Coverage & Access Call for Papers The Uninsured Chair: Kenneth Thorpe, Ph.D. Sunday, June 6 • 11:30 a.m.-1:00 p.m. • Case Management of Uninsured Emergency Department Patients: Results from an Economic Analysis of a Randomized Controlled Trial Susan Busch, Ph.D., Sarah Horwitz, Ph.D., Katie Balestracci, Ph.D., Jim Rawlings, M.P.H. Presented by: Susan Busch, Ph.D., Assistant Professor, Health Policy, Yale Medical School, 60 College Street, New Haven, CT 06520; Tel: 203.785.2927; Fax: 203.785.6287; E-mail: susan.busch@yale.edu Research Objective: To determine the economic costs and benefits of an intensive case management intervention. Study Design: Data comes from a randomized controlled trial which recruited patients visiting the Yale-New Haven Hospital ED during the period 4/22/02 to 7/21/02. Two hundred thirty six patients (123 intervention, 113 comparison) were enrolled. On intervention shifts all patients > 18 years who had no insurance and no primary care provider were recruited into the study. The patient was assisted in choosing one of 4 health care providers and the patient’s contact information was faxed to the case manager at the selected site. The case managers attempted to contact intervention patients within 72 hours. On comparison shifts care as usual took place. All comparison patients left the ED with a physician’s recommendation for follow up and an appropriate number to call. Population Studied: Uninsured Patients visiting the Yale-New Haven Hospital Emergency Department Principal Findings: In the six months post intervention, the intervention group had more primary care contacts, fewer inpatient admissions, but similar numbers of ED visits (Horwitz et al, 2003). Examining the economic implications of these changes we find that although total ED costs were similar in two groups, the net cost per ED visit was lower in the intervention group compared to the comparison group ($245 versus $312), suggesting these visits involved less intense care. Further, examining the per person net hospital costs for both groups, the intervention group was less expensive ($63 versus $242, p=.09). Although hospital costs were reduced, case management costs were substantial ($233 per intervention subject). Conclusions: An ED referral/intensive case management intervention can result in reduced patient care costs. Although the number of subsequent ED visits is similar in both groups, the intensity of these visits is less for the intervention subjects. Net inpatient costs were substantially reduced in the intervention group. Thus, this intervention has the potential to reduce health care costs. The short and long term effects on health have yet to be analyzed, but may be substantial. Implications for Policy, Delivery or Practice: Increases in the number of uninsured in the US coupled with the shrinking number of safety net providers have combined to make Emergency Departments (EDs) the care provider for increasing numbers of adults. ED care is expensive, often overly invasive and rarely able to attend to prevention. The economic benefits of case management interventions designed to establish medical homes for uninsured individuals visiting the ED have not been rigorously evaluated. This study provides evidence that hospitals and others can use to assess the feasibility of these programs. Primary Funding Source: HRSA • The Pent-up Demand for Health Care of the Uninsured Near Elderly When They Are Approaching Age 65 Li-Wu Chen, Ph.D., Wanqing Zhang, M.Ed., Phani Tej Adidam, Ph.D., Louis Pol, Ph.D., Keith Mueller, Ph.D., Dennis Shea, Ph.D. Presented by: Li-Wu Chen, Ph.D., Assistant Professor, Preventive and Societal Medicine, University of Nebraska Medical Center, 984350 Nebraska Medical Center, Omaha, NE 68198-4350; Tel: 402.559.7113; Fax: 402.559.7259; E-mail: liwuchen@unmc.edu Research Objective: To examine whether the pent-up demand for health care utilization exists for the uninsured near elderly who are approaching age 65, at which they will be eligible for Medicare enrollment. Study Design: Using a difference-in-difference model and logistic regression approach, we estimated and compared the likelihood of increasing physician visits and hospital inpatient stay from pre-Medicare period (i.e.,ages 63-65)to postMedicare period (i.e.,ages 65-67) between the uninsured near elderly and the insured near elderly. In addition, to address the possible self-selection bias related to the purchase of insurance, we tested whether the endogeneity of insurance coverage may exist for the observed insurance status variable at age 63. Population Studied: A sample of 640 near elderly people were obtained from Wave 1 to Wave 5 of the Health and Retirement Study (HRS). Longitudinal information related to health care use, insurance coverage, demographic, socioeconomic,health status, and life style of the study sample was used in the regression analysis. Principal Findings: Being uninsured (as opposed to being insured) at age 63 increases the likelihood of using 200% more or above of physician care after Medicare enrollment. On the other hand, being uninsured at age 63 does not increase the likelihood of using increasing hospital inpatient care after Medicare enrollment. Conclusions: Pent-up demand for physician care exists for the uninsured near elderly as compared with their insured counterparts. However, such phenomenon does not exist for the hospital inpatient care of the near elderly. One possible reason for this discrepancy may be that hospital inpatient care, in nature, is more acute and urgent when it is needed, thus reducing the chance of postponing care until the near elderly are covered by Medicare. Implications for Policy, Delivery or Practice: The delay in receiving physician care of the uninsured near elderly until they are covered by Medicare may impose a significant financial burden on Medicare program. With the possibility of developing a worsening health condition due to the postponement of care, the financial burden may become even larger in the long run when these previously uninsured Medicare beneficiaries are aging. Therefore, providing insurance coverage to the uninsured near elderly, especially those underserved people, should be a critical issue to consider when designing Medicare reform policies. Primary Funding Source: Economic Research Initiative on the Uninsured • The Accessibility of Specialty Care for California's Uninsured Suzanne Felt-Lisk, M.P.A., Megan McHugh, M.P.P., Melissa Thomas, M.A. Presented by: Suzanne Felt-Lisk, M.P.A., Sr. Health Researcher, Mathematica Policy Research, 600 Maryland Avenue, S.W., Suite 550, Washington, DC 20024; Tel: 202.484.4519; Fax: 202.863.1763; E-mail: SfeltLisk@Mathematica-mpr.com Research Objective: Many studies of the safety net have identified specialty care access as a major problem, but knowledge of the extent and nature of this problem has been limited. The objective of this study was to better understand the accessibility of specialty care for the uninsured in California, where they obtain specialty care, and how and why access varies across communities. Study Design: (1) Survey of all 101 FQHC medical directors in California regarding access to specialty care for uninsured patients across different specialties and where they refer patients for specialty care (response rate 76%); (2) survey of hospital outpatient department directors (from hospitals identified as referral organizations by the FQHC medical directors) regarding hospital policies for serving uninsured patients, wait times for appointments, and motivations for providing charity care (response rate 48%); and (3) case studies of four communities to identify barriers to care and learn why some communities are better able than others to meet the demand for specialty care from the uninsured; these included focus groups in each community with uninsured patients who have needed specialty care. Population Studied: Uninsured FQHC patients in California needing specialty care and the organizations that provide care to those patients. Principal Findings: Access problems for uninsured FQHC patients extend across a wide range of specialties, and are worse for adults than for children. Obtaining specialty care appointments typically requires time-intensive case-by-case efforts by primary care providers, adding cost to the safety net system. Hospitals are the major source of specialty care for uninsured FQHC patients; about half of those hospitals named by FQHC medical directors as key referral destinations focus their business on low-income populations and about half do not. In the case study communities, weak relationships between FQHCs and hospitals; lack of community support; having a large, geographically dispersed uninsured population; and having shortages of physician specialists were identified as barriers to access. Community efforts to address the problems include providing selected specialty services at the FQHCs, and covering more uninsured children through county-based programs. Conclusions: Access to specialty care is a major problem for uninsured Californians, particularly adults. Community efforts to address the problems seem poised to “chip away” at them but leave major gaps. Implications for Policy, Delivery or Practice: If it could be captured, the cost of time providers spend arranging specialty care for uninsured patients now could help offset the cost of standing formal referral arrangements or insurance coverage. Efforts to expand insurance coverage could have the added benefit of making it easier for the remaining uninsured to access care since providers appear to be more willing to provide charity care when the size of the uninsured population is small. State and federal funding streams that support uncompensated care are policy levers that could be fine-tuned to better counteract lack of community support for improvement at the local level. Primary Funding Source: California Healthcare Foundation • The Effect of Employment-Based Health Insurance on Wages and Returns to Tenure: Evidence that High- and Low-Wage Workers Differ Patricia Ketsche, Ph.D., M.H.A., William Custer, Ph.D. Presented by: Patricia Ketsche, Ph.D., M.H.A., Assistant Professor, Institute of Health Administration, Georgia State University, 33 Gilmer Street, S.E., Unit 4, (MSC 4A1473), Atlanta, GA 30303-3084; Tel: 404.651.2993; Fax: 404.651.1230; E-mail: pketsche@gsu.edu Research Objective: Employment-based insurance is credited with spreading risks but has been criticized because it limits choice of plans for most workers and fails to cover a large number of Americans. The extent to which risks are actually pooled depends upon how individual wages adjust for the costs of health insurance. These compensating differentials could undermine risk pooling if individual workers experience substantial wage penalties based on health risk. This paper contributes to the limited literature by demonstrating that adjustments to wages differ for different groups of workers. Study Design: Data are from the March CPS combined with the February Job Tenure and Occupation Mobility Supplements from 1996 and 1998. These data permit analysis of worker salary controlling for specific job tenure. Like other researchers we estimate the effect of coverage on wages using a dummy variable for the presence of health insurance. However, given that demand for health insurance is heavily dependent upon income and that low-wage workers face a different after tax price for coverage, analysis is performed separately for workers with wages above- and below the median for their geographic area. We first estimate the effect of coverage on wages and the differential effect of coverage on returns to tenure using a reduced form. We compare these results to a two part model that estimates the probability of coverage among workers as a function of worker demographic and firm characteristics. This estimated probability is used in a wage regression that includes variables that reflect individual productivity and firm characteristics. Identification is achieved through inclusion of state policy variables that are likely to have a strong effect on the price of coverage such as community rating requirements, guaranteed issue laws and the presence of risk pools in the coverage equation. Finally we estimate coverage and wages in a nonlinear OLS simultaneous equation model. Principal Findings: In each of the models estimated, wage adjustments for coverage and effect of coverage on returns to tenure are different in sign for high- and low-wage earners. Low wage workers with coverage experience increased wages but negative returns to tenure. Among workers with wages above the median, coverage depresses or has no effect on wages, depending upon model specification, but increases returns to tenure. Conclusions: In the short run, high-income workers with health insurance may bear a larger share of the cost of that coverage than do low-income workers with health insurance. However, the effect of coverage on worker retention may be different for high- and low-wage workers, with low-wage workers with coverage experiencing a greater degree of “joblock”. Implications for Policy, Delivery or Practice: If high-wage workers value coverage sufficiently to bear a greater share of the cost in terms of wage reductions, social welfare is affected by expanding coverage to those who otherwise might not purchase coverage. However, low-wage workers who value coverage may bear the cost of coverage through a reduction in wage increases associated with tenure. • The Demise of Oregon’s Medically Needy Program: Effects of Losing Prescription Drug Coverage and Pharmaceutical Company Drug Assistance Programs Judy Zerzan, M.D., M.P.H., Tina Edlund, M.S. Presented by: Judy Zerzan, M.D., M.P.H., Assistant Professor, Internal Medicine, Oregon Health & Science University, 3181 S.W. Sam Jackson Park Road, OP30, Portland, OR 97239; Tel: 503.768.9106; E-mail: zerzanj@ohsu.edu Research Objective: In January 2003, people covered by Oregon’s Medically Needy (MN) program lost their benefits due to financial shortfalls in the state budget. The MN program is a federally-matched optional Medicaid program in which states may chose to provide Medicaid coverage and/or Medicare premium assistance to certain groups not otherwise eligible for Medicaid with significant health care needs. This population is not well characterized either nationally or locally. In Oregon the MN program mainly provided a prescription drug benefit. The primary objective was to determine how the loss of benefits has affected this population's health. Secondary objectives include investigating how this population is getting their prescription medications and changes in use of prescriptions and health care. Most pharmaceutical manufacturers offer some of their medications free for low-income people through drug company assistance programs. These programs vary widely in the application process, drugs available and amount of drug supplied. After the loss of the MN program, many turned to drug company programs to obtain needed medications. This survey also provides information about use of these programs in this population. Study Design: A 49-question survey instrument was created to collect information about this population including demographic information, health insurance coverage, health conditions, access to care, prescription drug use, utilization of health care, and use of drug company assistance programs. A data collection contractor telephoned a sample of this population obtaining an overall completion rate of 58%. Medication use during the MN program was obtained from administrative data and current medication use was obtained from the interview. Results of this survey were compared to the Medicare Health Outcomes Survey and the National Health Interview Survey. Principal Findings: The 439 people surveyed included 36% men and 64 % women from age 21-91 with 70-75% at less than 133% of the federal poverty level. 2/3 of respondents rate their current health as poor or fair and compared to last year 44% rate their health as worse and 39% about the same. Only 7.5% currently have prescription coverage. In the six months since the MN program ended 61% have skipped doses of a medication and 64% have gone without filling a prescription. In order to pay for medications 60% of respondents have cut back on their food budget, 47% have borrowed money and 49% have skipped paying other bills. There was no significant difference in ER visits and a small statistically significant decrease in hospitalizations by self-report comparing the six months before losing the MN program and the six months after. Over half of respondents are aware of drug company assistance programs and currently 45% of those surveyed use these programs with most getting application assistance from a doctor’s office or clinic. 55% of people report using these programs is hard and 37% find it easy. Only 12% of respondents are very confident they can continue to use these programs, 27% are somewhat confident, and 52% are not at all or not very confident. 82% of people currently using these programs are only obtaining some of their drugs from this source because not all of their prescribed drugs are available. Drug use has shifted to more name brand drugs since these programs provide newer medications. Of the 55% who are not currently using drug company assistance programs, most have applied before or used these programs in the past. Conclusions: The MN program provided coverage for a lowincome, chronically ill population. Since its termination, there have been significant changes in prescription drug use and financial impact in their daily lives. This population is sicker with worse perceived health status than the general population over 65 and people eligible for both Medicare and Medicaid. As states make program changes, the effects on vulnerable populations must be considered. Losing prescription coverage has immediate impact on use of prescriptions and finances in this population. Drug company assistance programs are one way for people to obtain needed medications. About half of the MN population is now using drug company prescription assistance programs, but this does not appear to be a sustainable mechanism for obtaining prescription medications because it can be difficult to apply and not all medications are covered. With these programs people are using more brand name medications instead of generics which may have implications for shifting future prescription use. Primary Funding Source: Oregon Health Policy & Research via a RWJ State Coverage Initiatives Grant Call for Papers Health Insurance Changes Chair: Thomas Buchmueller, Ph.D. Sunday, June 6 • 3:00 p.m.-4:30 p.m. • Effects of an Economic Boom on Health Insurance Status Linda Blumberg, Ph.D., Lisa Dubay, Sc.M. Presented by: Linda Blumberg, Ph.D., Senior Research Associate, Health Policy Center, The Urban Institute, 2100 M Street, N.W., Washington, DC 20037; Tel: 202.261.5769; Fax: 202.223.1149; E-mail: lblumber@ui.urban.org Research Objective: This study assesses the effects of an economic boom on the health insurance coverage of modest income adults using the 1996 Survey of Income and Program Participation (SIPP). Our objective is to identify whether those whose incomes rose substantially over the period also benefited by gaining health insurance, or whether their coverage status was not affected significantly compared to those whose incomes were more stable. Study Design: Using a multivariate framework, we contrast observed changes in coverage over the course of the panel for adults who benefited from the economic expansion (those in families with significant increases in income between the first and last waves of the SIPP panel) with those for adults who did not benefit from the expansion. The specification of the multivariate models will take the form shown below: Probability(coverageiL = µ | coverageiF = unins) = a1 + a2 boomi + a3 individuali + a4 familyi + a5 regioni, where coverageiL = person i’s insurance coverage (private coverage, SCHIP coverage, or uninsurance) at the last wave of the survey; coverageiL = person i’s insurance coverage (private coverage, SCHIP coverage, or uninsurance) at the last wave of the survey; boomi = indicator for whether the individual is a non-elderly adult whose family income increased significantly between the 1st and last waves of the panel -- this is our key explanatory variable; individuali = vector of variables describing the characteristics of the individual (e.g., race, gender, health status , age); familyi = vector of explanatory variables depicting characteristics of the individual’s family (e.g., family type, number of working adults in the unit, family income in first wave, education of the high earner); regioni = vector of indicators for the region of residence. For example, a positive and significant coefficient on "Boom" in a regression where last wave coverage is private, implies that the those uninsured adults affected monetarily by the economic boom were more likely than their counterparts who were not directly affected to gain private insurance coverage. In other words, the boom not only increased incomes, it also increased private insurance coverage. We will also identify, descriptively, the characteristics of those with large income gains who remain uninsured. We will assess how they differ from those whose incomes increased but who also gained coverage. We will describe the characteristics of those who gained coverage but whose incomes did not increase during the period. Population Studied: We use a sample of non-elderly adults with first survey wave family incomes at or below 300 percent of poverty, focusing on those adults who were uninsured at the first interview. Principal Findings: Analysis has not yet been completed. Conclusions: Pending completion of the analysis. Implications for Policy, Delivery or Practice: Studying the effects of an economic boom not only gives us insight into whether such economic cycles benefit particular types of individuals, but it also assists us in understanding the potential ramifications of the reverse situation – an economic recession. If an economic boom significantly increases coverage among those who benefit most directly from it, one could also reasonably expect that a contraction of the economy would lead to a decline in coverage for those affected. Primary Funding Source: RWJF, The Urban Institute • Increasing Health Insurance Premiums and the Decline in Insurance Coverage Michael Chernew, Ph.D., David Cutler, Ph.D., Patricia Keenan, M.H.S. Presented by: Michael Chernew, Ph.D., Associate Professor, University of Michigan School of Public Health, Department of Health Management and Policy, 109 S. Observatory Street, Ann Arbor, MI 48109-2029; Tel: 734.936.1193; Fax: 734.764.4338; E-mail: ggbz@umich.edu Research Objective: This paper examines the determinants of declining insurance coverage during the 1990s, with a focus on the role of rising health care premiums, relative to other explanations such as changes in employment patterns or state policies. In contrast to substantial media coverage linking rising premiums to declines in coverage rates, empirical evidence quantifying the extent to which higher premiums deter coverage is limited. Study Design: We take advantage of wide geographic variation in changes in premiums and coverage rates to estimate the association between rising local health insurance costs and the falling propensity for individuals to have any health insurance coverage. We focus on coverage from any source (private or public), because some individuals may switch their source of coverage rather than become uninsured. Using probit regression models and instrumental variable techniques, we analyze changes in coverage between two periods, 1989-1991 and 1998-2000, using the March Current Population Survey (CPS). We estimate the impact of changing health care costs, tax subsidies, Medicaid reforms, other state regulatory reforms, a rise in spousal employment and general economic conditions on declining coverage, controlling for changes in population demographics. Our contributions, relative to the existing literature, include more detailed controls for non-premium explanations for declining coverage, focus on any coverage (as opposed to employer sponsored coverage), use of geographic variation in cost growth (as opposed to national trends in cost growth that may be confounded by other factors), a broader measure of health care costs, and recognition of the endogeneity of cost growth. Population Studied: The non-elderly population of the United States residing in 64 MSAs. Principal Findings: Over half of the decline in coverage rates experienced over the 1990s is attributable to the increase in health insurance premiums (1.6 percentage points of the 3.1 percentage point decline). Medicaid expansions led to a 1 percentage point increase in coverage. Changes in economic and demographic factors had little net effect. Effects of rising premiums on coverage declines are greater for individuals in families with lower as opposed to higher education, and for younger as opposed to older adults. Conclusions: As other studies have found, changing demographics and employment patterns explain little of the decline in coverage. Instead, much of the decline in coverage over the 1990s appears to be related to the rise in health care premiums. Implications for Policy, Delivery or Practice: The uninsured population is likely to increase further if health care cost growth continues to exceed income growth. Initiatives aimed at reducing the number of uninsured must confront the growing pressure on coverage rates generated by rising costs. At the same time, while policy has traditionally focused on the need to limit the growth of costs, this may not be desirable if medical cost increases are buying valuable services. A careful examination of how to encourage optimal design of insurance policies and delivery of medical care is warranted. Primary Funding Source: RWJF, RWJF's Economic Research Initiative on the Uninsured, National Institute on Aging • The Response of Small Business to Variation in the Price of Insurance: Results from a Randomized Controlled Trial Richard Kronick, Ph.D., Louis Olsen, Ph.D., Todd Gilmer, Ph.D. Presented by: Richard Kronick, Ph.D., Professor, Department of Family and Preventive Medicine, University of California, San Diego, UCSD School of Medicine, 9500 Gilman Drive, La Jolla, CA 92093-0622; Tel: 858.534.4273; Fax: 858 534.4642; Email: rkronick@ucsd.edu Research Objective: To determine what fraction of small businesses that do not currently offer insurance would purchase coverage when offered a 50% premium subsidy, and how that fraction varies as the subsidy is varied between 20% and 80% of premium. Study Design: A randomized trial conducted in San Diego. Businesses were sampled from a Dun&Bradstreet listing of small businesses, and randomized into 25 treatment groups. Each group was offered Sharp Health Plan, a San Diego-based HMO, and the price to employers and employees was experimentally varied. Subsidies for the difference between the full premium and the price offered the business were provided by the California HealthCare Foundation. Population Studied: Small businesses and their employees in San Diego. Principal Findings: Subsidized coverage was offered to 345 businesses that were eligible to purchase, and 71 of these businesses bought the product. The average subsidy was approximately 50% of premium; thus, at a subsidy of 50% of premium, approximately 20% of businesses purchased. As the subsidy was varied from 20% to 80% of premium, the percent purchasing varied from 10% to 40%. Conclusions: Small businesses that do not currently offer insurance are not very responsive to variations in the price of coverage. Even at an 80% subsidy, only 40% of businesses purchase coverage. It appears that many businesses operate satisfactorily without offering coverage, and are unlikely to start offering even when offered very large subsidies. Implications for Policy, Delivery or Practice: Increasing insurance coverage by offering subsidized coverage to small businesses is not likely to be effective in significantly reducing the number of uninsured. Primary Funding Source: California HealthCare Foundation • Uninsured Decliners of Employer Sponsored Health Insurance: How They Changed from 1997-2003 Peter Cunningham, Ph.D., Len Nichols, Ph.D., Len Nichols, Ph.D. Presented by: Len Nichols, Ph.D., Vice President, , Center for Studying Health System Change, 600 Maryland Avenue, S.W., Washington, DC 20024; Tel: 202.484.5269; Fax: 202.484.9258; E-mail: lnichols@hschange.org Research Objective: To assess if more or different people need subsidies to purchase insurance than in the mid-1990s. Prior work showed about 20% of the uninsured declined an offer of employer sponsored insurance (ESI). If workers, especially higher income workers, are increasingly likely to decline offers, then proposals to expand coverage could benefit from current data and analysis, especially in this election year. The 4th round of the Community Tracking Household Survey will be completed in January of 2004, with weighted analysis files ready by March, in time to produce reliable analyses by June. This survey and study can yield nationally representative estimates of coverage, offer, take-up, access to care, individual, family and job characteristics for calendar 2003, as well as support a multivariate model explaining the decision to accept or decline ESI. Study Design: Data are based on the 1996-97 and the 2003 Community Tracking Study household surveys, both of which are representative of the nation as well as of 60 randomly selected sites. We initially present descriptive results comparing 1996-97 with 2003 data to test if the likelihood of declining ESI and remaining uninsured increased and if so, for whom. We are able to study decisions made by workers in families with more than one ESI offer, which is increasingly common in a workforce dominated by two-earner couples. Next we develop a multivariate model of ESI take-up using both survey years to test if the likelihood of declining has increased and if so, why and/or for whom. Population Studied: The civilian non-elderly population of the United States, with particular emphasis on worker decisions, but with family members accounted for and tracked as well. Principal Findings: Previous studies and preliminary analysis of 1996-97 and 2000-01 data indicate that take-up rates declined for the lower income population, and that lower income, younger, and somewhat less healthy workers and their family members are more likely to decline ESI and remain uninsured. The 2003 CTS survey just completed will be the first to be conducted with weaker labor markets. With higher premiums, larger employee premium shares, and sharply higher co-payments at the point of service in recent years, we hypothesize and will test whether the average income of decliners has risen since 1996-97 while controlling for all other relevant factors, including the availability of SCHIP coverage for any children in the household. Conclusions: Depend on findings, to be determined. Implications for Policy, Delivery or Practice: If declining ESI is no more likely in 2003 than in 1996-97, then the relative importance of income-based subsidies for private group insurance is unchanged as well. But if indeed rising health care costs have increased the likelihood and income levels of decliners who remain uninsured, then policy makers may reconsider the wisdom of policy proposals that only subsidize those without ESI offers. Our study could improve the design and effectiveness of any coverage expansion proposal by indicating which populations should be targeted most directly. Primary Funding Source: RWJF Call for Papers Effects of Cost-Sharing & Reimbursement Chair: Willard Manning, Ph.D. Monday, June 6 • 8:30 a.m.-10:00 a.m. • A Medicaid Buy-in Program’s Effects on Costs and Earnings Robin Clark, Ph.D., Karin Swain, M.S., William Peacock, M.S. Presented by: Robin Clark, Ph.D., Associate Professor, Center for Health Policy and Research, University of Massachusetts Medical School, 222 Maple Avenue, Chang Building, Shrewsbury, MA 01545; Tel: 508.856.4226; E-mail: robin.clark@umassmed.edu Research Objective: To evaluate New Hampshire’s Medicaid for Employed Adults with Disabilities (MEAD) program, a “buy-in” program which allows persons with disabilities to maintain Medicaid coverage after their earnings exceed typical maximum amounts. Almost half of all states have a Medicaid buy-in program in place for persons with disabilities and others are in the early stages of implementation. Study Design: We measured changes in Medicaid payments, disability determination costs, earnings and total income for the 985 persons who enrolled during the first year of the program. Changes in provider payments and other state costs were compared to those for a similar group of 1,000 Medicaid beneficiaries who did not enroll in the program. Population Studied: All New Hampshire Medicaid beneficiaries enrolled in MEAD at any time between February 1, 2001 and January 31, 2002 were included in the evaluation. In addition, a group of persons with disabilities matched with MEAD participants on gender, types of medical diagnoses, age and community residence were selected for purposes of comparison. MEAD participants averaged 43 years of age and were equally divided by gender. More than one-third had a severe psychiatric disorder. Principal Findings: Average monthly income for MEAD participants increased by $218 after enrollment. More than 84% ($184) of the gain came from additional earnings, which increased from $292 to $476 per month. The total earnings increase for MEAD participants from the year before to the year following enrollment was approximately $1.4 million. Despite higher than expected enrollment during the first year, preliminary analyses show MEAD cost the state less than onethird of the $1,353,441 predicted in a prior fiscal impact study. The actual net state cost of MEAD was $435,950. The difference between actual and predicted costs was primarily due to lower than expected Medicaid payments for participants without prior Medicaid coverage and smaller than expected increases in payments for beneficiaries previously classified as medically needy. Provider payments for MEAD participants with prior Medicaid enrollment increased at a slightly higher rate than those for comparison group members. MEAD participants generated $3.28 in new earnings for each additional state dollar invested in the program. Conclusions: These findings suggest that many persons with disabilities can increase earnings significantly if the fear of losing health insurance coverage is removed. Further, additional state costs associated with the program were modest compared with benefits realized by participants. Implications for Policy, Delivery or Practice: Although further research in a variety of settings is needed to confirm the generalizability of these findings, Medicaid buy-in programs show significant promise as a way of increasing employment and earnings for persons with disabilities at relatively low cost. Many states currently have options similar to New Hampshire’s MEAD program or are planning to implement them. The long-term effects on participants personal and financial health remain to be studied as do costs for local, state and federal government. Additional work is also necessary to understand better the characteristics of persons who benefit most from buy-in programs and the additional supports that may be necessary to sustain gains in employment. Primary Funding Source: CMS • The Effects of Medicaid Reimbursement on Access to Care of Medicaid Enrollees: A Community Perspective Peter Cunningham, Ph.D., Len Nichols, Ph.D. Presented by: Peter Cunningham, Ph.D., Senior Health Research, Center for Studying Health System Change, 600 Maryland Avenue, S.W., Washington, DC 20024; Tel: 202.484.4242; Fax: 202.484.9258; E-mail: pcunningham@hschange.org Research Objective: Previous research has found that while higher Medicaid reimbursement levels increase physicians’ acceptance of Medicaid patients, reimbursement levels have little direct effect on access to care as reported by Medicaid enrollees. This research takes a community perspective, and examines the extent to which Medicaid fee levels and other local factors affect enrollee access indirectly by influencing the supply of physicians in a community who are willing to accept Medicaid patients. Study Design: Data are based on the 2000-01 Community Tracking Study physician and household surveys, both of which are representative of the nation as well as for 60 randomly selected sites. The survey data are supplemented by state-level data on Medicaid reimbursement levels. Using multivariate analysis from the physician survey data, we first examine the effects of varying Medicaid fee levels on whether physicians’ practices are open to new Medicaid patients. We then examine the effects of fee levels and the relative number of physicians accepting Medicaid patients (measured at the site-level) on measures of access to care for Medicaid enrollees, using the CTS household survey. Measures of access to care include usual source of care, use of outpatient physician and hospital care, and self-reported difficulties in getting needed medical care. Population Studied: U.S. physicians involved in patient care, and Medicaid enrollees. Principal Findings: The relative number of physicians who accept Medicaid patients varies considerably across communities. Multivariate analysis shows that variation in Medicaid reimbursement levels across states contributes to community variations in physicians’ willingness to accept Medicaid patients, but that many other factors are also important. These include the mix of practice type and specialty in a community, population characteristics, Medicaid managed care penetration, insurance rate in the community, and geographic area. As a consequence, Medicaid reimbursement levels have little direct effect on access to care as reported by Medicaid enrollees. Rather, reimbursement levels have an indirect effect insofar as they influence (but do not determine) the relative number of physicians in a community that accept Medicaid patients. Findings show that Medicaid beneficiaries in communities with a relatively high number of physicians accepting Medicaid patients are more likely to have a usual source of care, less likely to have difficulty getting medical care, and have fewer hospitalizations. Conclusions: Medicaid fee levels are just one of many factors that determine the “supply” of physicians in a community willing to accept Medicaid patients. For Medicaid enrollees, it is the availability of physicians who accept Medicaid patients—and not fee levels per se—that affect their access to care. Therefore, the effects of Medicaid reimbursement on enrollee access are indirect, and other market factors may outweigh fee levels in determining access to medical providers. Implications for Policy, Delivery or Practice: Rapidly escalating Medicaid costs are contributing to state budgetary pressures and deficits, which has led most states to adopt a variety of cost-containment strategies for their Medicaid program, including cuts in fees to providers. The results of this study suggest that such cuts are not as cost-free with respect to enrollee access as previous research implies. However, modest provider payment cuts may be less harmful to beneficiary access than other cost-containment strategies, such as a reduction in eligibility or benefits. Primary Funding Source: RWJF populations: one with commercial and one with Medicare insurance. We compared the relative rates of deaths using Cox-regression and hospitalizations, ICU admissions, and ED visits using a poisson mixed-effects model applied to monthly utilization counts, while adjusting for covariates. For subjects with commercial insurance, we classified ED copayments into five categories: free care or no copayment, $1–9, $10–19, $20– 35, and $50–100 per ED visit. For Medicare subjects, we classified ED copayments into three categories: free care or no copayment, $1-15, and $20-50 per ED visit. Population Studied: There were 2,257,445 subjects with commercial insurance and 261,091 subjects with Medicare insurance at the beginning of the study. At baseline, 23.4% of commercial subjects paid nothing for each ED visit; 34.8% paid $1-5; 22.8% paid $10-15; 13.6% paid $20-35; and 5.4% paid $50-100. Similarly, 24.9% of Medicare subjects paid nothing per visit; 70.7% paid $1-15; and 2.5% paid $20-50. In January 2000, 52.1% of the commercial subjects and 60.8% of the Medicare subjects experienced an increase in their copayment level. Principal Findings: In subjects with commercial insurance, there were reductions in ED visits of 12% with the $20-35 copayment (RR=0.88, 95% CI: 0.87–0.89), and 23% with the $50-100 copayment (RR=0.77, 95%CI: 0.76–0.77). Despite this reduction in emergency care, there were no increases in the rate of hospitalizations, ICU admissions, or deaths, e.g. relative hospitalization rates of 0.96 (95%CI: 0.94 – 0.98) and 0.90 (95% CI: 0.87 – 0.93) with the $20-35 and $50-100 levels respectively. In subjects with Medicare, there also were no increases in adverse health outcomes, e.g. relative hospitalization rate of 0.99 (95% CI: 0.97–1.02) with a $20-50 copayment. There also was a 4% reduction in ED visits (RR=0.96, 95%CI: 0.94–0.97) with a $20-50 copayment. Conclusions: Cost-sharing for emergency care appeared to reduce use of the ED without leading to more adverse health outcomes. Implications for Policy, Delivery or Practice: Modest levels (up to $50–100) of cost-sharing appear to be safe on average for patients with Medicare or commercial insurance. Additional research is needed to examine temporal trends in cost-sharing effects and the impact of higher amounts of costsharing. Primary Funding Source: AHRQ • Cost-Sharing for Emergency Care – Is It Safe? Findings on Health Outcomes from the Safety and Financial Ramifications of ED Copayments (SAFE) Study John Hsu, M.D., M.B.A., MSCE, Maggie Price, M.A., Richard Brand, Ph.D., Bruce Fireman, M.A., Joseph Newhouse, Ph.D., Joseph Selby, M.D., M.P.H. • Self-Reported Effects of Prescription Drug Cost-Sharing: Decreased Adherence and Increased Financial Burden Mary Reed, M.P.H., Vicki Fung, B.A., Bruce Fireman, M.A., John Hsu, M.D., M.B.A., M.S.CE Presented by: John Hsu, M.D., M.B.A., MSCE, Division of Research, Kaiser Permanente, 2000 Broadway, Oakland, CA 94612; Tel: 510.891.3601; Fax: 510.891.3606; E-mail: jth@dor.kaiser.org Research Objective: Despite the increasing use of patient cost-sharing, limited information exists about its clinical consequences. We examined the impact of cost-sharing for emergency care on adverse health outcomes. Study Design: In a large natural experiment, we examined the effect of emergency department (ED) copayment levels on adverse health outcomes between 1999-2001 for two Presented by: Mary Reed, M.P.H., Division of Research, Kaiser Permanente, 2000 Broadway, Oakland, CA 94612; Tel: 510.891.3808; Fax: 510.891.3606; E-mail: mer@dor.kaiser.org Research Objective: We investigated patient knowledge of their cost-sharing structure for prescription drugs and their self-reported behavioral responses, including drug adherence. Study Design: We conducted a cross-sectional telephoneinterview study among a random sample of adult integrated delivery system (IDS) members, with over-sampling of members age 65 years or older. Subjects reported how much they paid per drug, whether they paid different amounts for generic and brand-name drugs (tiers), and whether they had a drug benefit cap. Subjects then reported whether their cost- sharing amount affected their adherence to treatment (e.g. taking less than prescribed, and not filling or refilling prescriptions) or affected their behavior as a financial burden (e.g. switching to a cheaper medication, borrowing money to pay for medications). Population Studied: The 926 subjects (response-rate=72%) had a mean age of 67 years, tended to be female (57%), and to have Medicare insurance (71%). In addition, 74% reported White race; 44% reported having “excellent” or “very good” health status; 67% reported less than a college-graduate education; and 43% reported an annual household income of less than $35,000. Principal Findings: Among respondents, 94% reported having any cost-sharing; 60% reported having tiered copayments, and 31% reported having a benefit cap. Comparing with actual benefits, 86% of subjects accurately reported whether they had a cap or not; 75% were accurate about having tiered copayments or not; 70% were accurate about their generic copayment amount; and 44% were accurate about their brand-name copayment amount. Overall, 11% of all subjects reported not taking their medications as prescribed because of the cost. In multivariate logistic models, subjects who perceived having a generic copayment of >$10 (OR 2.7, 95%CI:1.2-6.1), perceived having tiered copayments (OR=2.3, 95% CI:1.2-4.4), or perceived having a cap (OR=1.9, 95% CI:1.1-3.5) were more likely to report being non-adherent, as were patients who took more medications, and those who had lower income. Furthermore, 20% of all subjects reported that cost affected their behavior as a financial burden. In the multivariate models, subjects who perceived having generic copayments >$5 (OR=1.9, 95% CI:1.1-3.5), or tiered copayments (OR=2.9, 95% CI:1.7-5.2) were significantly more likely to report a financial burden-related behavior, as were patients who took more medications, were <age 65, female, married, or had lower income. In contrast, patients’ actual copayment amount did not demonstrate any statistically significantly association with either of the self-reported behaviors. Conclusions: Patients appeared to have general knowledge of their cost-sharing structures, but less knowledge about the specific amounts. Patients’ perceived level of cost-sharing influenced whether the patients followed their prescription drug regimen, and changed the way they obtained their prescriptions. Implications for Policy, Delivery or Practice: Patient perceptions of their cost-sharing structure may strongly influence their behavior, yet many patients do not know their actual cost-sharing details. Moreover, some cost-sharing structures such as tiered copayments appear to create barriers to care, rather than shift resource use. Additional research is needed on the clinical and economic ramifications of prescription medication cost-sharing. Primary Funding Source: AHRQ, Kaiser Foundation Research Institute • The Impact of Premiums on Wisconsin’s BadgerCare Program Nathan West, M.P.A., Norma Gavin, Ph.D., Jamie Chriqui, Ph.D., Nancy Lenfestey, M.H.A. Presented by: Nathan West, M.P.A., Research Health Analyst, Division of Health Economics Research, RTI International, 3040 Cornwallis Road, Research Triangle Park, NC 27709; Tel: 919.541.6816; Fax: 919.990.8454; E-mail: nathanwest@rti.org Research Objective: As states have expanded health insurance coverage to uninsured, low–income children and families through Medicaid waivers and the State Children’s Health Insurance Programs (SCHIP), interest in beneficiary cost sharing though monthly premiums has grown. Premiums have both benefits and costs. While premiums make SCHIP programs more like private insurance and may reduce the social stigma sometimes associated with public programs, critics argue that premiums can reduce participation, increase churning and leave the uninsured without access to quality health care. This paper investigates the effects that premiums have had on program participation, enrollment patterns, and health care experiences for enrollees of BadgerCare, Wisconsin’s SCHIP program. Study Design: Information on the impact of premiums on participation was obtained from a computer-assisted household telephone survey conducted with BadgerCare families and eligible non-enrolled families, drawn from lists of families with children participating in the National School Lunch Program (NSLP) during the 2001/02 school year. Information on disenrollment and reenrollment rates for premium paying and non-premium paying BadgerCare enrollees were estimated from administrative enrollment records with Kaplan-Meier survival curves and Cox proportional hazard models. Information on the reasons for disenrollment among premium-paying families and on their experiences following disenrollment was obtained from a mail survey of premium-paying families with one or more members who had disenrolled from BadgerCare in the first half of 2002. Population Studied: BadgerCare participants whose family incomes were 150 – 200% percent of the federal poverty level (FPL) and therefore were required to pay monthly premiums approximately 3 percent of their income. Principal Findings: We found little evidence that premiums significantly reduced program participation; only 10 percent listed the premium amount being too high as a reason for not enrolling. Furthermore, 83 percent of the premium-paying families thought the premium amount was reasonable. We found little to no evidence that premium payments increased churning. In fact, because of the mandatory waiting period of 6 months for premium-paying families following disenrollment, these families were more likely than other BadgerCare families to delay reenrollment in the short term. Premiums were a major reason for disenrollment. Approximately 25 percent of disenrolled premium-paying families noted problems paying the premiums. When these families left the program most members experienced a period of uninsurance for several months, and those that obtained new coverage experienced a deterioration of access to health care and had greater unmet health care needs. Conclusions: BadgerCare’s modest premiums have not been a major deterrent to program participation nor do they appear to increase churning. However, a number of families had difficulties paying the premium and subsequently left the program, experiencing greater unmet needs. Implications for Policy, Delivery or Practice: Because of the adverse health care effects that can result, states should consider temporary assistance programs for families who have difficulties paying monthly premiums and are subject to being involuntarily disenrolled from the program. Primary Funding Source: CMS Call for Papers Access to Healthcare & Insurance Chair: Joel Cantor, Sc.D. Monday, June 7 • 10:30 a.m.-12:00 p.m. • The Disabled and Access to Care in Managed Care Teresa Coughlin, Sharon Long, Ph.D. Presented by: Teresa Coughlin, Principal Research Associate, Health Policy Center, Urban Institute, 2100 M Street, N.W., Washington, DC 20037; Tel: 202.261.5639; Fax: 202.223.1149; E-mail: tcoughli@ui.urban.org Research Objective: To gain an understanding of how the disabled individuals fare under managed care. Study Design: Using the 1997 to 2001 National Health Interview Surveys as the principal data source, we examine how managed care--relative to the more traditional fee-forservice reimbursement system—affects access to care and use of medical services among disabled, non-aged adults. For the analysis we focus on Medicaid beneficiaries and use a variety of measures to define disability—such as SSI enrollment or level of functional impairment. Employing a difference-in-differences approach, we compare changes in disabled Medicaid beneficiaries’ access and use in areas that shifted from FFS to managed care between 1997 and 2001 to changes in disabled Medicaid beneficiaries’ access and use in areas that remained FFS or had Medicaid managed care throughout the time interval. In making these comparisons, multivariate modeling was used to control for health status, individual and family characteristics. We also control for local health care market characteristics such as number of physicians and number of hospital beds in the counties. In addition, we account for the strength of the local health care safety net by including variables that measure county unemployment and poverty rates. Our outcome measures include emergency room use, number of inpatient care stays, number of encounters with health care professionals, extent of and reasons for unmet need, and presence and type of usual source of care. Population Studied: A national sample of disabled, non-aged Medicaid beneficiaries. Principal Findings: Preliminary findings suggest that managed care has limited effects on access to and use of health care among disabled Medicaid beneficiaries. For example, we find no differences between the share of managed care and FFS beneficiaries who had an encounter with a health care professional. Similarly, we found no evidence of a significant reduction in the proportion of managed care and FFS beneficiaries who had had a hospital stay. The results did show some positive effects in that managed care beneficiaries somewhat less likely to report have an unmet health care need. Conclusions: On balance, health care access and use for disabled Medicaid beneficiaries enrolled in managed care were not significantly different from disabled beneficiaries enrolled in FFS Medicaid. Implications for Policy, Delivery or Practice: While several recent studies have examined the impacts of managed care, few have focused on the disabled—a particularly vulnerable population under managed care given their diverse and complex health care needs. Given that states are increasingly looking at managed care programs for their disabled Medicaid enrollees as a way to stem program costs—with California’s Governor Schwarzenegger’s recent proposal to transition disabled Medi-Cal beneficiaries into managed care being especially noteworthy—the study findings suggest that cost savings may be realized without significantly compromising access to care. Primary Funding Source: CMS • Effects of the State Children’s Health Insurance Program (SCHIP) on Access to Care, Use of Services and Health Status Lisa Dubay, Sc.M., Genevieve Kenney, Ph.D., Amy Davidoff, Ph.D. Presented by: Lisa Dubay, Sc.M., Principal Research Associate, Health Policy Center, The Urban Institute, 2100 M Street, N.W., Washington, DC 20037; Tel: 202.261.5667; Fax: 202.223.1149; E-mail: ldubay@ui.urban.org Research Objective: To estimate the effects of the SCHIP program on access to care, use of services and health status at the population level. Study Design: We used a pre-post design, with a comparison group, comparing changes in access, use of services, and health status before and after SCHIP implementation. The treatment group consisted of newly (SCHIP) eligible children and the comparison group included children with incomes slightly above the eligibility thresholds for SCHIP. Estimated difference-in-difference regression models controlled for observable child, family, and area characteristics. Data were from the 1996, 1999, and 2002 National Survey of America’s Families (NSAF), and the 1997, 2000, and 2001 National Health Interview Survey (NHIS). Access and use measures included usual source of care, unmet need for medical and other services, number and type of provider visits, and out-ofpocket spending. We also examined general health status, change in health status, school days missed, and the presence of asthma, anemia, and frequent headaches or ear infections. Treatment and comparison groups were identified using an algorithm that incorporates federal and state specific SCHIP eligibility requirements to replicate the eligibility determination process. Population Studied: Children aged 0-17 years. Principal Findings: Estimates indicate that among eligible children, the SCHIP expansions were associated with a 4 to 6 percentage point increase in the probability of having a dental visit (NHIS, NSAF) and a 4 percentage point increase in the probability of an eye care visit within the year (NHIS). SCHIP also led to a decrease in the number of hospital visits and to lower unmet medical needs among eligible children (NSAF). There was a downward shift in out-of-pocket spending among eligible children, with a 5 percentage point reduction in the proportion with spending over $500 (NHIS). There was also a 2 percentage point increase in reported diagnosis of asthma, likely reflecting increased provider access (NHIS). An increase in satisfaction with care of about 4 percentage points was also observed (NSAF). The implied impacts are substantially greater for eligible children enrolled in the SCHIP expansions than for all children in the income eligible group since many of these children maintained their private coverage over this period. We did not find effects on use of other health services, or other measures of health status. Conclusions: The eligibility expansions improved some measures of access for children, and the out-of-pocket burden on families was reduced. Gaps remain for other components of access. Implications for Policy, Delivery or Practice: Further progress in improving access to care for SCHIP eligible children may require additional targeted enrollment of uninsured children and attention to potential barriers to care among enrolled children. Such changes are increasingly unlikely, given the dire fiscal situation in many states, which have led some states to establish enrollment caps and cut back on outreach. Under the current policy environment, the recent improvements in access documented in this study are at risk of being reversed. Primary Funding Source: , The Robert Wood Johnson Foundation and the Maternal and Child Health Bureau, HRSA, U.S. Department of Health and Human Services. • Physician Supply and Effectiveness of the Primary Health Care System James Laditka, D.A., Ph.D., Sarah Laditka, Ph.D., Janice Probst, Ph.D. Presented by: Sarah Laditka, Ph.D., Associate Professor and MHA Program Director, Health Services Policy and Management, University of South Carolina, Arnold School of Public Health, 800 Sumter Street, Columbia, SC 29208; Tel: 803.777.1496; Fax: 803.777.1836; E-mail: sladitka@gwm.sc.edu Research Objective: Federal policy works to increase physician supply in shortage areas, which can be rural or urban. The premise is that people in low supply areas lack access to primary care, and thus may have poorer health outcomes. Yet studies of physician supply effects have mixed findings. We study effects of supply on hospitalization for ambulatory care sensitive conditions (ACSH). Also called potentially preventable hospitalization, ACSH is a commonly used indicator of the effectiveness of primary care. The ACSH indicator informs federal health policy, and that of many states. Study Design: We used a dataset of county ACSH rates compiled by the Safety Net Monitoring Initiative of the Agency for Healthcare Research and Quality, using ACSH risks that were adjusted for physician practice style. The data were supplemented with data from the Area Resource File, the Behavioral Risk Factor Surveillance System, the Centers for Disease Control and Prevention, the Environmental Protection Agency, and the National Center for Health Statistics. We restricted the analysis to states in which the majority of counties had available data. Logistic regression estimated ACSH risk. Physician supply was measured, in separate analyses, as a continuous variable (MDs/100,000) and in quartiles. Rural areas were modeled separately, with separately defined supply quartiles. Controls included health system and use characteristics, demographic and socioeconomic characteristics, population health measures, air quality, and county/area factors. Population Studied: Children (0-17), younger adults (18-39), and middle age adults (40-64) in 664 counties of twenty U.S. states. Principal Findings: Controlling for all other factors, increased physician supply is associated with lower ACSH risks in urban areas for children (p=.003), younger adults (p=.019), and middle age adults (p<.0001). In all models, standardized estimates indicate that physician supply is one of the most important factors associated with ACSH variation. Results are more pronounced for urban than for rural areas. In the urban quartile analysis, areas with greatest supply had notably lower ACSH risks than those with least supply, for all three age groups (p=.039; p=.017, p<.0001, respectively). The greatest effect was for middle age adults, where there was doseresponse relationship between increasing supply and ACSH risk reduction. Conclusions: Children and working age adults in areas with greater physician supply had lower ACSH rates. For middle age adults in urban areas, there was dose-response relationship between supply and ACSH risk reduction. Weaker results for rural areas may be attributable to a smaller rural sample (n=216), less supply variation among rural areas, or the omission of counties with the lowest supply. Implications for Policy, Delivery or Practice: Primary care physician supply affects the performance of the primary care system positively. Policies promoting physician placement in underserved areas, including the National Health Service Corps, state loan repayment programs, and Title VII training programs, should receive continued support. Innovative local efforts to induce physicians to locate in poor and remote areas should be explored. Results suggest that increasing the number of primary care physicians may be beneficial at all supply levels. Primary Funding Source: South Carolina Rural Health Research Center • How Well Does Medicaid Work in Improving Access to Care? Sharon Long, Ph.D., Teresa Coughlin, M.P.H., Jennifer King, B.A. Presented by: Sharon Long, Ph.D., Principal Research Associate, Health Policy, Urban Institute, 2100 M Street, N.W., Washington, DC 20037; Tel: 202.261.5656; Fax: 202.223.1149; E-mail: slong@ui.urban.org Research Objective: To provide an assessment of how well the Medicaid program is working at providing access to and use of health care for low-income mothers. Study Design: Using instrumental variables methods to control for selection into insurance coverage, we estimate the effect of Medicaid coverage on access to and use of health care relative to private coverage and being uninsured. We use the 1997 and 1999 National Survey of America’s Families (NSAF), with state and county information drawn from the Area Resource File and other sources. Population Studied: Women with children in families with incomes below 200 percent of the federal poverty level. Principal Findings: We find that Medicaid beneficiaries’ access and use is significantly better than that obtained by the uninsured. By controlling for insurance selection, the analysis showed that the benefits of having Medicaid coverage versus being uninsured are substantially larger than what was estimated when selection is not accounted for. Our results also indicate that Medicaid beneficiaries’ access to care is comparable to that of the privately insured. Without controls for insurance selection, access for Medicaid beneficiaries is found to be significantly worse than for the low-income privately insured. Conclusions: Our results show that the Medicaid program improves access to care relative to being without insurance for low-income mothers. Moreover, our results indicate that, after controlling for selection into insurance status, Medicaid beneficiaries achieve access levels comparable to the privately insured. Our results also indicate that prior research, which has not controlled for selection into insurance coverage, has likely understated the gains of Medicaid relative to uninsurance and overstated the gains of private coverage relative to Medicaid. Implications for Policy, Delivery or Practice: Our findings indicate that the Medicaid program has substantially improved access to care for those who would otherwise be uninsured, bringing them up to the levels of access reported by those with private insurance coverage. Our results also highlight the importance of using appropriate analytic methods in assessing the impacts of policy changes. Primary Funding Source: , The Henry J. Kaiser Family Foundation (The Kaiser Comission on Medicaid and the Uninsured) • Ambulatory Care Sensitive Condition Hospitalizations among Elderly Medicare and Medicaid (Dual) Enrollees Marlene Niefeld, M.P.P. Presented by: Marlene Niefeld, M.P.P., Doctoral Candidate, Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, 272 Yale Court, Arnold, MD 21012; Tel: 410.647.6687; Fax: 410.647.6687; Email: mniefeld@jhsph.edu Research Objective: The first objective was to analyze the relationship between perceived access to care barriers and likelihood of preventable hospitalization. Out-of-pocket costs and organizational barriers to specialty and long-term care were expected to increase the likelihood of an ambulatory care sensitive condition (ACSC) hospitalization. The second objective was to analyze the association of prior ACSC to future ACSC hospitalization to examine care management over time. Study Design: Data are from a 1999 6-state survey of elderly dual enrollees (n=2,128) funded by The Robert Wood Johnson Foundation and The Commonwealth Fund,1999 and 2000 Inpatient Medicare Claims data matched to the population surveyed, and the Area Resource File. ACSC’s were defined by panels of physicians as conditions that should not require hospitalization if timely and appropriate ambulatory care was provided. The ICD-9-CM code reported as the primary diagnosis for each hospitalization was used to determine whether a hospitalization could be classified as potentially preventable. ACSC hospitalizations were included as a dependent variable in the analysis if they occurred within 12 months after the survey interview. Logistic regression analyses controlling for predisposing, enabling, and need characteristics were conducted on all dual enrollees to identify characteristics associated with ACSC hospitalization. Population Studied: Community-resident Medicare beneficiaries aged 65 and older who were continuously enrolled in Medicaid for the previous 12 months was the population of interest. Previous research has shown that dual enrollees were more likely than Medicare beneficiaries with private supplemental insurance to have no physician office visits in a one-year period, and were less likely to receive specific types of preventive care, follow-up care, and medical testing. Principal Findings: Twenty-five percent of all hospitalizations within one year were preventable for elderly dual enrollees. Dual enrollees with out-of-pocket costs were 1.45 times more likely than those without out-of-pocket costs to experience a preventable hospitalization, those with 5 or more chronic conditions were 1.64 times more likely than those with 0-2 chronic conditions to have a preventable hospitalization, and for each additional prescription medications taken in the previous month the likelihood of a preventable hospitalization increased by 12 percent. Prior preventable hospitalization also increased the likelihood of a subsequent preventable hospitalization by 9.00 times. Conclusions: Despite the comprehensive insurance coverage provided by the Medicare and Medicaid programs to elderly dual enrollees, and the fact that virtually all communitydwelling dual enrollees reported that they had a usual source of care, access to care and care management problems remain. The high rates of preventable hospitalization in this population, and the fact that persons with multiple chronic conditions, previous ACSC hospitalizations, and multiple prescription medications were more likely to experience subsequent ACSC hospitalizations, indicate a need for improving the current system of chronic disease management. Implications for Policy, Delivery or Practice: These results suggest that any out-of-pocket costs present a burden for lowincome populations. They also indicate the need for several innovations in the health care delivery system. It has been hypothesized that coordination of care between multiple providers and innovations in medical informatics to improve prescription drug management will improve care among vulnerable populations. In order to accomplish these innovations, reimbursement to physicians and hospitals may need to be changed. Primary Funding Source: AHRQ Call for Papers Insurance Coverage Issues & Effects Chair: Jessica Banthin, Ph.D. Tuesday, June 8 • 9:15 a.m.-10:45 p.m. • "Churning": SCHIP Coverage Discontinuity and Its Consequences Julia Costich, J.D. Presented by: Julia Costich, J.D., Assistant Professor, School of Public Health, University of Kentucky, 121 Washington Street, Lexington, KY 40536; Tel: 859.257.8709; Fax: 859.257.3909; E-mail: jfcost0@uky.edu Research Objective: This study assesses the extent to which children enrolled in the Kentucky Children's Health Insurance Program (KCHIP) or Kentucky Medicaid renew coverage in a timely manner, and the effect of discontinuities in coverage in the context of state health policy. Repeated loss of coverage followed by re-enrollment is referred to here as churning. Study Design: Data on coverage renewals and new enrollments were provided by the state Department for Medicaid Services. We analyzed KCHIP and Medicaid data in tandem because of prior findings that many families' incomes fluctuate between the eligibility criteria for the two programs. Coverage renewal can be accomplished by mail but initial application requires an in-person interview at the local eligibility determination office. We therefore tested the hypothesis that families would take the steps necessary to renew children's coverage in a timely manner. The study design compared timely renewal rates with the volume of "new" applications as a proportion of total enrollment. Our second hypothesis was that county rates of successful renewal would relate to the proportion of all children enrolled in KCHIP or Medicaid because information about renewal procedures would be more readily available in counties with high rates of enrollment. We compared enrollment rates by county with renewal rates and "new" approvals by county to test this hypothesis. Population Studied: Children and youth enrolled in KCHIP or Medicaid during the 2003 state fiscal year. Principal Findings: Statewide, only 56 percent of eligible children had timely coverage renewal. While far from optimal, this finding is consistent with renewal rates in other studies. New application approvals equaled 53 percent of total combined KCHIP and children's Medicaid enrollment, although total enrollment increased by only 4 percent. The clear implication is that about 87.5 percent (about 254,600) of those whose coverage was not renewed on time regained coverage through the new application process. The rate of successful coverage renewal ranged from 30 percent in the state's most affluent county to 78 percent in one of the poorest; the percentage of children enrolled in KCHIP or Medicaid was modestly reliable in predicting the rate of timely renewal. Conclusions: Despite the availabilty of a computer-generated mailed renewal form, timely renewal rates for KCHIP and children's Medicaid are unacceptably low. Previous studies have characterized the renewal form as hard to decipher, and families complained of not receiving renewal information in a timely manner. While the re-entry of most children through new applications is encouraging, it represents an inappropriate use of state agency resources. Implications for Policy, Delivery or Practice: The high volume of families whose children's coverage was not renewed in time and associated administrative cost imply that the state (as well as eligible children) would benefit by dedicating resources to improving the legibility of the renewal form and supporting families in the renewal process. Primary Funding Source: RWJF • Effects of the State Children’s Health Insurance Program (SCHIP) on Health Insurance and Access to Care for Children with Special Health Care Needs (CSHCN) Amy Davidoff, Ph.D., Genevieve Kenney, Ph.D., Lisa Dubay, Sc.M. Presented by: Amy Davidoff, Ph.D., Senior Research Associate, Health Policy Center, Urban Institute, 2100 M Street, N.W., Washington, DC 20037; Tel: 202.261.5259; Fax: 202.223.1149; E-mail: adavidof@ui.urban.org Research Objective: To estimate the effects of SCHIP expansions in public insurance eligibility on insurance coverage and access to care for CSHCN, compared to children without special health needs. Study Design: We used a pre-post design with a comparison group, to examine changes in coverage, access, use of services, and health status, before and after SCHIP implementation. The treatment group consisted of newly (SCHIP) eligible CSHCN. We used multiple comparison groups, including children with incomes slightly above the eligibility thresholds. Estimated difference-in-difference regression models controlled for observable child, family, and area characteristics. We also examined spillover effects for children previously eligible for Medicaid. Data were from the 1997, 2000, and 2001 National Health Interview Survey. Access and use measures included usual source of care, unmet need for medical and other services, number and type of provider visits, and out-of-pocket spending. Treatment and comparison groups were identified using an algorithm that incorporates federal and state specific SCHIP eligibility requirements to replicate the eligibility determination process. Population Studied: Children 0-17 years. CSHCN were identified based on reported diagnosis of a serious chronic condition (asthma, heart disease, diabetes, etc.), or reported limitation of activity caused by a chronic health condition. Principal Findings: Approximately 10 percent of newly eligible CSHCN enrolled in SCHIP. This was associated with a 3 percentage point decline in private coverage (ns), and a 6 percentage point or 27 percent decrease in the probability of being uninsured. Despite decreased uninsurance rates, 15 percent of SCHIP eligible CSHCN remained uninsured. The SCHIP expansions resulted in a 9 percentage point decline in any unmet need, most of which came from reductions in unmet dental care need, but there was also a negative effect on unmet prescription drug need. There was an increase in the probability of dental care and eye care visits, and a downward shift in family out-of-pocket spending. We failed to find effects on other types of provider visits or service use, and there were no significant effects on health status. For children without chronic conditions, we found greater declines in private coverage and smaller reductions in the probability of being uninsured. Impacts on access and use were similar. Conclusions: The SCHIP expansions reduced the proportion of uninsured CSHCN, but there is still a group of eligible children who lack coverage, thus additional progress is possible. The eligibility expansions improved some measures of access for CSHCN, but gaps remain for other components of access, and there is no evidence in the short run of improvements in health status. Implications for Policy, Delivery or Practice: Further progress in reducing uninsurance and improving access to care may require additional targeted outreach to CSHCN. Furthermore, SCHIP plans may need to enhance the scope of covered benefits or restructure contracts with provider plans, so as to facilitate access to care for enrolled CSHCN. These types of changes may be increasingly unlikely, given the poor financial environments in the states. In fact, many states have made various retrenchments in SCHIP coverage. As CSHCN have not generally received special protections, the recent progress documented in this study may be reversed. Primary Funding Source: HRSA • Costs of Enrolling Children in Medicaid and SCHIP Gerry Fairbrother, Ph.D, Melinda Dutton, J.D., Deborah Bachrach, J.D., Kerry-Ann Newell, M.S., Patricia Boozang, M.P.H., Rachel Cooper, M.A. Presented by: Gerry Fairbrother, Ph.D., Senior Scientist, Health and Science Policy, New York Academy of Medicine, 1216 Fifth Avenue, New York, NY 10029; Tel: 212.822.7398; Fax: 212.822.7369; E-mail: gfairbro@nyam.org Research Objective: The purpose of this study is to assess the administrative cost of enrolling children into public health insurance programs under the current enrollment process in New York City and to estimate how much of this cost could be saved if application requirements were simplified. Study Design: Interviews with key staff members were used to identify the main components of the enrollment process. Overall costs were assessed for these components using 2001 expenditure reports for three managed care organizations (MCO’s). Structured questionnaires were used to determine time spent in enrollment activities during the typical enrollment process and during a simplified post-September 11th process. Activities were organized into four functional categories: 1) outreach and health insurance education, 2) application completion/documentation, 3) quality assurance, and 4) application submission and troubleshooting. Costs and cost savings associated with these activities were determined using staff estimates of time spent and average salaries and benefits associated with each staff member. Population Studied: The population studied was children enrolled in Medicaid and SCHIP through New York City MCOs in 2000/2001. Principal Findings: The average administrative cost to enroll a child in health insurance was approximately $280. Time spent by enrollment workers was categorized into five major functional areas: eligibility screening (16%), gathering documents (20%), application process (27%), outreach and health education (18%) and other enrollment activities (19%). Up to 80% of the enrollment cost is associated with the complicated guidelines and calculations of the eligibility process while outreach and health insurance education account for approximately 20% of enrollment costs. Enrollment costs could be reduced by 40% ($282 to $166) under a simplified system. Conclusions: Our cost estimate is equivalent to more than two months premium paid to MCOs in New York. Complexities in the enrollment process result in higher administrative costs. Administrative costs absorb significant funds that could be realized as costs savings or distributed elsewhere. Implications for Policy, Delivery or Practice: Many states are responding to budget deficits by increasing administrative barriers to enrollment. Such barriers will decrease the number of people served by these programs and consequently reduce overall program costs in the short term. However, in the long run, these barriers will result in significant administrative costs making the case for simplification more compelling than ever. Primary Funding Source: , The Foundation for Child Development • Churning: Disenrollment and Reenrollment in Wisconsin’s Medicaid and BadgerCare Programs Nancy Lenfestey, MHA, Norma Gavin, Ph.D., Nathan West, M.P.A. Presented by: Nancy Lenfestey, MHA, Public Health Policy Analyst, RTI International, 3040 Cornwallis Road, Research Triangle Park, NC 27709; Tel: 919.316.3364; Fax: 919.990.8454; E-mail: nlenfestey@rti.org Research Objective: Churning, the process of repeatedly coming in and out of a health insurance program, can have adverse health and financial consequences. Discontinuities in coverage may lead to foregone preventive care, increased severity of illness and health care expenditures, and higher administrative costs. This study investigates disenrollment and reenrollment in Wisconsin’s Medicaid and BadgerCare programs to provide information on churning. Differences in patterns of disenrollment and reenrollment are examined for adult and child enrollees by eligibility group, cash assistance status, racial/ethnic group, and time period---pre- and postBadgerCare. Study Design: A record was created for each episode of enrollment, defined as a period of continuous enrollment in Medicaid/BadgerCare regardless of switches between the two programs or between eligibility groups within programs. Kaplan-Meier survival curves and Cox proportional hazard models of the length of enrollment episodes and the number of months between episodes were estimated for new and newly ended episodes, respectively, in two different time periods: 1) a 2-year period preceding BadgerCare implementation, and 2) a 2-year period following BadgerCare implementation. Survival curves and hazard rates were compared for the first 32 months from enrollment and disenrollment. Population Studied: Medicaid and BadgerCare enrolled adults (> 18 years) and children by age group (0 to 5 and 6 to 18 years) in three different enrollment categories ---AFDCrelated, Healthy Start, and BadgerCare. Principal Findings: Disenrollment and reenrollment rates were similar among BadgerCare and Medicaid beneficiaries. Enrollment episodes were substantially longer for Medicaid enrollees following BadgerCare implementation. This result was most dramatic for adult Healthy Start pregnant women; only 12% remained enrolled 12 months from enrollment prior to BadgerCare whereas 40% did so following BadgerCare. Children were less likely than adults and younger children were less likely than older children to disenroll early; cash assistance recipients were less likely to disenroll early than enrollees not receiving cash assistance; and Hispanics were more likely to disenroll early than white non-Hispanics. Furthermore, short periods of disenrollment between two enrollment periods were common among all enrollees, and were even more prevalent following BadgerCare implementation. As many as 15 to 20% of children reenrolled after only 1 month and 60 to 70% had reenrolled within the first 32 months following disenrollment. Children, cash assistance recipients, and minorities were more likely to reenroll after short periods of time relative to other enrollees. BadgerCare beneficiaries were somewhat less likely to reenroll than AFDC-related Medicaid beneficiaries. Conclusions: While evidence shows that churning is a substantial problem, particularly among children and minorities, continuity of coverage has improved significantly in Wisconsin’s Medicaid program since the implementation of BadgerCare. Implications for Policy, Delivery or Practice: The high amount of churning among children and minorities should be investigated further to determine possible adverse effects on compliance with recommended well-child visit and immunization schedules. Primary Funding Source: CMS Related Posters Poster Session A Sunday, June 6 • 6:45 p.m.-8:00 p.m. • S-CHIP and Children’s Usage in Georgia and Alabama: Does Structure Matter? E. Kathleen Adams, Ph.D., Janet Bronstein, Ph.D., Curtis S. Florence, Ph.D. Presented by: E. Kathleen Adams, Ph.D., Professor, Health Policy and Management, Rollins School of Public Health, Emory University, 1518 Clifton Road, N.E., Atlanta, GA 30322; Tel: 404.727.9370; Fax: 404-727-9198; E-mail: eadam01@sph.emory.edu Research Objective: State variation in benefits, eligibility, administrative features and provider networks under their Child Health Insurance Programs (S-CHIP) affects children’s access and utilization. We examine primary, preventive and emergency room (ER) utilization among children enrolled in a Medicaid look-alike versus a stand-alone CHIP program compared to Medicaid children. Study Design: Medicaid claims, enrollment and provider data are used to measure children’s utilization quarterly, 19992000, as a function of demographics, provider availability and enrollment in Medicaid versus CHIP. Children are categorized as: 1) CHIP only; 2) Medicaid only; 3) CHIP with Medicaid history; or 4) Medicaid with CHIP history. Provider availability reflects distance to closest participating providers and presence of certain provider types in their zip. Population Studied: Random samples of Medicaid children plus all PeachCare kids in Georgia (98,000) and ALLKids in Alabama (35,000) were studied. PeachCare, a Medicaid lookalike program, uses the same physician network and fee schedule as the state’s Medicaid’s primary care case management program (pccm). ALLKids subcontracts to Blue Cross and Blue Shield (BCBS) of Alabama, using their physician network and fee rates but does not assign a primary care gatekeeper. Principal Findings: Provider availability appeared greater for ALLKids where 75 percent had a large volume office physician in their zip versus 66 percent of PeachCare kids. Using state data and controlling for provider access, PeachCare kids were less likely than Medicaid children in the same delivery system to receive primary or well-child care as well as ER visits. This may reflect higher incomes and fewer health problems among CHIP kids but the lower use of preventive care is inconsistent. On the other hand, ALLKids had significantly higher odds of any primary care and preventive care visits among users of preventive care than Medicaid children under the Alabama pccm system. Unexpectedly, they were more likely than Medicaid children to use the ER at least once but users did have fewer visits. Pooled data better measure the delivery system effects under ALLKids since they can be compared to CHIP kids in pccm. Results indicate the patterns are similar though magnitudes are lower. A key difference is that ALLKids are not more likely to use the ER compared to Medicaid, but are compared to CHIP kids in pccm based on pooled data. Conclusions: If CHIP is integrated into an existing provider network and not accompanied by fee increases utilization may be constrained by overall participating provider availability. In contrast, there appears to be strong delivery system effects on utilization where CHIP kids have private-like coverage, greater provider availability and no managed care. The lack of gatekeeping, however, may result in unnecessary visits to the ER although appropriate triage may occur once children visit the ER. Implications for Policy, Delivery or Practice: These results indicate that the design features of CHIP matter. The private stand-alone system appears to provider better access in terms of not only provider availability but receipt of primary and preventive care. While the costs of such a system are higher these could potentially be constrained by primary care gatekeeping. Primary Funding Source: AHRQ, Child Health Insurance Research Initiative (CHIRI) • Physician Continuity and Association with Satisfaction, Trust and Preventive Services in the Rural Southeast Katrina Donahue, M.D., M.P.H., Evan Ashkin, M.D., Donald Pathman, M.D., M.P.H. Presented by: Evan Ashkin, M.D., Clinical Assistant Professor, Department of Family Medicine, University of North Carolina, 1008 Green Street, Durham, NC 27701; Tel: 919.361.2644; Fax: 919.484.0849; E-mail: evan_ashkin@med.unc.edu Research Objective: Physicians and patients report placing a high value on continuity in healthcare. In an era of rapid change in our health care system, these long term relationships can be affected. Our study examines (1) characteristics of rural patients who have longer relationships with their physicians and (2) whether the length of this relationship is associated with various aspects of satisfaction and trust, counseling on behavioral issues (smoking, physical activity, nutrition) and receiving preventive services (mammogram, pap, flu shot, colon screening and cholesterol). Study Design: Cross-sectional data on health care access indicators were obtained from participants in the Southern Rural Access Program (SRAP) using a computer-assisted telephone interview system (CATI) from October 2002 to July 2003. Population Studied: SRAP respondents (N=3176) who indicated they see a particular physician for their care. SRAP includes 4879 participants in non metropolitan counties of eight southern states (Alabama, Arkansas, Georgia, Louisiana, Mississippi, South Carolina, Texas and West Virginia. Principal Findings: 10.8% (N=319) had seen the same physician for the past 12 months, 11.8% (N=369) for the previous 13-24 months, 20.7% (N=669) for the past 25-60 months and 56.7% (N=1819) for more than 60 months. Compared to persons with 12 months or less continuity, a higher proportion of respondents with 5 years or more continuity were Caucasian (65.5% versus 59.8%, p <0.01), had insurance (77.2% versus 68.6%, p<0.01), had income greater than $25,000 ( 62.5% versus 48.5%, p< 0.001) and reported good to excellent health (76.0% versus 69.4%, p<0.05) . When participants saw a physician for more than 12 months, there were no significant differences. There were also no significant differences for age, gender or education. Fewer respondents with 5 years or more of continuity reported not being satisfied with their overall health care (3.7% versus 8.7%, p<0.01), but these differences disappeared with at least 24 months of continuity. There were no significant differences found for other satisfaction and trust variables or for any of the preventive and counseling services, except those who reported seeing a physician for 12 months or less were more likely to report being counseled in nutrition over the past year (55.8% versus 47.7%, p<0.05). Conclusions: Over half of the participants in the SRAP report seeing the same physician for more than 5 years. Participants who change (or who are forced to change) physicians after one 1 year have decreased satisfaction with overall healthcare and tend to report poor health status. Differences in demographic, satisfaction, and physician counseling resolved after a person had been seeing a physician for more than 2 years. Implications for Policy, Delivery or Practice: It is encouraging that there are little to no differences in preventive health services and counseling services by years of continuity. However, differences in reported health status indicate continuity may still be important for those who are ill. This group could be at risk when continuity is jeopardized. Primary Funding Source: RWJF • Design Considerations in the Evaluation of the Integrated Health Outreach System Project Craig Blakely, Ph.D., M.P.H., James Burdine, Dr.P.H., Alicia Dorsey, Ph.D., Petra Reyna, M Ed, Monica Wendel, M.S., M.P.H., Miguel Zuniga, M.D., Dr.PH Presented by: Craig Blakely, Ph.D., M.P.H., Head, Department of Health Policy and Management, School of Rural Public Health, 3000 Briarcrest Drive, Suite 300, Bryan, TX 77802; Tel: 979.458.0937; Fax: 979.458.4264; E-mail: blakely@srph.tamu.edu Research Objective: The Robert Wood Johnson Foundation in a joint effort with HRSA, funded an evaluation of the Integrated Health Outreach Project. The research objectives of this project are to evaluate the impact of the overall strategy as well as specific components in building community capacity, the capacity of lay health workers, and participating health care organizations to more effective collaborate and as a result, increase acccess to care for colonia residents Study Design: Seven different data collection activities were built into the design for this effort looking at: community leaders, the promotoras (lay health workers), health care organizations, and colonia residents. Population Studied: Health care organizations, promotoras, colonia residents and community leaders were all studied in the various aspects of this evaluation. Principal Findings: This paper will focus on the challenges of the overall initial evaluation design, and subsequently challenges to the implementation of that design. Significant factors included in this discussion are the impact of changes in federal, state and local health policy (e.g., reimbursement rates) on each of the study populations as well as the overall system of care. Conclusions: At the third year in a four year study, conclusions are primarily observations related to the success of the various data collection activities. However, while prelmiinary results suggest the project is on track in meeting its goals, early data also point toward the potential for substantial changes in collaborative community service delivery and other proximal outcomes projected by the project model. Implications for Policy, Delivery or Practice: Designing an evaluation of a complex community-based intervention to improve access to care to a medically indigent population yields important information for consideration in changes in health policy. Policy makers and service providers must recognize and consider this information in shaping future health policy. Further, as the field of community prevention research ages, the importance of translating research to practice grows considerably. Illustrations of community-based projects that directly affect the way communities address their own health problems provide important data for the field. Primary Funding Source: RWJF, HRSA • The Impact of SCHIP on Health Insurance Coverage Rates for Children Lynn Blewett, M.A., Ph.D., Michael Davern, Ph.D., Gestur Davidson, Ph.D., Holly Rodin, M.P.A. Presented by: Lynn Blewett, M.A., Ph.D., Assistant Professor, Health Services Research and Policy, University of MN, School of Public Health, 420 Delaware Street, S.E., MMC 729, Minneapolis, MN 55455; Tel: 612.626.4739; Fax: 612.624.2196; E-mail: blewe001@umn.edu Research Objective: Recent national data document the decrease in the number of uninsured children while the number of uninsured adults continues to rise. This study examines the impact of SCHIP and other public program expansions in reducing the number of uninsured children for each of the 50 states and the District of Columbia. Study Design: We use data on children age 18 and under from the CPS annual demographic supplement for two time periods: pre-SCHIP 1996-1998 and post-SCHIP 2002-2003. We estimated a multinomial logistic regression predicting whether a child was uninsured, covered by public insurance, or by private insurance for each time period. We include family economic and demographic characteristics along with the individual child’s demographic characteristics as covariates in the model. We use the parameter estimates generated by this model to “recycle” all the CPS cases (whether pre or post SCHIP) through two counterfactual scenarios: (1) all the respondents were from the pre-period, (2) all the respondents were from the post period. We compare the resulting predicted recycled probabilities for each state and calculate the change in health insurance coverage between the pre and post periods controlling for economic and demographic characteristics of respondents. Population Studied: Non-institutionalized children age 18 and under in each state and DC. Principal Findings: We found significant increases in the percent of children enrolled in public insurance in all 50 states (1.9% increase in Colorado to a 4.8% increase in New Mexico); significant decreases in the percent of uninsured children (from 0.7% in Iowa to 2.7% in New Mexico); and significant decrease in private health insurance coverage (from –3.5% in Vermont to -.7% in Colorado) in 35 of the states. We found that public program participation increased the most for children with incomes below 200% of the poverty. Conclusions: SCHIP and other public program expansions aimed at children have succeeded in enrolling children in public programs and reducing the numbers of uninsured for all states and the District of Columbia. One concern with these program expansions has been the amount of substitution of public for private coverage that would occur as a result. Our analysis is consistent with the expectation that a limited amount of substitution is occurring and public program expansions were the greatest for those with the lowest incomes. Implications for Policy, Delivery or Practice: At a time when health insurance coverage was declining for adults, significant progress was made in reducing the number of uninsured children. SCHIP and public program expansions were key to preventing a similar increase in the number of uninsrued children. This finding is robust across all states with different eligibility and other SCHIP program specifications and rates of enrollment, suggesting that additional factors are playing a role in getting more children enrolled in public health insurance. State outreach, administrative simplification, as well as larger outreach efforts such as “Cover All Kids” are all enabling families to enroll eligible children in public programs nationwide. Primary Funding Source: RWJF • Cervical Cancer Screening among Elderly Women: The Effects of Health Beliefs vs. Insurance Coverage Tyrone Borders, Ph.D. Presented by: Tyrone Borders, Ph.D., Assistant Professor and Chief, Health Services Research, Texas Tech University Health Sciences Center, 3601 4th Street, MS 8161, Lubbock, TX 79430; Tel: 806.743.6984; Fax: 806.743.1292; E-mail: tyrone.borders@ttuhsc.edu Research Objective: To examine the effects of emotional barriers, anxiety about cervical cancer screenings, and the perceived importance of cervical cancer screenings versus the effect of insurance coverage on the odds of utilizing Pap tests. Study Design: A retrospective cohort study in which older women were asked to report about their utilization of Pap tests over the previous two years, their beliefs and attitudes about Pap tests, and their insurance coverage for Pap tests. Multivariate logistic regression analyses were conducted to model the probability of receiving Pap tests. Population Studied: Data were collected through a population-based telephone survey of 1,338 elderly women (age 65 and older) residing in the 111-county region comprising West Texas. Principal Findings: Insurance coverage for Pap tests and perceptions of the importance of Pap tests had the largest effects on whether elderly women received a Pap smear in the previous two years. Women with insurance covering Pap tests had a much higher odds (OR=5.19, 95% CI=3.86, 6.80) of receiving a Pap test than those without insurance. Those who perceived that Pap tests were important for the prevention of cancer had a higher odds (OR=3.69, 95% CI=2.40, 5.67) of receiving a Pap test than those who perceived that they were unimportant. Emotional barriers and anxiety were not significantly associated with Pap test utilization. Conclusions: Among older women, insurance coverage for Pap tests and perceptions of the importance of Pap tests have a strong impact on Pap test utilization. Implications for Policy, Delivery or Practice: Health policy makers should consider further expanding health insurance coverage to older women to cover Pap smears. Among those women who have insurance coverage, medical care providers and public health workers should promote the benefits of Pap tests in preventing cervical cancer. Primary Funding Source: U.S. Administration on Aging • Most Veterans Obtain Cardiac Revascularization Outside VHA William Weeks, M.D., M.B.A., David Bott, Ph.D., David Bott, Ph.D., Dorothy Bazos, Ph.D., Michael Racz, M.A., Steven Wright, Ph.D., Elliott Fisher, M.D. Presented by: David Bott, Ph.D., Research Associate, Center for the Evaluative Clinical Sciences, Dartmouth Medical School, 7251 Strasenburgh Hall, Room 313B, Hanover, NH 03755-3863; Tel: 603.650.1958; Fax: 603.650.1935; E-mail: david.m.bott@dartmouth.edu Research Objective: VHA seeks to provide comprehensive, coordinated care through a regional referral network. However, many veterans use multiple healthcare systems to obtain care. We sought to determine whether veterans rely on VHA or private sector for cardiac revascularization. Study Design: Matching enrollment and utilization data from VHA to utilization data from New York’s Cardiac Surgery Reporting System (CSRS), we were able to create a comprehensive file of coronary artery bypass grafting (CABG) surgeries and percutaneous coronary interventions (PCI) provided to male enrolled veterans in New York in 1997. We determined the site of intervention (VHA or private sector) and risk-adjusted mortality quartile for private sector hospitals, and compared utilization rates by veterans to non-veterans for two age groups (45-64 and 65+). Population Studied: VA enrolled male veterans aged 45 or older with residence in New York during study period. Principal Findings: Male veterans aged 45-64 underwent more CABG (4.29/1000 vs. 2.82/1000, p<.001) and PCI (5.55/1000 vs. 4.32/1000, p<.001) than male non-veterans, while those aged 65+ underwent fewer CABG (7.57/1000 vs. 8.15/1000, p=.051) and PCI (6.59/1000 vs. 7.69/1000, p<.001). Older veterans who underwent private sector CABG were more likely to use hospitals ranked in the lowest versus highest quartile of risk-adjusted mortality (OR 1.30, 95% CI:1.03-1.64). Veterans aged 45-64 received 41% of their CABGs and 35% of their PCIs within VHA, while veterans 65 and older received 19% of their CABGs and 18% of their PCIs within VHA. Conclusions: Both younger and older veterans receive the majority of their cardiovascular interventions in the private sector compared to the VA. Implications for Policy, Delivery or Practice: Efforts to improve care for veterans will require collaboration with private sector care systems including Medicare. Primary Funding Source: VA • Changes in the Hospitalized Uninsured from 1997 to 2002 Ellen Campbell, Ph.D., Tifini Preliou, B.A., Suzette Grant, Pharm.D student Presented by: Ellen Campbell, Ph.D., Assistant Professor, Economic, Social & Administrative Pharmacy, Florida A&M University, College of Pharmacy - Dyson Building, Tallahassee, FL 32307; Tel: 850.599.3082; Fax: 850.412.7548; E-mail: ellen.campbell@famu.edu Research Objective: The U.S. health care system has undergone many changes in the past decade including an increased emphasis on cost cutting managed care, the passage of the Portability Act of 1996, and the passage and revision of the Balanced Budget Act. Health policies aimed at reducing the number of uninsured have been called fragmented and incremental. Overall, the 1990s enhanced access for some individuals through SCHIP and HIPAA, but tightened the financial situation particularly on safety net providers. Changes over this period have lead to a reduction in health benefits offered to low income workers, while increasing health insurance premiums make it more difficult for the uninsured to obtain coverage. Moreover, this period saw widespread changes in hospital affiliations with a growth in multihospital systems. How have these system-wide changes affected the composition and needs of the uninsured? In this paper we examine the characteristics of the uninsured and their safety net providers, and how they have changed during the recent period of health care system transition. Study Design: Using discharge abstracts from the State of Florida for uninsured patients hospitalized in the years 1997 and 2002, we compare differences in patient characteristics such as age, race and primary diagnosis. We also examine differences in safety net facility characteristics such as geographic location, ownership and teaching status. Since the data utilized here comprises the entire population of Florida hospitalizations, sample statistics will not be reported. All observed differences are exact for the population studied. Population Studied: Uninsured patients hospitalized in a Florida short-term general hospital in the years 1997 and 2002. Principal Findings: Preliminary analysis of over 4 million hospitalizations during the periods studied indicates that the proportion of uninsured has changed very little between 1997 and 2002, however the profile of the hospitalized uninsured has changed over this 5-year period. As expected with the implementation of SCHIP there was a reduction in the proportion of uninsured children, with adults (18-64) accounting for a larger proportion of the uninsured. As indicated by other studies, there was a large increase in the proportion of white Hispanics among the uninsured, while the proportion of non-Hispanic whites decreased. Finally, there appears to be increasing use of the emergency room by the uninsured indicating less access to primary care providers. Further analysis will be performed on changes in diagnoses and hospital characteristics. Conclusions: Although there has been little change in the proportion of uninsured in the U.S., the characteristics of this vulnerable population has changed over the past 5 years. Implications for Policy, Delivery or Practice: By identifying how the uninsured population has changed in recent years, this study can be used to guide future health care policy. • The Influence of Consumers’ Experiences with Care on Health Services Utilization in the Oregon Health Plan Matthew Carlson, Ph.D., David Mosen, Ph.D., M.P.H., Charles Gallia, M.A. Presented by: Matthew Carlson, Ph.D., Assistant Professor, Department of Sociology, Portland State University, P.O. Box 751, Portland, OR 97207; Tel: 503.725.9554; Fax: (503) 725.3957; E-mail: carlsonm@pdx.edu Research Objective: This paper examines disparities in patient satisfaction and access to health care and explores the net impact of satisfaction and access on use of preventive, outpatient and emergency department use. Study Design: Data are derived from Oregon's statewide Medicaid CAHPS survey (response rate=51%). Survey data were merged with health care claims for preventive, outpatient and ER use in the 6 months before and 6 months after survey administration. Population Studied: 512 adults enrolled in 1998-2000 in the largest Non-Profit Health Plan Serving Oregon Health Plan enrollees. Principal Findings: Predisposing and enabling characteristics explained satisfaction, but did not predict use after controlling for satisfaction. Need characteristics (health status) predicted poor perceived access but increased utilization. Health care ratings and prior health care use predicted outpatient and ER use. Conclusions: Health care utilization, including outpatient and ER use, is directly influenced by consumers' experiences with their care and indirectly influenced by predisposing and enabling characteristics. ER users who have positive experiences with their care are more likely to continue to use the ER. Implications for Policy, Delivery or Practice: Improving appropriate use of health care services and reducing innapropriate use among Medicaid enrollees requires understanding the complex interplay between individual characteristics, expectations, and experiences with care. • Welfare Reform’s Influence on Healthcare Use among Young Children Daniel Choi, B.A., Laura Amsden, M.P.H., M.S.W., Jane Holl, M.D., M.P.H. Presented by: Daniel Choi, B.A., Project Coordinator, Feinberg School of Medicine, Northwestern University, Institute for Health Services Research and Policy Studies, 339 E. Chicago Avenue, Chicago, IL 60611; Tel: 312.503.2834; Fax: 312.503.2936; E-mail: dkchoi@northwestern.edu Research Objective: To describe the healthcare use among young children whose mothers were affected by the 1996 “welfare reform” policies aimed at promoting work and reducing receipt of welfare. Study Design: The study is a five-year longitudinal panel study (1999-2004) consisting of annual in-person interviews, linked administrative data and medical chart reviews. Phase I: During the annual interview, mothers provided names/contact information of all of their child’s medical providers since the previous interview. Using this information, the child’s medical charts were obtained. Phase II: Abstraction and coding of medical chart data included information about wellchild and other visits (i.e., sick, emergency, follow-up, hospital) including ICD9 codes, treatment plans, immunization records, developmental milestones, indications of abuse/neglect and birth history. Phase III: Data was aggregated into appropriate timeframes, and bivariate and multivariate analyses were conducted. Population Studied: The study includes a random stratified sample of mothers (n=1,363) receiving welfare in the fall 1998 in Illinois and a sub-sample of their children (n=554), 0-3 years of age at Year 1 of the study. Principal Findings: Annual response rates were between 9095%. Medical charts have been retrieved for 91% of the child sample and to date, ~50% of charts have been abstracted (n=265). Approximately 80% of children remained covered by Medicaid during the study; average age of children in Year 3 was 4.6 years (1-10 years); average household income of children in Year 3 was ~$15,761. Children 1-2 years of age averaged 3.5 age appropriate well-child visits; children 2-5 years averaged 5.1 visits, and children 6-10 averaged 5.4 visits. Children of all ages received ~40% of the American Academy of Pediatrics recommended well-child visits. Children of working mothers had significantly more age appropriate wellchild visits (4.9) than children whose mothers were not working (4.0, p<.05). Children of either working or nonworking mothers had significantly more other visits (9.0 and 7.5, respectively) than well-child visits (4.9 and 3.9, respectively, p<.001). Conclusions: While all children in this population received an inadequate number of age appropriate well-child visits, work promotion policies may have contributed to improved wellchild care. Nevertheless, children of mothers affected by welfare reform, regardless of their mother’s work status, had more other visits than well-child visits. Implications for Policy, Delivery or Practice: Welfare reform appears to have a modest influence on well-child care visits. Further study is needed to examine the impact of the large number of children’s other healthcare visits on a mother’s ability to achieve work promotion/welfare reduction goals. Primary Funding Source: NICHD (UO1 HD39148) • Medicaid Patients Who Left Hospitals Against Medical Advice: Trend Analysis Kyusuk Chung, Ph.D., Ross Mullner, Ph.D., Sally Freels, Ph.D. Presented by: Kyusuk Chung, Ph.D., University Professor, Health Administration, Governors State University, 1 University Parkway, University Park, IL 60466; Tel: 708.534.4047; Fax: 708.534.801; E-mail: k-chung@govst.edu Research Objective: It has been well documented that, compared with patients who are regularly discharged, those who leave against medical advice (hereafter AMA) are more often Medicaid recipients. Patients leaving AMA were known to have multiple readmissions with longer lengths of hospital stay. There have been intervention programs and this study examines whether AMA discharges among Medicaid recipients have significantly decreased across time, taking into account confounding factors including shorter stays associated with managed care and a potentially overestimated AMA size implied by recent studies suggesting that hospitals tend to report transfers as AMA discharges to get reimbursed fully. Study Design: Data are from the National Center for Health Statistics’ National Hospital Discharge Survey. For each of the years 1980, 1985, 1990, 1995, and 2000, a separate logistic multiple regression model was fit using a binary indicator for AMA as the dependent variable. Covariates included diagnoses (substance/alcohol-related mental disorder and other mental disorders), length of stay, hospital characteristics, and patient characteristics. Length of stay was used as a covariate because with managed care, length of stay has been shortened and there are typically shorter stays of patients with AMA discharges. Length of stay can also control for the suspicious AMA discharges that have a same day admission to another hospital. The significance of changes in effects across time was evaluated by fitting a single model using data from all of the five years pooled together. The variable YEAR was entered as a single continuous variable, and interactions were tested between year and each of the other independent variables. If the effect of a given independent variable has increased or decreased systematically from 1980 to 2000, the trend is reflected in the corresponding interaction term with year. Population Studied: adult patients discharged alive from Non-Federal short-stay hospitals in the United States. Principal Findings: The relative odds ratio of leaving AMA for Medicaid discharges significantly increased during the study period compared to their counterparts who are privately insured (OR: 1.95, 2.12, 2.72, 2.60 and 3.19, respectively for the year 1980, 1985, 1990, 1995, and 2000) (p<.0001). As expected, substance-related mental disorders was a strong predictor of AMA discharges, but its relative odds ratios significantly decreased (OR: 23.62, 24.36, 21.02, 16.04 and 13.40) (p<.0001). Alcohol-related mental disorders and other mental disorders showed similar linear trends. Finally, the relative odds ratio for stays of less than one day compared with stays of 7 and more fluctuated (OR: 15.51, 11.88, 7.30, 10.00 and 23.41) (p=0.46). Conclusions: AMA discharge rates among Medicaid patients significantly increased relative to those that are privately insured. This happened despite a rapid decrease in AMA rates among those with mental disorders who tend to account for a majority of AMA discharges. Implications for Policy, Delivery or Practice: Our finding implies that intervention programs may not have been effective in reducing AMA discharges for Medicaid patients relative to those that are privately insured. For states that have grappled with rising Medicaid costs, it is necessary to review current Medicaid discharge planning policies to ensure that institutions participating in Medicaid provide more effective discharge planning. Primary Funding Source: GSU fund • Do Nonprofit Hospitals Pay Their Way and Justify Favorable Tax Treatment? Helen Citkina, Ph.D. Presented by: Helen Citkina, Ph.D., Postdoctoral Scholar, School of Public Health, University of California, Berkeley, 140 Warren Hall, MC7360, Berkeley, CA 94720; Tel: 510.849.2838; Fax: 510.643.1853; E-mail: hcitkina@uclink.berkeley.edu Research Objective: This study addresses the effect of hospital ownership on the delivery of services to the medically indigent patients and their communities. Thus, it defines and quantifies two measures of community benefits, uncompensated care and broad community dividends, to evaluate whether amount of community benefits provided by a tax-exempt hospital is at least equal to the value of the tax subsidy plus the amount of community dividends provided by an investor-owned hospital. We also try to address whether investor-owned hospitals locate in areas with lower need for community benefits. Lastly, the study investigates what community and hospital market characteristics (such as hospital competition, concentration, and managed care penetration) affect hospital supply of community benefits. Study Design: We use regression analysis, TSLS, to estimate the effect that hospital ownership has on hospital’s supply of community benefits. Population Studied: We use three years of data (1998-2000) to assess the impact of area, health care system and financial pressures on two measures of community benefits by California’s general acute care hospitals. Hospital-level data came from the Annual Hospital Disclosure Reports published by Caifornia’s Office of Statewide Health Planning and Development (OSHPD). The analyses exclude specialty hospitals, long-term care facilities as well as Kaiser hospitals since they treat Kaiser members only and do not report to OSHPD. Then the sample was reduced to 869 non-profit and investor-owned hospitals (government and district hospitals were excluded). Principal Findings: Using data from 1998 to 2000 from general acute care hospitals in California we find that the amount of community dividends provided by a tax-exempt hospital is equivalent to the value of the tax contributions plus the amount of community dividends provided by an investorowned hospital in a similar market although investor-owned hospitals indirectly locate in better-insured areas. These results are sensitive to the definition of community benefits. Hospitals located in more competitive markets with higher managed care penetration supply fewer community dividends. Conclusions: Non-profit ownership is related to the volume of community benefits supplied by a hospital. The answer to the question whether non-profit hospitals justify their taxexempt status depends on the definition of community benefits. Since investor-owned hospitals are located to face lower demand for community benefits, then non-profit hospitals provide community services by locating in areas with higher demand for charity care and by operating emergency room services that uninsured often use. Without tax subsidies such areas and uninsured population may be underserved. Therefore, tax-exemption is justified once we define community dividends broadly as well as in providing care in underserved areas and populations. Implications for Policy, Delivery or Practice: Results of this research suggest more explicit identification of the community benefits expected of non-profit hospitals in return for special tax treatment. • The Impact of Health Services Regulation on Insurance Coverage Christopher Conover, Ph.D. Presented by: Christopher Conover, Ph.D., Assistant Research Professor, Terry Sanford Institute of Public Policy, Duke University, Box 90253, Durham, NC 27708; Tel: 919.684.8026; Fax: 919.684.6246; E-mail: conoverc@hpolicy.duke.edu Research Objective: To assess the impact of health services regulation on the cost of health care and its ancillary impacts on number of uninsured Study Design: Literature synthesis of research on the benefits and costs of federal and state health services regulations, including regulation of health facilities, health professionals, health insurance, FDA and medical malpractice. Population Studied: Net impact of health services regulation was calculated as percent of personal health care expenditures and projected effect on number of uninsured estimated using standard elasticity estimates. Principal Findings: The net economic impact of health services regulations equals 9.6 percent of personal health care expenditures. Approximately 7 million uninsured (1/6 of uninsured total) may be attributable to regulatory costs. Conclusions: The net impact of regulation is sizable in the health care industry, in excess of $130 billion a year, contributing to the large number of uninsured and diverting resources that otherwise could more than finance universal coverage. Implications for Policy, Delivery or Practice: Reductions in health services regulations offer an important components of strategies to expand coverage. Primary Funding Source: AHRQ • A CAP that Works: The Southeastern Kentucky Community Access Program Julia Costich, J.D., Ph.D., Frances Feltner, R.N. Presented by: Julia Costich, J.D., Ph.D., Assistant Professor, School of Public Health, Univ. of Kentucky, 121 Washington Street, Lexington, KY 40536; Tel: 859.257.6712; Fax: 859.257.3909; E-mail: jfcost0@uky.edu Research Objective: The Southeast Kentucky Community Access Program (SKYCAP) has won national acclaim for its service to low-income uninsured residents of four Appalachian counties. This study assesses SKYCAP's effectiveness in reducing emergency department visits and hospitalizations in the target population and suggests directions for further inquiry to identify elements of successful CAPs. Study Design: We compared service utilization by SKYCAP enrollees in the 12 months before SKYCAP enrollment with utilization in the 12 post-enrollment months, noting variation by targeted health problem and type of service. We derived estimates of cost savings associated with the avoidance of unreimbursed hospital care using median reimbursement rates for the same or similar services. Population Studied: Uninsured, low-income clients of a fourcounty Community Access Program in southeastern Kentucky diagnosed with heart disease, diabetes, asthma, hypertension, or mental illness. Principal Findings: Most of SKYCAP's staff are lay health workers from the communities they serve, and this staffing contributes to both the cost-effectiveness of the program and its ability to serve hard-to-reach persons with limited trust in the health care system. SKYCAP enrollment was effective in reducing clients' need for hospital-based care. When utilization rates in the 12 pre-enrollment and 12 postenrollment months were compared, emergency department visits dropped by 92 percent and hospitalizations dropped by 88 percent across all diagnoses. The greatest cost savings were realized for clients with diagnoses that posed the highest risk of true health emergencies, notably heart disease, and with asthma, for which only one hospitalization occurred following program enrollment. Consistent access to prescription drugs and routine health monitoring were critical to SKYCAP's achievements. The savings in unreimbursed emergency department and inpatient care alone equalled 244 percent of HRSA's investment in the CAP grant for the year. SKYCAP's success is particularly noteworthy because it operates in a chronically under-resourced region. Conclusions: SKYCAP exemplifies the potential achievements of Community Access Programs: not only has it reduced the burden of uncompensated care in a region with high rates of uninsurance and chronic disease, but it has improved the quality of life for the large majority of its clients at strikingly low cost. This outcome has been achieved by attending to clients' essential health care needs, notably prescription drugs and outpatient services. Savings have been realized primarily by the hospital sector, while costs in the form of uncompensated care have been borne primarily by physicians in the SKYCAP service area. For those physicians who are not hospital employees, this pattern of cost-shifting may make SKYCAP difficult to sustain. Loss of federal funding in this under-resourced region is also a constant threat. Implications for Policy, Delivery or Practice: The striking success and low unit cost of the SKYCAP program argue in favor of continued federal and state funding for similar community-based initiatives. SKYCAP also illustrates the potential problems of a health care system that forces underresourced localities to take on the added burden of care for those who suffer the effects of longstanding lack of health care access. Findings from SKYCAP suggest that broad implementation of its model could have a significant impact on the burden of chronic disease care nationally. Primary Funding Source: HRSA • Hospital Use by Dual Medicare-Medicaid Eligibles in Risk Health Maintenance Organizations and Fee-for-Service, California, 1991-1996 Nasreen Dhanani, Ph.D., June O'Leary, Ph.D, Elizabeth Sloss, Ph.D., Glenn Melnick, Ph.D. Presented by: Nasreen Dhanani, Ph.D., Consultant, Health, RAND Corporation, 1700 Main Street, Santa Monica, CA 90401; Tel: 310.393.0411 Ext. 6214; Fax: 310.451.7061; E-mail: dhanani@rand.org Research Objective: To establish rates of inpatient hospital use among dual Medicare-Medicaid eligibles in Medicare risk health maintenance organizations (HMO) and provide comparisons with similar populations in Medicare fee-forservice (FFS) in California from 1991 through 1996 Study Design: Enrollment data on Medicare beneficiaries from the Centers for Medicare and Medicaid Services (CMS) were linked to inpatient discharge data from the California Office of Statewide Health Planning and Development for 1991-1996. Three measures of inpatient use were estimated: (1) admissions per thousand beneficiaries, (2) total days per thousand beneficiaries, and (3) average length of stay in days. Measures were also computed by age and gender. The unit of analysis is person-year. Population Studied: The sample included all dual MedicareMedicaid eligibles, 65 years of age and older, enrolled in a Medicare risk HMO or covered by Parts A and B of traditional FFS Medicare in California between 1991 and 1996. The sample for each year was restricted to Medicare beneficiaries who were Medicaid-eligible for at least one month during that year. Beneficiaries with end-stage renal disease were excluded. Principal Findings: In California, enrollment of dual eligibles in Medicare risk HMOs increased from 5.9% in 1991 to 13.1% in 1996 and the number of person-years among all dual eligibles increased from 415,739 to 452,129 during this period. At the beginning of the period, dual-eligibles in Medicare HMOs used slightly more than half the inpatient days per thousand beneficiaries as FFS dual-eligibles. The difference narrowed over time from 2228 (HMO) vs. 3982 (FFS) days per thousand beneficiaries in 1991 to 1792 (HMO) vs. 2775 (FFS) days per thousand beneficiaries in 1996. In 1991 and 1996, the differences were due to lower rates of admission and shorter lengths of stay for dual-eligibles in risk HMOs. Several socio-demographic differences persist between dual eligibles in FFS and HMO. Conclusions: Dual eligibles in Medicare risk HMOs had lower admission rates, shorter lengths of stay, and fewer inpatient hospital days than the dual eligibles in Medicare FFS. Implications for Policy, Delivery or Practice: To date, measures of hospital use by dual-eligibles while enrolled in HMOs have not been reported. These data on inpatient utilization among dual eligibles in Medicare risk HMOs provide important baseline information that can be used to monitor this group of vulnerable beneficiaries in managed care systems. Primary Funding Source: RWJF, ASPE/DHHS • Enrolling Children in Public Insurance: SCHIP, Medicaid and the Impact of State Policy Choices Karl Kronebusch, Ph.D., Brian Elbel, M.P.H. • Factors Influencing Medicaid Recipient Switching Behavior in Medicaid MCOs Cheryl Fahlman, Ph.D. Presented by: Brian Elbel, M.P.H., Doctoral Student, Health Policy and Administration, Yale University School of Public Health, P.O. Box 208034, New Haven, CT 06520; Tel: 203.809.0517; Fax: 203.785.6287; E-mail: brian.elbel@yale.edu Research Objective: The Balanced Budget Act of 1997 established federal grants to the states to create state children's health insurance programs. This presented the states with a number of implementation choices concerning administrative models for the new programs, as well as choices about eligibility standards, enrollment simplification, crowd-out and cost sharing requirements. At the same time, the states were also implementing welfare reform. We describe the most important of these implementation choices and, using data from the Current Population Survey, we estimate the impacts of these policy choices in this multiprogram environment. Study Design: Cross-sectional, using the 2001 Current Population Survey. The probability of enrollment in Medicaid, SCHIP, Private Insurance and any Insurance was estimated using a logit model. This allows us to assess the impact of state Medicaid, SCHIP, and TANF policy choices on children’s health insurance enrollment. Population Studied: Children 17 and under in the United States with family incomes of less than 400% of the Federal Poverty Level. Principal Findings: The results indicate that SCHIP programs that are administered as Medicaid expansions are more successful in enrolling children than either separate SCHIP plans or combination programs. States that remove asset tests and implement presumptive eligibility and selfdeclaration of income have higher enrollment levels. Continuous eligibility and adoption of mail-in applications have no effect on overall enrollment. Waiting periods and premiums reduce enrollment. Stringent welfare reform reduces children's enrollment, despite federal policy that was intended to protect children from the consequences of welfare reform. The negative impacts of a number of these policy reforms substantially reduce enrollment, potentially offsetting the more favorable impacts of other policy choices. Conclusions: We estimate that if all states adopted the policy options that facilitate program use, enrollment for children with family incomes less than 200 percent of the poverty line could be raised from the current rate of 42 percent to 58 percent. Implications for Policy, Delivery or Practice: Policy makers and researchers advocate enrollment simplifications as a means to increase enrollment. However, while some of these policy choices can lead to an increase in enrollment, some do not. In addition, many welfare policies actually lead to a decrease in enrollment. If the goal is to increase enrollment in children’s health programs, policy makers can and should adopt the successful policies and avoid those that reduce enrollment. Presented by: Cheryl Fahlman, Ph.D., Senior Research Associate, HRET, 325 Seventh Avenue, N.W., Washington, DC 20004; Tel: 202.638.1100; Fax: 202.626.2255; E-mail: cfahlman@netzero.net Research Objective: The primary aim of this study was to provide an understanding of factors influencing Maryland Medicaid recipients’ MCO switching behavior. Study Design: The research built upon a large natural experiment, in which approximately 319,000 Medicaid recipients were enrolled in 9 MCO plans between June and December 1997, in conjunction with the Maryland Medicaid HealthChoice program. The study used enrollment data, feefor-service claims, and MCO encounter data to follow the cohort through their transition from fee-for-service into HealthChoice. Utilization data, both medical and drug, were collected from July 1996 to December 2000. Logistic regression techniques were used to estimate the determinants of switching behavior. Threetypes of switches were investigated including those related to initial enrollment (free switch), an interruption in Medicaid eligibility, and anniversary switches. Domains analyzed were: the recipient’s stock of health capital, health shocks, perceived returns on health investment, and perceived quality of the source MCO. Population Studied: Five thousand eight hundred and four children qualified for the asthma cohort based on age (between 5 and 18 years old at time of MCO enrollment), a minimum of 3 months of continuous Medicaid enrollment during the baseline FFS period, and evidence of asthma at baseline. Principal Findings: The majority of recipients (60.0%) did not switch MCOs at any time during their HealthChoice enrollment. Yet 40.0% experienced at least one switch during their enrollment; with a small minority of recipients (0.3%) having four switches during the study period. The multivariate models suggest that claims/encounter and eligibility data can explain some of the individual decisions regarding staying or switching MCOs among pediatric Medicaid recipients with asthma. Only 252 (4.5%) of recipients elected to take a free switch unexpectedly the method of initial enrollment (auto vs. self) did not play a significant role in the decision process. Recipients with a usual provider relationship were actually more likely to undertake a free switch, yet asthma itself does not appear to play a role in the decision to undertake a free switch. Auto-enrolled recipients who switch due to an interruption in Medicaid eligibility almost 50% more likely to switch MCOs. The presence of a usual provider relationship was not predictive of whether recipients switched or not, but the directionality showed that recipients with a relationship were slightly (17%) less likely to switch. For recipients eligible for an anniversary switch, recipients with a usual provider relationship were 32% more likely to remain with their original MCO. System-level characteristics of MCO quality did not appear to influence provide additional information to recipients’ decision process. Conclusions: Individual health stock characteristics played an inconsistent role in switching decisions. Although a large segment of the study cohort was auto-enrolled by the State, and auto-enrollment played a part in some of decisions to switch MCOS, but it was not consistent. While the presence of a usual provider relationship was important in switching decisions. The source MCO plays an important role in the decision process, becoming more pronounced with later switching decisions. Since system-level characteristics did not influence the decision process, this suggests there is some unmeasured characteristic of the MCO that plays an important role. The relationship between the smaller MCOs and FreeState was consistent throughout all the switching decisions modeled in this study, where the smaller MCOs were more likely to have recipients who made the decision to switch. Implications for Policy, Delivery or Practice: States should reduce the level of auto-enrollment for Medicaid recipients. Using a variety of outreach methods, some states have reduced their auto-enrollment rate to between one and eleven percent. Certain marketing approaches work better in different populations, yet DHMH only used written materials, and significant problems were encountered during the process. Although the startup phase of HealthChoice is completed, the program is continually enrolling new recipients and re-enrolling those with interruptions in eligibility. The state should emphasize the development of a usual provider relationship for every recipient. The State, MCO, and provider can all play a role in the development of this relationship. When designing HealthChoice, Maryland Medicaid recognized the importance of patient-provider relationship by encouraging the inclusion of historic providers in the provider network. Yet for many recipients, access to these providers did not translate into explicit decisions to follow a particular provider. Moreover, the failure to do so appears based on this study’s findings, to predispose recipients to future switches among MCOs. Primary Funding Source: Dissertation • Effect of Stable Insurance on Access and Use of Health Care for Children in Low-Income Areas of New York City Gerry Fairbrother, Ph.D., Roberta Scheinman, M.P.H., Beth Osthimer, JD, Kerry-Ann Newell, M.S. Presented by: Gerry Fairbrother, Ph.D., Senior Scientist, Health and Science Policy, New York Academy of Medicine, 1216 Fifth Avenue, New York, NY 10029; Tel: 212.822.7398; Fax: 212.822.7369; E-mail: gfairbro@nyam.org Research Objective: The purpose of this study is to examine how access and use varies by length of time uninsured or insured through public coverage for low-income children. Study Design: A cross-sectional random-digit dial survey of parents of publicly insured or uninsured children was conducted in New York City from December 2002 to February 2003. Questions about access to and use of medical care were asked about a focal child (n=1172). Length of time uninsured or insured was ascertained. Access and use were compared for children in four categories: insured for more than 12 months; insured for less than 12 months; uninsured for less than 6 months; uninsured for more than 6 months. Population Studied: The population studied was children who were publicly insured or uninsured in New York City from December 2002 to February 2003 Principal Findings: Chi-square bivariate comparisons of access and use in the four time-insured categories revealed a gradient based on time insured. The likelihood of not having a usual source of care increased with time insured (p<.01) as did the likelihood of not seeing a particular medical provider (p<.01). Long term uninsured children were also less likely to have well care visits (p<.05), prescriptions filled (p<.01), and dental visits (p<.05) than shorter term uninsured and insured children. The level of delayed or unmet need for well care visits (p<.05), prescription medication (p<.01) and dental visits (p<.05) increased with time uninsured. Multivariate analyses controlling for demographic factors confirmed the gradient evident in the bivariate results. Using children insured for a year or more as a reference group, the odds ratios increased across the time insured categories for all access and use variables. Conclusions: In this sample, presence or absence of insurance predicted access and utilization. The longer the time insured, the greater the access to and use of care; conversely, the longer the spell of uninsurance, the poorer the access to and use of care. Thus, stability in coverage is important. Implications for Policy, Delivery or Practice: It is important to put in place policies that promote stability of coverage. Such policies would include those that promote retention of eligible children in public insurance programs through streamlined and simplified renewal processes. We need to ensure that gains in simplification made due to expansion of public insurance programs in the late 1990’s are not rolled back during more difficult economic times. Primary Funding Source: Altman Foundation and William T. Grant Foundation • Why Are Latinos the Most Uninsured Racial/Ethnic Group of US Children? Glenn Flores, M.D., Milagros Abreu, M.D., Sandra Tomany, M.S. Presented by: Glenn Flores, M.D., Director, Community Outcomes; Associate Director, Center for the Advancement of Urban Children, Pediatrics, Medical College of Wisconsin/Children's Hospital of Wisconsin, 8701 Watertown Plank Road, Milwaukee, WI 53226; Tel: 414.456.4454; Fax: 414.456.6385; E-mail: gflores@mail.mcw.edu Research Objective: Latinos continue to be the most uninsured racial/ethnic group of US children, with 23% (over 3 million) uninsured (compared with 14% African-American, 12% Asian/Pacific Islander, and 8% white children who are uninsured). Among poor uninsured children, Latinos outnumber all other racial/ethnic groups, including whites. Not enough is known, however, about the risk factors for and consequences of uninsurance in Latino children. The goal of this study, therefore, was to identify risk factors for and consequences of being uninsured in Latino children. Study Design: Cross-sectional survey of 1,100 parents (900 Latino) at urban, predominantly Latino community sites, including supermarkets, beauty salons, and laundromats. Subjects were asked 76 questions on access and health insurance. “Uninsured” was defined as the parent reporting that the child had no health insurance coverage at the time the parent was interviewed. Population Studied: Parents/legal guardians of 1,100 community-dwelling children in predominantly Latino communities in the greater Boston area known to have high rates of uninsurance. Principal Findings: Uninsured Latino children (n = 167) were significantly more likely than insured Latino children to be older (mean age: 9 vs. 7 years; P<0.001), poor (89% vs. 72%; P<0.001), and to have parents who are limited in English proficiency (86% vs. 65%; P<0.001), non-US citizens (87% vs. 64%; P<0.001), have lived in the US for < 10 years (59% vs. 39%; P<0.001), married and living with the spouse (67% vs. 51%; P<0.001), and both employed (35% vs. 27%; P<0.001). 33% of parents had applied for insurance for their child but were rejected or experienced problems, 51% had unsuccessfully applied for Medicaid in the past, and only 14% had heard of the State Children's Health Insurance Program (SCHIP) available in their state. Uninsured Latinos were less likely than their insured counterparts to have a regular physician (84% vs. 99%; P<0.001), and significantly more likely to not be brought in for needed medical care due to expense (22% vs. 9%; P<0.001), lacking insurance (25% vs. 12%; P<0.001), difficulty making appointments (32% vs. 19%; P<0.001), inconvenient office hours (18% vs. 11%; P<0.001), and cultural issues (13% vs. 4%; P<0.001). After adjustment in multivariate analyses, having parents who are undocumented (odds ratio [OR], 6.3; 95% CI, 3.3-12.2) or documented (OR, 2.1; 95% CI, 1.2-3.8) immigrants, poverty (OR, 2.7; 95% CI, 1.64.6), having both parents work (OR, 2.3; 95% CI, 1.1-4.7), and the child’s age (OR, 1.1; 95% CI 1.06-1.14) were the only factors significantly associated with a child being uninsured, with neither Latino ethnicity nor any other of 5 variables (the number of siblings, parental educational attainment, marital status, English proficiency, and the number of years parents have resided in the US) associated with lack of insurance coverage. Compared with insured Latino children, uninsured Latino children had 34 times the odds (95% CI, 12.7-88.8) of having no regular physician, and were significantly more likely to not be brought in for needed medical care due to expense (OR, 2.9; 95% CI, 1.8-4.7), lack of health insurance (OR, 2.7; 95% CI, 1.7-4.2), difficulty making appointments (OR, 2.0; 95% CI, 1.3-2.9), and cultural barriers (OR, 3.4; 95% CI, 1.8-6.5). Conclusions: After adjustment, parental non-citizenship, having two parents work, poverty, and older child age are associated with being an uninsured child, but Latino ethnicity is not. The higher prevalence of other risk factors appears to account for Latino children’s high risk of being uninsured. Uninsured Latino children are significantly more likely than insured Latino children to have no regular physician, and to not get needed medical care due to expense, lack of health insurance, difficulty making appointments, and cultural barriers. Implications for Policy, Delivery or Practice: Uninsured Latino children are at great risk for having no regular healthcare provider, and face multiple financial and nonfinancial access barriers to needed medical care. The study findings indicate specific high-risk Latino populations that might benefit most from targeted Medicaid and SCHIP outreach and enrollment, including children with immigrant parents and in families where both parents work, older children, and children living in poverty. • County-Level Approaches to Increasing Access to Care for the Uninsured: The California Story Annette Gardner, Ph.D., M.P.H., James Kahn, M.D., M.P.H. Presented by: Annette Gardner, Ph.D., M.P.H., Principal Public Administrative Analyst, Institute for Health Policy Studies, University of California, San Francisco, P.O. Box 0936, San Francisco, CA 94143; Tel: 415.514.1543; Fax: 415.476.0705; E-mail: algard@itsa.ucsf.edu Research Objective: In California, counties are responsible for addressing the health care needs of the uninsured, including reducing barriers to care and ensuring appropriate use of services. The last several years have witnessed significant initiatives in some counties to create and launch county-level health care access initiatives, e.g., Santa Clara County's Healthy Kids insurance program and service expansions such as new clinic facilities. In the face of limited federal and state reform, and with new funds (e.g., Tobacco Settlement), counties have demonstrated the willingness and ability to undertake these initiatives. UCSF has been conducting research on these programs since 2001, including administering a 58-county survey describing the distribution of innovative health-care access efforts in 2002 . We anticipate re-administering this survey in early 2004, deepening our understanding of county capacity to engage in significant reform under adverse economic conditions, particularly their willingness to undertake coverage expansions. Our research objectives are: 1. Update the inventory of county-level health care access programs and compare to 2002; 2. Assess the impact of key factors like rising number of uninsured and the state budget shortfall on county ability to undertake health care access programs, particularly coverage expansions. 3. Identify the resources required by counties to undertake these initiative Study Design: We intend to re-administer the internet-based 58-county survey that was administered in early 2004. The Survey will be amended to provide a more detailed inventory of coverage expansions, e.g., funding sources, key stakeholders. We had a very good response rate with the 2002 Survey (75%) and will work to increase this rate while addressing questions surfaced by the first survey such as type of coverage approaches being used and actual number of enrollees. Population Studied: 58 California counties Principal Findings: Our findings from the 2002 study indicated that California counties as a whole had made progress in connecting people to insurance and services. While coverage expansions are desired by many they were implemented by few counties, particularly those counties with a public Medicaid managed care plan. Since the 2002 study, in the face of state and local budget shortfalls, we have seen significant activity in the area of coverage expansions for youth using tobacco funds (Prop 10). Additionally, a local funder, the California HealthCare Foundation, is providing planning and implementation grants to numerous counties to launch coverage programs. The readministration of the survey would allow us to capture these new initiatives and better assess the likelihood of future local expansions Conclusions: TBD. We will focus on changes since 2002 in distribution of access initiatives and discuss the factors that facilitate or impede progress by counties. Implications for Policy, Delivery or Practice: Though county or local-level coverage expansions are underway in California and in other states, there are some who argue that true reform can only happen at the federal level. However, in the absence of federal or state reform, it behooves decisionmakers and researchers to better understand local efforts--capacity, best practices, outcomes--and consider how these efforts might inform state and federal policy. Second, we have found it useful to think of access to care for the uninsured as a "patchwork" of initiatives, with an emphasis on understanding how these approaches--services, coverage, outreach--can best work together. For some California counties, service expansions are easier to undertake, particularly if the county has a county-run health care system. This study will allow us to increase our understanding of how access programs work in tandem • Capitation Risk Adjustment for High Risk Populations: A Test of Alternative Predictive Functions Greg Gifford, B.A., M.A., Ph.D., Dave Knutson, Michael Finch, Ph.D. Presented by: Greg Gifford, B.A., M.A., Ph.D., Research Scientist, Health, State of Minnesota, 121 East Seventh, St. Paul, MN 55101; E-mail: greg.gifford@health.state.mn.us Research Objective: To evaluate the relative performance of an untransformed ordinary least squares (UNT) vs. a square root, “complexity adjusted” (SQR) prediction model for use in risk adjusting capitation rates for adults with physical disabilities, and other high risk populations. Study Design: . Concurrent simulation, comparative analysis. Two distinct capitation risk adjustment models based on the Chronic Illness and Disability Payment System (CDPS) were tested and compared for each of the Medicaid Only and Medicare/Medicaid Dually Eligible sub-populations (N = 10,847 and N = 6,315 respectively) of the larger population of adults with physical disabilities eligible for MA in CY 1997. The two competing models were compared to evaluate the relative predictive performance of alternative mathematical forms of the prediction functions. The two models tested were distinguished by the nature of the dependent variable – CY 1997 charges/payments per member per month (PMPM): (1) untransformed, vs. (2) the square root of total payments PMPM. Both models included a prediction error adjustment the “complexity” adjustment. The models were compared with respect to both individual and group level measures of predictive performance. Population Studied: This study was conducted as part of the evaluation of the feasibility of capitation risk adjustment for a voluntary, prepaid managed care demonstration designed for the adult Medicaid (MA) eligible population with physical disabilities that began September 1, 2001 – the Minnesota Disability Health Options (MnDHO) project. Secondary data representing the population of 17,162 adults with physical disabilities eligible for MA in calendar year 1997 was used for this study. Program eligibility and diagnosis data were used to explain/predict total charges/payments for calendar year (CY) 1997. Principal Findings: For both the MA Only and Dual Eligible sub-populations, the SQR model predicted mean charges/payments more accurately than the UNT model for most of the non-random groups studied. However, the individual level results (i.e., adjusted R2) were equivocal (MA Only: R2UNT = 0.363; R2SQR = 0.365; Duals: R2UNT = 0.179; R2SQR = 0.151). Conclusions: Since non-random group level performance was more important to the feasibility of capitation risk adjustment for MnDHO than individual level performance, we concluded that the SQR function outperformed the UNT function. A reasonable interpretation of the results is that the square root model better and more efficiently accommodates any nonlinearity and interactions involved in relationships among diseases and “costs” than an UNT model. We recommended the “square root complexity adjusted” model for MnDHO based on earlier similar tests, and these results support that recommendation. Implications for Policy, Delivery or Practice: The state planned to implement this prediction function in its program payment model. In addition, to evaluate the generalizability of the square root model, we recommend its testing for other high risk populations (e.g., Medicare frail elderly). • Spill-over Effects of Medicaid Managed Care on the Uninsured Jessica Haberer, M.D., Bowen Garrett, Ph.D., Loren Baker, Ph.D. Presented by: Jessica Haberer, M.D., Medicine Fellow, Center for Primary Care and Outcomes Research, Stanford University, 117 Encina Commons, Stanford, CA 94305; Tel: 650.723.1163; Fax: 650.723.1919; E-mail: haberer@healthpolicy.stanford.edu Research Objective: In addition to affecting the Medicaid population, many policymakers and clinicians are concerned that the rise of Medicaid managed care (MMC) may negatively affect the viability and functioning of the health care safety net. If so, the care of uninsured patients may suffer. This study aims to determine if the implementation of MMC from 19942001 had spill-over effects on perceived access to care for uninsured populations. Study Design: Data from the National Health Interview Survey (NHIS) was merged with county-level data on MMC plans. Data on MMC plans included the year of implementation, mandatory versus voluntary enrollment, use of primary care case management and use of health maintenance organizations. Five measures of access to care were investigated: delay in getting care due to cost, needing but not getting care due to cost, having a usual source of care, using emergency services as a usual source of care, and having any doctor visits in the past year. Multiple regression analyses were conducted using the first difference method to account for fixed county-level effects. All findings were controlled for gender, age, race, ethnicity, income, health status, and education. Because the NHIS was redesigned in 1997, all regressions were conducted on data from 1994-1996 and 1997-2001 separately, as well as from 1994-2001 together. Results were then compared to evaluate any influence of the redesign. Population Studied: The NHIS samples the noninstitutionalized, civilian population of the United States. This study investigated the uninsured population in the NHIS from 1994-2001. The uninsured were divided into three age groups: adults (25-64 years old), young adults (18-24 years old) and children (less than 18 years old). Uninsured individuals with chronic heart disease, diabetes and asthma were also studied separately, as were individuals with varying levels of selfreported health status. Principal Findings: None of the access to care measures indicated a consistent effect of any type of Medicaid managed care on the uninsured. Some statistically significant effects were seen in the age and chronic disease subgroups of the uninsured; however, they showed improvement in some measures and worsening in other measures within each subgroup. Further, no consistent differences were found between the uninsured with fair or poor health and the uninsured with good, very good or excellent health. Conclusions: The implementation of Medicaid managed care from 1994-2001 did not have a systematic effect on perceived access to care for the uninsured. Implications for Policy, Delivery or Practice: This study suggests that the implementation of Medicaid managed care has not negatively affected the ability of the health care safety net to care for the uninsured. This finding, however, only reflects perceptions of access to care. Further studies are needed to determine if the actual delivery of services has been affected. Primary Funding Source: AHRQ • Examination of Total Healthcare Expenditures for Patients on Long Acting Opioid Medications for the Treatment of Chronic Pain in a Medicaid Population Kevin Hawkins, Ph.D., Patricia Grossman, PharmD, Lucinda Orsini, DPM, M.P.H., Ron Ozminkowski, Ph.D., Onur Baser, Ph.D. Presented by: Kevin Hawkins, Ph.D., Senior Economist, Outcomes Research & Econometrics, The Medstat Group, 777 E. Eisenhower, Ann Arbor, MI 49230; Tel: 734.913.3145; Fax: 734.913.3200; E-mail: kevin.hawkins@medstat.com Research Objective: To identify factors associated with total annual healthcare expenditures for Medicaid patients prescribed a Long Acting Opioid (LAO) medication for chronic pain management. The LAOs considered were transdermal fentanyl (TF), controlled release morphine sulfate (CRMS) and controlled release oxycodone (CRO). Study Design: Medicaid beneficiaries from three states during 1999 - 2000 were studied. Patients were followed for at least one year, starting with their first LAO prescription in 1999. Patients who did not have a LAO in the six months prior to that index date were labeled ‘incident’, while patients with a history of LAO use were labeled as ‘prevalent’. Patients were then grouped by the opioid they received on the index date. Disease-type, demographics, health status, healthcare utilization, and expenditures were measured and compared among the three LAO cohorts. LAO costs and total annual healthcare expenditures by LAO cohort were compared using descriptive and multivariate analyses, for the incident and prevalent patient samples. Population Studied: 13,360 patients were included in the study. Of these, two thirds were incident and one third was prevalent. Forty nine percent of the patients were on CRO, 23% on TF, and the remaining 28% were on CRMS. Principal Findings: Descriptive results indicate that incident (I) and prevalent (P) LOA annual cost for CRO ($1,361 I; $4,146 P) were significantly higher than costs observed for TF ($1,202 I; $3,061 P) or CRMS ($919 I; $3,572 P) (p<0.01). After controlling for confounding characteristics, total annual health expenditures in the CRO incidence sample were similar to the CRMS sample. However, total annual healthcare expenditures for the CRO population were significantly lower than expenditures for the TF population, with an annual cost savings of about $916 (p<0.01). Economic differences were not noted among the prevalent LAO populations. Conclusions: : In the incident sample, patients on TF cost about $916 more per year (p<0.01), on average, compared to similar CRMS and CRO patients. Expenditure differences were not observed in the prevalent populations. Implications for Policy, Delivery or Practice: Total cost, not just pharmaceutical costs, should be considered when making policy decisions about insurance coverage for LAO drugs. Primary Funding Source: Purdue Pharma • Quality of Care Variations among Diabetic Medicaid Beneficiaries: What the Data Can Tell Us Ronda Hughes, Ph.D., M.H.S., R.N., Barbara Starfield, M.D., M.P.H., Helen Burstin, M.D., M.P.H. Presented by: Ronda Hughes, Ph.D., M.H.S., R.N., Health Scientist Administrator, Center for Primary Care, Prevention, and Clinical Partnerships, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20852; Tel: 301.427.1578; Fax: 301.427.1595; E-mail: rhughes@ahrq.gov Research Objective: This multi-state study evaluates the impact of Medicaid managed care on resource utilization among adult beneficiaries with diabetes by type of Medicaid eligibility and sets forth the methodological challenges in using claims data from multiple states. It is hypothesized that as patients move into managed care arrangements, the clinical detail of the administrative data diminishes, therefore limiting the ability to assess the quality of care. Study Design: This study uses State Medicaid Research Files (SMRF) maintained by the Centers for Medicare and Medicaid Services for 1996 through 1998, to assess outpatient and inpatient utilization among all adult Medicaid beneficiaries, age 18 and over. This includes patients that qualify for Medicaid services under the Aid for Families with Dependent Children (now Section 1931 eligibility) income provisions, the blind and disabled part including Supplemental Security Income (SSI), and the aged provisions of the Medicaid program. SMRF data were used to assess the diagnostic information associated with specific health care services, including rates of hospitalizations for diabetes and other conditions, diabetic measures including use of injectibles and oral hypoglycemics, utilization of outpatient services, and co-morbidities. Diabetic patients were identified if they had at least two diagnoses for diabetes during the year. Diabetic adult beneficiaries were further stratified by: 1.) length of enrollment in Medicaid (e.g., individuals enrolled in Medicaid 0 – 5 months, 6 to 12 months, and at least 12 continuous months); 2.) enrollment in managed care or fee-for-service (FFS); and 3.) type of eligibility (e.g., AFDC, blind and disabled non elderly, and elderly). Population Studied: Adult Medicaid beneficiaries with diabetes. Principal Findings: Using available data, this study found that frequencies of receipt of care for diabetics was limited by whether Medicaid patients were in managed care. Hospitalization rates with a primary diagnosis of diabetes ranged from 8.1 to 364.9 and the average number of emergency department visits was 2.6. Utilization differences in hemoglobin A1c measurement at least once in 12 months or a lipid profile in the past two years varied among subpopulations and states by as much as 22.6 percent. Hospitalizations, emergency department visits, and rates and having a retinal eye exam increased with age and number of co-morbidities. There were significant variations in number of outpatient visits and type of services utilized among the different enrollee groups, associated with the completeness of the clinical content of the data. Conclusions: As the clinical content of the SMRF for outpatient services was limited, findings must be interpreted cautiously. The methodological issues involved in using the Medicaid claims data for purposes of government oversight and quality of care improvement initiatives can prohibit use of the data for assessing the quality of care if outpatient data is available for only certain Medicaid sub-populations. Managed care penetration, effects the ability to examine utilization for diabetics due to data availability. Implications for Policy, Delivery or Practice: The implications of these findings for health care policy and future research will be discussed, including future research using state maintained Medicaid claims files as well as implications for investigation of utilization among dual eligibles and individuals with multiple chronic conditions. • Patterns of Health Insurance Coverage in a Rural Hispanic Border Community Tricia Johnson, Ph.D., Amy Bartels, M.P.H. Presented by: Tricia Johnson, Ph.D., Assistant Professor, Department of Health Systems Management, Rush University Medical Center, 1700 West Van Buren Street, TOB Suite 126B, Chicago, IL 60612; Tel: 312.942.7107; Fax: 312.942.4957; Email: tricia_j_johnson@rsh.net Research Objective: This study uses micro-data from a unique community health data system to study the patterns of insurance coverage for children who utilized health care services over a three-year period of time across race and ethnicity, age, and geographic proximity to health care providers. Study Design: We use data from health care providers and health-related organizations in Yuma County, Arizona to create the Yuma County Community Health Data System (YCHDS). The YCHDS merges data from the hospital system, federally qualified community health centers, private pediatric practices, state Medicaid and immunization data, employerbased health insurance and a survey of primarily Hispanic neighborhoods. We create seven main patterns of health insurance coverage experienced by children utilizing some form of health care in each of three years. The patterns are comprised of Medicaid or SCHIP, private insurance and spells without insurance. Population Studied: Yuma County is a sparsely populated county spanning more than 5,000 square miles in the southwestern corner of Arizona. The county’s proximity to the Mexican border; the presence of relatively large numbers of migrant workers; and the presence of a relatively small population in a large geographic area are characteristics that make it especially difficult for children to obtain adequate health insurance coverage and to access primary care. More than 60% of children age 0-19 are Hispanic. The sample includes 19,020 children ages 0 to 17 in 1999 who utilized at least one service in each of the three years of study, 1999 to 2001. Principal Findings: Nearly three-quarters of the children in the sample have a single consistent source of coverage, either Medicaid or private insurance, across the three years. An additional 16% switch between Medicaid and private insurance with no uninsured spells. Overall, 89% of the children experienced no uninsured spell of medical care during the three-year study period, while 11% of the children experienced at least one spell of medical care without insurance. Nearly 7% of the children switched between Medicaid and no insurance. Conclusions: We show that having no insurance coverage is not synonymous with receiving no medical care. In this very low income community, where nearly one in four families with dependent children live below the federal poverty line, more than one in ten children experienced at least one spell of medical care without insurance over three years. Most children seeking medical care, however, experience one year or less without insurance. Implications for Policy, Delivery or Practice: In an effort to reduce Medicaid expenditures, states such as Arizona are considering a shift to more frequent re-enrollment and eligibility checks. These changes, coupled with the lack of assistance with re-enrollment processes, are likely to increase churning between Medicaid and no insurance. Maintaining a consistent source of health care coverage is likely to reduce unnecessary and inappropriate health care utilization for lowincome children. • The Impact of Demographic and Labor Force Participation Changes on Health Care Utilization Tricia Johnson, Ph.D., William Johnson, Ph.D., Amy Bartels, M.P.H. Presented by: Tricia Johnson, Ph.D., Assistant Professor, Department of Health Systems Management, Rush University Medical Center, 1700 West Van Buren Street, TOB Suite 126B, Chicago, IL 60612; Tel: 312.942.7107; Fax: 312.942.4957; Email: tricia_j_johnson@rsh.net Research Objective: The first members of the baby boom generation are now approaching the ages when the incidence of disability increases. The increases are likely to overwhelm public and private insurance and the welfare plans that serve persons with disabilities unless the plans are significantly changed. This project examines the effects of the changing size and composition of the baby boom-aged labor force on the types of disabilities qualifying for Social Security Disability Insurance (SSDI) and Medicare for men and women age 40 to 64. In addition, we explore how health care utilization and expenditures differ across disabling conditions. Study Design: This study is a retrospective cross-sectional analysis of disability benefits and health care utilization by disabled persons who receive SSDI. The number of disabled persons age 40 to 64 who meet SSDI program requirements and receive benefits is estimated using the SIPP. The Medicare Current Beneficiary Survey (MCBS) is used to examine differences in the types of disabling conditions for men and women. The absolute growth in SSDI and Medicare participation by disabled persons is projected through 2025 by type of disabling condition, and a series of simulations are used to explore how changes in the rates and types of medical conditions, composition of the labor force and other factors will impact SSDI and Medicare. Population Studied: Respondents age 40 to 64 in the 1996 Panel of the SIPP and Medicare beneficiaries age 40 to 64 who participated in the MCBS between 1996 and 2000. Principal Findings: Less than one-third of disabled men and women receive SSDI and Medicare benefits, and a substantial proportion receive no benefit. The proportion of women eligible for SSDI will continue to increase, but remain constant for men through 2010, with the largest growth for women age 55 to 64. A smaller percentage of women will continue to qualify for SSDI and Medicare, due to lower labor force participation rates and lesser recent work experience. Results are compounded by the changing nature of disabilities, with a disproportionate share of disabled women suffering mental illnesses and musculoskeletal conditions. The inherent uncertainty in these types of illnesses results in a greater difficulty in demonstrating that the illness is permanently disabling and a greater likelihood of a disability application being denied. Conclusions: A shift to more chronic soft tissue illnesses and mental disorders could greatly impact the costs of health care for Medicare and other payers, particularly if these disorders are associated with more intensive service utilization over time. Changes in the types of disabling conditions, coupled with an increase in the population age 40 to 64, are likely to dramatically increase the costs and utilization of health care. Implications for Policy, Delivery or Practice: The number of disabled persons qualifying for SSDI and Medicare is expected to increase by 28 percent by the year 2010, and the impact will be even greater over the next 25 years, as the baby boom generation reaches the age of retirement. Gaps and duplications of coverage among programs providing disability benefits are potentially serious problems. Primary Funding Source: Disability Research Institute • Utilization of Medical Specialists by Immigrants in British Columbia, Canada Alice Chen, MC, M.S.c, Ph.D. Candidate, Arminée Kazanjian, DrSoc, Arminée Kazanjian, DrSoc, Marie Desmeules, M.Sc., Doug Manuel, M.D., Bilkis Vissandjée, Ph.D., Elizabeth Ruddick Presented by: Arminée Kazanjian, DrSoc, Professor, Health Care and Epidemiology, University of British Columbia, 5804 Fairview Avenue, Vancouver, B.C., V6T 1Z3; Tel: 604.822.4618; Fax: 604.822.4994; E-mail: a.kazanjian@ubc.ca Research Objective: This study looks at access to medical specialties by examining patterns of physician service utilization by immigrants in British Columbia Study Design: As part of a large Canadian immigrant health study, a database consisting of information on immigrants who landed in Canada between 1985 and 2000 and whose stated destination was British Columbia was linked to that province’s administrative health records for the same time period. Since most physicians in B.C. are paid directly by the province for each service provided, the provincial health records contain rather comprehensive data on physician services received by each immigrant. Using such administrative databases allows the analysis of the utilization records of the entire cohort of immigrants during the period under study while avoiding sampling and reporting biases inherent in other types of study design. Population Studied: All immigrants who landed in Canada between 1985 and 2000 and whose stated destination was British Columbia. Controls were randomly selected from the rest of B.C.’s population and individually matched to the immigrants by age, sex, health region and landing year. Principal Findings: Close to 420,000 immigration records were successfully linked to health records. Results show that, overall, immigrants made fewer visits to physicians than controls during the study period. Of the physician visits, the proportion of visits to specialists is lower for the immigrants. However, there is variation by specialty and the condition being treated. Immigrants are much less likely than controls to visit psychiatrists. Conclusions: In general, immigrants receive more of their health care from general practitioners than specialists. There are differences among specialties. Implications for Policy, Delivery or Practice: As an immigrant receiving country, Canada is becoming increasingly diverse. And since the majority of recent immigrants come from non-European regions, the challenge for health policy makers and service planners is to meet the health care needs of such a culturally diverse population. Under the universal health care principle operative in Canada, there should be few financial barriers to health services for immigrants. The differential rates of utilization of physician services, especially specialist services, between immigrants and non-immigrants in this study suggest that factors other than health care coverage affect utilization. Other barriers to services need to be identified in order to achieve equitable access. Primary Funding Source: Canadian Population Health Initiative • Impact of Health Insurance Benefits on Wages Patricia Ketsche, Ph.D., M.H.A., William Custer, Ph.D. Presented by: Patricia Ketsche, Ph.D., M.H.A., Assistant Professor, Institute of Health Administration, Georgia State University, MSC 4A1473, 33 Gilmer Street, S.E., Unit 4, Atlanta, GA 30303-3084; Tel: 404.651.2993; Fax: 404.651.1230; E-mail: pketsche@gsu.edu Research Objective: Economic theory suggests workers will opt for employment without health insurance if cash wages exceed the value the worker places on coverage plus wages paid in alternative employment with benefits. Therefore, there must be a negative wage differential for workers with coverage when compared to similarly productive workers without coverage. This wage penalty is difficult to identify empirically because of the correlation between productivity, income and coverage. Identifying the wage differentional is important to understanding risk pooling and estimating the effect of health policy. This study contributes to the existing literature by using a unique data set to estimate wages as a function of health insurance while controlling for total compensation. Study Design: We use Georgia’s 159 counties to isolate distinct labor and health care markets and identify distinct equations for total compensation and wages. Total compensation at the firm is estimated as a function of firm characteristics, the total productivity of the labor force as determined by their age and tenure, and the differential cost of labor in each market that is determined by the local cost of living, educational level, and population density. The composition of that compensation in terms of wages depends upon firm characteristics, worker characteristics such as age and gender that determine the demand for coverage, and local health care market characteristics that influence access and cost of care in the absence of coverage. The correlation between error terms implies joint estimation of the two equations. The key variable for this study is the effect of coverage offered on wages. Population Studied: We use survey data collected under a State Planning Grant project from over 1400 Georgia employers in 2002. The mail survey collected information about all benefits offered including health insurance and some work force characteristics. From the ES202 file it was possible to identify total wages paid, number of employees and industry. We imputed total average compensation for the workforce by adjusting wages for benefits using estimates of average costs obtained from the Bureau of Labor and Statistics. Principal Findings: Although a reduced form equation identifies a positive relationship between wages and coverage, simultaneous regression while holding total compensation constant identifies a significant negative effect of health insurance on average cash wages. We identify an average wage reduction of 9 percent at firms offering health insurance. Conclusions: Workers with employment-based health insurance receive bear the cost through reduced cash compensation. Implications for Policy, Delivery or Practice: Wages adjust downward at firms offering health insurance. Expanding coverage at firms currently not offering benefits will reduce wages to workers at those firms. In order for tax credits to expand coverage, they must be sufficiently generous that any wage reduction to employees is less than or equal to the value placed on coverage by those employees. This will create a potential windfall to firms where workers currently receive part of their compensation in the form of health insurance. Primary Funding Source: HRSA • Is Participation of Low-Income Elderly Cancer Patients in Medicaid Associated with Breast Cancer Diagnosis at Earlier Stages of the Disease? Siran Koroukian, Ph.D., M.S.N., M.H.A., Thomas Cook, M.P.H., Fang Xu, M.S., Kristen Mikelbank, M.A., David Litaker, M.D., Ph.D., Gregory Cooper, M.D. Presented by: Siran Koroukian, Ph.D., M.S.N., M.H.A., Assistant Professor, Epidemiology and Biostatistics, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, OH 44106-4945; Tel: 216.368.3197; Fax: 216.368.3970; E-mail: sxk15@case.edu Research Objective: Advanced-Stage Cancer (ASC) is a strong predictor of poor cancer prognosis, and, for cancers that are amenable to screening, ASC can be indicative of poor access to and/or utilization of cancer screening services. Although Part B Medicare provides coverage for many cancer screening services, the coinsurance and deductibles associated with the receipt of such services can be prohibitive to low-income Medicare beneficiaries (LIMBs). Medicaid, as a public source of supplemental health insurance coverage (SHIC) is likely to at least partially remove financial barriers, and improve access to and utilization of such services among LIMBs. The aim of this study is to determine whether participation of LIMBs in Medicaid is associated with decreased likelihood of ASC. Study Design: Older women diagnosed with breast cancer during the 3-year period 1996-1998 were identified through the Ohio Cancer Incidence Surveillance System (OCISS), a data source with an estimated 95% rate of completeness relative to breast cancer. The study cohort was limited to women 70 years of age or older to increase the likelihood they were enrolled in Medicare. OCISS records were linked with Ohio Medicaid enrollment files, using patient identifiers. Individuals also enrolled in Medicaid, or the dually eligible, were identified as such if OCISS and Medicaid records were linked successfully, and if the patient had been enrolled in Medicaid at least 3 months prior to cancer diagnosis. Individuals were categorized as having ASC, if the OCISS record carried a Surveillance, Epidemiology, and End Results (SEER) summary stage of Regional, Distant, or Unstaged/Unknown stage of cancer. The geocoded residence address of the patient at the time of cancer diagnosis was used to retrieve variables from the U.S. Census Bureau at the census tract level to act as a surrogate marker for cancer patients’ socioeconomic status (SES). Factor analysis of selected census indicators was used to create a smaller set of composite measures which reflect dimensions of concentrated disadvantage, socioeconomic status and educational attainment, residential stability and neighborhood age structure. Hierarchical logistic regression models were developed to account simultaneously for individual- and community-level attributes, and to assess the association of ASC with Medicaid status after adjusting for census tract SES. In particular, we tested an interaction between Medicaid status and SES to assess the effect of Medicaid status at various levels of social or economic disadvantage. Population Studied: Medicare beneficiaries diagnosed with breast cancer. Principal Findings: A total of 9,433 patients 70 years of age or older were identified through OCISS, of whom, 9.1% were identified as dually eligible. The proportion of patients diagnosed with ASC was 47.9% and 33.1% respectively in the dually eligible and all others (p < 0.001). Medicaid status was independently associated with ASC both in bivariate and multivariate models, adjusting for SES (adjusted odds ratio: 1.71, 95% confidence interval: 1.48-1.99). Of note, the interaction term of Medicaid status and income was negative and borderline statistically significant (coefficient: -0.14, p=0.053), suggesting a modest benefit of receiving SHIC relative to ASC, with increasing levels of poverty at the community level. Conclusions: The findings suggest that Medicaid status may have little or no additional benefits relative to access to and/or utilization of cancer screening services. However, due to the use of SES measures at the community level, we may not have been able to fully address residual confounding. Implications for Policy, Delivery or Practice: The findings do not support the notion that increased access to care may be associated with greater utilization of cancer screening services. Other dimensions of access to and use of cancer screening services in this economically disadvantaged population need to be evaluated. Primary Funding Source: NCI, American Cancer Society • Effects of Medicaid Managed Care and Medicaid Managed Care Penetration on Potentially Avoidable Maternity Complications Sarah Laditka, Ph.D., James Laditka, D.A., Ph.D., Kevin Bennett, M.S. Presented by: Sarah Laditka, Ph.D., Associate Professor and MHA Program Director, Health Services Policy and Management, Arnold School of Public Health, University of South Carolina, 800 Sumter Street, HESC Building, Room 116F, Columbia, SC 29208; Tel: 803.777.1496; Fax: (803) 7771836; E-mail: sladitka@gwm.sc.edu Research Objective: We examined differences in pregnancy outcomes associated with prenatal care in Medicaid managed care (MMC) and Medicaid fee-for-service (MFFS). MMC may enhance care coordination and improve outcomes. Alternatively, it may provide incentives for under-provision of services. We also examined the impact of MMC penetration. As penetration increases, outcomes may improve for MMC enrollees, due to knowledge and efficiency gains associated with increasing program size. Study Design: The outcome studied was Potentially Avoidable Maternity Complications (PAMCs). The indicator, defined by obstetricians/gynecologists, uses hospital discharge data to identify pregnancy-related complications that may often be prevented through accessible prenatal care of reasonable quality. We used year 2000 State Inpatient Data for Florida, Maryland, and New York, and also pooled 1995-2000 data for New York, supplemented with the Area Resource File and the American Hospital Association Annual Survey. To approximate population risks, data were restricted to delivery hospitalizations. Multi-level logistic analysis estimated PAMC risk, with a random effect at the level of patients’ counties of residence. MMC penetration was defined as the MMC rate among deliveries covered by Medicaid in the patient’s residence county in New York and Florida, and as the rate in the hospital’s county in Maryland. Individual controls included age, race, and comorbidities. Area controls included education, income, and health system measures. Population Studied: 74,627 delivery hospitalizations for women enrolled in Medicaid in Florida, 20,732 in Maryland, and 84,300 for year 2000 in New York. 547,077 such hospitalizations for New York, 1995-2000. Principal Findings: In Florida, 14.6% of Medicaid deliveries were in MMC. PAMC risk did not differ between MMC and MFFS. In Maryland, 77.5% of deliveries were in MMC, where PAMC risk was 52% lower than in MFFS (p<.0001). In New York, 16.2% of deliveries were in MMC, where PAMC risk was 26% lower than in MFFS (p=.0001). Penetration did not affect risk in annual data. In the pooled New York State analysis, each 1% increase in penetration was associated with an average 1% PAMC risk reduction for those in MMC (p<.001). There was also evidence that the risk reduction increased with increasing penetration (p<.0001). Conclusions: MMC was not associated with increased PAMC risk. In two states, PAMC risk was notably lower in MMC than in MFFS. MMC may offer successful care management, lowering the risk of adverse outcomes, or provide other services to produce accessible and successful care. Lack of evidence of an impact of MMC penetration in annual data may result from limited data. Pooling six years of New York State data produced results suggesting MMC becomes more effective as its penetration in patients’ counties increases. Implications for Policy, Delivery or Practice: MMC may provide comparable or better care for pregnant women than MFFS. Outcomes associated with prenatal care were improved for those in MMC as MMC penetration increased. Although based on the experience of only one large state, this result suggests that MMC performance may benefit from effects associated with program size. Primary Funding Source: HRSA • Distance to the Nearest Public HIV Test Site: Does It Matter? Arleen Leibowitz, Ph.D., Stephanie Taylor, Ph.D., M.P.H., Paul Ong, Ph.D. Presented by: Arleen Leibowitz, Ph.D., Professor, Policy Studies, UCLA, 6345 Public Policy Building, MC 165606, Los Angeles, CA 90095; Tel: 310.206.8653; Fax: 310.206.0337; Email: arleen@ucla.edu Research Objective: Efforts have been made to provide free HIV tests to higher-risk, uninsured or lower income persons by offering tests at publicly-funded clinics and locating test sites conveniently nearby. This has lead to geographic variations in HIV-testing site availability. Persons required to travel greater distances to publicly-funded test sites might forego HIV-testing because of the time costs involved. However, the relationship between test site proximity and individuals’ HIV-testing has gone relatively unexamined. This study examined the relationship between HIV-testing and the distance between individuals’ residences and the nearest publicly-funded HIV test site to address two questions, “Are people living further away from public HIV-testing sites less likely to test for HIV at public sites?”. If so, “Are those living further away more likely to forgo HIV testing entirely, or do they simply go to other types of test sites?” Study Design: Data was from the 1999 Los Angeles (L.A.) County Health Survey, a random digit-dialed population survey. Nested logistic regression analyses were conducted to examine if distance to the nearest public test site was related to: 1) testing/not testing, and, 2) among those testing, which of three types of sites were utilized: a) public b) physician office/laboratory or c) other sites such as at-home tests. ArcGIS software provided measures of distance between residences and nearest public test sites in miles. Travel time was estimated using the distance measures and reports of car ownership. Interactions terms were used to estimate if distance was a greater barrier to testing for persons: 1) without insurance, 2) having lower incomes (less than twice the federal poverty level), 3) having higher-risk sex behaviors (not always using condoms and having more than one partner), or 4) without regular sources of care. Models included weights and ten individual characteristics related to HIV-testing. Population Studied: The sample was a 1999 random probability sample of L.A. adults (n=5,414). A third were Latino, 10% were African American, 12% were Asian Pacific Islanders, and 4% were “Other”. Principal Findings: Persons living further away from public test sites were less likely to test for HIV. This effect was stronger for low income persons. Persons with higher-risk sex behaviors living further from public test sites were more likely to substitute testing at a physician office or other type of site. All these results held regardless of whether the mile or time measure was used. However, uninsured persons were affected only by their travel time and were affected in two ways – those living further away were more likely to go without testing than those with greater proximity to public test sites and also more likely to test at a non-physician site instead of a public site. Travel time for all was negatively related to the likelihood of testing, even for travel times of less than 20 minutes. Conclusions: The distance to free HIV-testing clinics matters. Lower income and uninsured persons appeared most vulnerable to test site location. Implications for Policy, Delivery or Practice: Providing very convenient testing opportunities to those who are poor, uninsured, or engaged in higher-risk behaviors might increase voluntary testing and counseling for HIV. • Medicaid Managed Care and Children’s Access and Use of Health Care Services Ana Lopez-De Fede, Ph.D Presented by: Ana Lopez-De Fede, Ph.D, Research Associate Professor, Institute for Families in Society, University of South Carolina, 937 Assembly Street, Carolina Plaza, Columbia, SC 29208; Tel: 803.777.5789; Fax: 803.777.1793; E-mail: adefede@sc.edu Research Objective: Although recent literature has posited numerous accounts of the effect of Medicaid managed care, there is no conclusive evidence that enrollment in Medicaid managed care leads to increased access to health care services and cost containment. The primary objective was to analyze the impact of managed care on cost and access to and use of health care services among Medicaid recipients. Study Design: Medicaid claims data were used to examine differences and changes in cost, access to care and use of health services among two groups of children, i.e., Medicaid managed care Physician Enhanced Managed Care Program (PEP) enrollees and recipients enrolled in Medicaid fee-for services (FFS). A retrospective quasi-experimental design with non-equivalent groups was used to examine costs, and access to and use of health care services. Multiple logistic regression analysis examined these factors and enrollment in Medicaid managed care. Population Studied: The study population consisted of children ages birth to eighteen continuously enrolled for a minimum of ninety days in either in the Medicaid managed care primary care case management initiative (PEP) or the Medicaid managed care fee-for-services initiative (FFSM). Medicaid administrative data was used to examine the costs and access to and use of health care services of children in the study population, with paid claims from July 1, 1999 to June 30, 2000. Principal Findings: Of the 34,926 Medicaid recipients 10,502 (30 percent) were identified as enrolled in managed care. Average annual costs ($1,167) were lower for children enrolled in Medicaid managed care compare to children enrolled ($1,893) in Medicaid fee-for-services. Younger children with chronic condition were more widely represented in Medicaid managed care and had higher rates of hospitalizations; visits to specialty care providers, and related costs. Conversely, lower adherence to primary care services was widespread among children in Medicaid managed care. Twelve percent had a primary care visit, and fewer than 50 percent had an annual routine physical exam. Overall, Medicaid managed care enrollment was associated with older children and lower access to and use of health care services Conclusions: The results of this study suggest that children in Medicaid managed care experience far more difficulties accessing health care than those in Medicaid fee-for-services. Prevalence and utilization of health care services in a Medicaid managed care population of children were lower than previous estimates reported nationally and by health maintenance organizations. Implications for Policy, Delivery or Practice: The study findings support the use of claims to measure health care utilization as a reasonable proxy for access to health care services. As a result of the Balanced Budget Act which eliminated the need for states to implement a waiver to institute mandatory managed care programs, the share of children enrolling in Medicaid managed care programs are likely to increase over time. Despite the implementation of voluntary managed care, knowledge is needed on its effect. This research addressed the information gap by examining the effect of managed care on access and use of care among children enrolled in Medicaid in South Carolina, Primary Funding Source: Science Department of Health and Human Services • Primary Care Practices May Lack the Capacity To Adequately Treat Asthma Robert Lowe, M.D., M.P.H., David Nicklin, M.D., Staci Maroney, M.P.H., A. Russell Localio, JD, M.S., Donald Schwarz, M.D., Sankey Williams, M.D. Presented by: Robert Lowe, M.D., M.P.H., Associate Professor, Center for Policy and Research in Emergency Medicine, Oregon Health & Science University, 3181 S.W. Sam Jackson Park Road, CR114, Portland, OR 97239; Tel: 503.494.7134; Fax: 503.494.4640; E-mail: lowero@ohsu.edu Research Objective: Asthma is prevalent among inner city, low-income residents, including many Medicaid beneficiaries, and efforts have been made to encourage these patients to seek care from their primary care providers (PCPs) rather than in the ED. The purpose of this study was to determine whether PCPs serving a Medicaid population had the equipment and capacity to treat asthma according to guidelines issued by the National Asthma Education and Prevention Program (NAEPP). Study Design: This observational study consisted of on-site data collection at primary care practice sites affiliated with a Medicaid HMO, conducted between June 1999 and February 2000. At each site, office staff were asked about availability of equipment recommended under NAEPP guidelines. Population Studied: Three hundred fifty-three primary care practices affiliated with a single Medicaid HMO were studied. Principal Findings: Of 337 sites treating children, 93 (28%) lacked nebulizers for bronchodilators. Of 338 sites treating adults, 135 (40%) lacked peak flow meters. Office staff reported having inhalers with spacers at only 262 (74%) of 353 sites. Of 325 sites asked if they had written, customizable action plans for asthmatics, 76 (24%) reported having them but only 49 (15%) were able to provide a copy. Conclusions: In this sample of primary care practices serving Medicaid patients, office staff reported that many practices did not have the equipment to comply with accepted guidelines for the management of asthma. This study is limited by the data collection process, in that physicians were not interviewed and some office staff may not have been aware that the equipment was available. Nevertheless, the results suggest that many practices may not be equipped to provide care for asthmatics according to NAEPP guidelines. Implications for Policy, Delivery or Practice: Availability of a "medical home" through Medicaid managed care is not sufficient to assure access to necessary services at that medical home. In addition to focussing on access to primary care, Medicaid managed care must focus on the comprehensiveness of services provided. Primary Funding Source: AHRQ • Has Managed Care Improved Accessibility of Urgent Care for Medicaid Patients? Robert Lowe, M.D., M.P.H., Staci Maroney, M.P.H., Sankey Williams, M.D., Donald Schwarz, M.D., M.P.H., M.B.A., Lucy Wolf Tuton, Ph.D., A. Russell Localio, JD, M.S. Presented by: Robert Lowe, M.D., M.P.H., Associate Professor, Center for Policy and Research in Emergency Medicine, Oregon Health & Science University, 3181 S.W. Sam Jackson Park Road, CR114, Portland, OR 97239; Tel: 503.494.7134; Fax: 503.494.4640; E-mail: lowero@ohsu.edu Research Objective: A 1994 study described barriers for Medicaid patients seeking urgent care from primary care providers. Since 1994, much of Medicaid has shifted from feefor-service to managed care, partly in the hope that managed care's assignment of patients to "medical homes" would improve access to care. The objective of our study was to determine the accessibility of urgent primary care visits among providers in one Medicaid managed care organization (MCO). Study Design: Following a protocol adopted from the 1994 study, research staff acting as simulated patients called primary care practices, randomly selected from sites contracting with the Medicaid MCO. Three different staff members called each practice on different weekdays, indicating that they were Medicaid patients assigned to the practice and needed urgent appointments. Staff followed a scripted algorithm, describing their symptoms and specifically requesting appointments within 24 hours and evening appointments. Calls were categorized on a 6-point scale, based on promptness of available appointments. If the 3 calls to a practice led to different results, the median result was used for analysis. Population Studied: Three hundred fifty-three primary care practice sites affiliated with the Medicaid managed care organization were randomly selected from the primary care practice roster of the MCO. Of these sites, 44 were excluded because they had closed permanently or for reasons precluding the simulation (e.g., office personnel asked for a valid Medicaid ID number), leaving 309 sites in the study. Principal Findings: Of the 309 eligible sites, no appointment could be made at 75 (24%) of sites - often because the practice was not open. Callers could be seen the same or next day at 168 (54%) of sites (compared to 35% in the 1994 study), and the same or next evening at 23 (7%) of sites (the same proportion as in the 1994 study). Conclusions: Because the 1994 study included many communities and our study includes only one, formal comparison of results is not meaningful. However it is clear that, in primary care sites affiliated with this Medicaid MCO, access to urgent care remains limited. Implications for Policy, Delivery or Practice: Providing a “medical home” to Medicaid beneficiaries through managed care has not solved the problem of access to care. Until policymakers succeed in implementing and enforcing requirements for prompt access to care, even Medicaid beneficiaries who have designated primary care providers may be forced to rely on emergency departments for urgent care. Primary Funding Source: AHRQ • Impact of Oregon Health Plan Cutbacks on Emergency Department Use Robert Lowe, M.D., M.P.H., K. John McConnell, Ph.D., Jodi Lapidus, Ph.D., Cody Weathers, BS, Annette Adams, M.P.H., Beverly Bauman, M.D. Presented by: Robert Lowe, M.D., M.P.H., Associate Professor, Center for Policy and Research in Emergency Medicine, Oregon Health & Science University, 3181 S.W. Sam Jackson Park Road, CR114, Portland, OR 97239; Tel: 503.494.7134; Fax: 503.494.4640; E-mail: lowero@ohsu.edu Research Objective: The Oregon Health Plan (OHP), begun in 1994, expanded Oregon's Medicaid program to cover an additional 100,000 Oregonians. In March, 2003, OHP cutbacks affected this "expansion population" in 3 ways: (1) premiums; (2) co-payments ($5 for primary care visits and $50 for emergency department visits); and (3) reduced scope of coverage (including loss of out-patient coverage for most mental health and chemical dependency problems). A useful measure of access to care for a population is emergency department (ED) use. The purpose of this study was to compare ED use before and after the OHP cutbacks. Study Design: This time series study compared ED use in an urban teaching hospital in the first 3 months after the OHP cutbacks with the same 3 months of 2002. Population Studied: We used computerized billing data to ascertain payment class and diagnoses for all ED visits. Percent changes in visits were calculated; 95% confidence intervals (CIs) are presented unless they include 0, making the change not significant (NS) at the 0.05 level. Principal Findings: There was a 20% (CI 16%-23%) decrease in ED visits by OHP beneficiaries after the cutbacks, an 8% (CI 3%-12%) decrease in visits by commercially insured patients, and a 17% (CI 10%-24%) increase in visits by uninsured patients. Alcohol-related visits by OHP beneficiaries fell by 5% (NS), while alcohol-related visits by uninsured patients rose by 136% (CI 55% -267%). Drug-related visits by OHP beneficiaries rose by 57% (NS), while drug-related visits by uninsured patients rose by 200% (CI 37%-625%). Other mental health visits by OHP beneficiaries fell by 12% (NS), while increasing by 37% (CI 1%-87%) for uninsured patients. There was a 3.5% reduction in the number of OHP beneficiaries in the hospital service area during the study period. Conclusions: The reduction in ED visits by OHP patients may be partially due to the new co-payments for ED visits, even though these co-payments were rarely collected. The rise in ED visits among the uninsured may be due to OHP cutbacks, but it may also be due to loss of commercial insurance or to decreased availability of non-ED "safety net" clinics serving uninsured patients. Implications for Policy, Delivery or Practice: Whatever the cause of increased ED visits among uninsured patients, this finding points to a worrisome reduction in access to medical care for uninsured Oregonians. Primary Funding Source: RWJF • An Assessment of Coordination Efforts between State SCHIP and Title V Programs in the Case of SCHIPEnrolled Children with Special Health Care Needs Anne Markus, J.D., Ph.D., Sara Rosenbaum, J.D., Soeurette Cyprien Presented by: Anne Markus, J.D., Ph.D., Assistant Research Professor, Health Policy, The George Washington University Medical Center, 2021 K Street, N.W., Suite 800, Washington, DC 20006; Tel: 202.530.2339; Fax: 202.296.0025; E-mail: armarkus@gwu.edu Research Objective: To explore the level of interaction and coordination between Title V Maternal and Child Health Services Block Grant programs that serve children with special health care needs (Title V CSHCN programs) and separatelyadministered SCHIP programs in the provision of services needed by CSHCNs. Study Design: Qualitative methods combining a review of research findings on states’ early experience with SCHIP, which together covered a majority of separately-administered SCHIP programs (51%) and SCHIP enrollees (74%), an analysis of the coordination and benefit provisions of separately-administered state SCHIP plans filed with CMS as of December 2000 (n=35), and a written survey of Title V agencies in states with separately-administered SCHIP programs conducted in 2001 regarding changes to their CSHCN program after SCHIP was enacted and benefits available to CSHCN under SCHIP and Title V (response rate=51%). Population Studied: 35 states with separately-administered SCHIP programs. Principal Findings: First, states have used the flexibility provided under SCHIP to adopt benefit packages that are generally less comprehensive than Medicaid. Although these benefit packages work well for the vast majority of children who are healthy, they can result in CSHCNs facing gaps in needed services. Second, a handful of states have used their Title V programs to attempt to fill the gaps in coverage for CSHCNs created by scaled-back SCHIP benefit packages. The vast majority of states, however, have not taken such steps. Third, even among the handful of states that have sought to coordinate their Title V and SCHIP programs to improve coverage for CSHCNs, some of these children - particularly those with extensive behavioral health needs - are likely to find it difficult to navigate the system and, once they do, to still face gaps in coverage. Conclusions: The limitations on SCHIP benefits are likely to have a disproportionate and potentially significant effect on CSHCNs. Although there are some exceptions, states generally have not used their Title V programs to fill effectively the gaps in care for CSHCNs created by a scaled back SCHIP benefit package. These children thus face the limitations of the SCHIP benefit package with nowhere else to turn for needed specialty care. Implications for Policy, Delivery or Practice: As states take advantage of renewed flexibility under the Health Insurance Flexibility and Accountability demonstration initiative to redesign Medicaid and SCHIP, they may want to pay particular attention to children and adults with special needs. Mobilizing the multiple state agencies whose mission is to serve such individuals at the program design stage to create a system where these individuals can be directed to the appropriate sources of care, and coordinating the delivery of services once the program is implemented would help ensure that fewer CSHCNs and other individuals with special needs fall through the cracks and more of them receive services filling in the gaps left by scaled-back benefit packages under reengineered public health insurance programs. Primary Funding Source: , Kaiser Commission on Medicaid and the Uninsured, Kaiser Family Foundation. • Insurers' Coverage Decision Making of Selected Services for SCHIP-Enrolled Children With Special Health Care Needs Anne Markus, J.D., Ph.D., Sara Rosenbaum, J.D., Jill Joseph, M.D., Ph.D., Ruth Stein, M.D. Presented by: Anne Markus, J.D., Ph.D., Assistant Research Professor, Health Policy, The George Washington University Medical Center, 2021 K Street, N.W., Suite 800, Washington, DC 21075; Tel: 202.530.2339; Fax: 202.296.0025; E-mail: armarkus@gwu.edu Research Objective: To explore how Medicaid- and SCHIPparticipating managed care organizations (MCOs) make coverage decisions for SCHIP-enrolled children with special health care needs (CSHCNs); to assess the impact of these decisions for pediatric access to care. Study Design: Case studies of coverage decisions combining (1) SCHIP plan and contract language analysis of the extent of coverage of benefits more critical to CSHCNs with (2) telephone interviews of the medical directors of Medicaid- and SCHIP-participating MCOs. Researchers presented interviewees with two hypothetical patient scenarios, asking them to decide whether they would have provided a select number of services, identical in each scenario, without restriction, limit or special review, with some limitations, or not at all. Where appropriate, respondents also provided the likely basis for their decision. Population Studied: Purposive sample of 71 medical directors of Medicaid- and SCHIP-participating MCOs in 14 states that issue separate managed care contracts for the two programs interviewed between March and June 2002 (13% participated from 6 states covering 46% of all SCHIP enrollees; 49% explicitly declined to participate; and 38% did not respond to repeated invitations to participate). Principal Findings: Insurers’ decisions varied by type of coverage, category of benefit, acuity of the children’s medical condition, and state of residence. First, insurers generally restricted certain services most needed by CSHCNs, such as speech and physical therapy, motorized wheelchairs, assistive communication devices, and enabling transportation, under SCHIP in ways not permissible under Medicaid. Second, with one exception, insurers were not in agreement on the coverage of any specific service, but overall they provided coverage beyond the limits and exclusions of the state program. Third, the less acute the condition experienced by the child, the more frequently insurers imposed exclusions, particularly in the coverage of speech therapy and hearing aids. Finally, while insurers would have provided more generous coverage than required by the state for the majority of services queried, in four states, there were discrepancies, notably in the area of assistive communication devices, speech therapy, hearing aids, and motorized wheelchairs, between some insurers’ decisions and the state requirements, with insurers excluding services when these services arguably should have been covered according to the plan and contract language. Conclusions: Our findings support our hypothesis that the variability and the limits in coverage and benefits found in separate SCHIP programs can influence access to care by children, including CSHCNs. Implications for Policy, Delivery or Practice: While SCHIP extends coverage to previously uninsured CSHCNs, the benefits offered by state programs vary greatly and are generally less comprehensive than those covered under Medicaid. Under SCHIP, coverage at current levels may not be sufficient to care for CSHCNs, making the availability of external reviews of insurers’ coverage decisions and the coordination with other sources of care important components of SCHIP program design. Additional research is needed to measure and compare actual use of services by CSHCNs who have SCHIP and Medicaid coverage. Primary Funding Source: HRSA • Exploring Cost as an Access Barrier to Mammography Screening Ann McAlearney, Sc.D., Katherine Whitney, M.P.H., Cathy Tatum, M.A., Electra Paskett, Ph.D. Presented by: Ann McAlearney, Sc.D., Assistant Professor, Division of Health Services Management and Policy, The Ohio State University, 1583 Perry Street, Atwell 246, Columbus, OH 43210-1234; Tel: 614.438.6869; Fax: 614.438.6859; E-mail: mcalearney.1@osu.edu Research Objective: Despite considerable advances in scientific understanding of the disease, breast cancer continues to be a significant public health problem in the United States, with estimates that in 2003 breast cancer claimed the lives of 39,800 women, and another 211,300 women were diagnosed with the disease. Although still controversial, it is generally accepted that mammography represents the best opportunity for early detection of breast cancer. However, in 2000, 24 percent of women over age 40 had not had a mammogram within the past two years. This research was designed to evaluate how issues related to cost affect mammography use within a rural population of lowincome women. Study Design: Baseline survey data for the full study population—897 women--were used to evaluate the impact of cost as a self-reported barrier to receiving mammograms, both historically and at follow-up. Chi square tests were performed to investigate bivariate relationships between independent variables and dependent variables related to cost. Logistic regression analysis was used to calculate odds ratios and 95 percent confidence intervals for multivariable associations. Multinomial logistic regression analysis was used in cases where the outcome variable had more than two response levels. Population Studied: The study population consisted of women who were not within guidelines for mammography and who were part of a randomized, controlled study to assess the efficacy of an individualized health education intervention delivered by a lay health educator on improving rates of breast cancer screening among women aged 40 and older who were patients of a county health clinic. This study took place in a predominantly rural, multi-racial county in North Carolina. Principal Findings: Over half of all respondents noted cost was a barrier to obtaining a mammogram, and cost was the most frequently cited reason for why women had not had a mammogram within the past two years—26 percent, or had never had a mammogram—20 percent. When asked about cost, the majority of women overestimated the cost of a mammogram, while only one quarter had an appropriate perception of this cost. Women who identified cost as a barrier were nearly nine times more likely to underestimate their level of insurance coverage; women with no insurance were almost 16 times more likely to note cost as a barrier. Among women who received an intervention to improve mammography rates, women who reported cost as a problem were just as likely to get a mammogram as those who did not report cost to be a barrier; in contrast, noting cost as a barrier was significantly associated with not receiving a mammogram in the control population. Conclusions: Cost reportedly made it difficult for over half of a population of rural, poor women to get a mammogram. However, only one-quarter of this population had an appropriate perception of the cost of a mammogram. Appropriate knowledge of mammogram cost and/or insurance coverage can reduce this barrier, as shown by no difference in mammogram rates among women with a cost barrier reported and those without this barrier in the intervention group. Implications for Policy, Delivery or Practice: Providing women with information about the actual out of pocket cost of mammograms may reduce the impact of cost as a barrier, as demonstrated within a population of low-income, rural women. • Treatment Stacking as a Response to Distance Barriers among VA Patients with Psychoses John McCarthy, Ph.D., Frederic Blow, Ph.D., John Fortney, Ph.D., John Piette, Ph.D. Presented by: John McCarthy, Ph.D., Research Investigator, HSR&D and SMITREC, Department of Veterans Affairs, P.O. Box 130170, Ann Arbor, MI 48113-0170; Tel: 734.769.7100 Ext. 6253; Fax: 734.761.2617; E-mail: John.McCarthy2@med.va.gov Research Objective: Patients and providers may reduce the negative impact of geographic accessibility barriers by combining clinic stops into single outpatient visits. As noted by Veterans Affairs (VA) administrators and the Government Accounting Office (GAO/HEHS-96-31), a consequence of such “stacking” is that analyses of visit day volume may overestimate distance effects. However, no studies have demonstrated a relationship between stacking and distance. This study evaluates distance as a predictor of clinic stop stacking among VA patients diagnosed with psychotic disorders. We also examine interrelationships among distance, psychosis type, and stacking. Study Design: We assessed straight-line distance to the nearest VA outpatient service. Using random intercepts hierarchical regression (cluster = nearest outpatient provider), we evaluated distance effects on treatment stacking (outpatient clinic stops per visit), adjusting for age, gender, ethnicity, marital status, VA service connection, and site and Charlson comorbidities at initial use. Psychosis and interactions with distance were also included. Population Studied: We categorized VA patients with psychoses in FY00 (N=192,982) as having schizophrenia, bipolar disorder, or other psychoses. Exclusion criteria were: homelessness; institutional stays of 150+ days; death; and missing data. The analytic dataset consisted of 142,055 patients. Principal Findings: Median clinic stops per visit were 1.45 among patients with schizophrenia, 1.55 among patients with bipolar disorder, and 1.60 among patients with other psychoses. Distance was associated with significant increases in stacking (p<0.0001). Controlling for other predictors, patients had 6% more clinic stops for every ten miles separating them from care. Those with bipolar disorders were most likely to stack generally and in response to distance. Separate analysis indicated that distance remains a barrier to clinic stop volume. Conclusions: With increased distance from care, patients with psychoses were more likely to stack clinic stops within visit days. Relative to patients with bipolar disorders, patients with schizophrenia or other psychoses were less likely to stack in response to distance barriers. While clinic stop stacking may reduce the impact of distance barriers, distance remains a barrier to utilization even when measured in terms of clinic stop volume. Implications for Policy, Delivery or Practice: This work presents national data on increased clinic stop stacking as a response to distance barriers among patients with psychoses. Treatment stacking may reduce accessibility barriers, yet distance remains a barrier to outpatient care. Primary Funding Source: VA • Treatment Retention among VA Patients with Psychoses: Health System Factors John McCarthy, Ph.D., Kristen Barry, Ph.D., Frederic Blow, Ph.D., Ellen Fischer, Ph.D., Teresa Hudson, PharmD, Rosalinda Ignacio, M.S. Presented by: John McCarthy, Ph.D., Research Investigator, HSR&D Field Program, Department of Veterans Affairs, P.O. Box 130170, Ann Arbor, MI 48113-0170; Tel: 734.769.7100 Ext. 6253; Fax: 734.761.2617; E-mail: John.McCarthy2@med.va.gov Research Objective: Despite high morbidity and ongoing treatment needs, patients with psychoses are difficult to retain in care. The purpose of this study was to evaluate the influence of geographic accessibility and VA and non-VA service availability on retention, adjusting for patient characteristics. Study Design: This was a three-year cohort study of 125,286 VA (Department of Veterans Affairs) patients with schizophrenia or bipolar disorder in FY98. Survival analysis was used to model time to first 12-month gap in VA care over three years, FY99-FY01. Covariates included age, gender, race/ethnicity, marital status, psychosis type, Charlson comorbidity, last treatment location in FY98, and days since last VA care (at end of FY98). Geographic accessibility was operationalized as straight-line distance to nearest VA medical center. Service availability was assessed using county-level VA hospital beds and non-VA beds, per 1000 residents. Patients who died before gaps were censored. Population Studied: Patients in the VA health care system diagnosed with schizophrenia or bipolar disorder in FY98 (N=125,286). Principal Findings: Over three years, 7585 patients (6%) had a 12-month gap in VA care. Individuals who were younger, male, non-white, unmarried, had bipolar disorder, low medical morbidity, an inpatient psychiatric treatment location, or resided farther from VAMCs were more likely to drop out of care. Greater VA bed availability improved retention, while non-VA availability was not significant. Conclusions: Among VA patients with psychoses, patient predisposing and need characteristics strongly predict retention in care. Distance increases risks of substantial gaps in treatment. VA service availability improves retention, however retention may not be sensitive to the availability of non-VA providers. Implications for Policy, Delivery or Practice: Geographic barriers remain a challenge for treatment retention. Moreover, when patients with psychoses fall out of care, nonVA services may not substitute for VA care. Primary Funding Source: VA • Disruptive Behaviors among VA Nursing Home Residents: Assessing the Impact of Psychoses and Dementia John McCarthy, Ph.D., Frederic Blow, Ph.D., Helen Kales, M.D. Presented by: John McCarthy, Ph.D., Research Investigator, HSR&D and SMITREC, Department of Veterans Affairs, P.O. Box 130170, Ann Arbor, MI 48113-0170; Tel: 734.769.7100 Ext. 6253; Fax: 734.761.2617; E-mail: John.McCarthy2@med.va.gov Research Objective: Nursing home providers may be reluctant to admit individuals with psychotic disorders because of concerns regarding disruptive behavior. While recent studies have examined behavior problems among residents with dementia, little is known about the relative prevalence of such problems among residents with psychoses. The purpose of this study was to assess the relative prevalence of disruptive behaviors among nursing home residents with psychoses, as compared to other subgroups of nursing home residents, especially those with dementia. Study Design: This uses cross-sectional administrative data regarding nursing home behaviors. Using diagnosis and utilization data from the Department of Veterans Affairs (VA) Patient Treatment File and Outpatient Care files, we evaluate whether residents received diagnoses of psychosis and/or dementia during the prior six months. In separate multiple regressions, we assess diagnostic correlates of verbal disruptive behavior, physical aggression, and socially inappropriate behavior. Analyses controlled for age, gender, length of stay, Resource Utilization Group, Activities of Daily Living limitations (eating, mobility, transfer, toileting), service connection, depression, substance abuse/dependence and coma status. We include interaction terms to assess the impact of having both psychosis and dementia. In sensitivity analyses, we compare residents with psychoses but without dementia to residents with dementia but without psychoses. Population Studied: We examine patient assessments for 9,661 VA nursing home residents in April, 2000. Principal Findings: Nearly half (46.2%) of all residents received psychosis or dementia diagnoses: 17.6% had psychoses, 28.6% had dementia without psychoses. Among all residents, 4.7% had both conditions. In all three models, psychosis and dementia were associated with increased behavior problems (p<0.0001). Supplemental models limited to individuals with psychoses (and without dementia) or dementia (and without psychoses) indicate that individuals with psychoses had increased risk of verbal disruptive behaviors, yet comparable risks of socially inappropriate and of physically aggressive behaviors. Conclusions: In general, nursing home residents with psychoses have elevated risks of disruptive behaviors, comparable to residents with dementia. Direct comparisons to residents with dementia indicate that residents with psychoses have increased risks of verbal disruptions, yet comparable risks of socially inappropriate and physically aggressive behavior. Implications for Policy, Delivery or Practice: Survey research indicates that nursing home administrators may be unwilling to admit individuals with serious mental illness due to concerns regarding disruptive behaviors. We find increased risk of disruptive behaviors among residents with psychoses, yet, with the exception of verbal disruptive behavior problems, risks among residents with psychoses were comparable to those among residents with dementia. Further research is needed to identify optimal delivery practices to meet the longterm care needs of individuals with serious mental illness. Primary Funding Source: VA • Enrollment Effects of Recent Changes in the Oregon Health Plan John McConnell, Ph.D., Neal Wallace, Ph.D., Charles Gallia, M.P.A. Presented by: John McConnell, Ph.D., Research Assistant Professor, Department of Emergency Medicine, Oregon Health & Science University, 3181 S.W. Sam Jackson Park Road, Mail Code CR114, Portland, OR 97239; Tel: 503.494.1989; Fax: 503.494.4640; E-mail: mcconnjo@ohsu.edu Research Objective: This study examines changes in rates of disenrollment among the Oregon Health Plan (OHP) expansion population. In February 2003, Oregon implemented OHP2 that significantly changed the conditions of coverage for individuals and families not otherwise eligible under traditional Medicaid eligibility rules. Families and the 10-50% FPL group had premiums increased and premium waivers were eliminated. New or increased service copayments and elimination of specific benefits were also implemented. In May 2003, a six-month enrollment lockout for premium non-payment was imposed. Subsequent to these changes enrollment in the OHP expansion population program declined by one-third. Study Design: The study estimates a binary choice model based on whether an individual or couple enrolled in one month continues enrollment into the next. A difference in difference approach is used to assess relative changes in month to month disenrollment using dummy variables for policy implementation time periods, enrollee characteristics, and time by characteristic interactions. Enrollee characteristics used in the study include FPL-based income groups (excluding 85-100% FPL as a reference), age, gender, nonCaucasian ethnicity, non-English speaking, couple vs. individual, and duration of enrollment greater than seven months. Population Studied: The study uses a sample of OHP “expansion group” enrollees from August 2002 through September 2003 who are either singles or couples and have monthly income that remains consistent within a defined FPL bracket. This sample contains over 83,000 individuals/couples and over 500,000 monthly enrollment observations. Enrollees are predominately single, Caucasian, and English speaking with an average age in the mid-thirties. Gender is nearly evenly distributed and slightly more than a third have continuous enrollment periods of over seven months. More than 60% of enrollees have income less than 50% FPL and two thirds of these have no reported income. Principal Findings: Both the initial OHP2 implementation and the subsequent lockout policy disproportionately increased disenrollment for the lowest income enrollees. The lockout policy had more than three times the effect on disenrollment than the original policy change. Notably its impact was fairly uniform across all income groups with the exception of a doubled effect on zero income cases. Disenrollment actually lessened for couples that experienced some of the largest premium increases, as well as non-English speaking individuals. Age and having longer prior enrollment were consistently and increasing protective to disenrollment. Males and persons of non-Caucasian ethnicity were generally at greater risk for disenrollment. Conclusions: The loss of premium waivers and the lockout policy appeared to have the greatest impact. Increased premiums had mixed effects given the results for couples. The strong and increasing protective effect for non-English speaking enrollees is consistent with anecdotal information indicating that safety-net providers supported premium payments in the face of lockout and waiver loss. Implications for Policy, Delivery or Practice: The availability of social or safety net support networks to help support extremely low income individuals with premium payments may differentiate those who need premium waivers and determine enrollment without them. Instability in income and/or expenses, including additional out-of-pocket medical expenses related to OHP2, may be key factors in the dramatic impact of the lockout policy. Primary Funding Source: Oregon Office of Health Plan Policy & Research • Changes in Longitudinal Medicaid Enrollment Patterns Following Welfare Reform Angela Merrill, Ph.D., Marilyn Ellwood, M.S.W. Presented by: Angela Merrill, Ph.D., Researcher, Mathematica Policy Research, 50 Church Street, 4th Floor, Cambridge, MA 02138; Tel: 617.491.7900 Ext. 237; Fax: 617.491.8044; E-mail: amerrill@mathematica-mpr.com Research Objective: This study examined patterns of retention of Medicaid among families leaving “cash-related” Medicaid coverage, and changes in these patterns following welfare reform. Retention of Medicaid provides a key support to families gaining employment, may help families remain off cash assistance, and reduces rates of uninsurance among lowincome families. Study Design: The study used Medicaid administrative data for a descriptive analysis of enrollment patterns among families enrolled in Medicaid before and after welfare reform. We followed cohorts of AFDC leavers in 1993 and TANF/Section 1931 leavers in 1998 and observed their enrollment in different types of Medicaid immediately following welfare exit and at one year after exit. Our data allowed us to examine continued enrollment in specific types of Medicaid, such as transitional medical assistance (TMA). Population Studied: Adults and children enrolled in AFDC/Section 1931 Medicaid in 1993-94 and 1998-1999 in nine states: Arkansas, California, Kentucky, Maine, Michigan, New Jersey, Pennsylvania, Washington, and Wisconsin. Principal Findings: On average across the study states, 55 percent of adults and 62 percent of children retained their Medicaid coverage after they exited TANF/Section 1931 in 1998. This was an improvement over the Medicaid retention rates for AFDC leavers in 1993. In 1998, 32 percent of adults who left TANF/Section 1931 coverage continued to qualify for Medicaid using the TMA provisions on average, an improvement over the TMA user rate in 1993. However, the average months on TMA declined, and less than one quarter remained enrolled in TMA for the full 12 months. On average, one-quarter of adults had returned to Section 1931 Medicaid coverage at one year after exit in 1998. Conclusions: Although families leaving welfare were more likely to retain their Medicaid coverage in 1998 than in 1993, many who disenrolled could have continued to qualify for coverage. Children, in particular, have many coverage options for continuing Medicaid coverage that are being underutilized. Although TMA use had improved in 1998, compared to 1993, few adults remained enrolled the full 12 months. Implications for Policy, Delivery or Practice: This study supports further efforts by states and the federal government to improve Medicaid retention for families leaving TANF and Section 1931 Medicaid when finding work. Since the period of study, many states have worked to educate workers and clients on coverage options, continued to de-link cash assistance and Medicaid enrollment systems, and expanded their use of disregards to extend Section 1931 and TMA coverage. Further steps at the federal level, such as easing TMA quarterly reporting requirements, may improve the proportion of adults who utilize this coverage for a full year. Primary Funding Source: CMS • Subsidizing Small Employers to Help Insure Low-Income Workers: The Massachusetts Experience Janet Mitchell, Ph.D., Joseph Burton, M.P.H., Deborah Osber, M.P.H. Presented by: Janet Mitchell, Ph.D., Research Director, RTI International, 411 Waverley Oaks Road, Suite 330, Waltham, MA 02452; Tel: 781.788.8100; Fax: 781.788.8101; E-mail: jmitchell@rti.org Research Objective: An increasing number of states are implementing premium subsidy programs to help insure lowincome working families. Massachusetts is unique in subsidizing not only low-income workers, but small employers as well. Employers participating in the state’s program (the Insurance Partnership, or IP) receive a fixed amount on behalf of each low-income employee, with the dollar amount varying by type of coverage (single vs. family). Employees receive an additional subsidy for their share of the insurance premium. A total of 3,500 small employers currently participate in the IP, with 10,000 adults and children covered through the program. However, little is known about other small employers in the state and how many offer health insurance. This study compares the characteristics of firms participating in the IP with those of small firms who have not joined the IP. Study Design: We conducted a mailed survey, with telephone follow-up, of a sample of IP participating firms and a sample of non-participating small firms in 2003. The response rate was 47 percent, for a total sample of 1,142. Chi-squares were used to compare characteristics of the two groups of firms. Logistic regression will be used to model the factors explaining participation in the IP. Population Studied: Small firms (50 employees or less) in Massachusetts. Principal Findings: IP participants and non-participating firms did not vary with respect to size, location, or industry type (based on SIC code). The majority were self-employed or very small (2-9 employees), and provided personal or business services. Prior to joining the IP, just over half (58%) of participating firms had offered health insurance to their employees. The subsidy allowed these firms to continue offering insurance (in the face of steep premium increases) and, in some cases, to expand benefits or coverage. For the remaining 42% of firms, the subsidy allowed them to offer health insurance for the first time. Offer rates were similar for non-participating firms (62%), with high premium costs cited as the main reason for not offering insurance. Despite a major media campaign, only one-quarter of these firms were familiar with the IP. The most frequent reason for nonparticipation (or lack of interest in participating, once the program was described) was that the firm had no eligible employees (i.e., no employees with incomes below 200% of FPL). Other reasons included: no need (already offer insurance) or subsidy not adequate. Conclusions: The lack of eligible low-income employees is a key factor in firm non-participation. While this could be interpreted as a lack of need for premium assistance among non-participating firms, these firms were not any more likely to offer insurance than were participants (prior to joining the IP). This suggests that the income eligibility threshold is too low relative to the cost of premiums in Massachusetts. Implications for Policy, Delivery or Practice: In order to further expand insurance coverage, state policymakers need to consider raising the income eligibility threshold for employees and disseminating information about the program more broadly. However, both of these strategies will add to the budget, at a time when policymakers are looking to save, not spend, money. Primary Funding Source: CMS • Hospital Staff Responses to Denials of Concurrent Utilization Review Mary Ellen Murray, Ph.D., R.N. Presented by: Mary Ellen Murray, Ph.D., R.N., Assistant Professor, School of Nursing, University of Wisconsin Madison, 600 Highland Avenue, K6/340 CSC, Madison, WI 53792-2455; Tel: 608.263.6945; Fax: 608.263.5332; E-mail: memurra1@wisc.edu Research Objective: To examine the responses and attitudes of hospital staff members who receive a denial of certification for reimbursement when providing clinical information for concurrent utilization review (UR). Study Design: The UR process requires that care providers, typically registered nurses employed by the hospital, report clinical information to payers, or their representatives. This information is related to the severity of illness of the patient as well as the projected plan of care. Reviewers then make a determination if the care meets criteria of medical necessity and appropriateness and is allowable under the terms of the hospital/payer contract and the individual insurance policy. Care that meets these criteria is certified for reimbursement. A qualitative design was used to study the responses and attitudes of hospital staff members who received a denial of certification during the 12-month study period. Each staff member was asked to complete an audiotaped interview with a research assistant. The taped interviews were subsequently transcribed, and analyzed using the affinity process. The analysis team consisted of 5 doctoral students in nursing and the principal investigator. Population Studied: The study was conducted at an Academic Health Center. Interviews with hospital staff were obtained for 99 of the 102 denials of reimbursement that occurred in the one-year study period. Hospital staff who conducted the UR process included nurse case managers (N=18), clinical social workers (N=9), case manager associates (N=8), and additional registered nurses (N=2) who provided part-time support for the UR function. The 102 denials of certification for reimbursement were associated with 26 different hospital clinical services and involved only adult inpatients. Denials occurred most frequently in psychiatry (21), orthopedics (11), neurology (9), transplant (7), neurosurgery (7), surgical trauma (6), rehabilitation (5), and peripheral vascular surgery (5). The remaining 18 clinical services had only 1-4 denials each over the 12-month study period. Principal Findings: Key themes that emerged from the data include: (1) Hospital staff refusal to accept responsibility for informing patient or family members of the denial, (2) Protective responses of patient by hospital staff when denial occurs, (3) MD with limited or no involvement in the utilization review process, (4) Initiation of adversarial payerprovider relationship when the payer makes a denial, (5) Occasional recognition that the denial decision is correct and payer/provider cooperative relationship is created, (6) Social issues related to denials, and (7) Feelings of powerlessness of UR staff to affect physician and payer decisions. Conclusions: Participation in the concurrent utilization review function produces conflict in hospital employees who perform this function. These employees identify themselves as patient advocates in opposition of payer pressures to reduce inpatient utilization. Implications for Policy, Delivery or Practice: This study raises important questions about the responsibility of health care providers to discuss financial issues related to patient care and the potential conflict this causes within the patient/healthcare provider relationship. Primary Funding Source: AHRQ • Emergency Department Utilization for Nonemergent Health Conditions Kyle Muus, Ph.D., Alana Knudson, Ph.D., Dmitri Poltavski, Ph.D., Michelle Bowles, M.P.A. Presented by: Kyle Muus, Ph.D., Assistant Professor, School of Medicine and Health Sciences, UND Center for Rural Health, P.O. Box 9037, Grand Forks, ND 58202; Tel: 701.777.3848; Fax: 701.777.6779; E-mail: klmuus@medicine.nodak.edu Research Objective: 1. Assess the extent to which North Dakota hospital emergency departments (EDs) are utilized for non-emergent conditions among (a) age cohorts, (b) payer groups, and (c) frequent ED users. 2. Examine the extent to which EDs are utilized for oral healthrelated conditions. Study Design: Descriptive; Secondary analysis Population Studied: All ED patient visits (calendar year 2002) from a randomly selected group of North Dakota hospitals (N=58,660 patient visits). Principal Findings: Among ED visits with classifiable principal diagnoses (N=25,871), most cases were categorized as ‘non-emergent’ (37%), followed by ‘emergent--preventable with primary care’ (36%), ‘emergent--ED care needed-preventable/avoidable’ (8%), and ‘emergent--ED care needed-not preventable/ avoidable’ (20%). Regarding ED use for nonurgent health conditions, commercial insurance (32.4%), self-payers (25.6%) and Medicaid (22.7%) comprised the largest shares of patient volume. Worker’s Compensation had the highest percent of its enrollees classified as non-emergent cases (68.6%). The number of ED visits per patient ranged from 1-52, with 3% of patients having four or more visits ('frequent ED users'). Medicaid enrollees comprised the largest volume of frequent ED users (34.1%). IHS/PHS (7.8%) and Medicaid (7.7%) had the highest percent of their enrollees as frequent ED users. Approximately 1% (N=658) of all ED visits had an (nontrauma) oral health-related principal diagnosis. Self-payers and Medicaid comprised nearly two-thirds (65.9%) of these cases. Conclusions: North Dakota EDs are principally utilized for nonurgent or primary care-treatable illnesses or injuries, with commercially insured individuals comprising the largest share of these ED visits. Health care costs could be curtailed by efforts to re-direct patients with minor afflictions from emergency care to outpatient ambulatory care settings. Implications for Policy, Delivery or Practice: Broader patient education efforts are needed regarding the roles of ambulatory office and ED care, and the importance of preventive care and continuity of care. In particular, state health policymakers within the Worker's Compensation and Medicaid programs should increase health education efforts among their enrollees to promote appropriate use of health resources for reducing costs. Future studies are needed on ND rates/patterns of uninsurance and underinsurance and barriers to adequate insurance coverage, including as assessment of potential avenues for increasing the affordability of health plans for individuals, families and employers. The findings on ED use for oral health-related conditions warrant closer examination of barriers to preventive dental care, particularly among Medicaid and uninsured North Dakotans. Primary Funding Source: Dakota Medical Foundation • Health Services Utilization among Homeless Individuals in Dallas, Texas Paul Nakonezny, Ph.D., Michael Ojeda, M.S. Presented by: Paul Nakonezny, Ph.D., Assistant Professor, Health Promotion, University of North Texas, P.O. Box 310769, Denton, TX 76203; Tel: 940.565.2651; Fax: 940.565.4904; E-mail: nakonezny@coe.unt.edu Research Objective: County hospitals need to explore alternative modes of health services delivery to reduce the high utilization of the hospital and emergency room setting for primary care among the homeless population. One alternative model being used by the Parkland Health and Hospital System in Dallas, Texas, a large county hospital, is the Homeless Outreach Medical Services (HOMES) program, which uses two mobile medical units and a fixed site outpatient clinic to provide primary health care to homeless individuals. The current study examined the relationship between outpatient utilization delivered via the HOMES program and inpatient utilization among a sample of the homeless population who were served by the Parkland Health and Hospital System. Study Design: A quasi-experimental design and health services utilization data from 293 male and 288 female homeless patients were used to address the research objective of this study. The dependent variables were measurements of inpatient and outpatient utilization of psychiatric, substance abuse, and musculoskeletal services. These three diagnoses were selected for analysis because they represent three of the top five disease conditions most frequently diagnosed among the homeless population being served by HOMES. Outpatient utilization was measured as the number of visits to the HOMES mobile medical units and fixed site clinic and inpatient utilization was measured as the number of visits to the Parkland inpatient units and emergency room between June 01, 1992, to June 30, 1999. That is, each homeless patient was matched on inpatient and outpatient utilization for each diagnosis (psychiatric, substance abuse, and musculoskeletal) and then utilization was measured across the seven year time period. Inpatient and outpatient utilization patterns by age, gender, and race were also assessed. Data were analyzed using a split-plot repeated measures MANOVA, one-way ANOVA, and simple linear regression. Population Studied: A sample of 581 homeless patients aged 20-64 years who used HOMES and the Parkland Health and Hospital System in Dallas, Texas. Principal Findings: Results suggested that, across the seven year time period, homeless individuals, regardless of gender and race, utilized the mobile medical units and outpatient clinic services more than inpatient services for psychiatric, substance abuse, and musculoskeletal conditions. Older homeless adults (50-64 yrs of age) compared with younger homeless individuals (20-49 yrs of age) utilized the mobile medical units and outpatient clinic services more than inpatient services for substance abuse and musculoskeletal conditions, but they utilized inpatient services the most for psychiatric-related conditions. This inpatient utilization is perhaps a result of greater severity or progression of mental illness among older homeless adults. Conclusions: The HOMES program is an example of a health care delivery model that addresses the complex and unique health care needs of individuals who are homeless. The outpatient utilization of health services is probably because the HOMES program provides accessible, continuous, comprehensive, appropriate, and sensitive health care to homeless persons, free-of-charge and on a walk-in basis, where homeless individuals congregate. The HOMES program mitigates many of the barriers that prevent homeless individuals from obtaining primary medical care. Implications for Policy, Delivery or Practice: A significant void exists between the medical needs of the homeless population and the current health care system in the United States. If hospital systems and the community are to be successful in meeting the health care needs of the homeless population, they must develop programs that are accessible-so that health problems do not go untreated, requiring more costly inpatient care--and they must build a social support network for homeless individuals to break down the social isolation barrier. These programs also must be designed with sensitivity to the multifaceted health problems of individuals who are homeless and to the special needs associated with homelessness. • Health Insurance and the Hidden Unemployed Lawrence Nitz, Ph.D., Gerard Russo, Ph.D., Sang Hyop Lee, Ph.D. Presented by: Lawrence Nitz, Ph.D., Professor, Political Science, Political Science, University of Hawaii, 2424 Maile Way, Room 640, Honolulu, HI 96822; Tel: 808.956.8665; Fax: 808.956.6877; E-mail: lnitz@hawaii.edu Research Objective: The objective is to determine the degree to which the employment status "delf-employed" hides a group of low income persons without health insurance, who may become unanticipated Medicaid participants in various proposals to increase the income limits for state Medicaid benefits. Study Design: The study uses a logistic regression of health insurance status against indicators for self-employment status, income, education, gender, and income-self employment interactions. The key questions are whether selfemployed individuals are less likely to be insured than those working for others or in other social statuses, whether there is an interaction between income and employment status that would support the notion that "self employment" is in many cases a brand of hidden unemployement, versus the notion that "self employment" involves the voluntary acceptance of health risks by rejecting health insurance. Population Studied: The study population was the BRFSS sampled from the Hawaii population in 2000, 2001, and 2002. The combined samples provide the basis for relatively fine divisions of demographic indicators, notably employment status, income and education. Principal Findings: The logistic regression of insurance status on the self-employment indicator clearly denotes more acceptance of uninsured status by the self-employed. The role of income suggests that higher income persons are more likely to be insured. The interaction between self-employment and income indicates that the higher the level of selfemployment income, the higher the likelihood of being uninsured. Conclusions: The picture of the self-employed as in part a hidden layer of the poverty population is only partially supported. Additional tabulations would be appropriate to distinguish between those self-employed persons who earn relatively well compared to their equally skilled compatriots who work for others, and those whose self employment income is markedly less than that expected for their level of human captial. Implications for Policy, Delivery or Practice: To the extent that the self-employed are risk takers on their own accounts and earn relatively well for their level of human capital, increasing the income floor for Medicaid coverage may be expected to draw few from this group as a woodwork effect. On the other hand, to the extent that the self-employed are in part voluntarily unemployed, increasing Medicaid floors may draw participants who would not be anticipated on income grounds alone. Primary Funding Source: HRSA • The Inpatient Utilization of California Medicare HMO Disenrollees to Fee-For-Service: Implications for M+C Enrollment Lock-In June O’Leary, Ph.D., Elizabeth Sloss, Ph.D., Nasreen Dhanani, Ph.D., Glenn Melnick, Ph.D. Presented by: June O’Leary, Ph.D., Consultant, RAND, 1700 Main Street, PO Box 2138, Santa Monica, CA 90407-2138; Tel: 310.393.0411 Ext. 6284; Fax (310) 393-4818, 310.451.7061 Email: oleary@rand.org. Research Objective: Previous research suggests that disenrollees from Medicare health maintenance organizations (HMO) are sicker and cost more on average than fee-forservice (FFS) beneficiaries or non-disenrollees. Whether disenrollees use a higher or lower level of services prior to disenrollment has different implications for why they use more services post-disenrollment. The utilization of services by Medicare beneficiaries while in an HMO has never been observed. We examine the use of inpatient services prior to and during enrollment in an HMO, and after disenrollment to FFS. Study Design: Enrollment data on Medicare beneficiaries from the Centers for Medicare and Medicaid Services were linked to inpatient hospital discharge data from the California Office of Statewide Health Planning and Development for January 1992 through December 1996. Three measures of inpatient utilization were estimated: (1) admissions, (2) total days, and (3) average length of stay. Since beneficiaries can enroll and disenroll on a monthly basis, utilization was calculated in person-months. Population Studied: Three subgroups of Medicare beneficiaries in California were identified: (1) beneficiaries who disenroll from an HMO to FFS anytime between January 1993 and December 1995; (2) a sample of beneficiaries continuously-enrolled in FFS; and (5) beneficiaries continuously-enrolled in HMOs. Beneficiaries who met any of the following criteria during any month were excluded: had end-stage renal disease; were not continuously-entitled to Medicare; did not have both Part A and B coverage; resided in counties with fewer than 1,000 HMO enrollees; moved out of California. Principal Findings: Inpatient use by disenrollees to FFS is higher on average than continuously-enrolled FFS or HMO beneficiaries. The admission rate per person-month after disenrollment was 0.086 versus0.043 before disenrollment for those 27,720 beneficiaries who disenrolled to FFS sometime between 1992 and 1995 and were enrolled in an HMO for at least three months and remained in FFS for at least three months or until death. A quarterly analysis of the year before and after disenrollment suggests that admissions begin rising prior to disenrollment. Conclusions: Disenrollees to FFS are sicker on average than continuously-enrolled HMO and FFS beneficiaries. Inpatient use is lower while enrolled but rises prior to disenrollment suggesting that a medical event might trigger disenrollment to FFS. Further research is needed to determine whether disenrollees experience difficulties obtaining care while enrolled leading to use of services in FFS. Implications for Policy, Delivery or Practice: Before the lock-in policy that will restrict how often and when beneficiaries can enroll and disenroll from HMOs is implemented, a better understanding of how HMOs manage the care of this population is needed. Primary Funding Source: AHRQ, HHS/ASPE • Do Parents’ Citizenship Statuses Predict U.S.-Born Mexican-Origin Children’s Access to Job-Based Health Insurance? Victoria Ojeda, Ph.D., M.P.H. Presented by: Victoria Ojeda, Ph.D., M.P.H., NIMH PostDoctoral Research Fellow, Health Care Policy, Harvard Medical School, 180 Longwood Avenue, Boston, MA 02115; Tel: 617.432.0819; Fax: 617.432.0173; E-mail: ojeda@hcp.med.harvard.edu Research Objective: Uninsurance is pervasive among Latino children, in part because of their lower rates of job-based coverage. This study reviews sources of health insurance coverage among a population of U.S.-born Mexican-origin children in working, two-parent families; data are disaggregated by parents’ citizenship statuses. This investigation examines whether parents' citizenship statuses predict disparities in children's job-based insurance coverage. Study Design: This is a cross-sectional study, based on data from the March 2001 supplement of the Current Population Survey, a monthly survey of about 50,000 households nationwide. The CPS sample is representative of the civilian non-institutionalized population. Statistical methods used include bivariate analyses and logistic regression techniques. This study focuses on parents’ citizenship statuses as determinants of Mexican-origin children’s coverage through an employer in the context of other sociodemographic characteristics. I tested the hypothesis that U.S.-born children in working two-parent families headed by any combination of noncitizen and citizen (i.e. naturalized citizens or native-born) parents would be less likely to have coverage through an employer than children in entirely native-born families. Population Studied: U.S.-born Mexican-origin children eighteen years or younger who are dependents and members of working two-parent families are the focus of this study. Native-born Mexican-origin children represent more than twothirds of all Latino children. I focus on children in two-parent families since the presence of two adults allows for potentially two workers in the workforce. Further, children in working families may have the best chances for obtaining employerbased insurance, the most common source of insurance for nonelderly populations, due to their families’ connections to the labor market. Principal Findings: Bivariate analyses show that U.S.-born Mexican-origin children in two-parent working families headed by at least one noncitizen parent are more likely to be uninsured and less likely to have job-based insurance than children in entirely native-born families. About one-third of children living with two noncitizen parents are uninsured, and 29% have job-based insurance. Uninsured rates drop to 22% for children in families with one citizen and one noncitizen parent; 55% have job-based insurance. Multivariate analyses suggest that parents’ citizenship statuses are less significant predictors of workplace coverage, after controlling for other sociodemographic and employment characteristics. Children in families headed by two noncitizen parents (OR=0.48) and children in families headed by one naturalized parent and one U.S.-born parent (OR= 0.52) are less likely than children in entirely native-born families to have job-based insurance. Low-incomes, firm size, and occupational distributions also play important roles in shaping children’s access to job-based insurance. Conclusions: Children in families headed by one noncitizen and one citizen parent (either naturalized or U.S.-born) benefit from having a citizen parent. Despite being in working twoparent families, children whose families are headed by two noncitizen parents remain at risk for lacking job-based insurance. Findings suggest that noncitizen parents face important barriers to obtaining workplace coverage. Job-based coverage is not always a successful coverage mechanism for children living with two noncitizens. Future research may expose barriers to coverage faced by subgroups of noncitizens, including legal permanent residents. Implications for Policy, Delivery or Practice: Policy solutions are needed to reduce uninsurance and disparities in job-based coverage among working families, including those with immigrants and low incomes. Continued support for safety-net programs may prove especially critical for working families if insurance is not offered in the workplace and the costs of health services, prescription drugs, and insurance in the private market continue to rise. Primary Funding Source: UCLA Dissertation Year Fellowship • Expanding Health Insurance Coverage through the States: Lessons from the California Health Insurance Act of 2003 Thomas Oliver, Ph.D., M.H.A. Presented by: Thomas Oliver, Ph.D., M.H.A., Associate Professor, Health Policy and Management, Johns Hopkins University, 624 N. Broadway, Room 403, Baltimore, MD 21205; Tel: 410.614.5967; Fax: 410.955.6959; E-mail: toliver@jhsph.edu Research Objective: To identify key factors in the health care system and in the political system that facilitate major expansions of health insurance coverage in the states. Study Design: The study reviews the history of past initiatives to secure universal or near-universal health insurance in California and three sets of variables affecting the prospects for new reforms: 1) the number and characteristics of the uninsured population and how they are understood in the policy community; 2) the basic design of proposals for expanding coverage, particularly financing and administration; and 3) political conditions, including the positions and resources of key elected officials and interest groups, state fiscal capacity, and federal support for major state expansions of insurance coverage. The study draws upon personal interviews with key participants in the policy community, archival documents, and health services research and analysis. Population Studied: N/A Principal Findings: The paper finds that several key factors changed to facilitate the enactment of the California Health Insurance Act of 2003. Organized labor joined with organized medicine to establish the core of a powerful coalition for reform; key political leaders sponsored the legislation and made it a high priority; the reform was viewed as a partial solution, not a contributor, to the state's fiscal crisis; and the recall election of the Democratic governor narrowed the window of opportunity for action on this issue. Conclusions: The conditions for adopting an employer mandate and other health care reforms in California in 2003 suggest that comparable reforms are viable in other states and that the issue of universal coverage is likely to achieve greater prominence on the national policy agenda regardless of the outcome of the 2004 presidential and congressional elections. Implications for Policy, Delivery or Practice: Primary Funding Source: California HealthCare Foundation • Short Tenures in Medicaid Managed Care Gerry Fairbrother, Ph.D., Heidi Park, Ph.D., Arfana Haidery, M.P.H., Bradford H. Gray, Ph.D. Presented by: Heidi Park, Ph.D., Research Associate, , The New York Academy of Medicine, 1216 Fifth Avenue, New York, NY 10029; Tel: 212.419.3525; E-mail: hpark@nyam.org Research Objective: A recent study shows that children's tenures in Medicaid managed care is shorter than in Medicaid itself. The purpose of this study is to examine the reasons for short tenures in Medicaid managed care and to better understand the practices and policies that lead to the gap between tenures in Medicaid and in health plans. Study Design: This qualitative study was conducted in five states: Arizona, Michigan, New York, Oregon and Pennsylvania. The states were selected to provide variation in geographical location and in experience with Medicaid managed care. Semi-structured in-depth interviews were conducted both face-to-face and by telephone with key individuals at each state, which included State Medicaid and Medicaid Managed Care officials, enrollment brokers, and appropriate health plan personnel, including Medical Directors and quality monitoring and enrollment staff. Population Studied: Medicaid managed care enrollees Principal Findings: Two factors emerged as the major contributors to short tenures in Medicaid managed care: (1) the gap between the time Medicaid coverage begins and enrollment in the health plan, and (2) the frequency and ease of Medicaid eligibility renewal and recertification. The gap in Medicaid coverage and enrollment in a health plan occurs because Medicaid coverage begins retroactively to the date of application and Medicaid eligibility must be determined before enrollment in a health plan. In our five study states, the interval between the start of Medicaid coverage and the start of health plan enrollment can be as long as 2 to 4 months. In addition, we found that the frequent and burdensome Medicaid renewal process often leads to breaks in enrollment, not only in Medicaid itself, but also at the health plan. Failure to renew on time can lead to another gap coverage. Care can be delayed during these gaps because enrollees do not know whether they have coverage. Conclusions: Current enrollment policies and practices lead to a gap in coverage in Medicaid and in Medicaid managed care organizations, and often loss of coverage at Medicaid renewal. This results in short enrollment tenures in health plans. Short tenures can affect continuity of care, quality monitoring, and the ability for health plans to manage costs. Implications for Policy, Delivery or Practice: Recent budgetary pressures have forced states to adopt policies that will shorten tenures in Medicaid. It is important to recognize that shortening eligibility periods and increasing hurdles in the enrollment process will exacerbate the problem of short tenures and make it more difficult to manage the care of these enrollees. Primary Funding Source: Center for Health Care Strategies • The Effects of Coverage on Barriers to Prenatal Care among Women Marsha Regenstein, Ph.D., Jennifer Huang, M.S., Linda Cummings, Ph.D., Donna Sickler, M.P.H. Presented by: Marsha Regenstein, Ph.D., Vice President of Research for National Association of Public Hospitals and Health Systems and Director for National Public Health and Hospital Institute, National Association of Public Hospitals and Health Systems, 1301 Pennsylvania Avenue, N.W., Suite 950, Washington, DC 20004; Tel: 202.585.0105; Fax: 202.585.0101; E-mail: mregenstein@naph.org Research Objective: To identify barriers to prenatal care and utilization among low-income women delivering at large safety institutions. Study Design: A self-administered survey was given to eligible patients post-delivery to identify barriers to prenatal care prior to and during pregnancy. The survey response rate was 76.1 percent, with a total of 2,509 respondents. Surveys were matched via unique identifiers to non-identifying information collected from labor and delivery records. These records included information about initiation of prenatal care, demographics, and birth-related outcomes. Population Studied: Twenty-six public hospitals and health systems belonging to the National Association of Public Hospitals and Health Systems were selected to participate. These hospitals were selected to participate based on number of expected births, geography, and patient demographics. Women delivering during a two-week period in fall 2002 were eligible to participate in the study. Principal Findings: In terms of race, 83.4 percent of respondents were non-white, compared to 42.2 percent of women delivering nationwide. Women delivering at study hospitals tend to be younger, less educated, and less likely to be married than women nationally. Women continuously uninsured throughout pregnancy were more likely to indicate they missed or delayed prenatal care, although women continuously insured throughout their pregnancy also reported barriers to prenatal care. Women who converted from uninsured to insured status identified barriers to care at higher rates than continuously insured. Coverage differed significantly by race of respondent, with Latina women at greatest risk for being uninsured. Medicaid was by far the dominant insurer of Black women, with 71.2 percent indicating that they were on Medicaid at the time of their delivery. Although women across all races reported reasons for missing or delaying care, for the majority of barriers, Latina women were more likely than women of other races to report barriers to prenatal care. Conclusions: Despite the availability of safety net providers, barriers to prenatal care are significant for both continuously uninsured and continuously insured women, with more obstacles for the continuously uninsured. Barriers remain for women without coverage at the outset of pregnancy. Implications for Policy, Delivery or Practice: The study findings demonstrate the value of programs that provide coverage to pregnant women and the effect that such coverage has on low-income women’s self-reported perceptions of barriers to care. Policies to increase coverage to women pre-pregnancy would have additional benefits for timely and adequate prenatal care. A combination of strategies, including more comprehensive coverage and more patient-centered care, may hold the greatest promise for improving access to timely prenatal care for low-income women. Primary Funding Source: March of Dimes; American College of Obstetricians and Gynecologists • The Structure, Financing and Accessibility of Safety Net Services in Ten U.S. Communities Marsha Regenstein, Ph.D., MCP, Lea Nolan, MA, Peter Shin, Ph.D., M.P.H., Marcia Wilson, M.B.A., Holly Mead, Bruce Siegel, M.D., M.P.H. Presented by: Marsha Regenstein, Ph.D., MCP, Assistant Research Professor, Department of Health Policy, George Washington University School of Public Health and Health Services, 2021 K Street, N.W., Suite 800, Washington, DC 20006; Tel: 202.530.2310; Fax: 202.296.6922; E-mail: marshar@gwu.edu Research Objective: The nation's safety net provides care for millions of uninsured and underserved individuals. Communities rely on safety net hospitals, community health centers, health departments, clinics and individual practitioners to provide primary and specialty care, high-tech emergency and trauma services, mental health and substance abuse care, dental health and prevention and screening services. Financing for these services is increasingly constrained by tight state and local budgets, despite increases in demand for care. This study assesses the structure, financing and availability of safety net services in U.S. communities and identifies opportunities for improving access and strengthening service delivery. Study Design: Ten communities were identified through a national competitive process to undergo an assessment of the local safety net. Data for the assessment were developed through site visits, interviews with local stakeholders, and review of secondary sources. Thirty focus groups were conducted in six languages with residents who were likely to use safety net services. Data from an emergency department in each community were analyzed using an established emergency department use profiling algorithm to determine the percentage of visits that were non-emergent and/or primary care treatable. A partner organization in each community convened meetings of stakeholders at the start of the assessment to describe the key issues affecting the safety net and at the conclusion of the project to disseminate findings from the assessment. Population Studied: The focus of the project was on uninsured and underserved residents in ten communities and the health care providers who serve them. A secondary focus was on patients using one safety net hospital emergency department in these communities. Principal Findings: All ten communities face substantial challenges in preserving the safety net and meeting the needs of residents. Experiences with state budget crises varied across communities but generally involved changes in the Medicaid program, decreases in state or local funding streams, and increased demand for services, especially from working uninsured or underinsured residents. Several of the sites lacked sufficient primary care and uninsured and underserved residents in most sites had significant difficulty accessing timely and affordable specialty, dental, and mental health services. Several sites had problems with coordination of services across multiple sites of care or within systems offering tiered services. Residents commonly used the emergency departments of large safety net hospitals for nonemergent or primary care treatable conditions. Conclusions: Communities would benefit from better coordination among safety net providers. Such coordination is essential to stretch limited resources across populations in need of care. Enhancing access through expansions to public programs may prove challenging, given budget constraints at state and local levels. Use of the emergency department for non-emergent or primary care treatable conditions varies across the sites and reflects a myriad of factors related to availability of alternative sites of care and patient preferences and awareness of options. Implications for Policy, Delivery or Practice: While some communities are in need of additional health service capacity, others may benefit from an emphasis on integration of existing resources. Primary Funding Source: RWJF • Hospital Concentration, DSH Payments, and Access for the Uninsured in Los Angeles County and the State of California Gerald Kominski, Ph.D., Dylan Roby, M.Phil. Presented by: Dylan Roby, M.Phil., Senior Research Associate, Center for Health Policy Research, UCLA, 10911 Weyburn Avenue, Suite 300, Los Angeles, CA 90024; Tel: 310.794.3953; Fax: 310.794.2686; E-mail: droby@ucla.edu Research Objective: This project examines how recent trends toward concentration within the hospital industry among hospital chains and the increase in for-profit ownership (specifically Tenet Healthcare Hospitals) affects access to affordable healthcare in Los Angeles. Specifically, it investigates the impact of hospital concentration on: (1) costs and revenues, which affects the affordability of employer-provided coverage for those that have it, and (2) the competitive position of the public hospitals, which have traditionally provided a safety net to the uninsured and underinsured. The primary focus of this study is Los Angeles County, since for-profit hospital growth has been greatest in this region. For comparison purposes, we also examined data for hospitals in the rest of California, excluding Los Angeles County. We examine how hospitals performed from 1995-2000 in Los Angeles County and in the rest of California in providing care to the uninsured, staying solvent, and taking advantage of funding designed for safety net providers. Study Design: The data sources used in this study are readily available from the California Office of Statewide Health Planning and Development (OSHPD) and the California Department of Health Services. The main data set used in this study is the Hospital Annual Financial Data files for 19952000. These data are reported on a hospital fiscal year basis by each acute care hospital in California, and include information on revenues, patient days, discharges, expenses and other variables at the hospital level. Acute care hospitals are defined by OSHPD as hospitals with at least half of their patient days classified as acute. The second source of data used in this study is Disproportionate Share Hospital (DSH) payment data collected by the Department of Health Services. These data track DSH payment information for each hospital, and also collects data on the transfer payments made by public hospitals into the DSH fund. These data were only available for 1998-2001. Descriptive statistics were developed based on revenue, profit margin, and patient mix variables for each group of hospitals (private, non-profit, government, and Tenet-owned hospitals). Population Studied: Acute care hospitals in California and the city of Los Angeles from 1995-2000. The definition of acute care hospitals does not include long-term care facilities, or psychiatric or rehabilitation hospitals. Principal Findings: There was an increase in the number of hospitals in the county owned by hospital systems. For example, Tenet Healthcare Corporation increased their market share from 8% in 1995 to 15% in 2000. In addition, Tenet’s concentration has greatly increased their payments from the Disproportionate Share Hospital (DSH) program. In California, Tenet owned 8 hospitals in 1995 that received DSH payments; by 2000, they owned 17. In Los Angeles County, Tenet went from owning one DSH hospital in 1995 to owning 6 by the year 2000. Overall, Tenet added 14 hospitals statewide within five years. This growth has occurred while Los Angeles County’s government hospitals have been facing financial pressures due to cutbacks and a growing uninsured population. The average total margin for Tenet hospitals grew at a healthy rate in Los Angeles County and the rest of California through 1999. From 1998 to 2000, total margins declined for all hospitals in Los Angeles County and for all hospitals in the rest of California except Tenet hospitals. Operating expenses in Tenet hospitals were fairly stable, while government hospitals in Los Angeles faced a serious problem of rising expenses without sufficient growth in patient revenue. One factor that is largely responsible for Tenet’s ability to achieve high profits in the current healthcare market is the Disproportionate Share Hospital (DSH) payment system. Conclusions: Tenet engaged in aggressive cost cutting, both in Los Angeles and the rest of California, and this contributed to its relatively high profit margins during 1995-2000. This is clearly one of the potential advantages of hospital chains, namely, that they can achieve greater efficiency than standalone facilities. Tenet’s profit margins grew during the 1995-2000 period because of its rapid increase in DSH payments, while other hospitals in Los Angeles County and the rest of California showed no growth in these payments. Yet during this same period, Tenet’s share of uninsured patient days declined and remained the lowest of any hospital group. Thus, at a time when the County’s health care system has faced substantial financial threats, Tenet was successful in rapidly increasing its DSH payments without any increase in uninsured patient days and with a large decline in the average severity of its patient mix. Implications for Policy, Delivery or Practice: Los Angeles County government hospitals continue to struggle financially while a greater portion of DHS payments continue to go to Tenet hospitals. Unless limitations in the formula for distributing DSH payments are addressed by the state legislature, the efficiencies associated with hospital concentration may continue to be offset by the costs imposed on the public sector by aggressive revenue maximization practices. In addition, DSH payments to Tenet hospitals are not being targeted toward hospitals where the poor and uninsured recieve most of their care. This shift in the distribution of safety net funding will affect the ability of government and non-profit hospitals to provide care to the underserved. Primary Funding Source: UCLA Institure for Labor and Employment • How Does Health Literacy Affect Individual Health Outcomes—Is Health Communication the Missing Link? Rafael Ruiz, Sc.M., Shoou-Yih Lee, Ph.D., Ahsan M. Arozullah, M.D., M.P.H. Presented by: Rafael Ruiz, Sc.M., Doctoral Student, Health Policy and Administration, University of North Carolina at Chapel Hill, School of Public Health, McGavran-Greenberg CB#7411, Chapel Hill, NC 27599; Tel: 919.960.8402; Fax: 919.966.6961; E-mail: rruiz@email.unc.edu Research Objective: Almost 25% of the U.S. population is functionally illiterate, raising concerns about those individuals’ ability to navigate the healthcare system. Indeed, studies have consistently shown that patients with a lower level of health literacy tend to have worse health status and display more risky health behaviors such as smoking and excessive drinking. Many researchers have suggested that such adverse outcomes may be due to inadequate communication between low health literacy patients and their healthcare providers. In other words, health communication may be a key mechanism that mediates the impact of health literacy on individual health outcomes. The study is intended to provide an empirical test of this hypothesis. Study Design: Data for this study were from a survey of patients, presenting from June to October 2001, at a Veterans Administration (VA) Hospital in Chicago. Four ordered probit regression models were run in order to analyze various selfreported measures of health communication. Health literacy was measured by the Rapid Estimate of Adult Literacy in Medicine (REALM). Control variables included measures of several socio-demographic factors and social capital. Population Studied: Male patients of the Veterans Administration Healthcare Systems. Principal Findings: Contrary to our expectation, health literacy did not show a statistically significant relationship with any of the four health communication variables examined in the analysis, suggesting that health communication did not mediate the effect of health literacy on individual health status and risky health behavior. In further analysis, we found that functional and structural social support was negatively correlated with health communication. Interestingly, having a regular provider was negatively correlated with most of the health communication variables. Conclusions: Health communication did not appear to explain the association of health literacy with health status and health behavior. Improvement of health communication alone may be ineffective in ameliorating the adverse health outcomes of individuals with low health literacy. Implications for Policy, Delivery or Practice: Programs designed to improve patient-provider communication may be inadequate in addressing the personal health problems of low health literacy patients. Further research is needed to understand how low health literacy is related to negative health outcomes. Primary Funding Source: Pfizer Foundation • The Impact of State Children’s Health Insurance Program (SCHIP) Expansion on Health Insurance Coverage in Hawaii Gerard Russo, Ph.D., Sang-Hyop Lee, Ph.D., Lawrence Nitz, Ph.D., Thamana Lekprichakul, Ph.D. Presented by: Gerard Russo, Ph.D., Associate Professor, Department of Economics, University of Hawaii, 2424 Maile Way, Saunders Hall 542, Honolulu, HI 96822; Tel: 808.956.7065; Fax: 808.956.4347; E-mail: russo@hawaii.edu Research Objective: The purpose of this study is two-fold. First, to assess the impact of the initial SCHIP expansion which occurred in Hawaii July 1, 2000 and extended free public insurance coverage to all children aged 0-18 years residing in households with incomes not greater than 200% of the Hawaii specific federal poverty level, FPL. Second, to predict take-up of public coverage and crowd-out of private insurance which may occur under a proposed expansion of eligibility to children aged 0-18 years residing in households with incomes between 200% and 300% of the federal poverty level FPL. Study Design: This study exploits the natural experiment of a July 1, 2000 SCHIP expansion, which is reflected in survey data, to estimate take-up and crowd-rates among newly eligible beneficiaries. Prior to the implementation of SCHIP, children in Hawaii were eligible for free medical assistance under Medicaid financing per the following age-income criteria. Children age 0 were eligible up to 185% of the Hawaii specific FPL, children aged 1-5 years up to 133% FPL and children aged 6-18 years up to 100% of the FPL. On July 1, 2000, SCHIP was implemented in Hawaii as a Medicaid expansion and extended coverage to all children age 0-18 up to 200% FPL. Thus, children under age 1 residing in families with incomes between 186% and 200% FPL, children age 1-5 in families with incomes between 134% and 200% FPL and children age 6-18 in families with incomes between 101% and 200% became newly eligible for free medical assistance under SCHIP July 1, 2000. This treatment group is tracked in 1998-1999 before eligibility and in 2001-2002 after eligibility and both direct and model-based estimates of insurance status are produced. Children aged 0-18 years residing in families with incomes between 201% and 250% the FPL have been ineligible for public insurance throughout the period 1998-2002 and are used as a control group. The study uses a difference-in-difference approach to sweep out non-policy related effects. Estimates are based on multiple years of the Hawaii Health Survey, confidential version and the Current Population Survey, Annual Social and Economic Supplement, public use version. Model-based estimates are also used to project a response to increased eligibility up to 300% of the FPL. Population Studied: The study is exclusive to Hawaii residents aged 0-18 years residing in households with incomes between 0% and 300% of the Hawaii specific federal poverty guidelines as published by the U.S. Department of Health and Human Services. Weighted population estimates are produced using the Hawaii Health Survey 1998-2002 and the Hawaii sample of Current Population Survey 1998-2003. Three groups are analyzed, children eligible for free public insurance throughout the period, children who began the period ineligible and became eligible in 2000 and children who were ineligible throughout the period. Principal Findings: The Initial SCHIP implementation of July 1, 2000, which brought public insurance eligibility to 200% FPL, enrolled approximately 6000-7000 children with rather modest displacement of private insurance. The proposed further expansion to 300%, however, is projected to generate dramatic crowd-out as households drop private family coverage to enroll children in free public insurance. Preliminary model-based estimates predict 29,000 new children enrollees coming from households with incomes between 201% and 300% of the FPL. Of these only 2000 would have been previously uninsured, while 27,000 would be drawn from private insured families--a staggering rate of crowd-out. Conclusions: Because of a number of factors, including Hawaii’s Pre-paid Health Care Act of 1974, which mandates private sector employment-based coverage, middle-class families are typically well insured. Of the approximately 50,000-55,000 children aged 0-18 years residing in families with incomes between 201% and 300% of the FPL, only 20003000 are without health insurance coverage. As a result, a generous public expansion which takes all comers up to 300% FPL is likely to yield a large number of privately insured along with the uninsured. Implications for Policy, Delivery or Practice: A Hawaii SCHIP expansion to 300% of the FPL will result in substantial crowd-out of private insurance. Whether viewed from the broad federal perspective or the narrow state budgetary perspective, such an expansion would generate high public expenditures per newly insured child and should therefore be pursued with caution. A successful and sustainable expansion of eligibility criteria must be accompanied by safeguards against excessive private insurance crowd-out. Primary Funding Source: HRSA, Hawaii State Department of Health • The Effect of Mandatory Employer-Sponsored Health Insurance on Insurance Coverage and Employment in Hawaii Sang-Hyop Lee, Ph.D., Gerard Russo, Ph.D., Lawrence Nitz, Ph.D. Presented by: Gerard Russo, Ph.D., Associate Professor, Department of Economics, University of Hawaii, 2424 Maile Way, Saunders Hall 542, Honolulu, HI 96822; Tel: 808.956.7065; Fax: 808.956.4347; E-mail: russo@hawaii.edu Research Objective: The purpose of this study is to examine the impact of Hawaii’s mandatory employer-provided health insurance on insurance coverage and employment structure in Hawaii. We hypothesize that mandated employer-sponsored health insurance has three effects on health insurance coverage and the labor market. First, it increases employerprovided health insurance coverage for full-time workers. That is, persons employed more that 20 hours per week. Second, it changes the distribution of equilibrium employment by hours worked as some employees seek and employers offer parttime employment to avoid the mandated benefit. Third, people who are not matched with full-time jobs with employerprovided health insurance will switch to other types of insurance, such as publicly provided health insurance or spousal benefits, as an optimal response. Study Design: Hawaii’s Pre-Paid Health Care Act of 1974 is a unique law which requires private sector employers to offer and employees to accept employer-sponsored health insurance as a fringe benefit. Hawaii is the only State in the Union with such a mandate as the Employee Retirement Income Security Act, ERISA, generally prohibits States from mandating health and retirement benefits. The Federal government reserves that right, except in the case of Hawaii which has a Congressionally sanctioned exemption. We produce model-based estimates of employment patterns and insurance coverage for working age adults for Hawaii, the U.S. as a whole and several comparative States including Nevada and Michigan controlling for worker characteristics and industrial structure. We treat Hawaii as the factual and the U.S., Michigan and Nevada estimates as the counter-factual. That is, if Hawaii did not have an employer-mandate, employment and coverage should mimic the rest of the United States controlling for other factors. Therefore, the difference in employment patterns and coverage can be inferred to be due the employer-mandate, if other factors have been properly controlled. Population Studied: We produce model-based estimates of employment by hours-worked and insurance coverage using the Current Population Survey, Annual Social and Economic Supplement 1988-2003. Estimates are for working age adults, 19-64 years, and are produced for Hawaii, Nevada, Michigan and the U.S. overall. Principal Findings: Our empirical evidence generally supports all three hypotheses. First, private employmentbased insurance coverage for full-time workers is more prevalent in Hawaii than the U.S. overall. Second, there is substantial avoidance of the employer-mandate in Hawaii by skirting the 20 hour rule. Third, switching to other types of insurance is significant compared to the other states and the U.S. overall. Conclusions: Hawaii’s Pre-paid Health Care Act of 1974 has increased insurance coverage overall. It has also generated labor market sorting, as employees who prefer cash over fringe benefits seek part-time employment and employers happily accommodate to avoid the employer-mandate. Implications for Policy, Delivery or Practice: Hawaii’s employer-mandate has allowed policy-makers to increase insurance coverage to Hawaii’s population while shifting a portion of premiums to federal tax-payers via the federal tax treatment of fringe benefits. While mandated coverage is generally considered distorting, Hawaii’s 20 hour rule provides a safety-valve which reduces any inefficiency generated. Primary Funding Source: HRSA, Hawaii State Department of Health • The Financial Feasibility of Healthcare System Reform Steven Schramm Presented by: Steven Schramm, Principal, Mercer Government Human Services Consulting, 3131 East Camelback, Suite 300, Phoenix, AZ 85016; Tel: 602.522.6500; Fax: 602.957.9573; E-mail: steve.schramm@mercer.com Research Objective: States provide critical laboratories to test health system reform proposals. Our objective was to develop a Statewide Health Care System Pricing Model that would provide a baseline cost for all health care expenditures in a state as well as a vehicle to price out the impact of various statewide health care reform proposals on those expenditures. Study Design: Actuarial analysis of four health care reform proposals was conducted to determine who would be covered, the cost per person and total cost. Mercer established baseline cost and population estimates by pre-defined insurance market segments under the current health care delivery model. For insured individuals, this was done by identifying their primary linkage to the health care coverage marketplace – who provided their coverage. For the uninsured, which by definition has no linkage to the primary health care coverage marketplace, Mercer determined their secondary linkage to the health care marketplace – where they could receive health care coverage. Mercer next developed baseline cost estimates for all individuals according to their primary and secondary linkages to the health care coverage marketplace. These baseline costs were further delineated by major product type and key category of service. Mercer then determined the funding source associated with the cost of coverage by the four major payor sources: employer, government/military, Medicaid, and the uninsured. The funding was further split into employer costs, out-of pocket costs, and administrative costs by payor source. For each health care reform proposal, Mercer shifted the populations according to the program design of each reform proposal. Mercer tracked the underlying risk of each population as it moved among payor sources and product types within each proposal. The impact of each health care reform proposal was determined by; • the number of Arizonans who would be covered by the various forms of public and private health insurance • the costs of extending coverage, and • the implicit shift in funding sources. Population Studied: All Arizonans, by health care costs and coverage category, excluding Medicare Principal Findings: Creating a state tax credit would provide coverage for 4% (38,000) of the currently uninsured population, increasing total health spending by $33.4 million annually. Expanding Medicaid through a sliding fee scale buyin would increase coverage by 14% (134,000), also at an annual cost of $33.4 million. An employer mandate would cover 48% of the uninsured (479,000) an increase of $400 million. Creating a ‘public utility’ to administer a universal, single-payor system would cover 100% of the uninsured at an increased annual cost of $600 million. Conclusions: Creating a public utility was the most costeffective model on a per person basis. Tax credits and the employer mandate were least cost-effective. Implications for Policy, Delivery or Practice: Our analysis has shown that health care reform proposals have widely divergent impacts from a cost and coverage perspective. Policy-makers need to be educated about the trade-offs among cost and coverage in general as well specifically by payor source. The health policy debate will be shaped not only by how much reform costs, but who must pay for it. Primary Funding Source: St. Luke's Health Initiatives • Prehospital Time and Outcome in Pediatric Trauma Roopa Seshadri, Ph.D., Denise Goodman, M.D., M.S. Presented by: Roopa Seshadri, Ph.D., Assistant Professor, Child Health Research Program, Northwestern University, 2300 Children's Plaza, #157, Chicago, IL 60614; Tel: 773.327.9646; Fax: 773.327.9688; E-mail: r-seshadri@northwestern.edu Research Objective: We sought to investigate the challenges to regionalization of trauma services for children by comparing outcomes between those treated in urban vs. nonurban settings. We hypothesized that children in non-urban areas would have longer prehospital (PH) times (sum of emergency medical services (EMS) response, scene, and transport times) than those in urban areas, and would have worse outcomes. Study Design: This was a retrospective data analysis using the Trauma Registry of the Illinois Department of Public Health, covering 1999-2001. Outcome was defined as (1) discharged from Emergency Department (ED); (2) further medical care needed; and (3) expired. PH time was studied using linear regression models, with urban vs. non-urban, patient age, and PH Glasgow Coma Score (GCS) as predictors, including the interaction effect of age and GCS. To determine the effect of urban vs. non-urban and PH time on ED and hospital outcomes, mixed-effect cumulative logit models were used, with hospital as a clustering effect. Population Studied: We studied children < 19 years of age. The city of Chicago and its 6 surrounding counties comprised the urban center, and was compared to the rest of the state. Principal Findings: Approximately 8000 observations were used. PH time was 35.2 +/- 17.2 mins non-urban vs. 29.7 +/10.3 mins urban (p < 0.001). Amongst the younger children (<12 yrs), those with GCS < 13 have shorter PH time (p < 0.001). PH time does not affect ED outcome but is related to ultimate hospital outcome (p < 0.001). If PH time is < 30 minutes, patients are 1.5 times as likely to have a better outcome (p < 0.001; 95 % CI (1.4, 1.8)). There was no difference in hospital outcome between urban and non-urban children. The hospital-level clustering effect was negligible. Conclusions: PH time is a significant determinant of hospital outcome. This marker of access to care is more important than is the distinction between urban vs. non-urban location. The influence of PH time transcends ED events and carries over into final hospital outcome. Implications for Policy, Delivery or Practice: Our results underscore the importance of early definitive hospital care in pediatric trauma. Further insight into the factors determining PH events may permit better regionalization of pediatric trauma care and improved access for children. • Disenrollment from Medicare Risk Health Maintenance Organization to Fee-for-Service among Dual MedicareMedicaid Eligibles and Other Vulnerable Subgroups, California, 1996-2001 Elizabeth Sloss, Ph.D., Melissa Lopez, Ph.D., June O’Leary, Ph.D., Nasreen Dhanani, Ph.D., Glenn Melnick, Ph.D. Presented by: Elizabeth Sloss, Ph.D., Natural Scientist, Health, RAND, 1200 South Hayes Street, Arlington, VA 22202; Tel: 703.413.1100 Ext. 5633; Fax: 703.414.4717; E-mail: sloss@rand.org Research Objective: To describe changes in rates of disenrollment from Medicare risk health maintenance organizations (HMO) to fee-for-service (FFS) from 1996 through 2001 among dual Medicare-Medicaid eligibles and other vulnerable subgroups of Medicare beneficiaries in California. Monthly and annual disenrollment patterns, as well as rates for demographic and geographic subgroups, are described for beneficiaries in Medicare risk HMOs. Study Design: Disenrollment rates were computed for all Medicare beneficiaries, dual eligibles, and other vulnerable subgroups in California using enrollment data from the Centers for Medicare and Medicaid Services (CMS) for 19962001. Other vulnerable beneficiary subgroups included females, African-Americans, disabled, and 85 years or older. Given the occurrence of monthly changes in entitlement and HMO status, rates were calculated on a monthly basis and then aggregated over 12 months to yield a disenrollment rate per member-year. Disenrollment to FFS was defined as a change from a risk HMO to FFS after which the beneficiary remained in FFS for at least three months. Population Studied: All Medicare beneficiaries eligible for Part A and Part B, regardless of age, who were enrolled in a Medicare risk HMO in California for at least one month between January 1996 and December 2001. Beneficiaries enrolled in cost-reimbursed HMOs, health care prepayment plans or demonstrations were excluded. Principal Findings: From 1996-2001, dual-eligibles disenrolled from Medicare risk HMOs to FFS at annual rates varying from 14% in 1996 to 18% in 2001. Disenrollment from Medicare risk HMOs was lower among other vulnerable subgroups, including females (2.5% in 1996 to 4.6% in 2001), African-Americans (4.3% to 7.0%), disabled (4.3% to 8.0%), and those 85 years or older (3.5% to 4.9%). The disenrollment rates among dual eligibles were much higher than among all Medicare beneficiaries (2.4% to 4.9% over 1996-2001). Rates among other vulnerable subgroups were closer to all Medicare beneficiaries. Monthly disenrollment rates peaked in December almost all years and subgroups, with dual-eligibles having higher rates than other vulnerable subgroups in all 12 months. Conclusions: Annual disenrollment rates from Medicare risk HMOs to FFS are higher among vulnerable beneficiaries than among other Medicare beneficiaries, with the highest rates among dual-eligibles. Disenrollment rates are considerably higher in 2001 than in 1996. Implications for Policy, Delivery or Practice: With high rates of disenrollment from Medicare risk HMOs, providing managed care options for vulnerable subgroups, especially dual eligibles, will continue to be a challenge. If higher disenrollment indicates lower satisfaction, it will be important to monitor access to care and quality of care provided by risk HMOs to these vulnerable beneficiaries. Updated studies of the reasons for disenrollment would assist CMS in its efforts to manage the care of vulnerable beneficiaries, and to coordinate Medicare and Medicaid benefits for dual-eligibles. Primary Funding Source: AHRQ • A Wealth of Local Expertise – Putting Together a Statewide Health Research and Evaluation Collaborative to Encourage Evidence-Based Decision-Making in Oregon Jeanene Smith, M.D., M.P.H., Lisa Krois, M.P.H., Tina Edlund, M.P.H. Presented by: Jeanene Smith, M.D., M.P.H., Deputy Administrator, The Office for Oregon Health Policy & Research, 800 N.E. Oregon Street, RM 607, Portland, OR 97232; Tel: 503.731.3005 Ext. 652; Fax: 503.872.6832; E-mail: Jeanene.Smith@state.or.us Research Objective: To inform policymakers as the restructured Oregon Health Plan (OHP) faces significant cuts in an ongoing budget crisis, the State of Oregon brought together local health service researchers representing managed care organizations, hospital systems, behavioral health, advocates and interested stakeholders to form the Oregon Health Research and Evaluation Collaborative (OHREC). This unique collaborative provides the opportunity for evaluative research to be conducted with local working knowledge of Oregon’s health services system but independent of the state. OHREC’s goals are to: 1.Facilitate research efforts to inform the legislature and stakeholders on OHP issues 2.Procure grants to fund collaborative research projects to study long-term impacts to Oregon Health Plan changes 3.Establish a working information network among health services researchers and the state for improved communication and efficiency of research efforts 4. Create a streamlined process for health services research on the Oregon Health Plan. Study Design: Multiple studies at the system, program and individual level have been conducted with the focus of health care delivery, benefit redesign, and the financial delivery systems of the Oregon Health Plan. All studies where conducted through the collaborative with results disseminated back to Oregon’s decision makers. Population Studied: Oregon’s Medicaid program (OHP); vulnerable populations Principal Findings: Research must be conducted in a timely manner to remain relevant to issues facing decision-makers. OHREC was successful in accomplishing this with several initial projects, with its’ survey of Oregon’s Medically Needy Program beneficiaries as a prime example. By conducting the survey shortly after the program was cut due to the state’s budget crisis, the results were quickly available and used by policymakers as they designed a scaled-down, replacement pharmacy assistance program for this vulnerable population. Research can provide valuable information to decision-makers but the key is getting it heard. Results must be disseminated using appropriate language and in accessible formats in order to be considered by policy makers. OHREC has used a variety of communication strategies including, short public presentations by researchers to policymakers and advocates, two-page research briefs, Listservs and verbal dissemination by staff to key policymaking bodies. The success of a collaborative depends on building relationships between policymakers and the researchers while balancing the competing needs of academic and political processes. The vision/mission of a collaborative is as important to retaining participating researchers as the potential for grant monies. Conclusions: The collaborative has proven to be a successful model for facilitating research important to state health services with scarce resources. Multiple strategies must be employed in order to connect research to policy makers. A collaborative’s success depends upon a delicate balance of researchers’ and policymakers’ needs Implications for Policy, Delivery or Practice: Timely and relevant research is important for evidence-based decision making on health care policy. Linking policymakers to researchers is key in order for this to happen. Primary Funding Source: RWJF • How Do Closures of Labor and Delivery Services Affect Community Access to Care? Joanne Spetz, Ph.D., Lisa Simonson Maiuro, Ph.D., Paul Kirby, M.A. Presented by: Joanne Spetz, Ph.D., Assistant Professor, Community Health Systems, University of California, San Francisco, 3333 California Street, Suite 410, San Francisco, CA 94118; Tel: 415.502.4443; Fax: 415.502.4992; E-mail: jojo@alum.mit.edu Research Objective: There is general consensus that hospitals face many challenges in today’s health care environment. Financial pressures may induce hospitals to drop certain categories of services that are perceived as unprofitable. As a result, some communities may lose access to needed services, and this loss of service may affect access to care. Study Design: The study design relies on a combination of quantitative and qualitative analysis to identify how hospital services and access to these services has changed among California hospitals. To begin, we examined changes in the availability of hospital services using patient discharge data. We grouped discharges into 48 categories, and then identified hospitals that had a decline of 85% or more in any service category. We did not examine closures of emergency rooms, because this issue has been addressed in other studies. Using the discharge data, we also created hospital service measure of concentration, similar to a Herfindahl (HHI) index, to reflect changes in the range of services provided by hospitals. After identifying types of services most likely to close, we conducted case studies to provide insight into how communities are affected by service changes. These case studies involved both data analysis and key informant interviews. Population Studied: All California hospitals that reported patient discharge data to the Office of Statewide Health Planning and Development (OSHPD) in 1995 and 2002. Principal Findings: The service mix among hospitals that have remained in operation over the time period was remarkably stable as reflected by a hospital concentration index. Among inpatient-based services, obstetric and gynecological care was most affected. Less than 10% of hospitals offering obstetric and gynecological care in 1995 eliminated this service by 2002. Rural hospitals were disproportionately affected. In most cases, the hospitals had low volumes of deliveries and did not offer labor and delivery care for high-risk births. Case studies of hospitals that closed their obstetric services found that there was little impact on access to care in urban and near-urban communities, but significant effects in rural areas. Conclusions: Although there has been concern that financial pressures are leading hospitals to eliminate critical community services, we found little evidence of this occurring on a widespread scale. Obstetric services were most commonly eliminated, with a disproportionate share of obstetric closures being in rural hospitals. Closure of obstetric services in rural communities has significant effects on access to care for community residents. Implications for Policy, Delivery or Practice: There was not a widespread closure of hospital services in the late 1990s. This suggests that the financial pressures faced by hospitals rarely leads to service closures. However, a number of hospitals eliminated their labor and delivery services. Closures of labor and delivery services have little effect on access to care in urban communities. In rural communities, however, labor and delivery closures result in significant barriers to access to care for patients both in travel for delivery and also in reduce community presence of obstetriciangynecologists. Policymakers and health care providers need to pay particular attention to closures of services in rural regions. Primary Funding Source: California HealthCare Foundation • Geographic Access to Hospital-Based Perinatal Care for At-Risk African American Women in South Carolina John Stewart, MS, M.P.H., Ana Lopez-DeFede, Ph.D. Presented by: John Stewart, MS, M.P.H., GIS Manager, Institute for Families in Society, University of South Carolina, 937 Assembly Street, Suite 618, Columbia, SC 29208; Tel: 803.777.5516; Fax: 803.777.1793; E-mail: jstewart@gwm.sc.edu Research Objective: Identify and map African American women at risk for poor perinatal outcomes in a South Carolina perinatal care region. Evaluate appropriate health care accessibility by estimating travel distance to hospital-based perinatal care for at-risk women. Study Design: Data from the 2000 Census were used to determine the number of African American women at risk for poor perinatal outcomes (women ages 15 to 17 and 35 to 44) in 24,949 census blocks comprising the Pee Dee perinatal care region in South Carolina. Using a geographic information system (GIS), area hospitals providing basic, specialty, and/or subspecialty perinatal care were geocoded at the street address level. A GIS network tool then was used to determine actual road distance from each populated block centroid (point representing the geographic center of a census block with one or more at-risk women) to the nearest hospital(s) providing basic perinatal care, specialty care, and subspecialty care. Population Studied: African American women ages 15 to 17 (N=5,355) and 35 to 44 (N=15,989) in an eight-county perinatal care region in South Carolina. Principal Findings: Geographic access to hospital-based perinatal services varies markedly by level of care. On average, African American women ages 15 to 17 live 9.3 miles from basic perinatal services, 18.4 miles from specialty care, and 31.8 miles from subspecialty care. Similarly, African American women ages 35 to 44 reside, on average, 9.2 miles from basic care, 18.0 miles from specialty care, and 31.2 miles from subspecialty care. Notably, 19.6% of younger women and 19.0% of older women live 50 miles or more from hospitalbased subspecialty perinatal care. Conclusions: Distance from care may limit access to hospitalbased perinatal services (especially specialty and subspecialty care) for at-risk African American women in the Pee Dee region. Efforts to identify and serve geographically remote subpopulations may improve perinatal outcomes and lower health care costs. GIS technology can be used to locate geographically remote at-risk subpopulations, identify community-based perinatal care specialists (e.g., obstetricians, gynecologists) serving remote areas, coordinate neonatal and maternal transportation to and from appropriate perinatal care sites, and target perinatal health education initiatives. Implications for Policy, Delivery or Practice: Enhanced coordination of regional perinatal care to improve perinatal outcomes for geographically remote at-risk African American women. Primary Funding Source: SC Department of Health and Human Services • ZIP-Code Characteristics Correlated with Individuals' HIV-Testing Stephanie Taylor, Ph.D., M.P.H., Arleen Leibowitz, Ph.D., Paul Simon, M.D., M.P.H. Presented by: Stephanie Taylor, Ph.D., M.P.H., Associate Social Scientist, Health, RAND, 1700 Main Street, P.O. Box 2138, Santa Monica, CA 90407; Tel: 310.393.0411 Ext. 7965; Fax: 310.451.7062; E-mail: staylor@rand.org Research Objective: To understand HIV-testing impediments, most studies focus on individuals’ characteristics, ignoring the potential effect of individuals’ residential areas. This paper explores eleven ZIP code characteristics as correlates of HIV-testing, controlling for individuals’ characteristics. It is important for understanding HIV-testing decisions because it documents the role of individuals’ residential areas and individuals’ risk and demographic factors. Identifying ZIP code-level correlates of low testing rates, particularly in higher-risk areas, is potentially important to public health agencies nationally. They can use these insights to increase their outreach efforts in areas where many higher-risk people reporting relatively low testing rates reside. Study Design: Individuals’ data was from the 1999 Los Angeles (L.A.) County Health Survey, a random digit-dialed population survey. A multi-level logistic regression analysis of HIV-testing was conducted using two samples: 1) respondents in all 233 ZIP codes and 2) the subset of persons in 20 regions having larger proportions of respondents reporting “higherrisk” sex (defined as, in the prior year, not always using condoms and having more than one partner). Regions were merged, contiguous ZIP codes containing over 30 respondents each. Higher-risk regions were examined because, given limited resources, it is important to focus HIV prevention efforts where they could have the greatest impact – where concentrations of residents at higher-risk for HIV live. MLwiN software with full and restricted maximum likelihood estimation methods was used. Population Studied: The sample was a 1999 random probability sample of L.A. adults (n=5,267). A third were Latino, 10% were African American, 12% were Asian Pacific Islanders, and 4% were “Other”. Principal Findings: African American and Latino individuals were more likely to test for HIV than Whites, consistent with national trends and HIV/AIDS prevalence rates. However, a different pattern of testing variation was evident at the ZIPcode level. Throughout L.A. and in higher-risk regions, persons living in ZIP codes with greater proportions of African Americans residents were more likely to test for HIV than in persons living in predominately White or Latino areas, regardless of individuals’ characteristics or neighborhoods’ number of AIDS cases and testing sites. However, residents of predominately Latino areas were no more likely to test than residents of mostly White neighborhoods. Furthermore, persons living in neighborhoods with greater percentages of married residents had lower testing rates, regardless of individuals’ marital status or risky behaviors. Conclusions: Latino individuals were more likely to test for HIV than Whites. However, the average testing rate for all residents of predominately Latino neighborhoods was lower than that for all residents of African American neighborhoods. The mechanism for this effect remains unknown. Implications for Policy, Delivery or Practice: Future research might attempt to elicit the HIV test-promoting mechanisms present in African American neighborhoods but absent from Latino communities. These might include replicable mechanisms such as increased HIV-testing outreach or nonreplicable mechanisms such as social networks that spread news of newly infected neighbors, subsequently prompting persons to test. African Americans’ networks may be more likely to affect individuals’ testing decisions than Latinos’ because the African American community has more openly and aggressively addressed the HIV/AIDS issue than the Latino community. • Use of Chiropractic Services in the Rural South Samruddhi Thaker, MBBS, M.H.A., Donald Pathman, M.D., M.P.H. Presented by: Samruddhi Thaker, MBBS, M.H.A., Research Associate, , Cecil G. Sheps Center for Health Services Research, 725 Airport Road, Chapel Hill, NC 27516; Tel: 919.966.4371; Fax: 919.966.3811; E-mail: thaker@mail.schsr.unc.edu Research Objective: The use of chiropractic services has grown over the past three decades; use in the South is lower than in other parts of the U.S. Less is known about how often rural individuals seek chiropractic services. This study examines the use of chiropractic services among adults in the rural South. We assess how chiropractic use varies with the availability of chiropractors across states and relates to individuals’ opinions of and access to physician care to assess if chiropractors are used as substitutes for physician services. Study Design: This is a secondary analysis of cross-sectional data collected through a random-digit dialing survey in 150 rural counties in eight southern states—AL, AR, GA, LA, MS, SC, TX, and WV—participating in the Robert Wood Johnson Foundation’s Southern Rural Access Program. The survey was conducted to determine access to outpatient services in the rural south. The primary outcome variable for the present study is subjects’ response to the question “In the past 12 months, have you been to see a chiropractor?” Analyses were weighted to adjust for sampling probabilities and demographic group response rates. Population Studied: Respondents were randomly selected from eligible adults living in the households contacted. Eligible participants were age 18 and older who had lived in the sampled county for at least a year and spoke English or Spanish. 4,879 respondents completed the survey for a response rate of 50.7%. Principal Findings: Of the 4,869 adults answering the question on chiropractic use, 353 (6.54%) reported seeing a chiropractor in the previous year. Chiropractor use was more common among whites than blacks (8% vs. 4%, p=0.001) and among the married, more educated, employed, and higher income individuals (p<0.05 for all). Chiropractic service use varied from 3.2% to 9.8% across states (p=0.0001) and tended to correlate with states’ chiropractorto-population ratios (r=0.61; p=0.1). 9% percent of those who said the cost of physician care was problematic used chiropractors compared with 6% who said the cost was not problematic (p=0.005). Other barriers to physician care—lack of health insurance, not having a usual source of care, dissatisfaction with medical care, confidence in doctor’s ability, and satisfaction with concern shown and getting questions answered—were unrelated to the use of chiropractors. Conclusions: 6.54% of individuals in this southern rural adult population reported chiropractic service use in the previous year. Use was higher among those who were white, married, more educated, employed and had higher incomes. Use varied greatly across states in association with chiropractor-topopulation ratio. Except cost of physician care, people’s experiences and perceptions of the care they received from physicians were unrelated to their likelihood of having received care from a chiropractor. Implications for Policy, Delivery or Practice: These findings provide information on the demographic characteristics of individuals who use chiropractic services in the rural south. We find some evidence that recent efforts of the National Health Service Corps and Veterans Administration to enhance chiropractor availability in underserved areas and in veteran administration treatment settings, respectively, will increase chiropractic service use by these populations. Primary Funding Source: RWJF • The Cost of State-Based Universal Health Care Coverage Teresa Waters, Ph.D., Peter Budetti, M.D., JD, Linda Perloff, Ph.D. Presented by: Teresa Waters, Ph.D., Associate Professor, Center for Health Services Research & Department of Preventive Medicine, University of Tennessee Health Science Center, 66 N. Pauline Street, Suite 463, Memphis, TN 38163; Tel: 901.448.5826; Fax: 901.448.8009; E-mail: twaters@utmem.edu Research Objective: To analyze costs and cost savings associated with various state-level options for universal health care coverage. Study Design: Data for the state of Illinois are used to examine reform options. A model of current state health expenditures was developed as a baseline. Four general reform options were postulated and the financial impacts of these changes derived using existing data and databases. Population Studied: The population and expenditure patterns of individuals in Illinois. Principal Findings: Using decision analysis models, we find that expanding coverage to all residents of Illinois would cost between $2.2 and $4.5 billion in the first year, but that incremental costs would diminish in subsequent years under the single payer approaches, principally due to savings in insurance overhead and administrative costs. By the 5th year of implementation, single payer models providing universal coverage would cost less than the current health care system with 1.7 million uninsured. Using probabilistic sensitivity analyses to allow key parameters to vary simulaneously, we find that single payer reforms are a least-cost strategy in more than 90 % of Monte Carlo iterations by the 8th year of implementation. Conclusions: State-level reforms appear to be a financiallyviable option for attaining universal health care coverage. Implications for Policy, Delivery or Practice: The cost savings associated with single payer models are too significant to be ignored. Policies and programs that seek to capture these savings could significantly curb health care costinflation and produce sufficient additional funding to provide coverage for the uninsured. Primary Funding Source: Illinois Physician Committee for the Defense of Medicine • How Many Emergency Department Visits Are There? Robin Weinick, Ph.D., Pamela Owens, Ph.D., Roxanne Andrews, Ph.D., John Sommers, Ph.D., Steven Machlin, M.S. Presented by: Robin Weinick, Ph.D., Senior Advisor for Intramural Research/Senior Advisor for Safety Nets and Low Income Populations, OPART, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850; Tel: 301.427.1573; Fax: 301.427.1792; E-mail: rweinick@ahrq.gov Research Objective: Early in 2003, Wears commented on an article published in the Annals of Emergency Medicine that used data from the Medical Expenditure Panel Survey (MEPS) to study emergency department (ED) visits in the U.S. He notes the large difference between the MEPS estimate of the number of ED visits (43.8 million in 1998) and estimates from the National Hospital Ambulatory Medical Care Survey (NHAMCS, 100.4 million in 1998). While known methodological differences between the surveys may contribute to this difference, he concludes that such differences account for only a small proportion of the disparity. In this paper, we add comparisons with the American Hospital Association Annual Survey (AHA) and the National Health Interview Survey (NHIS). In addition, we use administrative data from the Healthcare Cost and Utilization Project (HCUP) to help improve these estimates for 2001. Study Design: We compare national estimates derived from two household-based surveys (the Medical Expenditure Panel Survey (MEPS) and the National Health Interview Survey (NHIS)) and two facility-based surveys (the National Hospital Ambulatory Medical Care Survey (NHAMCS) and the American Hospital Association Annual Survey (AHA)). We also examine ED visits from the universe of discharge records contained in the State Emergency Department Databases (9 states), State Ambulatory Surgery Databases (additional 9 states), and the State Inpatient Databases (additional 15 states), all part of the Healthcare Cost and Utilization Project (HCUP). We compare the number of ED visits per hundred population to better understand the differences between the various data sources and to provide state and regional comparisons. Population Studied: Users of EDs in the U.S. Principal Findings: Survey estimates of the number of ED visits made by major data sources differ by as little as 1.5% to as much as 92.8% (e.g., NHAMCS 107,490,000, AHA 105,865,778, and MEPS 55,758,852 visits), with data collected from facilities providing higher estimates than that collected from households. However, estimates from the AHA data closely approximate state estimates collected from the universe of ED discharges in HCUP (77 to 100%). ED visit rates vary from 25 to 50 visits per 100 persons across the 9 states participating in all 3 HCUP databases. Conclusions: Major data sources provide widely differing estimates of the number of ED visits in the U.S. Householdbased surveys may suffer from underreporting due to respondent bias. Implications for Policy, Delivery or Practice: Knowing how many ED visits occur in the U.S. annually is important for health care planning, including both bed and staffing requirements, and for monitoring access to care and utilization patterns. In addition, a better understanding of the types of ED visits that are missing from the NHIS and MEPS data can help improve future research. Primary Funding Source: AHRQ • Who Can't Pay for Health Care? Robin Weinick, Ph.D., Sepheen Byron, M.H.S., Arlene Bierman, M.D., M.S. Presented by: Robin Weinick, Ph.D., Senior Advisor for Intramural Research/Senior Advisor for Safety Nets and Low Income Populations, Office of Performance Accountability, Technology, and Resources, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850; Tel: 301.427.1573; Fax: 301.427.1792; E-mail: rweinick@ahrq.gov Research Objective: In an era of rising health care costs and budget constraints, an increasing number of Americans may have difficulty paying for needed health care services. In this study, we make use of a unique data set to estimate the number of American adults who have problems paying for health care and their characteristics, including health insurance coverage and health status. We also present information on the extent to which people who have private health insurance coverage nonetheless report such difficulties. Study Design: The Commonwealth Fund 2001 Health Care Quality Survey collected detailed information about experiences with health care from a nationally representative sample of 6,722 adults ages 18 and older living in the continental U.S. Measures used in this analysis include having put off, postponed, or not sought medical care due to cost in the past 12 months; not filling a prescription medicine due to cost; having difficulty or not seeing a needed specialist because could not afford to; not following the doctor’s advice or treatment plan, or not getting a recommended test or seeing a referred doctor because it cost too much; having used alternative care because it is a cheaper way of getting care; or having any of these problems. Bivariate comparisons and logistic regressions were performed using Stata 7 to account for the survey design. Population Studied: Americans ages 18 and over. Principal Findings: The proportion of American adults having any one specific problem obtaining health care due to cost is small, with estimates ranging from 2.1 to 6.9 percent. However, the proportion of adults experiencing any of these problems affording care is more substantial (16.9%), including 14% of the privately insured and 22% of nonMedicare publicly insured. Of individuals reporting a functional impairment, 25% reported at least one problem affording care. Overall, Medicare beneficiaries reported fewer financial barriers to care. These findings remained significant in logistic regressions adjusting for sociodemographic characteristics. Conclusions: Some of our findings are encouraging: Medicare enrollees on average report relatively lower frequency of these problems and, net of other characteristics, there are no racial or ethnic disparities in problems affording health care. Other findings, however, are cause for concern. Substantial proportions of Medicaid enrollees as well as certain groups among those with private coverage, such as those with low incomes, report problems paying for needed health care. Such problems among those with substantial functional impairments are also cause for concern. We estimate that the number of American adults experiencing any of these problems obtaining health care because of cost is between 38.1 and 41.4 million. Implications for Policy, Delivery or Practice: Some individuals continue to encounter significant financial barriers to care despite being privately or publicly insured. Individuals with disabilities also experience problems. Proposals to expand health insurance must address these barriers as well as the specific issues of individuals with greater need. Primary Funding Source: AHRQ • The Impact of an Intervention Promoting Mass Transit Use on Access to a Medical Home for Low-income, Minority Urban Children Serena Yang, M.D., Sujata Tipnis, M.D., M.P.H., Celina Saenz, B.S., Nancy Kelly, M.D., M.P.H. Presented by: Serena Yang, M.D., Clinical Post-doctoral Fellow, Pediatrics, Baylor College of Medicine, 6621 Fannin Street, MC-1540.00, Houston, TX 77030; Tel: 832.822.1549; Fax: 832.825.3435; E-mail: sxyang1@texaschildrenshospital.org Research Objective: To determine the impact of a mass transit program on missed well child appointment rate for children at risk for missing appointments at an urban clinic. Study Design: This is a randomized controlled trial of an intervention. Caregivers are randomized to 2 groups: those who receive routine discharge instructions at the end of each clinic visit (control), or those who receive bus tokens and transit education in addition to routine discharge instructions at all well-child visits (intervention). Participants are followed for 6 months. Population Studied: The Texas Children's Hospital Residents' Primary Care Group (RPCG) Clinic provides primary health care in an urban setting to a population that is primarily minority and low-income with significant transportation difficulties. This study targets caregivers of RPCG patients less than 12 months old who are at risk for missing appointments and who live within the mass transit service area. Principal Findings: 136 participants have been enrolled. At baseline, 55% reported they used their own car to reach clinic, 72% had never used public transit to clinic, and 25% reported missing an appointment due to transportation. When asked about their English-speaking ability using a 5-point rating scale, 36% responded not well (4 out of 5) or not well at all (5 out of 5). A preliminary analysis of the 60 participants who have completed the program reveals no significant difference in missed well-child appointment rate between the groups. However, within the intervention group, the receipt of multiple intervention boosters (p<0.001) and a higher income level (p=0.003) were associated with a lower missed appointment rate. Conclusions: Preliminary results show that the number of exposures to the mass transit program may be important in decreasing missed appointments, and that income level appears to be an important factor associated with missed appointments. Further data collection and analysis of the remaining study participants will help clarify the impact of mass transit coordination on alleviating access barriers for low-income, minority families with transportation difficulties. Implications for Policy, Delivery or Practice: Federallyfunded non-emergency medical transportation programs which promote the usage of mass transit have demonstrated significant cost savings for their respective State agencies. Additional research is needed to demonstrate the impact of the promotion of mass transit use on health outcomes of vulnerable populations who have the greatest medical transportation needs. Primary Funding Source: American Academy of Pediatrics • Vaccination Coverage in Employer-Sponsored SelfInsured Health Plans Fangjun Zhou, Ph.D., Mark Messonnier, Ph.D., Jeanne Santoli, M.D., Anjella Vargas-Rosales, M.S., Abigail Shefer, M.D. Presented by: Fangjun Zhou, Ph.D., National Immunization Program, Centers for Disease Control and Prevention, 1600 Clifton Road, N.E., MS E-52, Atlanta, GA 30333; Tel: 404.639.8251; Fax: 404.639.8614; E-mail: faz1@cdc.gov Research Objective: To estimate the Diphtheria-tetanuspertussis (DTP) vaccination coverage based on vaccinations administered in employer-sponsored self-insured health plans. Study Design: We conducted a retrospective study of the Medstat MarketScan database, which include enrollees (children and adults) of approximately 100 self-insured health insurance plans of 45 large employers each year. We selected all children born between July 1, 1997 and June 30, 1998 who were continuously enrolled until the end of 1999. The DTP vaccination coverage among children aged 18 to 29 months was estimated based on in-plan administration of 3 or more doses of DTP vaccine (DTP3) and analyzed by characteristics of the families and their health insurance plans. Population Studied: Enrollees in the MarketScan database Principal Findings: Overall DTP3 in-plan vaccination coverage was 47.3%. The coverage rates for children residing in metropolitan statistical areas (MSA) and non-MSAs were 52.7 and 31.1%, respectively; for children who enrolled in health maintenance organization, point of service, preferred provider organization and fee for service coverage rates were 79.3%, 77.3%, 32.3% and 18.9%, respectively; for children who enrolled in capitated insurance plans coverage was 75.1% vs non-capitated plans (34.3%). Of children with well-baby coverage in their plan, 77.2% were vaccinated, while 26.3% without that benefit were vaccinated. Conclusions: Among employer-sponsored self-insured health plans, vaccination coverage based on vaccinations administered in-plan was highest in those that were capitated and for those children in plans which had well-baby coverage. Implications for Policy, Delivery or Practice: Knowledge of how vaccination coverage is related to families and their health plan characteristics may provide guidance for policymakers to improve vaccination among children in the U.S. • Health Status Measurement During a Household Evaluation of an Integrated Health Outreach Project Targeting Colonia Residents in Hidalgo County, Texas Miguel Zuniga, M.D., Dr.P.H., Craig Blakely, Ph.D., M.P.H., Martha Tromp, M.P.H., James Burdine, Dr.PH Presented by: Miguel Zuniga, M.D., Dr.P.H., Assistant Professor, Health Policy and Management, Texas A&M University System School of Rural Public Health, 3000 Briarcrest Drive, Bryan, TX 77802; Tel: 979.862.4142; Fax: 979.862.8371; E-mail: mzuniga@srph.tamu.edu Research Objective: To measure Colonia resident baseline health status in project intervention to improve the ability of lay health workers (promotoras) to impact health behaviors of Colonia residents and to integrate their activities with the actions of health providers in the area to change access and utilization rates. Study Design: Adult resident health status was measured using the Short Form 8 Health Survey. This survey measures physical and mental health domains and produces eight health scales. These scales are Physical Functioning, Role Physical, Bodily Pain, General Health, Vitality, Social Functioning, Role Emotional, and Mental Health. Respondents were selected randomly through selection of block groups, then households, then residents within households. Population Studied: A household survey targeted 643 residents in two-intervention project Colonias and a nonintervention control Colonia in Hidalgo County. Principal Findings: An overall response rate of 89% was accomplished. The adult respondents had a mean age of 40 (18-86), 56% were female, 94% were Hispanic, and 68% were married. The health status (SF-8) physical and mental summary scales were statistically different (worse) than national norms. With the exception of the mental health scale, the remaining seven health scales were statistically different (worse) than national norms. Health scale specific comparisons for disease burden and insurance status are presented. Conclusions: The measurement of health status is a useful tool for evaluation of community-based health interventions. The health disparities in the mostly Hispanic population are evident in their worse health scale score when compared to the US national norms. Implications for Policy, Delivery or Practice: The measurement of health status is a practical strategy to assess progress to the elimination of health disparities. Primary Funding Source: RWJF