for expanding local, state and federal support for key policy issues. For example, UCSF evaluation findings indicate that state-level clinic grant programs were expanded or maintained by consortia in spite of a weakening economy and state budget. In an era of decentralized decision-making and increased grassroots responsibility for addressing the needs of California’s vulnerable populations, clinic consortia have great potential for mobilizing resources and participating in policymaking in ways that are beyond the means of many individual clinics. Second, this funding approach and the activities undertaken by grantees are applicable to other settings and membership organizations. The results—expanded policy advocacy— speaks to the opportunities for increased representation in the policymaking process. For researchers, the evaluation provides some important lessons and tools in conducting a multi-site evaluation and developing applicable outcomes and measures. Primary Funding Source: The California Endowment ●Welfare Reform, Changes in Insurance Coverage, and the Timing of Prenatal Care Initiation Norma Gavin, Ph.D., E. Kathleen Adams, Ph.D., Willard G. Manning, Ph.D., Cheryl Raskind-Hood, Ph.D., Matthew Urato, MA Presented By: Norma Gavin, Ph.D., Senior Research Economist, , RTI International, 3040 Cornwallis Road, Research Triangle Park, NC 27709-2194; Tel: (919) 541-6432; Fax: (919) 990-8454; Email: gavin@rti.org Research Objective: Welfare reform had the unintended effect of decreasing health insurance coverage among lowincome women of child-bearing age. Thus, in the postreform era, more low-income women may be uninsured prior to pregnancy, enroll in Medicaid later in their pregnancies, and delay initiating prenatal care (PNC). Welfare reform also restricted welfare tenure and imposed work requirements and sanctions for non-compliance, thereby potentially increasing income and time barriers to PNC. This study investigates the impact of welfare reform on prepregnancy insurance coverage and first trimester PNC initiation among pregnant women financially eligible for Medicaid. Study Design: We used data for 1996 through 1999 from the Pregnancy Risk Assessment Monitoring system for eight states (Alaska, Florida, Maine, New York, Oklahoma, South Carolina, Washington, and West Virginia). We tested the endogenity of Medicaid and private insurance coverage in the model for first trimester PNC initiation and estimated a twopart logistic model of prepregnancy insurance coverage and first trimester PNC initiation. The impact of welfare reform was measured through the coefficients of prepregnancy insurance coverage and an interaction term for the postreform period and welfare-related eligibility. We compared the estimates from the two-part logistic model with a simple logistic model, ordinary least squares, and two-stage least squares models of first trimester PNC initiation. Population Studied: We defined the target group affected by welfare reform as welfare-related Medicaid eligible women and identified them as single women with children and family incomes equal to or less than the state’s prereform welfare level. We defined the comparison group as expansion-related Medicaid eligible women, which we identified as all other women with family incomes under 185% of the federal poverty level. Principal Findings: We found welfare reform to have a significant negative impact on prepregnancy Medicaid coverage but no significant impact on private insurance coverage prepregnancy. Private insurance coverage prior to pregnancy was endogenous to the equation for first trimester PNC initiation but Medicaid coverage prepregnancy was exogenous. Estimates of the effect of a reduction in Medicaid coverage on PNC initiation was consistent over the single and two-stage models, showing a reduction of approximately 10 percentage points in the percentage of welfare-related Medicaid eligible women initiating PNC in their first trimester. We found no evidence of an effect of welfare reform on first trimester PNC coverage beyond the effect on prepregnancy coverage. Conclusions: Welfare reform had an adverse effect on the attainment of the Healthy People goal of 90% of all pregnant women initiating PNC in the first trimester. Implications for Policy, Delivery, or Practice: Offering lowincome women Medicaid coverage once they become pregnant is not sufficient to attain national goals for PNC use and thereby reduce maternal and infant morbidity and mortality. Better coverage of women prior to pregnancy and better outreach to eligible women are needed. Primary Funding Source: RWJF ●Patient Perceptions of a Community-Based Care Coordination System Nora Gimpel, M.D., Alice Marcee, DVM, MS Presented By: Nora Gimpel, M.D., Assistant Professor, Family and Community Medicine, UTSouthwestern Medical Center, 6263 Harry Hines Blvd, Dallas, TX 75390; Tel: 214 648 0768; Fax: 214 648 1307; Email: nora.gimpel@utsouthwestern.edu Research Objective: The efficient management of a community’s health needs is the role of a community health worker (CHW). This role for a community member to act as a case manager is common in developing countries. CHWs serve as the link between a healthcare system and an individual’s needs. They contribute to the continuity and quality of healthcare for individuals within a community. This study is a qualitative evaluation of the impact of CHWs in a case management model designed to provide access to healthcare services to uninsured low-income patients. It evaluates patient perceptions of the effect of CHWs on their lives and on their ability to independently manage their health needs. Study Design: Focus groups were designed to evaluate three areas: patient empowerment, health education, and the healthcare system. The community-based program selected for this study is Project Access (PA). Three different samples of patients were selected from the primary components of the program: diabetes, depression and social services. Population Studied: Ninety-five adults receiving services through the program were invited to participate in the focus groups. Seventy three percent were contacted and 35.7 % (n=25) attended the focus group sessions. Principal Findings: Interaction with CHWs proved to be an invaluable asset in learning how to navigate the health care system and in obtaining appointments. In terms of empowerment, participants considered the assistance they received from the CHWs allowed them to better care for themselves. In addition to the emotional and psychological support, patients received very practical help such as assistance with transportation and language translation. However, patients with diabetes stated that receiving more specific education on self management would improve their capability in managing their health problems. Whereas some patients in the depression group had trouble coming into contact with their CHW, others had no problems and maintained good communication with their CHW. In exploring expectations, all participants were most grateful for the services they received from PA. They expressed the feeling that through the CHWs, PA provides medical health services as well as support for problems not directly related to their physical health. Conclusions: This study suggests that CHWs in a case management model results in improvement in patient comprehension of health issues, in patient navigation through a health care system, and in the ability of patients to independently manage health issues. Implications for Policy, Delivery, or Practice: Efforts to provide healthcare services to uninsured, low-income populations, not otherwise eligible for governmental assistance, should be coordinated through CHWs serving as case managers. It is also suggested from the work that patient input should be routinely assessed as both a way to evaluate program services and optimize patient empowerment. It is widely accepted that health care resources are limited, and this especially pertains to community efforts. CHWs build relationships between patients, healthcare providers, and community members which improve the effectiveness of health services delivered. Providing health services to underserved populations must include individual members of the population as stakeholders for the efficient delivery of limited healthcare resources. Primary Funding Source: CDC ●The Effect of Care Seeking on Transitions to Medicaid or SCHIP Mary Harrington, M.P.P., Ph.D. candidate Presented By: Mary Harrington, M.P.P., Ph.D. candidate, Deputy Director, Economic Research Initiative on the Uninsured, 555 South Forest, Ann Arbor, MI 48104; Tel: (734) 936-1302; Email: mehn@umich.edu Research Objective: To determine whether uninsured children eligible for Medicaid or SCHIP should be considered insured because they will become enrolled when they seek care. Study Design: Multiple panels of linked child-parent data from MEPS, 1998-2002 are used to model the effect of presenting for care on transitions from uninsured to insured among low-income children. Because comparisons of children who do and do not present for care is problematic due to the potentially endogenous relationship between enrollment and care seeking, the analysis uses an instrumental variables approach. An accident indicator variable is used to instrument for presenting for care to identify the effect of care on transitions from uninsured to insured. Comparisons are made between Medicaid or SCHIP eligible but uninsured children who do and do not have accidents during the study period. The analysis controls for a variety of other factors known to influence enrollment, including changes in parental employment and marital status, and various demographic indicators. Population Studied: The analysis uses monthly data on 4 cohorts of uninsured eligible-but-unenrolled children (roughly 700 children per cohort). Demographic, coverage, condition and selected utilization data for each child are linked with parent demographic and employment-related information. Principal Findings: Preliminary results indicate that care seeking does not have a significant effect on transitions from uninsured to insured, nor more specifically on transitions from uninsured to Medicaid or SCHIP, among eligible-butunenrolled low income children. Conclusions: Some have argued that uninsured children who are eligible for public programs should not be counted in the pool of the uninsured because they can apply for coverage and become enrolled when they present for care. This argument is made in particular about children who are eligible for traditional Medicaid because that coverage extends retroactively to the three month period prior to when an application is made. Others argue that we should still be concerned about eligible-but-unenrolled children because they are less likely to seek out care, especially care that is more discretionary. Findings from this study suggest that many uninsured children who are eligible for public coverage are not becoming insured when they present for care, leaving them vulnerable to negative effects of lacking coverage similar to other uninsured groups. Prior research has shown that children with some form of coverage are more likely to utilize care, particularly preventive care, have fewer access problems, and have fewer unmet health care needs. Implications for Policy, Delivery, or Practice: Policies to expand coverage for the uninsured should not ignore the needs of the substantial number of low income children who are uninsured despite being eligible for public coverage. Outreach and enrollment assistance efforts to increase the take up of public coverage among eligible families should be strengthened. Primary Funding Source: RWJF ●Toward Universal Child Coverage in Los Angeles County-Early Findings from the Healthy Kids Program Evaluation Ian Hill, M.P.A., M.S.W., Brigette Courtot, BA, Anna Sommers, Ph.D., Josh McFeeters, M.P.P., Embry Howell, Ph.D. Presented By: Ian Hill, M.P.A., M.S.W., Principal Research Associate, Health Policy Center, The Urban Institute, 2100 M Street, NW, Washington, DC 20037; Tel: 202/261-5374; Fax: 202/223-1149; Email: ihill@ui.urban.org Research Objective: Numerous counties in California have developed Children’s Health Initiatives (CHI) with the intent of providing universal health coverage to children. The Healthy Kids Program in Los Angeles County, funded primarily by tobacco tax revenues and philanthropic donations, is by far the largest CHI in the state, targeting children in families living below 300% of poverty who are ineligible for Medi-Cal and Healthy Families (California’s SCHIP). This four-year evaluation is assessing the implementation and impacts of the program on children’s coverage, access to care, and health status. Study Design: Begun in May 2004, the evaluation comprises multiple qualitative and quantitative components, including: case studies of implementation; focus groups with parents; monitoring of administrative outreach, enrollment, and utilization data; analyses of the effects of Healthy Kids on uninsurance and Medi-Cal and Healthy Families enrollment; and longitudinal household surveys of enrollees to assess impacts on children’s access to and use of care, and health status. Population Studied: Children in the Los Angeles Healthy Kids Program Principal Findings: Because of the program’s eligibility rules, Healthy Kids primarily serves a population of undocumented immigrant children. Since its inception in July 2003, the program has enjoyed the strong support of a broad range of key stakeholders in Los Angeles County. A large number of community-based organizations conduct intensive, culturally appropriate outreach and hands-on enrollment assistance with families in a wide array of settings to enable them to obtain coverage under all available programs. This approach is apparently effective, as it has helped over 45,000 children obtain Healthy Kids coverage in just two years. Case studies and focus groups with parents suggest that the program’s application process is simple, that the benefit package (based on Healthy Families) is meeting the needs of enrolled children, that the managed care delivery system is affording good access to care, and that sliding-scale premiums and copayments are not posing barriers to enrollment or service use. Parents of children with special health care needs have, however, indicated that multiple copayments do create some financial burden. Early program data suggest low utilization rates, however, measures are confounded by the likelihood that enrollees are also obtaining services under Emergency Medi-Cal and the Child Health and Disability Program. Preliminary results from the first household survey will be available in June 2006 and will report on whether Healthy Kids reduced unmet need for health and dental care, increased access to care, and improved utilization. Conclusions: Early implementation of Healthy Kids has been quite successful. Qualitative findings and program administrative data indicate that the program is effectively targeting and serving a population of undocumented immigrant Latino children that have been shown to be particularly vulnerable and at risk. Implications for Policy, Delivery, or Practice: Early evidence suggests that a SCHIP-like program, free of “public charge” stigma, providing broad benefits and utilizing communitybased, culturally-appropriate outreach and health care providers can succeed in extending comprehensive coverage to uninsured, largely undocumented children. Primary Funding Source: First 5 LA and The California Endowment ●A Tale of Two PPOs: Competitive Strategy and Prior Insurance Choices of New Enrollees Richard Hirth, Ph.D., Kyle Grazier, Ph.D., Michael Chernew, Ph.D., I-Cha Lee, MS, Arthur Hong, BS, Edward Okeke, M.D. Presented By: Richard Hirth, Ph.D., Associate Professor, Health Management and Policy, Univ. of Michigan School of Public Health, 109 S. Observatory, Ann Arbor, MI 48109-2029; Tel: (734)936-1306; Fax: (734)764-4338; Email: rhirth@umich.edu Research Objective: Preferred Provider Organizations (PPOs) have become the dominant employer-sponsored plan type, growing from 11% of enrollment in 1988 to 55% in 2004. Initially, PPO enrollment growth coincided with growth of other forms of managed care, coming at the expense of feefor-service (FFS) plans. Commonly it was felt that PPOs were a transition product leading individuals to stronger forms of managed care. However, since 1999 PPO growth has come at the expense of other forms of managed care, calling the transition product interpretation into question. We use a natural experiment at the University of Michigan (UM) to explore the impact of PPO availability on the share of other plan types. Study Design: In 2005, UM offered PPOs for the first time. Two PPOs were offered by vendors who offer other plans to UM employees. Vendor A also offers a FFS plan that, like the national experience, has lost most of its membership in this group over the past decade. Vendor B also offers HMO and point-of-service (POS) plans, the sorts of plans recently losing enrollment to PPOs. We complement plan-level analyses with an employee-level, multinomial logit model of the choice of a particular PPO relative to a non-PPO plan. Population Studied: UM employees at all 3 campuses eligible for benefits in both 2004 and 2005 (n=23,865). Employees were offered 6 plans from 5 vendors in 2004. These plans remained available in 2005 along with the new PPOs. Principal Findings: Overall, 3.1% of enrollees chose Vendor A’s PPO in 2005 and 4.8% chose Vendor B’s. Less than half (45.7%) of A’s PPO enrollment was “cannibalized” from its own FFS offering while the vast majority (84.3%) of B’s PPO enrollment was cannibalized from its other plans (64.4% from POS and 19.9% from HMO). A’s share of UM enrollment in its products rose from 4.3% to 5.2% following introduction of the PPO. B’s share increased from 68.4%to 70.7%. Employee-level analysis revealed that female employees and those previously enrolled in FFS or POS were more likely to choose either PPO. Relative to those choosing a non-PPO plan, enrollees in A’s PPO were older and more likely to cover dependents, have higher compensation, and reside outside the main campus’ county, and be employed outside the health system. B’s PPO enrollees were younger and more likely to reside in Ann Arbor. Conclusions: PPO enrollment drew from both FFS and HMO/POS. Selection of each plan was consistent with a model that the unique provider networks matter. There is no evidence that the plans were simply vehicles to transition people to more managed products. Implications for Policy, Delivery, or Practice: Because PPOs drew substantial enrollment from more managed plans, even after controlling for a range of factors at the micro level, the interpretation that these plans are merely transition products is likely incorrect. Future visions of the health insurance market must recognize the salience of this insurance form and policy makers must recognize that offering PPOs in public programs such as Medicare may not simply transition people to HMOs and may actually have the opposite effect. Primary Funding Source: AHRQ ●Covering more Lives by Improving the Public Safety Net Jennifer Houlihan, M.P.H. Presented By: Jennifer Houlihan, M.P.H., Director of Research and Planning, Research and Planning, Health Care District of Palm Beach County, 324 Datura Street Suite 401, west palm beach, FL 33401; Tel: 561-659-1270; Fax: 561-6594620; Email: jhoulihan@hcdpbc.org Research Objective: To understand the challenges and benefits of designing and implementing a distinct, comprehensive safety-net program outside the realm of the traditional hospital and public health clinic setting. Study Design: Data analyis will be presented that highlights trends in demographics, medical trends, and costs of program membership over a ten year period layered with the impact changes in the Medicaid and Medicare programs have had on respective program members. Case examples. Population Studied: 20,000 + membership pool of safety-net program. Confined to residents of Palm Beach County. Principal Findings: Will present trends in demographics, medical trends, and costs of members to the provider over a ten year period. Conclusions: The issue of health insurance coverage is essentially a local problem subject to demographic and employment trends unique to each community; solutions are also local. Implications for Policy, Delivery, or Practice: Defines the structure of a public safety-net system, including the benefit structure, which meets the needs of the uninsured. Primary Funding Source: No Funding ●A New Uninsured Population: Retirees Under Age 65 at Risk of Losing Health Coverage Paul Hughes-Cromwick, BS, MA, ABD, Charles Roehrig, Ph.D. Presented By: Paul Hughes-Cromwick, BS, MA, ABD, Senior Analyst, Health Solutions Division, Altarum Institute, 3520 Green Court, Ann Arbor, MI 48105; Tel: (734)302-4616; Fax: (734)302-4994; Email: paul.hughes-cromwick@altarum.org Research Objective: Develop a profile of early retirees with insurance coverage granted as a retirement benefit from a previous employer; estimate the subset at risk of losing coverage because of sponsor financial distress; and describe the policy implications if such losses occur. Study Design: We refer to individuals who retire prior to Medicare eligibility, and depend on health insurance from a previous employer, as early retirees. Many large employers, public and private, have accumulated substantial numbers of early retirees and are struggling to meet the resulting financial obligations. Delphi’s bankruptcy exemplifies a threat to retiree security – the potential loss of health insurance. We estimate the number of individuals dependent upon early retiree coverage using the Current Population Survey (CPS), and develop health and health-care related descriptors for this group using the Medical Expenditure Panel Survey (MEPS) and Health and Retirement Survey (HRS). The subset at severe risk of losing coverage is estimated through examination of financial statements using the Security and Exchange Commission’s EDGAR system. These statements include legacy costs broken out by pensions and other postemployment benefits – largely healthcare liabilities. These data allow us to approximate the numbers of persons covered and also include information on the financial health of the sponsor. A macroeconomic health model we have developed estimates the potential impact on health care utilization, uncompensated care, cost shifting to commercial premiums, and provider incomes. Population Studied: This is a national study of early retirees estimated to be at risk of losing health insurance and possibly having pensions reduced. Early retirees are predominantly aged 55-64 and the critical subset involves those whose previous employers are financially stressed. Principal Findings: This research is partially complete and is being internally funded. The number of early retirees aged 5564 who receive health coverage from a former employer or union is approximately 3.6 million. The per capita needs of this population are roughly 1.7 times that of the currently insured population. Thus, one million newly uninsured early retirees would increase the percentage of uninsured by about 2.2% (from 47 million), but on a needs-adjusted basis the increase would be about 3.7%. Timely research is continuing on the numbers and characteristics of those at risk for losing coverage and the expected impact on utilization, health status, uncompensated care, and potential increase of commercial premiums through cost shifting. Conclusions: Wages can be adjusted to compensate for current worker benefits, but not for retirees. These legacy costs directly displace investment and profits, making companies globally uncompetitive. Legacy costs also decrease the ability of states and localities to maintain core programs. Thus, early retirees are at risk of losing coverage. This has the potential to increase the ranks of the uninsured with a disproportionate effect on access and utilization due to the relative age of this population. Implications for Policy, Delivery, or Practice: Rising premiums and stagnant wages have resulted in increasing numbers of uninsured in this country. Sudden loss of coverage for early retirees could result in a quantum leap in this problem. Primary Funding Source: Internal Funding ●Health Care Access in Three Nations: Canada, Insured America, and the Uninsured America Jae Kennedy, Ph.D., Steve Morgan, Ph.D. Presented By: Jae Kennedy, Ph.D., Associate Professor, Health Policy and Administration, Washington State University, POB 1495, Spokane, WA 99210; Tel: (509)368-6971; Email: jjkennedy@wsu.edu Research Objective: To assess variation in health access and utilization by nationality and insurance status. Study Design: Secondary analysis of the 2002/2003 Joint Canada/US Survey of Health (JCUSH), a cross-national household survey conducted by the National Center for Health Statistics and Statistics Canada. Population Studied: Weighted prevalence rates were estimated for Canadians (sample n=3,503), currently insured Americans (sample n=4,673), and currently uninsured Americans (sample n=510). Principal Findings: Uninsured Americans are more likely to report unmet health care needs than insured Americans or Canadians (35.8% vs. 10.4% vs. 10.7%, P<.001). Among insured (40.2%) and uninsured (84.4%) Americans, cost was cited as the main reason for not receiving needed care, while Canadians with unmet needs were most likely to cite long wait times (32.5%). Uninsured Americans are less likely to have a regular physician than insured Americans or Canadians (39.8% vs. 84.5% vs. 84.9%, P<.001), and more likely to not fill a prescription due to cost (28.2% vs. 7.7% vs. 5.1%, P<.001). Conclusions: The fact that a substantial number of Canadians and Americans report access difficulties and dissatisfaction with the health care they receive affirms McGlynn’s observation that, at least from the consumer’s perspective “there is no perfect health system.” However, despite significant differences in the financing and administration of their health systems, most Canadians and Americans with health insurance are able to get the care they need. In contrast, this analysis indicates that approximately one in three uninsured Americans report unmet health care needs. Implications for Policy, Delivery, or Practice: The most important difference between insured Americans and Canadians is that the latter group is guaranteed access to health care regardless of age, employment, income, or insurability. Due to increasing private insurance costs and changing labor-markets, an increasing number of Americans are losing their health care insurance. In this sense, uninsured America is the fastest growing, as well as most vulnerable, of the three nations studied. Primary Funding Source: No Funding Principal Findings: A majority of job endings, among both those who leave and lose jobs, involve workers who do not have own ESI through the job that ended. Among those who do have own ESI, fewer than one in five (17.1 percent) of their job transitions lead to becoming uninsured. For those who have own ESI, losing a job poses a greater risk of losing health insurance than changing jobs. Those who lose their jobs are nearly twice as likely to lose coverage and become uninsured as those who leave a job to take another job (24.9 percent of job losers v. 13.4 percent of job to job movers.) However, job to job movements are more common than job losses among those who have own ESI. The higher risk of losing coverage associated with job loss thus translates into a much smaller gap in what preceded losing coverage: 41 percent of those who lost own ESI lost their health insurance coverage when they lost a job, while 32 percent lost own ESI when they moved from job to job. While fear of health insurance loss ("job lock") may keep some workers from changing jobs, other workers move and experience loss of own ESI. More than one in ten (13.4 percent) of those who move from job to job lose coverage when they change jobs. Conclusions: While a majority of those who are uninsured became uninsured as a consequence of a job transition, the majority of job transitions do not result in a change in health insurance coverage. Not surprisingly, job loss is more likely to lead to coverage loss than is job changing. Implications for Policy, Delivery, or Practice: Despite the enactment twenty years ago of COBRA, federal legislation to give those who lose coverage from employment the opportunity to continue it, a large share of the uninsured are people who could have continued to enroll in an employmentbased plan but did not. Primary Funding Source: RWJF ●Health Insurance Consequences of Employment Transitions Hanns Kuttner, MA ●Late Payments and SCHIP: Results from a Disenrollee Survey Glenn Landers, M.B.A., M.H.A., Patricia Ketsche, Ph.D., M.B.A. Presented By: Hanns Kuttner, MA, Senior Research Associate, Economic Research Initiative on the Uninsured, University of Michigan, 555 S Forest Street, Ann Arbor, MI 48104; Tel: (734) 936-1308; Fax: (734) 936-6341; Email: hkuttner@umich.edu Research Objective: There is a substantial body of research explaining the reasons for transitions in and out of employment and a growing interest in transitions in and out of insurance coverage. Among those uninsured at a point in time, loss of employment-sponsored insurance coverage in one's own name (own ESI) is the most common transition to being uninsured. This paper ties the loss of own ESI to several labor market experiences --changing jobs, losing a job, and experiencing unemployment. Study Design: Secondary analysis of the 1996 panel of the Survey of Income and Program Participation (SIPP), a survey of the non-institutionalized US population. The panel nature of the SIPP allows us to control for person-specific characteristics, both those that are observed (e.g., gender) and unobserved (e.g., attitudes and preferences towards health insurance.) Population Studied: Adults (ages 18 to 64) who either changed jobs or lost a job in the period covered by the panel (1996 to 2000). Presented By: Glenn Landers, MBA, MHA, Senior Research Associate, Georgia Health Policy Center, Georgia State University, 14 Marietta St., Suite 221, Atlanta, GA 30303; Tel: 404-463-9562; Fax: 404-651-3147; Email: glanders@gsu.edu Research Objective: Georgia’s S-CHIP Program (PeachCare for Kids) requires families to pay an income adjusted premium between $10 and $35 for 1 child and $15 and $70 for two or more children age six or older. On July 1, 2004, the Georgia Department of Community Health (DCH) implemented a three-month waiting period for coverage if the premium was not received, processed, and posted by the first of the month prior to the month of coverage. This study gathered information about the experiences of children during and immediately after the three-month waiting period. Study Design: We conducted a bilingual telephone survey of parents or guardians of PeachCare children affected by this policy. Population Studied: Completed surveys were obtained from 1,009 respondents for children who experienced a threemonth waiting period and subsequently re-enrolled and an additional 647 respondents for children who did not re-enroll when eligible. Principal Findings: Among all children who experienced the three-month waiting period, 89 percent said they had no other coverage during that time. A large number (48 percent) of children sought care during their lapse in coverage - most obtained that care in a doctor’s office, and most paid for the care at the time of service. About one-quarter of respondents report that their child was taking a prescription drug prior to their lapse in coverage. Most (63 percent) of these children were able to continue their medications during their gap in coverage. The majority (59 percent) of children who did not reenroll ultimately obtained other coverage, generally through Medicaid or an employer-sponsored plan. Among respondents who did not have any current coverage for their children, over half said they could not afford it. Conclusions: Almost half of surveyed children needed health care services during coverage gap. The majority that needed care did not seek care through an emergency department: half went directly to a physician’s office. Almost all (88 percent) of respondents who sought care stated that it was not provided free of charge, and 57 percent paid for the care themselves. Thirty-two percent planned to pay for it in the future. Unfortunately, 19 percent of the respondents who stated their children needed care did not get it. Approximately nine months after experiencing a waiting period, 16 percent of children previously covered by PeachCare had no insurance coverage, while 84 percent had reenrolled in PeachCare or obtained alternative coverage. Implications for Policy, Delivery, or Practice: Further research might investigate whether or not there was pent up demand for services for those children who returned to the program after the waiting period, particularly those who were not able to obtain care while not covered by PeachCare. Furthermore, we suggest additional analysis of the link between utilization of services and the likelihood of experiencing a lapse in coverage due to non-payment of premium. Primary Funding Source: Georgia Department of Community Health ●Community-based Coverage Options Glenn Landers, M.B.A., M.H.A., Karen J. Minyard, Ph.D., Lindsey Lonergan, M.H.S. Presented By: Glenn Landers, M.B.A., M.H.A., Senior Research Associate, Georgia Health Policy Center, Georgia State University, 14 Marietta St., Suite 221, Atlanta, GA 30303; Tel: 404-463-9562; Fax: 404-651-3147; Email: glanders@gsu.edu Research Objective: With support from the HRSA State Planning Grant program, the Georgia Governor's Office, with the Georgia Health Policy Center, sought to assist four community-based pilot projects in leveraging public-private partnerships to expand health care coverage among low-wage workers at small firms in four distinctly different communities across Georgia. Successful pilots would serve as models for replication state- and nationwide. Study Design: Communities were invited to participate based on the existence of mature community collaboratives that had a strong history of success implementing programs focused on health care access. Design support included an insurance survey of Georgia businesses with over-sampling in the communities of interest, focus groups with small employers and low-wage workers of small firms, econometric modeling of the impact of uninsurance in the pilot communities, and on-site technical assistance. Population Studied: Four communities in Georgia: Atlanta metro - Fulton and DeKalb counties, other metro - the Macon area, rural north - Dalton, and rural south - Brunswick. The four distinct profiles (Four Georgias) were revealed in a 2002 survey of 10,000 Georgia households. The four communities represent approximately 267,000 small firm workers and their dependents, 160,000 of which are estimated to be uninsured. Principal Findings: •Employers offering coverage dropped three percent from 2002, with firms of 25 – 99 employees the most affected. • The erosion in offer rates was focused on urban firms – although rural firms were still less likely to offer. • Between 2002 and 2004, the total average premium cost for an individual increased 27 percent. • Average monthly wages increased only for firms that do not offer coverage. • Those without health insurance spend about 60% per year in health care costs compared with those with private coverage. • About 5% of Georgia’s health expenditures are for care of the uninsured. • Georgians spend about $1,063 annually for each uninsured Georgian. • Working with key community leaders who are willing to commit significant time and effort to the design project was vital to the success of program development. • Having accurate, detailed data on a variety of indicators related to the uninsured and local employment, as well as detailed information on best practices, available at the outset was critical. • Communities required frequent interaction with outside facilitators to ensure project work progresses. • Leveraging local, public financial participation was the greatest obstacle for the three non-Atlanta communities. • The state plays a critical role in facilitating the departmental, regulatory, or legislative change needed to enact new coverage models. Conclusions: The creation of community-based public-private coverage models is technically complicated and largely dependent on local level leadership with demonstrated success in community efforts focused on health care access. Identification of a public share is more challenging in lean economic times. Implications for Policy, Delivery, or Practice: State governments play a critical role in facilitating the departmental, regulatory, or legislative change needed to enact community-based coverage models. Due to the unique circumstances that exist in each community, adaptation rather than replication may be the order of the day when such models are considered for duplication. Primary Funding Source: HRSA ●Health Insurance and the Transition to Financial Adulthood Helen Levy, Ph.D. Presented By: Helen Levy, Ph.D., Research Assistant Professor, Institute for Social Research, University of Michigan, 555 S. Forest, Ann Arbor, MI 48104; Tel: 734 615 9587; Fax: ; Email: hlevy@umich.edu Research Objective: To analyze the high rates of uninsurance among young adults and the reasons for gaps in coverage among in this population. Study Design: A multivariate analysis of determinants of insurance coverage and how they vary with age. I also estimate the distribution of remaining uninsured spell length for individuals by age and the one-year probability of exit from an uninsured spell by age. Population Studied: A nationally representative sample of young adults ages 16 - 30 in the 1996 and 2001 panels of the Survey of Income and Program Participation. Principal Findings: The probability of being uninsured peaks in young adulthood. More than 40 percent of men and 30 percent of women in their early 20s are uninsured. For both men and women, this peak is driven by a gap of several years between the rapid drop-off of parental coverage after about age 17 and a more gradual increase in employer-sponsored coverage through one’s own or a spouse’s job that does not begin until several years after parental coverage has declined. Public coverage also declines sharply for men at age 18 but only gradually for women, resulting in a higher fraction uninsured among young men than young women. About half of the lag in gaining employer-sponsored coverage among young adults is explained by their relatively tenuous attachment to the labor force (in particular, part-time or shortterm employment). Compared with older uninsured adults, young adults who are uninsured experience shorter spells of uninsurance on average and are more likely to gain insurance within a year. Conclusions: Uninsurance among adults is a high-probability but relatively short-term event that is related to other predictable dimensions of economic immaturity like job instability. Implications for Policy, Delivery, or Practice: Different public policies are indicated for uninsured young adults than for older, longer-term uninsured adults. Encouraging young adults to buy low-cost catastrophic insurance policies in the individual market would mitigate the consequences of employment instability for their insurance coverage. Other policies that target young adults, such as extending Medicaid coverage through the age of 25, have the potential to reduce substantially the number of uninsured. Primary Funding Source: No Funding ●Health, Utilization, and Cost Impacts of Policy Changes Regarding Post-Cesarean Length of Stay Jeanne Madden, Ph.D., Dennis Ross-Degnan, ScD, Stephen B. Soumerai, ScD, Tracy A. Lieu, M.D./M.P.H., Fang Zhang, Ph.D. Presented By: Jeanne Madden, Ph.D., Instructor, Dept of Ambulatory Care and Prevention, Harvard Medical School & Harvard Pilgrim Health Care, 133 Brookline Avenue, 6th floor, Brookline, MA 02215; Tel: 617-509-9953; Email: jeanne_madden@hphc.org Research Objective: Legislation at the federal level and in most states mandates that insurers cover 96 hours of postcesarean hospitalization for mothers and newborns. AAP and ACOG guidelines also recommend 96 hours. Nevertheless, there are no large-scale studies evaluating the safety of shorter post-operative stays or the impact of these mandates for mothers and newborns with cesareans. This research is a comprehensive examination of a double natural experiment: an early post-cesarean discharge protocol in an HMO and a state 96-hour coverage mandate. Study Design: This retrospective cohort study encompasses 7.5 years of relatively uncomplicated cesarean deliveries subject to two well-defined policy reversals regarding length of post-operative stay (“on-off-on”). Study variables based on claims data and medical records include maternal and neonatal rehospitalizations and ER visits, office and home nurse visits, post-discharge maternal complications, feeding practice and feeding problems, neonatal jaundice diagnoses, mental health diagnoses and therapies, utilization of antiinfectives, and HMO payments for hospitalization and home services. We use both pre-post contingency table analyses with chi-square tests and interrupted time series regression analyses for all outcomes. Population Studied: Cesarean deliveries (n=4,559 mothers and 4,698 neonates) in the population of a large multi-center staff-model HMO in Massachusetts (the former Harvard Community Health Plan) between October 1990 and March 1998, excluding dyads with discordant stays or stays >7 nights. Principal Findings: Post-cesarean stays of 3 nights or less accompanied 16% of births during 9 baseline quarters and 77% of births during the 11 quarters that followed the new 3night HMO standard protocol. After the later state mandate, short stays dropped back to 26% of births. Fewer than 1 in 10 dyads received a home visit by a nurse during the baseline period and the 3-night protocol, but home visits rose to over 44% post-mandate. There was no evidence of impact due to either the 3-night protocol or the mandate on post-discharge hospital-based services for mothers or newborns, breastfeeding rates, or other therapeutic and diagnostic measures. HMO payments for hospital stays and home visits declined by about $400 per woman after the shift to a 3-night protocol, but the later minimum-stay mandate had a negligible impact on HMO outlays, apparently because of simultaneous declines in prices within all length-of-stay groups. Conclusions: The attempt by this HMO to reduce costs by discharging women and newborns sooner after cesarean delivery appears to have been safely implemented. A later “protective” government mandate conferred no detectable health benefit in this population, although a secondary provision in the law resulted in post-cesarean nurse home visits becoming commonplace for the first time. Implications for Policy, Delivery, or Practice: Case-by-case physician discretion, as practiced in this setting, may be key to ensuring the safety of early post-cesarean discharge. However, early discharge is clearly a safe option for many patients, so healthy families who prefer to return home should be encouraged when follow-up care is accessible. Cost outcomes of discharge policies in a context of price volatility may defy expectations. Primary Funding Source: RWJF, AHRQ, MCHB, and HPHC Foundation ●Who Disenrolls from Community-Based Health Care Programs for the Uninsured? Catherine McLaughlin, Ph.D., Anne Warren, M.P.P., Paula H. Song, M.H.S.A., M.A.E., Erin Fries Taylor, Ph.D., M.P.P. Presented By: Catherine McLaughlin, Ph.D., Professor, Department of Health Management and Policy, University of Michigan, 555 S. Forest St., Ann Arbor, MI 48104; Tel: (734)615-9586; Fax: (734)998-6341; Email: cmcl@umich.edu Research Objective: Community-based health care programs are gaining popularity as a local solution to address the needs of the uninsured. Using evidence from three community programs, this paper investigates who disenrolls from these programs, how they differed from those who stayed, and whether disenrollees gain coverage after leaving. Study Design: We conducted a multi-wave survey of participants in three Communities in Charge (CIC) programs, a Robert Wood Johnson Foundation funded initiative aimed at improving access for low-income, uninsured adults. Participants were surveyed three times, at enrollment and six and twelve months after enrollment. The program sites were Alameda County, California; Austin, Texas; and Southern Maine. We compare baseline, i.e. time of enrollment, demographic, employment, and insurance characteristics of those continuously enrolled in the program to those who disenrolled during the first twelve months. We assess disenrollees' insurance status at twelve months to determine whether they gained insurance and, if so, the type of coverage. Further, we group respondents based on enrollment status and health insurance type at twelve months to compare demographic and employment characteristics and insurance histories of these groups. Population Studied: The sample consists of participants in CIC programs who responded to all survey waves: Alameda County, n=371; Austin, n=253; and Southern Maine, n=233. Principal Findings: Disenrollment rates varied significantly by site, ranging from 11 percent in Alameda County to 62 percent in Southern Maine. Characteristics of disenrollees also differed across sites, e.g., disenrollees in Southern Maine and Austin were significantly younger and more likely to have been insured than continuous enrollees. Only in Southern Maine were disenrollees more likely to be employed, although in all three sites, disenrollees were more likely to work for an employer who offers coverage. In Alameda County and Southern Maine, approximately three-quarters of disenrollees reported disenrolling due to a job that offered health insurance or eligibility for another program. In Austin, approximately 40 percent cited these reasons, while another 40 percent reported problems with paperwork. At twelve months after enrollment, approximately half of disenrollees were insured, of which half had employer sponsored insurance (ESI). Approximately 30 percent of disenrollees in Southern Maine became publicly insured. In Austin and Southern Maine, disenrollees who gained ESI tended to have higher incomes, more years of education, and were more likely to have had ESI coverage in the last year than continuous enrollees and other disenrollees. Conclusions: We found evidence that those who stay enrolled in community-based programs for the uninsured are predominantly individuals who need access to care for longer than just short periods. Fewer than half of those who disenroll are returning to some kind of coverage. The majority are remaining uninsured, having obtained short term access to care through the program. Implications for Policy, Delivery, or Practice: Disenrollment from community-based programs for the uninsured is not necessarily problematic. If disenrollment occurs because participants gain insurance, then these programs enable participants to access stop-gap care. However, if participants disenroll but remain uninsured, then programmatic design should be reevaluated. Primary Funding Source: RWJF ●The Myth of Emergency Department Diversion and the Uninsured: Evidence from Three Community-based Programs Catherine McLaughlin, Ph.D., Paula H. Song, M.H.S.A., M.A.E., Anne W. Warren, M.P.P. Presented By: Catherine McLaughlin, Ph.D., Professor, Department of Health Management and Policy, University of Michigan, 555 S. Forest St., Ann Arbor, MI 48104; Tel: 734-6159586; Fax: 734-998-6341; Email: cmcl@umich.edu Research Objective: n an effort to control health care costs, organizations and community programs try to divert emergency department (ED) use by the uninsured. Rarely does this approach produce the expected results in part because the underlying assumptions about utilization, appropriateness of use, and costs of ED services are not well understood. This paper provides a closer evaluation of the myths surrounding ED utilization within the context of community-based programs for the uninsured. Specifically, we assess the claim that providing the uninsured with a medical home will reduce ED utilization and therefore result in substantial cost savings. We propose that effective ED diversion of the uninsured hinges on three assumptions: (i) the uninsured utilize the ED inappropriately, (ii) given a medical home, patients will use an alternative source of care, and (iii) costs of ED services are high. Study Design: We utilize survey data from three sites that participated in the Communities in Charge (CIC) program, a grant initiative funded by the Robert Wood Johnson Foundation that focused on changing the care-seeking behavior of low income, uninsured adults. Surveys of program participants were conducted at enrollment and again at 6 months and 12 months after enrollment in Alameda County, CA; Austin, TX; and Portland, ME. We analyze survey data of all participants to compare demographic characteristics, health status, ED and physician utilization patterns for the year before enrollment and during the first year of enrollments for all participants. In addition, we compare those who changed their ED use to those who did not change ED use in terms of their overall health care utilization patterns for the year before enrollment and during the first year after enrollment. Population Studied: Program participants from three CIC programs who participated in all three waves of the CIC survey (Alameda County, n=303; Austin, n=164; Portland, n=89). All participants were low-income adults and were ineligible for public insurance at the time of enrollment. Principal Findings: Results from our survey analyses find that in each site, the number of people who used the ED only after enrollment was approximately the same as the number who only used the ED beforehand. Thus, the net change in utilization is negligible. Participants who used the ED both before and after were more likely to have a chronic condition or report fair or poor health status. These results suggest that much of the ED use by these groups of low income, uninsured adults was appropriate and therefore not likely to be reduced through ED diversion programs. Conclusions: Providing a medical home to participants in community-based programs for the uninsured did not change net ED utilization. This "steady state" of ED utilization suggests that participants are utilizing the ED appropriately. Provided the first two assumptions are not met, the marginal cost of the minimal number of “diverted” ED visits would have to be extraordinarily high to result in cost savings. Implications for Policy, Delivery, or Practice: In order to achieve cost savings, local communities or programs attempting to control costs through diverting ED utilization should evaluate the underlying assumptions necessary to achieve cost savings. Primary Funding Source: RWJF next year was consistently larger, between 7 and 14 percentage points. Conclusions: The overall improvement in health insurance stability from 1984 to 2001 did not come with a narrowing of gaps between vulnerable populations and the population overall. The relative rates of coverage remained constant for most comparisons. The exception was for adults in chronically poor families. Their coverage rate became relatively lower over this time period. Implications for Policy, Delivery, or Practice: Those who examine health insurance coverage differentials between vulnerable populations and the population overall should be aware that in addition to having lower rates of coverage, vulnerable populations are also more likely to experience loss of coverage. Primary Funding Source: RWJF ●Coverage Transition in Vulnerable Populations Catherine McLaughlin, Ph.D., Matthew Ruttledge Presented By: James Moser, Ph.D., Senior Health Policy Economist, , American College of Radiology, 1891 Preston White Drive, Reston, VA 20191; Tel: 703-648-0690; Fax: 800832-9227; Email: jmoser@acr.org Research Objective: To compare the use of hospital inpatient radiology services by the uninsured with those of persons with health insurance. Study Design: Statistical analysis primarily of National Hospital Discharge Survey, a data source that facilitates identification of both type of service and source of payment. The uninsured are defined as persons whose principal source of payment is "self pay" or "no charge." Population Studied: Hospital inpatients under age 65 who received one or more radiology services during their episode of care. Principal Findings: Uninsured radiology inpatients received 12% fewer services than did their insured counterparts. By radiology modality, the uninsured tended to receive a larger proportion of lower-cost modalities (e.g., x-rays) and a smaller proportion of higher-cost modalities (e.g, computed tomography). Conclusions: Uninsured inpatients receive fewer, lower-cost radiology services compared with insured inpatients. These differences are due in part to decisions by patients, payers, hospital management, and attending and/or referring physicians, but the separate contributions of each to differences by insurance status are unknown. Implications for Policy, Delivery, or Practice: If worse health outcomes for uninsured inpatients can be shown to be associated with differences in receipt of radiology services, action may be required on the part of policymakers, payers, and providers to address these disparities. Primary Funding Source: No Funding Presented By: Catherine McLaughlin, Ph.D., Professor, Health Management and Policy, University of Michigan, 555 S. Forest, Ann Arbor, MI 48104; Tel: (734) 615-9586; Fax: (734) 998-6341; Email: cmcl@umich.edu Research Objective: For decades, national surveys recording health insurance coverage levels have consistently estimated lower levels of coverage for certain so-called vulnerable populations. We evaluate changes in those differentials over the last 20 years, in particular, changes in the rate at which individuals in these population groups gain and lose health insurance. Study Design: We conduct a secondary data analysis of multiple panels of the Survey of Income and Program Participation (SIPP). We analyze changes in the probability of beginning and ending a spell without health insurance in the first year of each of the eleven SIPP panels begun between 1984 and 2001 Population Studied: Random national sample of noninstitutionalized Americans age 18-64. Principal Findings: While health insurance status became more stable between 1984and 2001, the difference in coverage rates between vulnerable groups and the population overall has shown little change, with the notable exception of adults from low income families (annual income <200 percent of poverty). An analysis of potentially vulnerable subgroups, including African-Americans, Hispanics, and those in low income families shows the difference in the probability of becoming uninsured between the vulnerable group and the population overall has changed little for African-Americans and Hispanics, but increased over the past decade for those who are in chronically poor familes, that is, below 200 percent of the poverty level all four years of the panel. The gap in the probability of losing coverage between each of these three vulnerable populations and the overall population is greater than the gap in the probability of being without coverage. For example, while the difference between whites and -AfricanAmericans in insurance coverage was between 7 and 9 percentage points at the outset of each of the eleven waves, the difference in the probability of losing coverage over the ●Use of Inpatient Radiology Services by the Uninsured James Moser, Ph.D. ●Use of Prescription Medications and Medical Costs in Multiple Sclerosis Katia Noyes, Ph.D., M.P.H., Steven Schwid, M.D., Robert Holloway, M.D., Ray Dorsey, M.D., Chunyu Li, MS, Huai-Che Shih, MS Presented By: Katia Noyes, Ph.D., M.P.H., Assistant Professor, Department of Community and Preventive Medicine, University of Rochester School of Medicine, 601 Elmwood Avenue Box 644, Rochester, NY 14642; Tel: (585) 275-8467; Fax: (585) 461-4532; Email: katia_noyes@urmc.rochester.edu Research Objective: Multiple sclerosis (MS) affects nearly 400,000 Americans, but data on the medical expenditures in this population are limited, especially since the introduction of expensive disease modifying agents (DMAs) in the 1990s. DMAs (Avonex, Rebif, Betaseron and Copaxone) are highly expensive biological medications with a potential to slow MS progression, may have significantly affected costs and patterns of medical care in MS. Because the age of onset in MS is around 30 years of age, costs due to lost productivity are particularly important in MS. This study examines health services use and expenditures in MS population, with the focus on indirect, out-of-pocket (OOP), and prescription medication expenses. Study Design: We used 1992-2000 Medicare Current Beneficiary Survey (MCBS), 1996-2003 Medical Expenditure Panel Survey (MEPS) and Medicare claims data. MS cases were identified based on participant self-report and diagnostic codes (ICD-9 CM 340). Using log transformed CPI-adjusted dollars, multivariate linear models were developed to estimate total healthcare costs, Medicare-reimbursed total costs, OOP costs, indirect costs, and total and OOP costs of prescription medications. Risk adjustment factors included age, gender, race, residence area, health status (co-morbidities and severity of MS), and year of services use. Similar negative binomial models were developed for appropriate utilization. All costs were adjusted to 2004 U.S. dollars. Analyses incorporated complex survey design. Population Studied: 112,871 MCBS person-years representing Medicare fee-for-service beneficiaries, including 427 from MS patients. Data from the MEPS included 135,496 person-years, with 161 from MS patients. Principal Findings: MS patients had higher costs in all examined categories. Annual unadjusted total costs were $9,752 for MS and $2,229 for non-MS in general population, and $15,866 for MS vs $7,242 for non-MS in the Medicare beneficiaries. The differences in costs were partially explained by differences in socio-economic and demographic characteristics since MS patients were more disabled, less wealthy, and more likely to be on Medicaid. On average, MS patients lost more days lost than patients without MS (8 vs 3 days a year). MS patients used significantly more prescription medications and spent a significantly larger portion of their income on prescription medications compared to non-MS patients. MS patients who used DMAs were more likely to have private insurance coverage, were wealthier and more likely to be males, and spent more OOP on pharmaceuticals than MS patients who did not use these agents. Significantly fewer Medicare patients used DMAs compared to nonMedicare beneficiaries. Conclusions: Annual direct and indirect medical costs were much higher for individuals with MS in both the general and Medicare populations compared to non-MS population. Use of DMAs was associated with having private health insurance and higher socio-economic status. Implications for Policy, Delivery, or Practice: MS places a substantial financial burden on individuals, private payers, Medicare, and Medicaid. Factors associated with utilization of DMAs and potential impact of DMAs on the economic consequences of MS require more investigation. Primary Funding Source: NIA, National Multiple Sclerosis Society (NMSS) ●Did Medicaid/SCHIP Crowd-Out Private Insurance Coverage among U.S. Low-Income Children?: A Multilevel Approach Adetokunbo Oluwole, MS, Dennis Shea, Ph.D. Presented By: Adetokunbo Oluwole, MS, Graduate Assistant (Doctoral Student), Health Policy and Administration, The Pennsylvania State University, 116 Henderson Building, University Park, PA 16802; Tel: (814) 863-0875; Fax: (814) 8632905; Email: axo129@psu.edu Research Objective: The objective of this paper is to measure the extent of crowd-out of private insurance coverage, among American low-income children, that is attributable to Medicaid/SCHIP expansion. Study Design: We combine event history analysis and multilevel modeling techniques to analyze data from the 1996 panel of the Survey of Income and Program Participation (SIPP), which covers the period 1996 to 2000. The combined approach will estimate discrete-time logistic regression models. Events to be modeled are the following transitions in insurance coverage: (1) private coverage to Medicaid; (2) private to uninsured; (3) uninsured to private coverage; (4) uninsured to Medicaid; (5) Medicaid to private; and (6) Medicaid to uninsured. In addition to addressing the question of whether or not children have particular types of health insurance coverage, event history analysis also incorporates the timing of such coverage and transitions. Multilevel modeling accounts for cross-state variations in coverage transition rates as well as corrects for within-group correlation and non-constant error terms. State-level data from other sources such as the Area Resource Files, Regional Economic Information Services, and the Centers for Medicare and Medicaid Services will be appended to the SIPP dataset. To address the issue of heterogeneity among states, the models estimated in this study control for state- and program-specific factors, which many previous studies failed to account for. Population Studied: The paper focuses on children living in low-income families in the United States. The sample for the study consists of unmarried children nineteen or younger, and living in families with incomes at or below 300 percent of the federal poverty level (FPL). The analytic sample is expected to comprise about 8,000 low-income children. Principal Findings: Preliminary analyses show that the conditional probability of transition from private insurance to Medicaid/SCHIP started to increase shortly after states began to implement their Medicaid/SCHIP expansion programs. Conclusions: Given the controversial estimates of crowd-out in previous studies, this study’s use of combined techniques of event history and multilevel modeling will provide more accurate and updated estimates using the 1996 panel of the SIPP dataset, which no other study has analyzed. Implications for Policy, Delivery, or Practice: The implementation of the Medicaid/SCHIP, which began shortly after the major welfare reform of 1996, represented a more recent effort to expand health insurance among lower income populations in the U.S. Previous studies have shown that earlier Medicaid expansions significantly increased public program participation and have reduced the number of uninsured individuals. However, empirical evidence indicates that the reduction in the number of uninsured has been offset by declines in private insurance coverage. Because the Medicaid/SCHIP expansions targeted a different population than earlier expansions and took a different form, a detailed study of these more recent efforts is needed. Results from this study will provide more insight on how best health insurance coverage policies can be directed to have maximum positive effect on the targeted (uninsured) population. Primary Funding Source: Graduate Assistantship ●A Card and A Home: Types of Insurance Coverage and Usual Sources of Care Stephen Petterson, Ph.D. Presented By: Stephen Petterson, Ph.D., Senior Health Policy Researcher, Robert Graham Center, American Academy of Family Physicians, 1350 Connecticut Avenue, N.W., Suite 201, Washington, DC 20036; Tel: (202)331-3360; Fax: (202)3313374; Email: SPetterson@AAFP.org Research Objective: The main objective is a thorough analysis of the relationship between type of health insurance coverage and usual source of care. Prior studies have conclusively shown that the uninsured and underinsured are substantially less likely to have a usual source of care. Little is known about the extent to which the likelihood of having a usual source of care varies between public and private insurance coverage as well as between traditional fee-forservice and managed care insurance. Likewise, little is known about the relevance of insurance types on the character of a person’s usual sources of care, including the specialty of the usual provider of care and whether relies on a facility rather than a particular provider. Study Design: This study is based on secondary data using a quasi-experimental design. We use data from the 2002 Medical Expenditure Panel Survey (MEPS). The association between insurance type and usual source of care type is examined in models that control for demographic characteristics as well as measures of physical and mental well-being. Population Studied: The MEPS sample is representative of the civilian non-institutionalized population. In our study, separate analyses are carried out for children, non-elderly adults and the elderly. Principal Findings: Adjusting for age and health, Medicare recipients are by far the most likely to have a usual source of care. Among insured non-elderly adults, those covered by private or Medicaid fee-for-service insurance are the least likely to have a usual source of care. Moreover, the nonelderly in private HMOs are significantly more likely to have a usual source of care than their counterparts in Medicaid HMOs. Children covered by private insurance are more likely to have a usual source of care than children covered by public insurance. The nonelderly with private insurance are significantly less likely than their counterparts with public insurance to rely on a facility than a particular provider. Finally, among those who can name a particular provider as their usual source of care, persons in HMOs and in other managed care plans are more likely to have a family physician or general practice physician than a provider from another specialty. Conclusions: Type of insurance of insurance coverage is strongly associated with type of usual source of care for both children and adults. With important exceptions, persons covered by HMO and managed care plans are more likely to have a usual source of care and more likely to have a family physician. Medicaid recipients often do not have a usual source of care and, among those that do, they are the most likely to rely on a facility rather than a particular provider. Implications for Policy, Delivery, or Practice: Previous work has shown that insurance coverage and having a usual source of care are important and independent predictors of access to necessary medical care. Our findings suggest that policies designed improve health outcomes should address the implications of potential changes for the combined impact of health insurance coverage and providing all individuals with a medical home. Primary Funding Source: Other ●Emergency Department Visits: Is This the Right Place? Mary Price, MA, Richard Brand, Ph.D., Joseph Newhouse, Ph.D., Joseph Selby, M.D., M.P.H., John Hsu, M.D., MBA, MSCE Presented By: Mary Price, MA, Sr. Consulting Data Analyst, Division of Research, Kaiser Permanente Northern California, 442 Brimhall St., St. Paul, MN 55105; Email: Maggie.Price@kp.org Research Objective: Previous studies suggest that patients often seek care from the emergency department (ED) unnecessarily, resulting in overcrowding and inefficient resource use. We investigated the severity of ED visit diagnoses, and the association between visit severity and subsequent hospitalization or deaths. Study Design: For all ED visits (1999–2001) in a large prepaid integrated delivery system (IDS), we used the New York University ED Diagnosis Algorithm to classify visits as low or high-severity. This algorithm assigns probabilities to four ordered categories for each visit’s ICD-9 diagnosis; the sum of the four probabilities equals 100%. We classified visits as lowseverity when there was a >=75% probability that the visit diagnosis either did not require emergency care or could have been treated in a primary-care setting. We classified the visit as high-severity when there was a >=75% probability that the visit diagnosis did require preventable or unpreventable emergency care. We used the most severe diagnosis when there were >1 visit diagnosis; visits were indeterminate if the probability was <75%. Using GEE models, we examined the patient and system factors associated with low-severity ED diagnoses, stratified by insurance type (commercial or Medicare). We also examined the association between visit severity and subsequent hospitalization or death, adjusting for year, month, age, gender, neighborhood socio-economic status, comorbidity, prior utilization, having a regular physician, cost-sharing, and medical center. We repeated the analysis using proportional hazard models and 50% and 90% probability thresholds. Population Studied: All 628,272 subjects were IDS members in January 1999 with ?1 ED visit in 1999-2001. Principal Findings: There were 1,085,229 ED visits (19992001): 29% low and 20% high-severity visits, with 52% indeterminate visits. Of the low-severity visits, 4% resulted in a subsequent non-elective hospitalization within one-day and 0.7% resulted in death within 30 days. In addition to several patient characteristics, factors associated with low-severity visits included not having a regular physician and low ED copayments in both insurance groups, and high drug copayments in the commercial group. Compared with highseverity visits, low-severity visits were less likely to result in a subsequent non-elective hospitalization (OR=0.14, 95%CI: 0.13—0.14) or in death within 30 days (OR=0.21, 95%CI: 0.19—0.23) for subjects with commercial insurance; the results for hospitalizations (OR=0.21, 95%CI: 0.20—0.21) and deaths (OR=0.30, 95%CI: 0.28—0.31) were similar for Medicare subjects. The findings were robust across analytic approaches. Conclusions: Almost one third of ED visits were for lowseverity conditions, which are substantially less likely to result in subsequent hospitalizations or death. Several system factors, such as having a regular physician and higher ED copayments, are associated with decreased use of the ED for low-severity conditions. Implications for Policy, Delivery, or Practice: Use of the ED for low-severity conditions is potentially inefficient. These data suggest that even in an insured population within an integrated system, many patients use the ED for low-severity conditions. More research is needed to examine the factors that encourage patients to make more efficient choices about where to seek care and the clinical and economic ramifications. Primary Funding Source: AHRQ, The Commonwealth Fund ●Safety Net on the Edge: Growing Pressures on Health Centers' Ability to Meet Demand Michelle Proser, M.P.P., Daniel R. Hawkins Presented By: Michelle Proser, M.P.P., Director of Policy Research, National Association of Community Health Centers, 1400 Eye Street, NW, Suite 330, Washington, DC 20005; Tel: (202) 296-1960; Fax: (202) 296-3526; Email: mproser@nachc.com Research Objective: To determine national trends in community health center patient characteristics and the affects of such trends on the financial stability of health centers nationwide. Study Design: Analyzed health center patient and financial data from the Uniform Data System (UDS) for calendar years 1999-2004. All federally-funded health centers must report UDS annually to the federal Bureau of Primary Health Care. In 2004, 914 health center organizations from across the country reported UDS. UDS data were compared to US Census and other national data sources where possible in order to contrast health center and US trends. Population Studied: Analyzed data on all patients visiting federally-funded health center organizations during study years (13.1 million patients in 2004 UDS). Patients tend to be low-income, and be uninsured or publicly insured. Also analyzed health center costs of care, third-party charges and collections, and other revenue reported in UDS. Principal Findings: Between 1999 and 2004, the number of patients treated by health centers increased by 46%. Between 1999 and 2003, the number of low-income patients who sought care at health centers grew four times as fast as the number of low-income Americans. Over the same time period, the number of uninsured patients grew nearly three times as fast as the number of uninsured nationally. Medicaid continues to be the dominant insurer of health center patients, covering 36% of patients in 2004. Federal health center grant revenue per patient has not kept up with cost per patient ($272 vs. $504 in 2004). Concurrently, payments from Medicare, other public insurance, and private insurance are on the decline. While Medicaid is the most reliable payer, paying an average of 87% of health centers’ costs, private insurance is the worst payer, covering only 57% of costs. Although average operating margins hover around 1% annually for the study period, those health centers with negative or zero margins on average collect less from Medicaid, Medicare, and other public insurance programs compared to health centers with positive operating margins. Conclusions: The demand for health center services is on the rise, especially among low-income and uninsured populations that may be costlier to treat. Simultaneously, health centers are experiencing rising costs of care also seen across the health care system, and they continue to rely on a diverse mix of revenue sources to deliver their current scope of services. Although Medicaid is currently the most reliable payer, program cuts endanger health centers’ ability to meet their patients’ needs. Implications for Policy, Delivery, or Practice: Given rising numbers of uninsured and those in poverty, health centers likely will continue to experience growing demand for their services. As the largest insurer of health center patients, adequate Medicaid payments and sustained Medicaid enrollment are essential to a health center’s financial viability. Diminishing revenues from other sources exacerbate the financial pressures on health centers. While health centers on average have slim positive operating margins, one unexpected emergency or change in payment policy could jeopardize their financial stability, threatening the existence of primary care for an entire community. Primary Funding Source: No Funding ●Insurance Instability Among a National Sample of Children in the Child Welfare System Ramesh Raghavan, M.D., Ph.D., Gregory A. Aarons, Ph.D., Laurel K. Leslie, M.D., M.P.H. Presented By: Ramesh Raghavan, M.D., Ph.D., Policy Director, The National Center for Child Traumatic Stress, University of California, Los Angeles, 11150 W Olympic Blvd Suite 650, Los Angeles, CA 90064; Tel: (310) 235-2633 x233; Fax: (310) 235-2612; Email: rraghavan@mednet.ucla.edu Research Objective: To estimate the prevalence, patterns, and predictors of health insurance instability among a national sample of children in the child welfare system. Study Design: We analyzed child-level data from four waves of the National Survey of Child and Adolescent Well-Being (NSCAW), the first nationally representative study of children coming into contact with child welfare agencies. Outcomes were types of current insurance status – Medicaid, private, federal, or uninsured – at each wave. Predictors included child-level sociodemographic characteristics, maltreatment history, placement status, Child Behavior Check List (CBCL) scores, caregiver education, and log of caseworker experience in months. We performed weighted descriptive and bivariate analyses of insurance change over time within types of insurance categories, and conducted survey-weighted logistic regressions to estimate odds of being uninsured, adjusting for child-, caregiver-, and caseworker-level characteristics. Population Studied: 3041 children aged between 2 and 16 years investigated by child welfare agencies for suspected child maltreatment. Data was used from Wave 1 (November 1999 to April 2002), and 12-, 18-, and 36-month follow up waves (Waves 2 through 4). Principal Findings: Overall, 52% of children and adolescents were stably insured across all 4 waves, and less than 1% were never insured. Proportions of children enrolled in Medicaid increased across waves (from 62% at Wave 1 to 69% at Wave 4), proportions of those uninsured declined (from 12% to 7% over the same time period), and proportions of those in private or federal insurance plans were stable at around 25%. Acquisition of insurance coverage by the previously uninsured occurred steadily across waves, with 5.2% becoming insured between Waves 1 and 2, 4% between Waves 2 and 3, and 4.5% between Waves 3 and 4. However, losses in insurance coverage also occurred over time, with 4.5% losing coverage between Waves 1 and 2, 2.4% between Waves 2 and 3, and 3.4% between Waves 3 and 4. On logistic regression older age, African-American or Latino ethnicity, having caregivers with a high school or less education, and having less experienced caseworkers were all associated with higher odds of being uninsured at each wave. Children and adolescents who were placed in homes other than their own displayed a greater degree of insurance stability than children maintained within their own homes. Conclusions: Children within the child welfare system exhibit a lesser degree of churning when compared to Medicaid- or SCHIP-enrolled children in the general population. Such instability occurs immediately following entry into the child welfare system. Children in foster care placement (which usually makes such children Medicaid-eligible) are more stably insured than those maintained in-home. Implications for Policy, Delivery, or Practice: Expansion of categorical Medicaid eligibility benefits to children in the child welfare system maintained within their own homes is likely to be successful at preventing insurance instability. Child welfare agencies should consider health needs in addition to assessing risk while making placement decisions. Caseworkers should pay particular attention to the health needs of children maintained in-home, and work with caregivers to secure insurance coverage for these children. Primary Funding Source: No Funding ●Use of Preventive Care by the Working Poor Joseph S. Ross, M.D., Susannah M. Bernheim, M.D., Elizabeth H. Bradley, Ph.D., William T. Gallo, Ph.D. Presented By: Joseph S. Ross, M.D., Robert Wood Johnson Clinical Scholar, Internal Medicine, Yale University, Room IE61 SHM, P.O. Box 208088, New Haven, CT 06520-8088; Tel: (203) 785-5920; Fax: (203) 785-3461; Email: joseph.s.ross@yale.edu Research Objective: Although lower-income adults are less likely to receive preventive care when compared with higherincome adults, it is not clear whether working adults living in poverty — the working poor — are similarly at risk. Our objective was to examine the association between poverty and use of preventive care among older working adults. Study Design: We performed a cross-sectional analysis of data from the 1996, 1998 and 2000 waves of the Health and Retirement Study (HRS), a nationally-representative sample of U.S. adults aged 51-61 in 1992. We studied five main outcome measures: self-reported use of (1) Pap smears for cervical cancer screening, (2) mammograms for breast cancer screening, (3) serum prostate-specific antigen measurement for prostate cancer screening, (4) serum cholesterol screening for cardiovascular risk reduction, and (5) influenza vaccination. Adults whose annual household income was within 200% of the household-size adjusted federal poverty level were defined as living in poverty. We used 2 stages of multivariable analysis to examine the association between working poverty and each outcome, with all analyses taking into account the complex survey design and weighted sampling probabilities of the data set. The first stage analysis adjusted for socio-demographic covariates including age, gender, race/ethnicity, marital status, household size, and net worth, as well as self-reported health status, insurance status, and annual physician visits. In order to understand how education may moderate the effect of working poverty, we performed a second-stage analysis that additionally adjusted for education. Population Studied: 6,835 community-dwelling working adults participated in at least one wave of the HRS from 1996 through 2000. Repeated assessment yielded a total of 10,088 observations of working adults: 1,196 (12%) living in poverty and 8,896 (88%) not living in poverty. Principal Findings: Among eligible adults, use of preventive services ranged from 30% for serum cholesterol screening to 76% for breast cancer screening. First-stage multivariable analysis demonstrated that working adults living in poverty were significantly less likely to receive each preventive service when compared with working adults not living in poverty (Pvalues < 0.05). Second-stage multivariable analysis demonstrated that working adults living in poverty were significantly less likely to receive breast cancer screening (OR 0.70, 95% CI, 0.58-0.85, P<0.001), prostate cancer screening (OR 0.72, 95% CI, 0.58-0.91, P=0.005), and cholesterol screening (OR 0.74, 95% CI, 0.64-0.86, P<0.001), while education moderated the effect of working poverty on use of cervical cancer screening (OR 0.85, 95% CI, 0.70-1.03, P=0.10) and influenza vaccination (OR 0.88, 95% CI, 0.75-1.03, P=0.11). Conclusions: While use of preventive care varied markedly by service and was below recommended targets for every service except breast cancer screening, the working poor were less likely to receive important preventive care services. Implications for Policy, Delivery, or Practice: Difficult choices between purchasing daily necessities and preventive care leave the working poor at risk for preventable deaths and avoidable health care costs. Reducing economic and noneconomic barriers to preventive care, by easing access to safety-net providers, providing means-tested preventive care vouchers, or expanding Medicare eligibility, could yield significant health benefits. In addition, the role of education as a moderator suggests that health literacy may influence use of preventive care. Primary Funding Source: No Funding ●Arkansas and Massachusetts: a Comparison of Health Insurance Status in Two States Joseph Thompson, M.D., M.P.H., Kevin W. Ryan, JD, MA, Jennifer L. Shaw, M.P.H., MAP, Kirk K. Larsen, MA, Patricia M. Gallagher, Ph.D., Anthony Roman, MA Presented By: Joseph Thompson, M.D., M.P.H., Director, Arkansas Center for Health Improvement, 1401 West Capitol, Suite 300, Little Rock, AR 72201; Tel: (501) 526-2244; Fax: (501) 526-2252; Email: thompsonjosephw@uams.edu Research Objective: The increasing number of uninsured has created significant challenges for public and private stakeholders in states across the nation. Operating within balanced budget mandates, and with no federal solution on the horizon, states have had various responses and reallocated resources to optimize financial coverage and meet the basic health needs of their citizens. Given current budget constraints and increased healthcare utilization, the ability of states to continue to adequately address healthcare needs is in question. Study Design: We undertook this study of two states (Arkansas and Massachusetts) in order to generate state level economic and healthcare system profiles. Information was obtained from multiple data sources in order to create these state profiles. Sources included US Census Bureau, the Agency for Healthcare Research and Quality Medical Expenditure Panel Survey, state Departments of Health and Human Services, and state-specific household surveys conducted in 2004. Importantly, the same household survey instrument, which collected detail information about health insurance and healthcare system utilization in each state, was fielded in both states by the University of Massachusetts Center for Survey Research using random digit dialing methodology. Population Studied: The population studied consisted of Arkansas and Massachusetts household members. Respondents numbered 6,684 and 12,749 in Arkansas and Massachusetts respectively. A weighting system was applied to estimate the number of people in households for both states. Principal Findings: Analyses of national datasets and results from the household surveys yielded substantively disparate profiles of the states under review. Estimates indicate that 17.2% of all Arkansans were uninsured in 2004, while only 7.4% of all Massachusetts residents were without coverage that year. More than twice as many Arkansas adults aged 1964 were uninsured (24.4%) as their counterparts in Massachusetts (10.6%). Similarly, over three times as many Arkansas children aged 0-18 were uninsured (10.4%) as Massachusetts children (3.2%). The 2004 median household income in Arkansas was estimated at $32,983 and $55,658 in Massachusetts. Additionally, in 2003 only 42.2% of Arkansas private sector firms offered health insurance to their employees while approximately 66% of private firms in Massachusetts offered health insurance that same year. Comparison of public sector eligibility reveals greater programmatic eligibility for adults aged 19-64 years in Massachusetts than in Arkansas (e.g. disabled adult income eligibility threshold of 133% to receive Medicaid benefits as contrasted with 72% in Arkansas and existence of premium assistance purchasing programs for working and non-working Massachusetts adults that do not exist in Arkansas). Conclusions: The ability for low-income states to insure the 19-64 year old adult population is limited. In each of these states, neither private nor public sectors have adequate resources to achieve high and consistent levels of insurance coverage across all ages. This comparison of Arkansas and Massachusetts demonstrates that increased household income, private sector offering and public sector program availability are insufficient in and of themselves to reduce uninsurance rates to acceptable levels. Implications for Policy, Delivery, or Practice: The federal Medicaid matching assistance program, based upon average per capita income alone, may not sufficiently level the playing field for all states. Primary Funding Source: HRSA, RWJF ●Dual Use of Medicare and the Veterans Health Administration: Are There Adverse Health Outcomes? Fredric Wolinsky, Ph.D., Thomas Miller, MBA, Hyonggin An, Ph.D., Paul Brezinski, MBA, Tomas Vaughn, Ph.D., Gary Rosenthal, M.D. Presented By: Fredric Wolinsky, Ph.D., John W. Colloton Chair, Health Management and Policy, The University of Iowa, 200 Hawkins Drive, E-205GH, Iowa City, IA 52242; Tel: 319384-5129; Fax: 319-384-5125; Email: fredric-wolinsky@uiowa.edu Research Objective: Millions of veterans are eligible to use the Veterans Health Administration (VHA) and Medicare because of their military service and age. This paper examines whether an indirect measure of such dual use increases the risk of mortality. Study Design: Data from the Survey of Assets and Health Dynamics among the Oldest Old (AHEAD) were linked to Medicare claims and the National Death Index. Dual use was indicated when the self-reported number of hospital episodes in the 12 months prior to baseline was greater than that observed in the Medicare claims. The independent effect of dual use on mortality was estimated using proportional hazards regression. Population Studied: Nationally representative sample of 1,566 self-responding men who were 70 years old or older at the time of their baseline interviews in 1993. Principal Findings: 97 (11%) of the veterans were classified as dual users. 817 (52%) men had died by December 31, 2002, including 67% of the dual users and 51% of all others (p < .001). Adjusting for demographics, socioeconomics, comorbidity, hospitalization status, and selection bias at baseline, as well as subsequent hospitalization for ambulatory care sensitive conditions, the independent effect of dual use was a 61.8% increased risk of mortality (AHR =1.618; p = .02). Conclusions: An indirect measure of dual use of VHA and Medicare services in veterans was associated with an increased risk of death. Implications for Policy, Delivery, or Practice: Dual use is significantly associated with increased mortality, and the likely etiological mechanism involves poor coordination between the VHA and Medicare health care systems. Although the focus of this paper was on dual use of the VHA and Medicare at the system level, the real problem is the increasingly fragmented and uncoordinated nature of health care at all levels of delivery. Therefore, improvement of communication tools that facilitate better patient care management under a single primary care practitioner is warranted. Primary Funding Source: NIA ●The Rural Safety-net: Collaborations between Rural Community Health Centers and Critical Access Hospitals Sudha Xirasagar, MBBS, Ph.D., Michael Samuels, DrPH Presented By: Sudha Xirasagar, MBBS, Ph.D., Research Assistant Professor, Health Services Policy and Management, University of South Carolina, Arnold School of Public Health, 800 Sumter St., Columbia, SC 29208; Tel: (803) 576-6093; Fax: (803) 777-1836; Email: sxirasagar@sc.edu Research Objective: Because rural areas have limited healthcare resources, existing institutions need to coordinate and complement one another’s services to maximize their impact. This study sought to: 1. Identify model collaborations between between rural Community Health Centers (CHC) and critical access hospitals (CAH), and the barriers and facilitators for collaboration. 2. Assess the prevalence and functionality of collaborative relationships at a national level Study Design: Qualitative study of five model collaborations through site visits (in 2003), followed by national survey of eligible CAHs (in 2004) to assess the prevalence and functionality of collaborations, as well as directions for policies concerning improving rural health safety networks. Population Studied: Phase 1: Site visits of five model CHCCAH collaborations identified by directors of state primary care associations, state offices of rural health, and state hospital associations. Phase 2: Survey of CEOs of all 386 CAHs in the US with a CHC within a 60-mile radius (n=161, response rate 41.6%). Principal Findings: Qualitative: 1. Movement towards an integrated delivery system model 2. Service consolidation: Mutual cross-referral, sharing physicians, shared laboratory and radiology services and MIS improved the continuum of care, prevented duplication of services, and saved costs 3. Shared HR practices including recruiting, credentialing, joint training, and medical malpractice insurance for mutual benefit 4. Mutual leveraging of strengths for jointly realizing benefits of HRSA’s 340-B pharmacy program. 5. Lack of competition favors collaboration 6. Major barriers to collaboration: Competition, Lack of trust, Different corporate cultures, Preoccupation with routine responsibilities, Conflicting regulations. 7. Major facilitators: Vision about safety-net role; Funded, separate entity to establish the collaboration; Agree to compete on certain aspects while collaborating on others; Shared patient information system facilitates long term collaboration. National survey findings: CAH responses to survey: 161/386 (41.7%) Report collaborations: 24 (14.9% of respondents, or 6.2% of total CAHs with a CHC within 60 miles) Service areas: 15 radiology, 13 specialty care, 12 laboratory, 9 primary care, 7 pharmacy, 6 wellness programs Satisfaction with collaboration: - Clinical: Primary Care 100% Wellness Programs 50% - Human Resources: Training 90.9%` Shared Staff 72.7% - Administrative: Malpractice Insurance 80% Quality assurance 80% Reasons for lack of collaboration (n=137 CAHs without collaboration): - Unaware of CHC in vicinity – 20.4% - No proposal from CHC – 40.9% - Not considered it – 37.2% - Other priorities – 39.4% Barriers to collaboration (n=24 with collaboration): - Competition – 58.3% - Different corporate cultures – 50% - Lack of trust – 45.8% Facilitators for collaboration: Seed money – 48.9% Funded unit – 38.7% Funded facilitator – 36.5% Conclusions: Lack of awareness/information is a key reason for lack of CAH-CHC collaborations, despite HRSA’s FLEX program emphasis on rural networks. Lack of funding mechanisms to establish the collaboration is the next major reason Implications for Policy, Delivery, or Practice: Policy recommendations - National/state level conferences and meetings between CHCs and CAHs - Expert-driven workshops (national and state level) drawing on expertise from model partnerships - Funding mechanisms to facilitate collaborative relationships Future research - Cost-benefit and effectiveness studies of functional areas of collaboration - Comparative study of “corporate culture” differences between rural CHCs and CAHs - Evaluate barriers in CHC and CAH legislation and program regulations. Primary Funding Source: National Rural Health Association ●The Great Divide in Access to Health Care in Suburban and Inner City Communities: Assessment of Hospitalizations for Ambulatory Care Sensitive Conditions Gary Young, MA, Jo Anne Romano Presented By: Gary Young, MA, Executive Vice President, Strategic Planning and Government Relations, Cooper University Hospital/Robert Wood Johnson Medical SchoolCamden, One Cooper Plaza, Camden, NJ 08103; Tel: 856-3423002; Fax: 856-342-3299; Email: Younggary@cooperhealth.edu Research Objective: To assess the level of access to primary health care services and the impact of the health care delivery system at work in inner city and suburban environments by an empirical study of age- and sex-adjusted rates of hospitalization for ambulatory care sensitive conditions. Study Design: A post-hoc, population-based study focused on a comparative analysis of UB-92 hospital abstract data (20002003) for Camden, New Jersey—the second poorest city in the United States—with the southern New Jersey region focusing on age and sex-adjusted rates of hospitalizations for 19 ambulatory care sensitive conditions (preventable, rapid onset, and chronic conditions). The study replicates the methodology for studying ambulatory care sensitive conditions employed in McClellan, M.D., et al., Refinement of the Healthcare Cost and Utilization Project (HCUP) Quality Indicators- AHRQ, 2001. Rates of hospitalization are presented for age/sex cohorts and odds-ratios are utilized to present statistically meaningful variations in rates of hospitalization for Camden City residents as compared with the suburban southern New Jersey region. Population Studied: Hospitalization abstract data for residents of the City of Camden (pop. 75,000) and the seven counties of southern New Jersey (1.72 million). Principal Findings: 17% of the hospitalizations for the City of Camden were found to be for preventable, ambulatory care sensitive conditions. The payer mix of these hospitalizations were predominately Medicaid, Managed Medicaid and charity care, while the cost of this care was estimated to be $25MM. Hospitalizations for ambulatory care sensitive conditions ranged from 20% to 360% higher in the City of Camden than in the suburban New Jersey region, with the highest rates associated with diabetes, asthma, hypertension. Other findings presented detail specific conditions and the population cohorts most at risk in the different communities. Conclusions: Data suggests that there is a fundamental breakdown in the health care delivery system in the City of Camden with clear evidence of a lack access to quality primary care and management of chronic conditions and primary care follow-up. Implications for Policy, Delivery, or Practice: A number of implications for policy are developed in the paper including the following: Medicaid, managed Medicaid, and New Jersey’s system of charity care reimbursement needs to be restructured to incentivize physicians to provide greater access to primary care and disciplined follow-up of chronic health conditions. Hospitals need to leverage their resources to create alternative health care delivery mechanisms focused on increased access to primary care and disease management. Primary Funding Source: No Funding ●Access to Eye Care in the United States, 2002 Xinzhi Zhang, M.D., Ph.D., Sanjat Kanjilal, M.P.H., Michael R. Duenas, OD, K.M. Venkat Narayan, M.D., Jinan B. Saaddine, M.D., M.P.H. Presented By: Xinzhi Zhang, M.D., Ph.D., Senior Service Fellow, , Centers for Disease Control and Prevention, 4770 Buford Hwy, N.E. (K10), Atlanta, GA 30341; Tel: 770-488-4877; Fax: 770-488-1148; Email: xbz3@cdc.gov Research Objective: Approximately 1 in 28 Americans older than 40 years have visual impairment and many more have asymptomatic eye diseases that need eye care services. Access to eye and vision care services has become a major public health concern. We aim to estimate the level of access to eye care, and explore barriers to eye care in the nation. Study Design: We used the 2002 National Health Interview Survey (NHIS), an annual cross-sectional population-based survey of households across the United States. We analyzed data from 30,920 adults aged 18 and older using a modified well-established behavioral access model. We evaluated indicators of both potential access and realized access and their relationship. Indicators of potential access to eye care included age, sex, marital status, race/ethnicity, education, income, health insurance, region, diabetes, and vision problems. Indicators of realized access to eye care included visiting eye doctor in the past 12 months, having a dilated eye exam in the past 12 months, and the inability to afford eyeglasses when needed. We used Multivariate logistic regressions for the three dichotomous realized access variables to estimate odds ratios (OR), conditional margins (CM), and predictive margins (PM). All analyses were weighted to adjust for the complex sample design and to make estimates representative of the U.S. civilian noninstitutionalized population. Population Studied: U. S. Population Principal Findings: On average, 60% (95%CI: 58%-61%) of people with vision problems visited eye doctor in the past 12 months and 59 % (95%CI: 57%-60%) had dilated eye exam in the past 12 months, compared to 32% (95%CI: 31%-33%) and 34% (95%CI: 33%-35%) of people without vision problems respectively. Additionally, 11.1% (95%CI: 10.1%-12.0%) of people with vision problems cannot afford eye glasses compared to 4.3% (95%CI: 4.0%-4.6%) of people without vision problems. Controlling for all other potential access variables in the model, the probability of visiting eye doctor in the past 12 months and the probability of having a dilated eye exam in the past 12 months increased with age, education, and income (P <0.01). Moreover, having health insurance, being female, having diabetes, or having vision problems increased the probability to use eye care services for each (P <0.01). Inability to afford eye glasses was significantly associated with being unmarried, being female, having low income, without insurance, having diabetes, or having vision problems (P <0.01). Conclusions: There is inequity of access to eye care in the United States. These findings will help target limited resources, identify new interventions, and set a baseline in the effort to eliminate disparities in access to eye and vision care. Implications for Policy, Delivery, or Practice: Conditions causing vision impairment and blindness are often asymptomatic in the early and treatable stages. Populationbased survey is an effective way to systematically monitor and evaluate access to eye care. Public health policy needs to be developed to increase access to eye care, allocate scarce recourses, and achieve a population with healthy vision in the U.S. Primary Funding Source: No Funding ●Uninsured Families Erika Ziller, MS, Andrew F. Coburn, Ph.D., Nath Anderson, M.P.H. Presented By: Erika Ziller, MS, Research Associate, Muskie School of Public Service, University of Southern Maine, PO Box 9300, Portland, ME 04104-9300; Tel: (207) 780-4615; Email: eziller@usm.maine.edu Research Objective: Although numerous studies have examined the prevalence and characteristics of the U.S. uninsured, research into the coverage of entire families has been limited. The purpose of this study was to examine health insurance coverage at the family level, to determine the extent to which U.S. households are fully covered by health insurance, fully uninsured, or have a mix of insured and uninsured members. Among partially uninsured families, we also identified the source(s) of coverage for insured members, including Medicare, Medicaid/SCHIP, private insurance, and a public/private mix. With these analyses we offer critical new information about the extent of uninsurance within U.S. families. Study Design: Employing a combination of bivariate and multivariate techniques, we use a pooled sample of the 2001 and 2002 Medical Expenditure Panel Survey (MEPS) to estimate family health insurance coverage rates, and to identify the characteristics associated with different family coverage mixes. To address known sampling errors in the MEPS, we apply family-level weights and use statistical techniques to adjust for data clustering. Using multinomial logit (MNL) modeling, we estimate the likelihood of being fully or partially uninsured, compared to fully insured, based on family characteristics. Among partially uninsured families, our MNL models measure the relationship between sociodemographic and employment characteristics and source(s) of coverage among insured family members, with fully uninsured families as the referent. Population Studied: We examined the family insurance coverage status of a nationally representative sample of U.S. households with two or more members, based on the Current Population Survey (CPS) definition of family. Because of nearuniversal Medicare coverage of the senior population, we excluded families where every member was 65 or older. Principal Findings: While 17.5% of U.S. residents under age 65 were uninsured in 2002, we found that 30% of all nonelderly families had an uninsured member. This difference can be explained by the fact that the bulk of uninsured families have at least one member with health insurance coverage. In only 7% of U.S. families is every member uninsured. The characteristics of uninsured families are generally comparable to those of uninsured individuals (lower income and education, poorer health, unmarried, racial/ethnic minority). However, we find some differences in the characteristics associated with being fully versus partially uninsured. For example, partially uninsured families are much more likely to be headed by someone who has retired or is out of the labor force, compared to fully uninsured families. Among families with at least one uninsured member, only 22% are fully uninsured, while 78% have at least one member with health insurance coverage. The most common source of coverage is private insurance (35% of uninsured families), followed by Medicaid or SCHIP (19%). Fourteen percent of uninsured families have at least one member with Medicare, while 10% have a blend of public and private coverage. Multivariate modeling is underway to determine the independent contribution of specified characteristics to a family's predicted insurance coverage mix. Conclusions: Our findings on family insurance coverage suggest that the uninsured problem in the United States is understated if we only measure coverage at the individual level. While fewer than 20% of non-elderly individuals are uninsured, nearly one in three families has at least one member that lacks coverage. The most common source of coverage for partially uninsured families is private insurance, although this group represents only one-third of all uninsured families. Further research is needed to determine if uninsured members of these families are ineligible for this private coverage, or if the cost of dependent coverage is prohibitively high. Additionally, the relatively high rate of Medicaid/SCHIP coverage among uninsured families (19%) suggests that expanding coverage for Medicaid-eligible individuals to their entire families could substantially reduce the uninsured population. Implications for Policy, Delivery, or Practice: In an era where comprehensive health reform is unlikely, this study will assist policymakers interested in incremental strategies for reducing the uninsured. By understanding within-family patterns of health insurance coverage, policymakers can develop expansion strategies that build on current health insurance arrangements (e.g. by subsidizing dependent coverage or expanding public coverage to families). In this way, resources can be targeted so that they achieve the greatest coverage increases while supporting the development of cohesive family insurance units. Primary Funding Source: HRSA, Office of Rural Health Policy