Coverage & Access Call for Papers Coverage Expansions & the Uninsured Chair: Catherine McLaughlin, University of Michigan Sunday, June 26 • 5:30 pm – 7:00 pm ●Who Benefits When States Expand Medicaid Coverage to Parents? Susan H. Busch, Ph.D., Elizabeth Vigdor, Ph.D. Presented By: Susan H. Busch, Ph.D., Assistant Professor, Health Policy, Yale Medical School, 60 College Street, New Haven, CT 06520; Tel: (203)785-2927; Email: susan.busch@yale.edu Research Objective: One common metric used to assess competing health insurance reform proposals is the increase in insurance rates caused by the program. Yet policy makers may also be interested in other outcomes, including any changes in health that arise from increased insurance coverage. Few studies have examined differences between proposals in who becomes insured. We expect that those with either chronic conditions or health risk factors will be more likely to benefit from becoming insured. We examine who becomes insured under one proposed policy for increasing insurance coverage rates – expanding Medicaid to low income parents. Study Design: We consider whether low income individuals with chronic conditions (diabetes), health risk factors (smoking, obesity) or those in fair or poor health were more likely to become insured or use needed health services than other low income individuals under the parent Medicaid/SCHIP expansions. We use instrumental variable regressions to study the effects of these policies. We instrument for actual eligibility using a simulated eligibility measure for each state and year. Population Studied: National Data come from the Behavioral Risk Factor Surveillance System (BRFSS) and the Current Population Survey (CPS) from 1996-2002. Principal Findings: We find that when states expand income eligibility levels for parents those most likely to take up coverage have chronic conditions or additional risk factors for disease. Additionally, we find that parents with these conditions are less likely to report foregoing needed care due to cost in expansion states. For example, we find that 20 percent of newly eligible, uninsured smokers enrolled in public coverage after the expansions, compared with only 10 percent of newly eligible, uninsured non-smokers. Examining use of services, there was no change in foregoing needed care due to cost in the non-smoking group, but a 25 percent reduction in foregoing needed care due to cost for the smoking group. Unlike previous research, we find that Medicaid expansions at these income levels did not have differential effects by race/ethnicity on health care utilization. Conclusions: In considering public coverage expansions, the fact that vulnerable groups were more likely to take up coverage suggests these programs will have important health benefits to parents. Implications for Policy, Delivery, or Practice: This work suggests that health outcomes should be considered as well as the number of newly insured individuals when comparing competing health insurance proposals. The recent Medicaid expansions appear to have improved coverage disproportionately for individuals who are most likely to benefit from having health insurance: low income people with chronic conditions or health risk factors. Policies that use a different mechanism for increasing coverage may be more or less successful at targeting these groups, which has implications for the efficiency of health system reforms. Primary Funding Source: No Funding ●Does Imputation Bias Lead to Finding Significantly More Uninsured in the Current Population Survey’s Estimates of Health Insurance Coverage? Michael Davern, Ph.D., Lynn A. Blewett, Ph.D., Holly Rodin, MS, Kathleen Thiede Call, Ph.D. Presented By: Michael Davern, Ph.D., Assistant Professor, SHADAC, University of Minnesota, 2221 University Avenue SE, Suite 345, Minneapolis, MN 55414; Tel: (612) 625-4835; Fax: (612) 624-1493; Email: daver004@umn.edu Research Objective: Determine whether the imputation procedure used by the US Census Bureau to replace missing data produces bias in the estimates of health insurance coverage in the Current Population Survey’s Annual Social and Economic Supplement (ASEC). Study Design: Twelve percent of the ASEC health insurance coverage values are missing and imputed by the US Census Bureau. We compare the health insurance coverage of imputed respondents in the 2003 Current Population Survey’s ASEC to those without imputed health insurance coverage. We compare demographic characteristics of the two groups and model the likelihood of having insurance coverage given the data are imputed controlling for the same demographic characteristics. Population Studied: The non-institutionalized population in 2003. Principal Findings: Although properly specified imputation can alter basic distributional summary statistics (e.g., means, rates and variances) from the statistics calculated using complete cases only, it should not transform the relationships among variables. In other words, imputation should not create significant covariances between variables that were not there before the imputation, nor should it reduce the magnitude of significant covarainces between variables that were there prior to imputation. In the 2003 ASEC, 59.7% of 18-64 year old adults have commercial health insurance coverage if they have imputed data. However, 72.4% of the non-full supplement imputations have commercial health insurance coverage. Furthermore, imputed cases have a 26.7% uninsurance rate while all other 18-64 year old adults in the CPS have an uninsurance rate of 14.6%. Part of this differential is due to demographic difference between those with imputed data and those without, but having imputed data remains a significant predictor of health insurance coverage in multivariate models controlling for the demographics. Conclusions: The imputed ASEC data seem to be coding too many people to be uninsured and too many people to have both public and private insurance coverage. The US Census Bureau should consider reviewing and altering its imputation specifications. Implications for Policy, Delivery, or Practice: The ASEC is likely to be imputing too many people to be uninsured, resulting in a count of the total number of people who are uninsured being too high. We attempt to quantify how much higher the estimates from the ASEC are due to this imputation problem and depending on the simulation its anywhere from 1-4 million people too high. Primary Funding Source: RWJF ●How Long Do People Remain Medically Uninsured? Julia Prentice, Ph.D. Presented By: Julia Prentice, Ph.D., Health Services Research Fellow, Bedford VAMC, Center for Health Quality, Outcomes and Economic Research, 200 Springs Road (152), Bedford, MA 01730; Tel: (781)687-2882; Email: jprentic@bu.edu Research Objective: It is essential to know how long people remain uninsured because policy options to decrease the number of uninsured will differ if people remain uninsured for short versus long periods. However, due to limitations in the type of data collected, past research estimating the length of uninsured spells has only included spells that begin during a defined observation period, i.e. new spells, and has excluded spells already in progress at the beginning of this observation period. Only including new spells may exclude the chronically uninsured leading to an underestimate of the average length of uninsured spells. This project uses a new data set that has duration information on both new spells and spells in progress to examine how the average length of uninsured spells varies if 1. only new spells are included, 2. only spells in progress are included and 3. all spells are included. Study Design: Data comes from the Los Angeles Family and Neighborhood Survey, L.A. FANS, which interviewed 2,623 adults from 65 different census tracts in Los Angeles County. L.A. FANS collected monthly information on health insurance coverage for the two-year period before the interview. Unlike earlier studies, L.A. FANS collected information on the start date of insurance spells already in progress at the beginning of this two-year period. Thus, both new spells that began during the two year period and spells in progress at the beginning of the two year period can be included in analyses estimating duration of uninsured spells. The median duration of uninsured spells were calculated by survival analysis using Kaplan-Meier estimation. The median duration of uninsured spells was compared between a sample that included only new spells, a sample that included only spells in progress and a sample that included all spells. Population Studied: Working age adults (18-64 years old) who experienced at least one uninsured spell during the two year period before they were interviewed. Principal Findings: Of the 946 uninsured spells experienced by respondents during the two-year EHC, 745 (79%) are spells in progress. Only 201 spells (21%) would be included in analyses that are limited to new spells. The median duration of an uninsured spell was 17 months when only including new spells, 43 months when including both new spells and spells in progress and 59 months when including only spells in progress. Conclusions: The exclusion of spells in progress at the beginning of an observation period severely underestimates the median duration of uninsured spells. Future research should include both spells in progress and new spells. Implications for Policy, Delivery, or Practice: Surveys that collect information on health insurance over time, such as the Survey for Income and Program Participation or the Medial Expenditure Panel Survey, should ask respondents how long they have been in their current health insurance spell at the start of the observation period. This would allow researchers to include both spells in progress and new spells in analyses providing a more accurate estimate of how long people remain uninsured. Primary Funding Source: VA ●Use of Preventive Services by the High-Income Uninsured Joseph Ross, M.D., Susan H. Busch, Ph.D. Presented By: Joseph Ross, M.D., Robert Wood Johnson Clinical Scholar, Department of Internal Medicine, Yale University, IE-61 SHM, P.O. Box 208088, New Haven, CT 06520; Tel: (203) 785-5920; Fax: (203) 785-3461; Email: joseph.s.ross@yale.edu Research Objective: Uninsured adults are less likely than insured adults to receive appropriate preventive care and disease treatment. The disparity between insured and uninsured is greater for more costly services compared to less costly services. It is unknown if the growing numbers of uninsured adults with higher incomes experience these same negative consequences or if their higher incomes allow them to purchase appropriate health care. Our objective was to compare nationally representative estimates of the proportion of uninsured and insured adults, categorized by annual household income, receiving clinically indicated cancer prevention services. Study Design: We used the 2002 Behavior Risk Factor Surveillance System, a random household telephone survey, to obtain data on health insurance, health status, use of preventive services, and socio-demographic information. Annual household incomes were categorized as less than $35,000, $35,000-$50,000, $50,000-$75,000, and greater than $75,000. Clinically indicated cancer prevention services examined include rates of Pap smears within three years for women aged 18 to 64 with an intact uterus, mammogram and clinical breast exam within two years for women aged 40 to 64, and fecal occult blood testing within two years and sigmoidoscopy or colonoscopy within five years for adults aged 50 to 64. Pap smear and fecal occult blood testing were considered less costly, while mammogram, sigmoidoscopy, and colonoscopy were considered more costly. We determined proportions of respondents who had received these preventive services, adjusting for socio-demographic characteristics and health status. Population Studied: A representative U.S. sample of 194,943 adults aged 18 to 64 years. Principal Findings: Fifteen percent of our sample lacked insurance. The uninsured were more likely to be young, male, non-white, poor, not employed for wages, not college educated, single, and report fair or poor health status. Uninsured adults were less likely than insured adults to receive fecal occult blood testing (20% vs. 36%, p<0.01) and sigmoidoscopy or colonoscopy (17% vs. 36%, p<0.01). Uninsured women were less likely than insured women to receive a Pap smear (78% vs. 91%, p<0.01) and a mammogram and clinical breast exam (45% vs. 75%, p<0.01). Within each income category, the uninsured remained less likely than the insured to receive any of the examined clinically indicated preventive services (p<0.01). When compared across annual household income categories, adjusted odds of the insured receiving any of the examined preventive services (compared to the uninsured) were not significantly different. Conclusions: Uninsured adults were less likely to receive clinically indicated cancer prevention services than insured adults. Compared to insured adults, uninsured adults from wealthier households were just as unlikely as uninsured adults from poorer households to receive clinically indicated cancer prevention services, regardless of the cost of the service. Implications for Policy, Delivery, or Practice: Policy-makers must assume that all uninsured adults experience negative consequences from lacking health insurance. Importantly, greater household income does not diminish this risk. This study raises questions about the perceived value of cancer prevention services and whether individuals who make discretionary purchases of health care, such as those enrolled in Health Savings Accounts, will obtain them appropriately. Primary Funding Source: No Funding ●Are Adults Benefiting from State Coverage Expansions? Stephen Zuckerman, Ph.D., Sharon Long, Ph.D., John Graves, BA Presented By: Stephen Zuckerman, Ph.D., Principal Research Associate, Health Policy Center, The Urban Institute, 2100 M Street NW, Washington, DC 20037; Tel: (202)261-5679; Fax: (202)223-1149; Email: szuckerm@ui.urban.org Research Objective: During the late 1990s, a small number of states implemented major expansions of public coverage for adults. Some states focused on the parents of children eligible for Medicaid and SCHIP, while others expanded coverage to broader groups of low-income adults. Moreover, income eligibility standards varied across states. The programs we examine are from California, Massachusetts, New Jersey and Wisconsin. Our goal is to study the effects of these programs on eligible adults with respect to public coverage, displacement of private coverage and health care access and use. Study Design: We use a difference-in-differences (DD) framework that compares changes among eligible adults in states with expansions to near-eligible adults within each state as well as to -adults with comparable characteristics in states without expansions. Such comparison groups provide the counterfactual for what would have happened if a state’s coverage rules had remained fixed. Because DD estimates may be sensitive to the choice of the comparison states, we draw on several different comparison groups to explore the robustness of our estimates. Population Studied: Low-income adults eligible for public insurance coverage in California, Massachusetts, New Jersey and Wisconsin under the program rules in place in each state in 2002. Data on low-income adults are drawn from the 1997, 1999 and 2002 rounds of the National Survey of America’s Families (NSAF), a nationally representative survey of children and non-aged adults. The survey oversamples families with incomes below 200 percent of the federal poverty level and families in 13 study states (including the four study states) and allows us to compare outcomes prior to and following the coverage expansions. Principal Findings: All four of these coverage expansions significantly increased public coverage for adults, with more coverage obtained in states with more generous expansion efforts. Private coverage eroded (especially in New Jersey), but the findings are not consistently significant across all four states. The only state in which adult uninsurance rates fell significantly was Wisconsin. We find little impact on overall access and use, although there is some evidence of greater effects in states with more generous expansions in coverage. The consistency of the impact estimates across multiple states using multiple comparison groups provides support for the reliability of the conclusions from this study. Conclusions: Of the four states studied, Wisconsin’s Badgercare program had the greatest impact on insurance coverage among low-income adults. Although private coverage offset some of the increase in public coverage, the results show that the overall gains were large enough so as to translate into improved access in Wisconsin. Implications for Policy, Delivery, or Practice: Public programs that expand eligibility for insurance coverage among low-income adults should expect to experience increased enrollment. The results from this study show that many adults are willing to enroll when given the opportunity. Policymakers need to recognize that crowd-out of private coverage should be expected, but the extent is likely to depend on both the features of the public program and the local health care market. Access to and use of care may improve if overall insurance coverage is increased. Primary Funding Source: RWJF Call for Papers Public & Private Health Insurance Chair: Jose Escarce, University of California, Los Angeles Monday, June 27 • 2:30 pm – 4:00 pm ●Changes in Health for the Uninsured after Reaching Ageeligibility for Medicare David W. Baker, M.D., MPH, Joseph M. Feinglass, Ph.D., Ramon Durazo-Arvizu, Ph.D., Whitney Perkins Witt, Ph.D., Joseph J. Sudano, Ph.D., Jason A. Thompson, BA Presented By: David W. Baker, M.D., MPH, Chief, Division of General Internal Medicine, Medicine, Northwestern University, 676 Saint Clair, Suite 200, Chicago, IL 60611; Tel: (312)696-0917; Fax: (312)695-0951; Email: dbaker1@nmff.org Research Objective: Uninsured adults in late middle age are more likely to have a health decline than the privately insured. It is unclear if this increased risk of adverse health outcomes persists for a period of time even after gaining insurance or whether gaining health insurance leads to improved health. This study examined these issues among a cohort of U.S. adults who gained Medicare coverage based upon ageeligibility. Study Design: Prospective cohort study. Population Studied: Population Studied: We used data from the Health and Retirement Study (HRS), a prospective study of U.S. adults aged 51-61 in 1992 with interviews every 2 years. We analyzed the `96, `98, `00, and `02 data files and identified the first interview at which participants reached age 65 and reported having Medicare (t0). Changes in health were defined for 1) the two-year period during which subjects transitioned to Medicare (t-2 to t0), and 2) the subsequent two year period (t0 to t2). Self-reported overall health was measured as excellent, very good, good, fair, or poor. Physical difficulties were measured with 4 dichotomous items assessing participants' difficulty walking or climbing stairs (“Mobility”) and 6 items measuring difficulty peforming instrumental activities of daily living (“Agility”). We used multivariate logistic and multinomial regression to analyze whether being uninsured prior to Medicare was associated with worse outcomes after adjusting for baseline health, sociodemographics, and health behaviors. Principal Findings: During the period in which subjects transitioned to Medicare coverage (t0 to t2), previously uninsured individuals (N=441) were more likely than those who previously had private insurance (N=2978) to have a major decline in overall health (adjusted relative risk [ARR] 1.46; 95% CI 1.03 – 2.04) and a new physical difficulty affecting Mobility (ARR 1.24; 95% CI 0.96 – 1.56) or Agility (ARR 1.33; 95% CI 1.12 – 1.54). In contrast, during the subsequent two years after the first HRS interview when participants had gained Medicare coverage (t0 to t+2), the previously uninsured no longer had worse outcomes compared to those who previously had private insurance: ARR major decline in overall health 1.14 (95% CI 0.76 – 1.68; p = 0.52), ARR major decline or death 1.21 (95% CI 0.85 – 1.70; p=0.52), ARRs new Mobility difficulty 0.93 (95% CI 0.67 – 1.26), and ARR new Agility difficulty 1.07 (95% CI 0.85 – 1.31). During both periods, adjusted rates of improvement in overall health and physical functioning were not associated with insurance coverage prior to Medicare. Conclusions: Gaining Medicare coverage does not lead to immediate health benefits for individuals who were uninsured prior to reaching age 65, but after two or more years of continuous coverage the increased risk for adverse health outcomes among the previously uninsured disappears. Implications for Policy, Delivery, or Practice: Lowering the age of eligibility for Medicare could yield significant health benefits that could partially offset the cost of expanding coverage. Earlier studies of the health effects of expanding insurance probably underestimate these health benefits because previously uninsured individuals continue to have higher risk for adverse health outcomes for a period of time after gaining coverage. Primary Funding Source: AHRQ ●The Impact of Increased Cost Sharing on Adults Enrolled in Medicaid: Early Results from a Prospective Cohort Study Bill Wright, Ph.D., Matthew J. Carlson, Ph.D., Tina Edlund, M.S., Jennifer DeVoe, M.D., Ph.D., Charles Gallia, Jeanene Smith, M.D. Presented By: Matthew Carlson, Ph.D., Assistant Professor, Sociology, Portland State University, PO Box 751, Portland, OR 97207; Tel: (503) 725-9554; Fax: (503) 725-3957; Email: carlsonm@pdx.edu Research Objective: The objectives of this ongoing study are to assess the short and long-term impact of policy changes, including increased premiums and co-pays, on individuals’ insurance coverage, access to and utilization of health care, financial solvency, and health status. Study Design: Baseline survey findings from a prospective cohort study are reported. Mail-return surveys were obtained from a stratified random sample of Oregon Health Plan Members (n=1783, RR=34%). In order to assess impact of increased cost sharing, two groups are compared: those who reported leaving OHP because they could not afford increased costs and those who left the OHP for other reasons including increased income, obtaining other insurance coverage, or failing to turn in application. Population Studied: Adults age 19 and older who were enrolled in the Oregon Health Plan(OHP)for at least thirty days prior to February 15, 2003, just prior to the initial wave of program changes for the OHP Standard population. Principal Findings: Baseline results from an ongoing cohort study suggest that even modest increases in cost sharing led to widespread disenrollment, especially among the poorest OHP members. More than 40% of OHP enrolled adults reported losing coverage in the 6 months following program changes. Of those leaving, 44% reported leaving because of increased costs. Compared to those who left for other reasons, those who left OHP because of increased cost sharing reported significantly higher unmet health care needs (71% vs. 47%), were more likely to report no usual source of care (28% vs. 21%), or the emergency department as the usual source of care (12% vs. 4%) and were more likely to have an emergency department visit (42% vs. 19%). Additionally, those who left because of cost had higher levels of medical debt, and were more likely to reporting being refused care due to unpaid medical bills (20% vs. 6%). Conclusions: These findings suggest that Medicaid cost saving strategies requiring higher out-of-pocket expenditures for low-income individuals are likely to have cascading effects including decreased enrollment among the very poor, increased unmet need, reduced primary care use, and increased emergency department utilization. Implications for Policy, Delivery, or Practice: Observed short-term consequences may have long term implications for increased health care costs and poor health outcomes among low income adults. Primary Funding Source: RWJF ●Charity Care, Risk Pooling, and the Decline in Private Health Insurance Michael Chernew, Ph.D., David Cutler, Ph.D., Patricia Keenan, Ph.D. Presented By: Michael Chernew, Ph.D., Professor, Health Management and Policy, University of Michigan, 109 Observatory, Ann Arbor, MI 48109-2029; Tel: (734)936-1193; Fax: (734)764-4338; Email: mchernew@umich.edu Research Objective: Textbook economic theory suggests that rising medical expenditures ought to increase the utility of health insurance coverage because insurance mitigates risk. As medical costs increase, the variability of spending will usually rise as well, and so the insurance product becomes more valuable. Yet a growing body of empirical work finds an inverse relationship between health insurance premiums and the propensity for individuals to have coverage. We examine possible explanations for these empirical regularities, focusing on risk pooling and availability of charity care. Study Design: We use linear probability models and IV methods to relate changes in insurance premiums at the MSA level to coverage. Explanations for declining coverage are assessed by estimating models separately by sub-populations and by including an interaction between premiums and charity care availability. The model uses extensive individual and market covariates to controls for a range of other explanations for declining coverage. Population Studied: We analyze two cohorts of non-elderly individuals from the Current Population Survey: one cohort from 1988-1990 and the second from 1997-1999. Principal Findings: Like existing literature, we find that rising insurance premiums are associated with declining coverage. The effects were strongest among the young, which is consistent with the risk pooling explanation for the relationship between premiums and coverage. The effect of rising premiums was greatest in markets with greater availability of charity care, which is consistent with the view that the availability of charity care, shown by other researchers to be associated with lower coverage rates, also increases the sensitivity of individuals to premiums. We estimate that if no charity care had been available the effect of rising premiums on coverage in the 1990s would have been half as large as was observed. Conclusions: The decline in coverage associated with rising coverage likely reflects several phenomenon. Our results suggest one important effect is that as health care costs rise, the implicit subsidy from low risk to high risk consumers rises, which appears to lead some low risk consumers to opt out of coverage. Moreover, charity care provides and alternative to coverage for some individuals and as the costs of coverage rise the relative attractiveness of charity care may also increase, causing further deterioration in the coverage rate. Implications for Policy, Delivery, or Practice: The new era of rising medical spending we have recently entered could have a major impact on private insurance coverage. Moreover, the decline in coverage caused by rising premiums will place a greater burden on charity care providers. Though important in a time of declining coverage, bolstering the strained charity care system may further exacerbate the decline in coverage, posing a policy dilemma in responding to increases in the uninsured population. Primary Funding Source: RWJF, NIA, Sloan Foundation ●A Statewide Study of Frequent Users of Emergency Departments in Massachusetts Kathleen Fuda, Ph.D., Rachel Immekus, MS, MPH Presented By: Kathleen Fuda, Ph.D., Data Analysis Manager, Health Policy Institute, Boston University, 53 Bay State Road, Boston, MA 02215; Tel: (617)353-8900; Fax: (617)353-6393; Email: fuda@bu.edu Research Objective: Frequent use of the emergency department (ED) is often attributed to lack of insurance or otherwise reduced access to primary care. Most studies of frequent users of ED visits in the United States rely on data from only one or at most several hospitals, or include only ED visits resulting in discharges to the patient's home. We combined statewide data for Massachusetts inpatient discharges, outpatient observation stays, and outpatient ED visits, in order to include all visits to all Massachusetts EDs regardless of disposition. We analyzed these data to characterize ED frequent users and to look for evidence that frequent use of the ED is attributable to lack of access to primary care. Study Design: Statewide encounter-level data for outpatient ED visits was combined with data on ED visits that resulted in outpatient observation stays or inpatient admissions to produce a database of all ED visits to Massachusetts hospitals in FY2003 (2.78 million records). Using encrypted social security numbers (SSNs), or a constructed patient identifier when SSNs were missing, records were aggregated to the patient level. Individuals with five or more ED visits were considered frequent users. Frequent and non-frequent users were compared on demographic, payer, and clinical characteristics. This study was supplemented with analysis of additional Massachusetts-specific data from the National Survey of American Families (NSAF), which included data on primary care use by ED visitors. Population Studied: All patients who made at least one visit to an emergency department in Massachusetts in FY 2003, with a special emphasis on Massachusetts residents. Principal Findings: Only one percent of Massachusetts residents made five or more ED visits in FY 2003, but they comprised 3.7% of all ED visitors and made 17.1% of all ED visits that year. Over half of frequent users visited multiple EDs. Frequent users appeared to be sicker than non-frequent users, as they were more likely to be elderly, to be admitted to an inpatient stay, to have a disposition of death at the last ED visit, and to receive relatively intense levels of service in the ED. A mental health- or substance abuse-related primary diagnosis was found for 11.3% of visits by frequent users compared to 3.7% of visits by non-frequent users. However, frequent users are not a homogeneous population, and frequent users among the enrollees of different payer groups have distinctive diagnostic profiles. A larger percentage of patients insured by either Medicaid or Medicare are frequent users than are the uninsured or the privately insured. It was also found that only 28% of frequent users in FY02 were also frequent users in FY03. NSAF data indicated that frequent ED users were heavier users of primary care visits than were nonfrequent users. Conclusions: The large proportion of frequent users that use multiple EDs indicates the importance of using data that draws from all EDs in a large area. Although frequent users are a very small subset of the population as a whole, they account for a substantial fraction of ED visits. Frequent users are not a homogenous population in terms of clinical characteristics, but mental health and substance abuse diagnoses are much more common among frequent users as a whole. The publicly-insured are more likely to be frequent users than are the uninsured or those with private insurance. Overall, frequent users in Massachusetts do not appear to be substituting ED visits for primary care visits. Implications for Policy, Delivery, or Practice: Interventions to reduce frequent ED visits, to the extent they are needed at all, should be targeted to the subset of frequent users that is most likely to continue to use the ED at high rates. The heterogeneity of frequent users makes it important to carefully analyze the intended targets to assess what is driving their high utilization rates, and to tailor programs accordingly. Efforts to merely link frequent users to primary care providers are likely to be unsuccessful. Primary Funding Source: MA Division of Health Care Finance and Policy ●The Effects of Welfare Reform on Health Insurance of Immigrants and their Children – Differences by Immigration Status Ithai Zvi Lurie, MA Presented By: Ithai Zvi Lurie, MA, Research Associate, Institute for Policy Research, Northwestern University, 2040 Sheridan Road, Evanston, IL 60208-4100; Tel: (847) 733-7425; Email: i-lurie@northwestern.edu Research Objective: Investigate the impact of the welfare reform legislation on public and private health insurance coverage for non-citizens by examining two related issues. First, the study explores whether health insurance coverage of non-citizens fell as a consequence of the 1996 Personal Responsibility and Work Opportunity Reconciliation Act (PROWRA). The reduction in public insurance eligibility should have a direct effect on non-citizen public coverage. However, private insurance may have increased to partly or completely offset the drop in public coverage (reverse crowdout effect). Second, the study identifies whether the reduction in eligibility for non-permanent residents and permanent residents had an unexpected effect on others who did not lose eligibility (chilling effect). Study Design: The 1996 Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) represented the most profound change in the welfare system since the Great Society programs were launched. As part of this sweeping change, eligibility requirements for a number of government programs were fundamentally altered. The changes made to eligibility requirements had particularly important implication for immigrants. Specifically, immigrants without permanent resident status and immigrants with permanent resident status who arrived after PRWORA was enacted lost eligibility for a number of programs, including Medicaid. Using data from the Survey of Income and Program Participation (SIPP) from 1996 and 2001 in order to explicitly identify who is a permanent resident and who is not. This distinction allows this study to better identify individuals who should and should not have been affected by PRWORA. Population Studied: This study uses data from the Survey of Income and Program Participation (SIPP) from 1996 and 2001, a national representative sample. Overall there are about 141 thousand non elderly individuals. Foreign born adults represent about 13% of the adult sample and their children represent and about 15% of the children sample. Principal Findings: Non-permanent residents experienced a large drop in public coverage, while permanent residents did not. The drop in public coverage comes with a significant increase in private coverage. This reverse crowd-out effect eliminates most of the decline in coverage. In addition there is a significant chilling effect for citizen children of nonpermanent residents’ parents. These children lost almost 12 percentage points of public coverage relative to citizen children of natives and gained very little in private coverage. Conversely, there is no evidence of a chilling effect for citizen children of permanent resident parents. As well there is some evidence of a chilling effect for foreign-born adults that lost federal eligibility, but who could have obtained Medicaid from their respective states. Conclusions: This study shows the importance of identifying precisely the individuals targeted by policy changes. In the context of PROWRA, evaluating the effects of the policy for non-citizens as a group without distinguishing between permanent residents and non-permanent residents can result in misleading conclusions. Implications for Policy, Delivery, or Practice: The Medicaid outreach programs that states enacted after SCHIP implementation should put more emphasis on attracting citizen children of non-permanent resident parents. Primary Funding Source: No Funding Call for Papers The Health Care Safety Net Chair: Michael Chernew, University of Michigan Tuesday, June 28 • 10:30 am – 12:00 pm ●Which Safety Net Services Make a Difference in Health Care Use? Janet Bronstein, Ph.D., Shailender Swaminathan, Ph.D. Presented By: Janet Bronstein, Ph.D., Professor, Health Care Organization and Policy, University of Alabama at Birmingham, 1350 3rd Avenue South, RPHB 330, Birmingham, AL 35209-0022; Tel: (205) 975-8962; Fax: (205) 934-3347; Email: jbronste@uab.edu Research Objective: Due in part to empirical challenges, studies have not consistently identified a positive relationship between safety net services in a community and access to care. This study assesses (1) whether different types of safety net services have different effects on access for adults and children, (2) whether these effects differ across insurance groups, and (3) whether these effects vary across different aspects of health care access. Study Design: Using survey data we control for a range of predisposing, need and enabling characteristics, along with some community context measures, to assess whether community level measures of physician acceptance of Medicaid patients, provision of charity care, federally supported community health centers (CHC’s) and public/teaching hospital emergency departments and clinics (public EDs) have an impact on usual source of care, unmet needs and actual use of physician and emergency department services. Impact is measured as the differential association between access to care and these safety net services for publicly insured and uninsured individuals (separately) compared to their impact on the privately or Medicare insured population. Population Studied: Our study population of 44,439 adults and 9,085 children were respondents to the 2000-2001 round of the household component of the Community Tracking Study (CTS). Data from the CTS physician survey, the 2000 census, AHA survey and Area Resource File are used for safety net and community context measures for the 60 communities represented in the CTS sample. All safety net measures are adjusted for the size of the relevant populations and then ranked across the 60 communities; quintile of relative availability of the service is used as the indicator in the regression analyses. Standard errors in regressions are adjusted for the stratified sample and for the clustering of respondents into community sites. Principal Findings: Uninsured, but not publicly insured adults and children have poorer access to care by all measures than the privately insured, and other individual characteristics are associated with access and use as would be expected. Controlling for these and for community level factors, we find that uninsured adults have more physician visits where there are more CHC’s and more emergency visits where there are more public EDs. Both uninsured and publicly insured adults report fewer unmet needs in communities with more public EDs. Publicly insured children are more likely to have a usual source of care where more doctors participate in Medicaid but where fewer provide charity care. Publicly insured adults are more likely to have a usual source where more doctors provide charity care and there are fewer community health centers and Medicaid participating physicians. No safety net services improve the likelihood of a usual source of care for uninsured adults or children. Conclusions: This study suggests that different components of the health care safety net tend to serve different components of the vulnerable population, both by age and by insurance status, and to meet different aspects of the health care needs of these different groups. Implications for Policy, Delivery, or Practice: Components of the safety net are not interchangeable; communities need a diverse mix of services to meet the access needs of vulnerable populations. Primary Funding Source: UK Center for Poverty Research ●Awareness of Safety Net Providers by Uninsured Persons Peter Cunningham, Ph.D., Jack Hadley, Ph.D., Genevieve Kenney, Ph.D., Amy Davidoff, Ph.D. Presented By: Peter Cunningham, Ph.D., Senior Health Researcher, Center for Studying Health System Change, 600 Maryland Avenue, SW, Suite 550, Washington, DC 20024; Tel: (202)484-4242; Fax: (202)484-9258; Email: pcunningham@hschange.org Research Objective: It is generally acknowledged that health care providers who offer services for free or at reduced cost play an important role in providing at least a minimum level of access to care to uninsured persons. However, the effectiveness of these safety net providers may be limited to the extent that many uninsured aren’t aware of these providers. This research examines the extent to which uninsured persons are or are not aware of safety net providers in their communities; examines a variety of health system, policy, and sociodemographic characteristics associated with safety net awareness; and explores the implications of awareness or lack of awareness for access to care. Study Design: Data are based on the 2003 Community Tracking Study household survey. The survey included questions asked of uninsured persons and their family members as to whether or not they paid full or reduced cost at their usual source of medical care, or if they were aware of providers in their community who offered affordable care to uninsured persons. Uninsured respondents who were aware of safety net providers were also asked the type of place (e.g. physician’s office, hospital-based facility, clinic). A multinomial logistic regression analysis is conducted to examine the effects of various health system, policy, and sociodemographic characteristics on awareness of the different types of safety net providers. We focus especially on the effects of local safety net capacity on awareness, such as safety net hospitals, community health centers, and physician charity care, as well as the impact of Community Access Program (CAP) grants, all obtained through secondary data and linked to the CTS household survey. We also examine whether awareness of safety net providers is associated with fewer unmet medical needs. Population Studied: Nonelderly uninsured persons. Principal Findings: Less than half of the uninsured (47.5 percent) reported that they used or were aware of a safety net provider in their community that offered subsidized services to the uninsured. Among those uninsured who were aware of a safety net provider, most identified clinics and health centers (44.6 percent) and physician’s offices (26.1 percent) as the safety net provider they were aware of. Fewer uninsured identified hospital outpatient clinics (12.9 percent) and emergency departments (8.3 percent) as safety net providers. The results of the regression analysis show that higher capacity of safety net hospitals, CHCs, and physician charity care increases awareness of safety net providers. The presence of CAP grants in the community also increases safety net awareness. Nevertheless, many uninsured remain unaware of safety net providers despite living in close proximity to a safety net provider. Awareness of safety net providers was also associated with fewer unmet medical needs. Conclusions: Many uninsured are not aware of safety net providers, including many who live near a major safety net provider such as a public hospital or community health center. Greater awareness of safety net providers is associated with improved access, although further research is needed to more firmly establish the causal link between safety net awareness and access. Implications for Policy, Delivery, or Practice: Increasing awareness of safety net providers may be as important in improving access as increasing capacity in the system. Just as outreach efforts have been critical for enrolling eligible children in public insurance programs, outreach activities may be important in increasing awareness of safety net providers by uninsured persons. Primary Funding Source: RWJF ●Copayments and Emergency Departments: Effective Incentives or Effects Unintended? Mary Price, MA, Richard Brand, Ph.D., Joseph Newhouse, Ph.D., Joseph Selby, M.D., MPH, John Hsu, M.D., MBA, MSCE Presented By: Mary Price, MA, Senior Consulting Data Analyst, Division of Research, Kaiser Permanente Northern California, 442 Brimhall Street, St. Paul, MN 55105; Tel: (651) 690-1981; Fax: (651) 690-1981; Email: Maggie.Price@kp.org Research Objective: Cost-sharing theoretically encourages patients to make more efficient choices about where to seek care. We investigated the effect of emergency department (ED) copayments on ED visits for non-emergency and emergency reasons. Study Design: We examined the association between the size of ED copayments and the rate of non-emergent and emergent ED visits (1999–2001) in a large prepaid integrated delivery system. We classified ED copayments into five categories: no copayment, $1-5, $10-15, $20-35 and $50-100 per visit. We used the New York University ED Diagnosis Algorithm to classify visits as non-emergent or emergent. This algorithm assigns probabilities to one non-emergent and three emergent categories for each visit’s ICD-9 diagnosis; the sum of the four probabilities equals 100%. We classified visits as non-emergencies when the probability was at least 75% that the visit diagnosis was in the non-emergent category, and as emergencies when the sum of the probabilities for the three emergent categories was at least 75%. Visits with both emergency and non-emergency diagnoses were classified as emergencies; visits also could be indeterminate if the probabilities were below 75% for either type of visit. We compared relative rates of non-emergent and emergent ED visits by copayment level using a negative binomial mixedeffects model applied to annual utilization counts, adjusting for year, and for age, gender, SES, comorbidity, prior utilization (ED, office, hospital), having a regular physician, drug cost-sharing level, and medical center using a propensity score. We repeated the analysis using a Poisson mixed-effects model applied to monthly utilization counts. Population Studied: All subjects were commercially insured members of the IDS in January 1999. Of these 2,257,445 subjects, 23.4% paid no copayment, 34.8% paid $1-5, 22.8% paid $10-15, 13.6% paid $20-35, and 5.4% paid $50-100 per ED visit in 1999. In January 2000, 52% of subjects had their copayment increase. Principal Findings: There were 1,041,719 ED visits (19992001): 51,692 and 299,005 were for non-emergent and emergent diagnoses respectively. Compared to no copayment, the relative rate of non-emergent visits decreased 3% (RR=0.97, 95% CI: 0.94–0.99), 6% (RR=0.94, 95% CI: 0.91– 0.98), 14% (RR=0.86, 95% CI: 0.83–0.88), and 28% (RR=0.72, 95% CI: 0.69–0.76) with the $1-5, $10-15, $20-35, and $50-100 copayments respectively. For emergent ED visits, compared with no copayment, there was no effect with the $1-5 and $1015 copayments, and visit rates decreased 6% (RR=0.94, 95% CI: 0.93–0.95) and 13% (RR=0.87, 95% CI: 0.85–0.89) with the $20-35 and $50-100 copayments respectively. The findings were robust across analytic approaches. Conclusions: Patients with higher copayments were less likely to seek care from the ED for non-emergency reasons; there was a smaller copayment effect on visits for emergency reasons. Implications for Policy, Delivery, or Practice: These preliminary data suggest that copayments do influence patient decisions about seeking care from the ED. In particular, these incentives appear to encourage patients to reduce use of the ED for non-emergencies. Further research is needed to assess the clinical and economic effects, especially with large copayments. Primary Funding Source: AHRQ ●Healthcare Markets, the Safety Net and Access to Care Among the Uninsured Carole Roan Gresenz, Ph.D., Jeannette Rogowski, Ph.D., Jose J. Escarce, Ph.D., M.D. Presented By: Carole Roan Gresenz, Ph.D., Senior Economist, RAND Corporation, 1200 South Hayes Street, Arlington, VA 22202; Tel: (703)413-1100 x5419; Fax: (703)414-4726; Email: gresenz@rand.org Research Objective: We use nationally representative data including observations on more than 12,000 uninsured adults to provide new insight into the relationship between access to care and not only the structure and capacity of the safety net but also the healthcare market more generally. Study Design: We use Medical Expenditure Panel Survey (MEPS) data linked with data from multiple secondary sources. To describe the healthcare safety net and market structure in each individual’s location, we derive variables from the American Hospital Association Annual Survey of Hospitals, Area Resource File, the InterStudy Regional Market Analysis database, the Bureau of Primary Healthcare (BPHC) Physicians Uniform Data System, the Current Population Survey, Census of Governments, and the Census Bureau’s Annual Survey of State and Local Government Finances. A major innovation of this research is the calculation of detailed measures of safety net availability: We calculate the distance between each uninsured individual and the nearest BPHC provider, public hospital and emergency room. We analyze outpatient care utilization (office-based physician and nonphysician visits, emergency department visits) and medical expenditures. We simulate utilization for a number of health care market and safety net values using standardized predictions. Population Studied: Uninsured adults in the U.S. Principal Findings: Distances between the rural uninsured and safety net providers such as hospital emergency rooms, public hospitals, migrant health centers, public housing primary care programs, and community health centers are significantly associated with utilization of a variety of healthcare services. In urban areas, we find that the percentage of individuals in the area who are uninsured and the pervasiveness and competitiveness of managed care have a significant relationship with healthcare utilization. Conclusions: Facilitating transport to safety net providers and increasing the number of such providers are likely to improve access to care among the rural uninsured. Our findings for urban areas suggest that particular attention be paid to the uninsured living in areas where many of those insured are covered by managed care, and especially so where little competition among managed care organizations exists. Ironically, the “backlash” against managed care may result in improved access to care for some uninsured, although the salutary effects would be offset to the extent that the backlash also results in increasing healthcare costs, greater numbers of uninsured, and more competition for healthcare resources. Implications for Policy, Delivery, or Practice: Absent the universal provision of health insurance, policy approaches to alleviating the barriers to access facing the uninsured include incremental efforts to increase the affordability and availability of public or private health insurance as well as measures to increase the accessibility of healthcare for the remaining uninsured. Our findings shed light on areas of focus for the latter class of measures. Primary Funding Source: AHRQ ●The Shortage of On-Call Specialist Physician Coverage in U.S. Hospitals Benjamin Vanlandingham, M.D., MPH, Neil R. Powe, M.D., MPH, MBA, Barbara Marone, Marie Diener-West, Ph.D., Haya R. Rubin, M.D., Ph.D. Presented By: Benjamin Vanlandingham, M.D., MPH, Fellow, Robert Wood Johnson Clinical Scholars Program, Johns Hopkins University, 600 North Wolfe Street, Carnegie 291, Baltimore, MD 21287; Tel: 410-614-6037; Email: bvanlandingham@jhmi.edu Research Objective: Anecdotal reports suggest that hospitals are losing the support of physician specialists to provide oncall coverage as a backup to physicians providing care to patients in emergency departments (ED) and in the hospital. The objective of this study was to measure the extent to which hospitals lack on-call coverage by specialists and to explore the factors that might be associated with on-call coverage shortages such as location, the community-level supply of specialists, the cost of professional liability insurance and the penetration of ambulatory-surgery centers (ASCs) and specialty hospitals. Study Design: This was a national cross-sectional study in which we administered a standardized questionnaire to hospital ED directors asking them whether they have a problem with inadequate on-call coverage. For each of 11 specialties, respondents also reported 1) the percentage of days their hospitals had formal on-call coverage during a preceding three-month period and 2) the perceived degree of importance in having on-call coverage with respect to the needs of the patient population at their hospital. Data were also collected on payer mix for ED patients, community-level estimates of specialist supply, the cost of professional liability insurance and the penetration of ASCs and specialty hospitals. Population Studied: Medical directors of emergency departments at acute-care general hospitals in the U.S. (N=4428). Specialty and federally-owned hospitals were excluded. Principal Findings: Two-thirds [68% (95% CI, 66-70%)] of respondents reported that on-call coverage is inadequate to meet the needs of their patients. On-call coverage problems were reported in more often in urban [73% (95% CI 70-75%)] than rural [60% (95% CI, 57-64%)] hospitals and were similar in geographic regions of the country. The greatest shortage of specialists was in hand surgery: among hospitals where hand surgery coverage is perceived to be “very or extremely important for overall patient outcomes,” 69% (95% CI, 6572%) of hospitals has less than full-time coverage. Hospitals also have less than full-time coverage for plastic surgery (52%), neurosurgery (49%), ENT (44%), psychiatry (42%). On-call coverage shortages were related to the proportion of uninsured patients in the hospital ED, but not to the supply of specialists. Conclusions: A large proportion of hospitals have unmet need for on-call specialist coverage based on ED directors’ perceptions of coverage requirements. On-call coverage shortages are found for several specialties and in both urban and rural hospitals and in all regions of the country. The problem is related to the lack of health insurance and does not appear related to physician supply Implications for Policy, Delivery, or Practice: The shortage of on-call coverage is an emerging trend that threatens the integrity of the health care safety net, placing patients at potential risk for injury. This is one of the first studies at the national level demonstrating the scope of this problem. Until now, responsibility for adequate on-call staffing has rested with individual hospital administrators, but their efforts appear to be failing. Further study about the effect on patient outcomes warranted. Primary Funding Source: RWJF Related Posters Poster Session B Monday, June 27 • 6:15 pm – 7:30 pm ●Price Elasticity of Demand for Non-Group Health Insurance David Auerbach, Ph.D., Sabina Ohri, Ph.D. candidate Presented By: David Auerbach, Ph.D., Associate Analyst, Health and Human Resources, Congressional Budget Office, 2500 Davis Avenue, Alexandria, VA 22302; Tel: (202) 2262669; Fax: (202) 225-3149; Email: david.auerbach@cbo.gov Research Objective: To estimate the price elasticity of demand for nongroup health insurance, both as a pure elasticity and as the more useful "percent of the uninsured would take up coverage in response to a given premium subsidy". Two secondary objectives are to investigate whether this elasticity varies among poor or sick subpopulations and whether state regulations concerning nongroup underwriting impact insurance takeup in expected ways. Study Design: We extracted a subsample of nonelderly without access to ESI from the SIPP (data from 2002). We then imputed offered nongroup premiums, focusing on as much exogenous variation as possible (tax deductibility among the self-employed and state-level premium compression/community rating regulations prevailing in 2001). We use the latter variation in an instrumental variables probit model that yields a similar response to the premium as does the simpler reduced form probit equation. Population Studied: Primarily, 1,840 single, working, nonelderly respondents in the 17th month of the 2001 panel of the Survey of Income and Program Participation who do not have offers of employer-sponsored insurance. Principal Findings: Elasticity of demand is roughly -.56, indicating that a 10% reduction in premiums would cause a 5.6% increase in coverage rates. Since initial coverage rates in our sample were low, however, this premium reduction would result in just under a 1% reduction in the uninsured. State premium compression and community rating regulations of nongroup insurance do appear to improve coverage among the unhealthy at the slight expense of coverage among the healthy. Conclusions: Our effect of premiums on coverage rates is consistent with a recent study analyzing working families in California that used geographic variation in offered premiums. Analyses (including our own) that use primarily endogenous sources of premium variation (e.g., age, sex, health status) appear to obtain larger premium responses. Implications for Policy, Delivery, or Practice: Our findings support the contentions of some modelers that subsidies or tax credits for nongroup insurance would probably not generate a large response among the uninsured. If our results generalized to the entire population of uninsured without offers of employer-sponsored insurance (~30 million at a point in time) then a 40% premium subsidy would result in roughly 1 million uninsured taking up coverage. Primary Funding Source: No Funding Source ●Access and Travel to Cancer Services Among Rural Colorectal Cancer Patients Laura-Mae Baldwin, M.D., MPH, Eric Larson, Ph.D., Yong Cai, Ph.D., George Wright, Ph.D., Sharon Dobie, M.D., Gary Hart, Ph.D. Presented By: Laura-Mae Baldwin, M.D., MPH, Professor, Family Medicine, University of Washington School of Medicine, Box 354982, Seattle, WA 98195; Tel: (206) 685-4799; Fax: (206) 616-4768; Email: lmb@u.washington.edu Research Objective: To examine the degree to which rural, Medicare-insured colorectal cancer patients have cancer care services available locally, and to determine rural patients' travel patterns to these services. Study Design: Cross-sectional, descriptive study using Surveillance, Epidemiology, and End Results (SEER) program data for colorectal cancer patients diagnosed between 1992 and 1996 linked with their 1991-1998 Medicare claims. Travel distances between patients' ZIP codes and the ZIP codes of their closest and actual cancer care physicians were calculated using software developed at the Dartmouth Center for the Evaluative and Clinical Sciences. Population Studied: 27,143 stage I-III colorectal cancer patients age 66 and older reported to one of 11 SEER cancer registries between 1992 and 1996. Rural-Urban Commuting Area (RUCA) codes were used to classify patients' residence locations as urban (n=22,366), large rural (n=1696), small rural (n=1575), and remote rural (n=1506). Principal Findings: Over 90% of patients in all rural and urban areas live within 30 miles of a surgeon who performed colorectal resections, and close to or over 90% of patients in all areas live within 30 miles of a physician who administered chemotherapy. Radiation therapy is much less available to rural patients, with only 21% of remote rural, 31% of small rural, and 58% of large rural area residents within 30 miles of a physician who administered radiation therapy. Despite the availability of surgeons and chemotherapy administrators in all geographic areas, substantial proportions of rural colorectal cancer patients travel to urban or less rural areas for these services. For example, 16.9% of colorectal cancer patients living in large rural areas, 37.6% living in small rural areas, and 45.4% living in remote rural areas received their cancer surgery in an urban area. These patients receiving care in urban areas traveled a median of 70, 50, and 48 miles, respectively, to their surgical care. Conclusions: Access to surgical and chemotherapy services for colorectal cancer patients is roughly equivalent for urban and rural residents, regardless of their degree of rurality. Despite this relatively uniform access, large proportions of cancer patients bypass services in their local areas, traveling great distances to surgical and chemotherapy care. Radiation therapy, unlike other cancer care services, is much less available in rural areas. Implications for Policy, Delivery, or Practice: The distance to radiation therapy, often requiring daily treatment, likely represents a substantial barrier to its use by many rural elderly, for whom transportation may be not be readily available. Supporting transportation services for these elderly may be needed to ensure equitable access to this important cancer service. At the same time, this study demonstrates that many rural elderly are bypassing local surgical and chemotherapy services, raising questions about whether the local resources are saturated, or whether patients are choosing more distant services. Further research exploring the reasons for these travel patterns will elucidate whether rural areas need to recruit additional cancer care specialists, or ensure that patients are aware of and satisfied with the services available to them locally. Primary Funding Source: NCI ●The Impact of Surgical Infection on Hospital Resource Utilization and Patient Access to Care in Orthopaedic Surgery Kevin Bozic, M.D., MBA, Michael D. Ries, M.D. Presented By: Kevin Bozic, M.D., MBA, Assistant Professor in Residence, Institute for Health Policy Studies, University of California, San Francisco, 500 Parnassus, MU 320W, San Francisco, CA 94143-0728; Tel: (415) 476-3900; Fax: (415) 4761304; Email: bozick@orthosurg.ucsf.edu Research Objective: Acute and chronic deep surgical infections can have significant clinical and economic implications for patients, surgeons, hospitals, and payors. Surgical infections often results in the need for multiple reoperations, prolonged use of intravenous and oral antibiotics, extended inpatient and outpatient rehabilitation, and frequent follow-up visits. Futhermore, the lack of incremental reimbursement associated with treating surgical infections has resulted in strong financial disincentives for hospitals and surgeons to provide care for these patients. This purpose of this study was to assess the impact of surgical infections associated with a common orthopaedic procedure, total hip replacement, on hospital and surgeon resource utilization and patient access to care. Study Design: Clinical, demographic, and economic data were obtained for a cohort of patients who underwent revision total hip replacement (THR) for a diagnosis of infection (Group 1) during a 2 year period at a tertiary care referral hospital. Similar data were collected during the same time period for a cohort of patients who underwent revision THR for aseptic causes (Group 2) and a cohort of patients who underwent primary THR (Group 3). Quantitative variables were compared between patient groups using the Kruskal- Wallis nonparametric rank test. Categorical variables were compared using the Fisher’s exact test. Population Studied: Patients who underwent primary or revision total hip replacement at a tertiary care referral hospital during a two year period. Principal Findings: Septic revision THR procedures were associated with longer operative time, more blood loss, and a higher number of complications than aseptic revisions or primary THR. Septic revisions were also associated with a higher total number of hospitalizations, total hospital days, total number of operative episodes, total hospital costs, total outpatient encounters, and total outpatient charges during the 12 month period following the index procedure. During the time period of the study, referrals for patients with infected total hip replacements increased from 3% to 24% of all orthopaedic referrals to our tertiary care hospital. Conclusions: The treatment of patients with infected total hip replacement is associated with significantly greater hospital and physician resource utilization than aseptic revision THR or primary THR. The lack of incremental reimbursement associated with these procedures has resulted in strong financial disincentives for physicians and hospitals to provide treatment for patients with infected total hip replacements. Implications for Policy, Delivery, or Practice: Surgical infections pose significant clinical and economic challenges for surgeons, patients, and hospitals. The results of our study suggest that fewer hospitals and surgeons in our region are willing to provide care for patients with infected total hip replacements, which has resulted in a dramatic increase in the number of patients with infected THRs who are referred to our institution for treatment. Given the additional resources required to provide care for these challenging patients, these findings have significant implications with respect to hospital financial viability, patient access to care, and ultimately patient outcomes. Primary Funding Source: Orthopaedic Research & Education Foundation Health Services Grant ●The Clustering of Safety Net Services and their Resource Supports Across Communities Janet Bronstein, Ph.D. Presented By: Janet Bronstein, Ph.D., Professor, Health Care Organization and Policy, University of Alabama at Birmingham, 1350 3rd Avenue South, RPHB 330, Birmingham, AL 35209-0022; Tel: (205) 975-8962; Fax: (205) 934-3347; Email: jbronste@uab.edu Research Objective: Case studies and reviews indicate that a heterogeneous mix of health care services serve as the safety net for publicly insured and uninsured vulnerable populations. However, it is not known whether communities themselves tend to support a diverse mix of safety net service types or whether communities with different types of resources tend to “specialize” in different service types. This study has two objectives: (1) To describe the typical clustering pattern of different types of safety net services across communities, and (2) to assess whether different safety net services are linked to different types of resources. Study Design: Data are compiled on resources and safety net services available in 60 U.S. communities. Factor analysis is used to describe the clustering pattern for seven safety net services in these communities. A comparison of the means of available resources in communities with above or below the median capacity of each safety net service indicates which services are most commonly linked to which types of resources. Population Studied: Data from the physician survey component of the Center for Studying Health System Change Community Tracking Study (CTS) are used to estimate measures of physician acceptance of Medicaid and physician charity care hours provided in the 60 communities included in the CTS. AHA, ARF and census data are used to measure the number of Community Health Centers (CHC’s), public and teaching and children’s hospital beds and emergency departments and clinics. All measures are adjusted for the size of the relevant vulnerable population. Principal Findings: Factor analysis indicates three overlapping clusters of safety net services across communities. Physician participation in Medicaid and provision of charity care tend to co-occur with children’s hospital beds and outpatient facilities; these communities tend not to have federally subsidized CHC’s. Public/teaching hospital beds occur in the absence of children’s hospital resources, but in the presence of physician participation in Medicaid. Public/teaching hospital outpatient facilities cooccur with beds in these hospitals but in the absence of strong participation in Medicaid by physicians. Physician based safety net services are much more common in higher income communities where more of the population is privately insured and there are more physicians per capita. Public/teaching beds are more common where more of the otherwise uninsured population is covered by public insurance and where there are a larger proportion of Blacks in the population. Public/teaching outpatient facilities are more common in communities with fewer Hispanics and with less income inequality across the population. Conclusions: Communities do not typically maintain a broad array of safety net services, but tend to have either physician based, public/teaching hospital based or CHC based services. Different types of resources support different sets of services. CHC’s are the least vulnerable to changes in local resources but tend to be isolated from other types of safety net services. Implications for Policy, Delivery, or Practice: Failures of the safety net are unlikely to occur across the board, but more likely to be isolated in communities dependent on services supported by highly stressed resources. Safety net monitoring systems must be localized. Primary Funding Source: UK Center for Poverty Research ●Health Uninsurance in North Dakota: Demographic, Economic and Geographic Predictors Mike Cogan, Ph.D., Kyle Muus, Ph.D., Alana Knudson, Ph.D., John Baird, M.D. Presented By: Mike Cogan, Ph.D., Assistant Professor, Center for Rural Health, University of North Dakota, PO Box 9037, Grand Forks, ND 58202; Tel: (701)777-3848; Fax: (701)7776779; Email: mcogan@medicine.nodak.edu Research Objective: Specific research questions included: *What percent of North Dakotans are uninsured? *Are there significant differences in uninsured rates by geographic residential location (small rural, large rural, urban)? *Are there significant differences in uninsured rates by key demographic factors, including gender, age group, income level and race/ethnicity? *Which factors are significant predictors of health insurance status? Study Design: In spring 2004, data were collected through telephone interviews of a random selection of North Dakota households. A total of 3,199 interviews were completed from 4,641 households contacted, yielding a response rate of 68.9%. Of particular interest was to examine health insurance coverage by demographics (e.g., age, gender, race, income), employment-related factors, and rural-urban residential location. Population Studied: North Dakota household residents. Principal Findings: Findings indicated an overall state uninsured rate of 8.2%, or approximately 52,000 persons. This figure compares favorably to the national estimated rate of 15%. By geographic region, it was found that residents of small rural areas were most likely to be uninsured (9.1%), compared to large rural (7.7%) and urban areas (7.4%). Further breakdown indicated that 44.5% of the uninsured resided in small rural areas (population less than 5,000). This number is more than twice the national percentage as reported in the landmark study Health Insurance Coverage in Rural America (2003). In addition, Native Americans (31.7%) and those reporting more than one race (11.5%) were significantly more likely to be uninsured than Whites (6.9%). Binary logistic regression was conducted to determine which of these independent variables were significant predictors of health insurance status. Results indicated: *Males were 1.2 times more likely to be uninsured than females. *Married persons were 2.2 times more likely to have health insurance than unmarried North Dakotans. *Those who had attended college were more than three times more likely to have health insurance than those with a high school diploma or less. *Employed persons were 4.5 times more likely to have health insurance than unemployed persons. *Those who did not serve in the military were 1.8 times more likely to have health insurance than those who did serve in the military. *Persons working in firms with 11-50 employees were 1.4 more likely to be insured than persons working in oneperson firms. *Persons residing in households with an income above 200% of the federal poverty level were 6.8 times more likely to have health insurance than persons residing in a household with an income below 100% of the federal poverty level. Conclusions: Results assisted health department personnel and policy makers in devising and reviewing policy options for increasing health insurance access among at-risk subgroups of North Dakotans. Similar to other states, North Dakota’s budgetary woes restricted the nature, breadth and scope of these options. The presentation will include discussion on a series of 'budget-neutral' options for increasing access to health insurance coverage among rural/underserved populations. Implications for Policy, Delivery, or Practice: Extending health coverage for low-income residents and children in North Dakota is an important policy and health objective for state health department personnel and policy makers. Efforts to extend coverage should include consideration of building upon the state's current public programs (Medicaid and SCHIP) to increase the number of eligible persons. For Native Americans, additional efforts are needed to increase awareness of and enrollment in public programs designed to increase access to health care coverage among indigent adults and children. Finally, additional federal funds should be directed to strengthen the ability of the Indian Health Service to address problems of health care access among Native people, particularly those in isolated, remote areas, such as rural North Dakota. Primary Funding Source: HRSA ●The Southeastern Kentucky Community Access Program: How it Worked and Why Julia Costich, JD, Ph.D. Presented By: Julia Costich, JD, Ph.D., Director, Kentucky Injury Prevention & Research Center, 333 Waller Avenue, Lexington, KY 40504; Tel: (859) 257-6712; Fax: (859) 257-3909; Email: julia.costich@uky.edu Research Objective: To contribute to the replication of a successful program to increase health care access in a chronically underserved population by identifying critical program components and strategies. Study Design: Interviews with program staff and clients, analysis of enrollment and utilization data, and comparison with similarly structured but less successful programs are used to link specific program elements with outcomes. Population Studied: Adult residents of the Appalachian region of southeastern Kentucky who enrolled in the program for management of one of 5 targeted chronic conditions: severe mental illness, diabetes, heart disease, hypertension, and asthma. Principal Findings: The program served 15 percent of the 4county service area's population in its 4 years of funding, suggesting a broad potential impact. Critical elements identified by staff and clients include cultural competence, establishment of trust, and dependability of service provision. Prescription drug access had the greatest impact for those with severe mental illness, while lifestyle changes were important for those with cardiovascular diagnoses. Conclusions: These data demonstrate the burden of chronic illness in the population served and the benefits derived from relatively minor expense to address long-unmet needs. The termination of federal funding for Community Access Programs, while appropriate in many cases, leaves the chronically underserved at risk for relapse into the costly and burdensome illness from which this program offered some relief. Implications for Policy, Delivery, or Practice: The reasons why a few Community Access Programs succeeded when most did not meet objectives should receive closer attention to avoid further deterioration of already challenged populations and promote successful replication. Primary Funding Source: HRSA ●Results of a Randomized Trial of Transtheoretical ModelBased Interventions for Informed Choice in the Medicare Population Carol Cummins, MEd, MLIS, Deborah Levesque, Ph.D., James Prochaska, Ph.D., David Miranda, Ph.D., Amy Heller, Ph.D. Presented By: Carol Cummins, MEd, MLIS, Senior Project Manager, Pro-Change Behavior Systems, Inc., PO Box 755, West Kingston, RI 02892; Tel: (401) 874-7023; Fax: (401) 8744103; Email: ccummins@prochange.com Research Objective: To assess the efficacy of a Transtheoretical Model-based, or stage-based, manual and computer-administered expert system intervention developed for the Centers for Medicare and Medicaid Services to increase participation in informed health plan choice among new Medicare enrollees. Study Design: In Study One, the efficacy of the manual alone was tested primarily among 900 individuals with no access to a computer. In Study Two, the efficacy of the manual plus computer expert system was assessed among 451 individuals who did have computer access. Individuals in each study were randomly assigned to an intervention group that received the intervention materials, or to a control group that did not. The interventions were delivered to participants’ homes via mail. About one month later, individuals in the intervention groups were mailed a brief survey assessing the “acceptability” of the materials. A follow-up survey was administered six months post-baseline to assess the efficacy of the interventions. Outcomes included Medicare knowledge, stage of change for comparing Medicare health plans, information-seeking, and satisfaction with Medicare information. Population Studied: Participants were randomly selected from the Center for Medicare and Medicaid Service’s Initial Enrollment File, a data repository for persons who were going to turn 65 in the next three months. The sample was 83 percent white and 57 percent female. Principal Findings: The intervention materials, designed to be user-friendly and appropriate for individuals in all stages of change for comparing plans, were well received by study participants. About three-quarters of Study One participants who returned the acceptability survey had read more than half the manual. A majority of respondents found the manual to be easy to use and understand, attractive, and informative; 87 percent still had the manual, 68 percent thought they’d use it again, and 85 percent reported that they would recommend it to a friend. At follow-up, compared to participants in the control group, Study One participants in the intervention group had higher scores on objective and subjective measures of knowledge about the Medicare program, were more likely to think that the Medicare program was understandable, and were more likely to be satisfied or very satisfied with Medicare information. Study Two participants in the intervention group had higher scores on objective measures of knowledge and where about twice as likely to have used the Medicare program’s educational materials. Effect sizes were similar in magnitude to effects found in randomized trials of Transtheoretical Model-based programs for health behavior change. Conclusions: The Transtheoretical Model-based manual and expert system interventions were received positively by users and increased informed choice among new Medicare enrollees. Implications for Policy, Delivery, or Practice: Like other programs based on the Transtheoretical Model, the stagebased interventions designed to increase informed choice among new Medicare enrollees can have an enormous impact when applied on a population basis. In the next phase of this research, Pro-Change will work with the Centers for Medicare and Medicaid Services to enhance the interventions to increase their impact and facilitate dissemination. For example, the materials may be adapted for dissemination by employers. Primary Funding Source: CMS ●Validating Health Insurance Coverage Survey Estimates: A Comparison Between Self-Reported Coverage and Administrative Data Records Michael Davern, Ph.D., Kathleen Thiede Call, Ph.D., Jeanette Ziegenfuss, Gestur Davidson, Ph.D., Timothy Beebe, Ph.D., Lynn A. Blewett, Ph.D. Presented By: Michael Davern, Ph.D., Assistant Professor, SHADAC, University of Minneosta, 2221 University Avenue SE, Suite 345, Minneapolis, MN 55414; Tel: (612) 625-4835; Fax: (612) 624-1493; Email: daver004@umn.edu Research Objective: To examine the validity of estimates of health insurance coverage obtained from self-report survey data by comparing them to a health plan’s enrollment data to (1) assess whether health plan enrollees self-report being uninsured and (2) asses whether health plan enrollees are able to correctly identify the type of health insurance coverage they are enrolled in. Study Design: We included a health insurance coverage screener within a smoking survey administered to adult enrollees in the Blue Cross and Blue Shield of Minnesota Health Plan (BCBS-MN). The survey had a 61.5% response rate. From these responses we are able to validate selfreported health insurance status against the actual type of coverage using BCBS enrollment files. Population Studied: The population of BCBS health plan enrollees represents approximately one third of Minnesota adults. We drew a stratified random sample of 4,575 adult members (18 and over) of BCBS-MN representing four major payers: (1) Medicaid enrollees, (2) MinnesotaCare enrollees (a Minnesota-specific public health insurance program), (3) commercial enrollees and (4) commercial enrollees 65 years and over with Medicare supplemental. Principal Findings: Consistent with past research, we found a small upward bias—of 0.4% points—in the rate of uninsurance arising from individuals with insurance coverage answering the survey as though they were uninsured. No one type of enrollment group was more likely than any other to falsely report not having coverage; specifically, Medicaid enrollees were not more likely to answer they were uninsured than commercial enrollees. We also found that “falsepositive” Medicaid responses by those not on Medicaid accounted for approximately 75% of the total survey estimate of those enrolled in Medicaid. Conclusions: This study suggests that the Medicaid “undercount” does not arise in any quantitatively meaningful way from those on Medicaid answering health insurance surveys as though they were uninsured. As a result, we suggest that making post-survey adjustments to the number of uninsured to make the survey estimate match the Medicaid administrative data may be inappropriate. In addition, we found that a large proportion of the total self-reported count of Medicaid enrollees from the survey were contributed by individuals who were enrolled in primarily MinnesotaCare and commercial insurance. Implications for Policy, Delivery, or Practice: Current health insurance survey designs are good for determining whether someone is or is not uninsured. The current designs, however, are not as good at determining what type of insurance coverage a person has, and are especially poor in determining the specific type of public coverage a person has. This information is useful to understanding the potential bias in estimates of uninsurance and provides information to those evaluating current models designed to adjust for the “Medicaid undercount” when estimating rates of uninsurance based on survey data. Attention should be given to redesigning surveys of health insurance coverage in ways that might reduce the amount of “false-positive” responses given to public programs such as Medicaid and MinnesotaCare. Primary Funding Source: Blue Cross and Blue Shield of Minnesota ●Immigrants May Extend The Impending Aging BabyBoomers’ Effect In This Country Further Than Anticipated. Martey Dodoo, Ph.D., Robert L. Phillips, M.D., MSPH, Jessica L. McCann, MA, Ginger Ruddy, M.D., Larry A. Green, M.D., Lisa S. Klein Presented By: Martey Dodoo, Ph.D., Senior Economist, The Robert Graham Center, 1350 Connecticut Avenue N W, Suite 201, Washington, DC 20036; Tel: (202)331-3360; Fax: (202)3313374; Email: mdodoo@aafp.org Research Objective: To analyze the effect of US immigration on the age distribution of the US population, the impending baby-boomers’ aging, and its implications for Medicare and the US health care system. Study Design: Analyzed the components of population change in the U.S. over the last 24 years, particularly the contribution of net immigration to that change. Assessed the size and age-distribution of the US population including the baby-boom age-group cohort. Assessed the size and age distribution of legal and official estimates of illegal immigration into the US relative to the baby-boom cohort. Undertook computer simulations of the period and cohort effects on the US population of the current rate of annual influx of the immigrants, and the influx of immigrants when the baby-boomers retire and labor markets tighten, using Census Bureau projections. Population Studied: Age-specific data on the US population, immigration to the U.S., and the U.S. labor force from 1980 to 2003 Principal Findings: New immigrants make up almost half of the net annual increase in the U.S. population. Immigrants are always younger than the general U.S. population. About half of new immigrants are 25 to 44 years old with an average age below 37, compared to less than 40 percent of the US population between 25 to 44 and an average age above 37. Over time the demographic and cumulative effect of these immigrants will result in the relative swelling of the ranks of the age groups following the baby-boomers, extending the effect of the aging of a large segment of the population beyond the year 2031. Conclusions: The future population of the United States may include large cohorts of older people, not only as the babyboomers age, but if current trends persist, continuing past the baby-boomers with new cohorts of immigrants exacerbating our health resource demands. Implications for Policy, Delivery, or Practice: Policy makers and health planners should take into account these immigration population effects as there is likely to be an increased need for medical care by an older population that will probably continue for a longer time than previously anticipated. These effects are likely to have policy implications for Medicare, Social Security, the medical infrastructure, and the healthcare workforce. Primary Funding Source: No Funding Source ●Differences Between Rural and Urban Communities on Referral and Utilization Patterns in Cardiovascular Patients Dawn Finnie, MPA, James Naessens, MPH, Stephanie Anderson, BS, Rosa Cabanela, Ph.D. Presented By: Dawn Finnie, M.P.A., Health Services Analyst, Health Care Policy and Research, Mayo Clinic, 200 2nd Street SW, Rochester, MN 55905; Tel: (507) 288-3346; Fax: (507) 2841731; Email: finnie.dawn@mayo.edu Research Objective: Assess the effect of the urban-rural continuum on cardiovascular referral patients Study Design: Secondary administrative data from a large, Midwestern multi-specialty group practice was used for patients seen in 2003. These patients were classified by distance traveled from home: local(<=50 miles), regional (51120 miles), and distant (121+ miles). Patients were also identified as physician-referred or self-referred. Classification of patient residence on the urban-rural continuum was determined using the 2003 Rural-Urban Continuum Codes developed by the Economic Research Service of the USDA. These continuum codes were sub-grouped into three categories: metro, non-metro, and rural. Population Studied: Data on all patients (n=96,703) seen on an outpatient basis by a cardiovascular provider at a large Midwestern multi-specialty group practice in 2003 were identified. An episode of illness classification system was applied to all services at the practice during 2003 for these patients. Episodes in which an urban-rural continuum code could not be determined (n=3462), such as for international patients, were excluded. Information was also obtained on whether the patient was physician or self referred and the distance traveled for care. Patients were classified on the rural urban continuum using the 2003 Rural-Urban Continuum Codes developed by the Economic Research Service of the USDA. Physician utilization and referral patterns were compared among the different rural-urban classifications. Analysis incorporated adjustments for distance traveled, patient age and source of payment. Principal Findings: A total of 83,476 episodes were generated on patients in which an urban-rural continuum code could be determined. Of these episodes, 19.1% were cardiovascular-related. 61.0% of patients originated from metro counties, 32.0.% from non-metro counties, and 7.0% from rural counties. Patients from rural areas were less likely to be self-referred (22.1%) than patients from metro areas (26.8%). A higher percentage of episodes were classified as Other Cardiovascular Symptoms (includes tachycardia, palpitations, and murmurs) among non-rural (6.5%) patients vs. rural (4.7%) patients. Data will be presented which investigates the differences on type and severity of cardiovascular problem and extent of comorbidities among metro, non-metro and rural patients. Furthermore, the relationship between referral source and rural status will be presented. Conclusions: Episode patterns of self-referral and physician referral to cardiovascular providers appear to differ among patients originating in metro, non-metro and rural environments. Whether these differences are due to differing prevalence of disease, availability of medical care or provider and patient preferences needs to be studied further. Implications for Policy, Delivery, or Practice: Limited availability of primary and general medical care may affect access to cardiology services. Patients from rural areas appear to be referred to cardiology with different cardiovascular disease severity and type than patients from metro and non-metro areas. More care is needed in rural areas. Further studies are needed to assess other specialties. Primary Funding Source: No Funding Source ●Racial Disparity in Cardiovascular Procedures Within the Veterans Health Administratio: The Role of Hospitals Peter Groeneveld, M.D., MS, Gregory Kruse, MSc, Zhen Chen, Ph.D., David Asch, M.D., MBA Presented By: Gregory Kruse, MSc, Health Care Systems Department, Wharton School, University of Pennsylvania, 1909 Pine Street; Philadelphia, PA 19103; Tel: (215)868-2553; Email: gkruse@wharton.upenn.edu Research Objective: Prior studies suggest blacks receive fewer cardiac procedures than whites in Veterans Affairs Medical Centers (VAMCs), but the reasons for this are unclear. Variation in procedure use among VA hospitals may play an important role, particularly if VAMCs with larger black patient populations are less likely to perform procedures than VAMCs with predominantly white patient populations. Study Design: We examined hospital discharge data from the VA's Patient Treatment File, a database that comprises all VA healthcare utilization nationwide. Assuming that rapid-growth procedures would be the most likely to demonstrate variation in use, we selected nine predominantly inpatient cardiovascular procedures that rapidly increased in volume throughout the VA healthcare system between 1998-2003. Records from all VAMC hospitalizations between 1998-2003 were examined for codes indicating receipt of the target procedures. We also used the recorded race of each admitted patient to determine the racial composition of each VAMC's inpatient population for each year. Patient-level logistic regression models were used to generate propensity scores for each cardiac procedure based on patients' diagnoses, comorbidities, age, race, sex, and year of admission. The annual sums of hospitalized patients' propensity scores for each VAMC were included in a series of nine hospital-level, negative binomial regression models predicting utilization of each of the nine technologies. Covariates included each VAMC's inpatient racial composition, academic affiliation, and urban/non-urban location. Model parameters were estimated using generalized estimating equations. Population Studied: All patients admitted to Department of Veterans Affairs Medical Centers from 1997-2003. Principal Findings: Cardiac procedures were performed in 119 VA hospitals during 1997-2003, with the percentage of black inpatients at each center varying from <0.1% to 77%. Our regression models indicated that a 10% increase in a VAMC's black patient population was associated with lower rates of open mitral valvuloplasty (rate ratio [RR] 0.80, 95% confidence interval [CI] 0.69-0.92) and bioprosthetic aortic valve replacement (RR 0.81, 95% CI 0.68-0.98). However, there were no significant differences between hospitals with large and small black patient populations in the use of coronary arterial stents (p=0.99), coronary angioplasty without stenting (p=0.67), non-coronary arterial stents (p=0.66), dualchamber pacemakers (p=0.62), implantable cardioverterdefibrillators (p=0.70), transmyocardial revascularization (p=0.55), or negative contrast cardiograms (p=0.55). Conclusions: VAMCs with larger black inpatient populations performed new cardiac procedures at essentially the same frequencies as VAMCs with predominantly white inpatient populations. The only procedures that were more commonly performed in predominantly white VAMCs were cardiac valve replacement surgeries. Implications for Policy, Delivery, or Practice: These results suggest that racial differences in procedure utilization within the VA system are more likely attributable to local, withinhospital factors rather than clinical performance differences between VAMCs. Primary Funding Source: University of Pennsylvania Research Foundation ●Standard and Rapid HIV Testing Practices in Safety Net Hospitals Romana Hasnain-Wynia, Ph.D., Gretchen Williams Torres, MPP, Heidi Whitmore, MPP, Jeremy Pickreign, MS, JK Stanger, BA, Ray Kang, BA, Gloria J. Bazzoli, Ph.D. Presented By: Romana Hasnain-Wynia, Ph.D., Senior Director, Health Research and Educational Trust, Health Research and Educational Trust, One North Franklin, 30th floor, Chicago, IL 60606; Tel: (312)422-2643; Fax: (312)4224568; Email: rhasnain@aha.org Research Objective: The CDC estimates that 1 in 4 people living with HIV in the US are unaware of their HIV status. The use of rapid HIV testing in hospitals yields an opportunity to readily identify HIV infections and refer these patients to care. There continue to be missed opportunities for early HIV testing in hospital settings that often serve as a primary access point for many at-risk or HIV-infected patients. The US health system relies on the hospital safety net to provide care to the uninsured and indigent, however little is known about safety net hospitals’ HIV testing practices. The objective of this paper is to examine standard and rapid HIV testing practices in safety net and non-safety net hospitals. Study Design: Surveys on HIV testing were mailed to the universe of 4,497 non-federal, short-term hospitals. 1,230 hospitals completed the survey for a 27.4% response rate. Data were weighted to adjust for non-response. Focusing on the 623 metro hospitals, we calculate safety net status from the American Hospital Association database. Population Studied: We identified and classified hospitals into three categories by their safety net status, core safety net (4.8%), voluntary safety net (15.9%) and non-safety net (79.3%), based on the actual amount of uncompensated care provided. Principal Findings: 98% of core safety-net hospitals offer HIV testing compared to 94% of non-safety net hospitals. 10% of core safety net hospitals offer inpatient HIV testing compared to 1% of voluntary safety-net hospitals. 10% of core safety net hospitals offer HIV testing in their emergency departments compared to 0% of voluntary safety-net hospitals. 34% of labor and delivery units in core safety net hospitals offer HIV testing compared to 18% in non-safety net hospitals. 7% of outpatient centers in core safety net hospitals offer HIV testing compared to 2 % in non-safety net hospitals (all values significant at p<0.05). Knowledge of rapid HIV testing is similar across core safety net, voluntary set net and non-safety net hospitals. 47% of non-safety net hospitals do not use rapid testing versus 28% of safety net hospitals. 41% of emergency departments in core safety net hospitals provide rapid testing versus 21% in non-safety net hospitals. 32% percent of labor and delivery departments in core safety net hospitals provide rapid testing versus 17% in non-safety-net hospitals. 14% of outpatient centers in safety net hospitals use rapid testing compared to 5.5% in non-safety net hospitals (all values significant at p<.05). Conclusions: Safety net hospitals are more likely to provide HIV testing compared to non-safety net hospitals. However, given the CDC recommendations for universal testing for (1) all pregnant women and (2) all patients in hospitals serving high prevalence communities (greater than 1%), overall testing rates are low. Implications for Policy, Delivery, or Practice: HIV testing is more likely to take place in core safety net hospitals. Given the important role safety net hospitals play in providing care to the indigent and to at-risk populations, policymakers should provide incentives to support HIV testing and follow-up care. Primary Funding Source: CDC ●Rurality and Disparities in Mental Health Treatment Emily Hauenstein, Ph.D., LCP, APRN, BC, Stephen Petterson, Ph.D., Douglas Wagner, Ph.D., Virginia Rovnyak, Ph.D., Elizabeth Merwin, Ph.D., MSN, Barbara Heise, MSN, NP Presented By: Emily Hauenstein, Ph.D., LCP, APRN, BC, Associate Professor, School of Nursing, University of Virginia, PO Box 800782, Charlottesville, VA 22908; Tel: (434)9240093; Fax: (434)982-1809; Email: ejh7m@virginia.edu Research Objective: To assess urban-rural differences in mental health treatment rates and pathways to care among non-elderly adults using a refined measure of rurality. Study Design: We use data from the Medical Expenditure Panel Survey (MEPS) to examine the relationship between rurality and mental health treatment. We exploit the longitudinal design of the MEPS to examine treatment over four-month intervals subsequent to interviews, using up to three observation periods per respondent. Three measures of mental health treatment are examined: any medical visit related to a mental health condition, visits associated with specialized care and the number of visits in the observation period. We use the Urban-Rural Continuum as an ordinal measure of rurality, which classifies metropolitan counties on the basis of the size of the metropolitan area and nonmetropolitan counties on the basis of their urbanization and their adjacency to metropolitan areas. Binary logistic and linear regression analyses are performed to estimate differences in mental health treatment, adjusting for demographic, need, and access variables. Variance estimation is based on the clustered survey design of the MEPS; the weights and design variables take multiple observations of an individual into account. Population Studied: The sample consists of adults aged 18 to 64 in the first four panels of the MEPS, covering a period from 1996 to 2000. The MEPS is a large nationally-representative panel survey of households designed to provide estimates of the use of health services, medical expenditures and sources of payment. The sample size is 34,801 respondents, yielding 100,339 person-interview observations. Of these respondents, 10% reside in the three most rural types of counties on the urban-rural continuum. Principal Findings: Rural residents are significantly less likely to obtain mental health treatment, especially specialized treatment, than urban residents despite evidence of greater need among rural residents. These disparities persist after controlling for measures of reported mental health, age, marital status, region, schooling, income, and insured status. Other things being equal, the odds of any type of mental health treatment for an urban person were 35% greater than the odds for a rural person; the odds of receiving mental health specialty care were 37% greater for urban residents than rural residents. By contrast, there are no substantial urbanrural differences in the number of visits for the subset of respondents with at least one visit. Our results also suggest different pathways to care. In particular, the association between reported mental health and receipt of treatment is stronger in more urbanized areas, while the association between reported physical health and treatment is stronger in more rural areas. Conclusions: Limited access to mental health services and other characteristics of rural areas reduces the rural population's receipt of mental health visits, despite higher needs. Implications for Policy, Delivery, or Practice: Residents living in rural counties may have unique problems that put them at greater risk for mental health problems. Inadequate services and mental health delivery mechanisms may heighten that risk by reducing overall access to treatment. Primary Funding Source: NIMH ●Access to Care and Utilization among Children: The Effects of Public and Private Coverage Julie Hudson, Ph.D. Presented By: Julie Hudson, Ph.D., Economist, Center for Financing, Access, and Cost Trends, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850; Tel: ( 301)427-1683; Fax: (310)427-1277 Research Objective: We examine the relationship between health insurance coverage and children’s access to and utilization of medical care. We present results for a wide range of access measures, including having a usual source of care (USC) and delaying or going without needed medical care for financial reasons. We also examine indicators for ambulatory visits, well-child visits, dental visits, emergency room use, and inpatient hospital stays. Study Design: We pool data from the first seven years (19962002) of the Medical Expenditure Panel Survey (MEPS). Pooling yields a large sample of children, enabling us to analyze access and utilization using simple descriptive statistics, multivariate analysis, and instrumental variables estimation (IV). In particular, our IV estimates offer insights into the widespread concern than conventional estimates overstate the effect of coverage, because children with coverage may have greater health risks than those without coverage. Population Studied: Children in the civilian noninstitutionalized population of the United States who are covered by private, public, or no health insurance. Principal Findings: As previous studies have found, public and private coverage are both associated with large increases in access and utilization. Simple mean comparisons suggest that private coverage has a larger effect than does public coverage. This difference is reduced (and often changes sign) when we control for other characteristics of children and their families. IV coverage effect estimates are larger than conventional estimates, mitigating concerns that conventional estimates are overstated. Conclusions: Public coverage and private coverage are both associated with large increases in access to and utilization of medical care. Implications for Policy, Delivery, or Practice: Our results highlight the importance of reducing uninsurance among children. Primary Funding Source: AHRQ first 6 months of 1999. Of the 38,358 colorectal cancer patients, 2,887 (7.5%) had colectomy surgery in the first six months of 1999. Evidence of chemotherapy administration was found in 1999 Medicare or VA files for 887 (30.7%) of the colectomy patients. Mean time between surgery and start of chemotherapy, for patients who received chemotherapy after surgery, was 57.4 days (approximately 8 weeks). Mean time between surgery and chemotherapy was longer for patients who used both VA and Medicare providers than for “Medicare only” users (60.5 days v 53.6 days; P<.05). Mean time between surgery and chemotherapy was similar for dual system users and for “VA only” users (60.5 days v 56.2 days; P<.5). It is not clear if these differences are clinically different. Multivariate analysis across provider types is continuing. Conclusions: These early results suggest that elderly colorectal cancer patients using more than one system of care may experience delays in receipt of adjuvant therapy. Studies of outcomes that adjust for cancer stage and comorbidity will be needed to determine if the delays are clinically significant. Implications for Policy, Delivery, or Practice: While it is well documented that initial surgery and adjuvant therapy are key to disease-free and long-term survival in colon cancer, variations in care exist. Patients using more than one healthcare system are particularly vulnerable, because coordination of care across systems of care may be lacking, resulting in delays in care. Primary Funding Source: VA ●Timing of Surgical and Adjuvant Therapy for Colorectal Cancer Denise Hynes, Ph.D., MPH, RN, Ruth A. Perrin, MA, Qiuying Zhang, MS, Kristin Koelling, MS, Min-Woong Sohn, Ph.D. ●Expanding Health Insurance to Cover at-Risk Populations in a Rural State: Qualitative Results from the Nebraska State Planning Grant. Katherine Jones, Ph.D., Erin Carlson, MPH, Roslyn FraserMaginn, BA, Brandi Shay, BA, Katherine Jones, Ph.D. Presented By: Denise Hynes, Ph.D., MPH, RN, Senior Health Scientist, Midwest Center for Health Services and Policy Research, Veterans Administration, PO Box 5000 (151H), Hines, IL 60141; Tel: (708)202-2413; Fax: (708)202-2415; Email: Denise.Hynes@med.va.gov Research Objective: To assess and compare colorectal cancer initial surgical and adjuvant treatment patterns for elderly patients across the Veterans Administration and Medicare systems of care Study Design: Based on a quality of care theoretical model, the study links data from multiple comprehensive data sources for a retrospective cohort of colorectal cancer patients to characterize and compare patient demographic characteristics, comorbidity, timing and type of cancer surgery and timing and type of adjuvant therapy. Provider type (VA or Medicare) and geographic region are compared. Calendar year 1999 comprises the study period. Population Studied: A retrospective cohort of 1999 incident and prevalent colorectal cancer patients, identified from VA and Medicare data, who are at least 66 years old and eligible to use both VA and Medicare healthcare Principal Findings: Using an ICD9CM code algorithm to identify colon and rectal cancer patients in Medicare claims and VA utilization files, we identified 38,358 unique patients. Of those, 3,882 (10.1%) died in 1999 and by the end of 2003 a total of 17,166 (44.8%) of the patients had died. To increase the likelihood that evidence of adjuvant therapy would appear in the 1999 medical record, further analysis was restricted to colorectal cancer patients who had a colectomy during the Presented By: Katherine Jones, Ph.D., Assistant Professor, Preventive and Societal Medicine, University of Nebraska Medical Center, 984350 Omaha, NE 68198-4350; Tel: (402) 559-8913; Fax: (402)559-7259; Email: kjonesj@unmc.edu Research Objective: To inform policymakers about the lived experiences of Nebraskans at risk of being uninsured and underinsured, and the factors small employers consider when deciding whether to offer health insurance. Study Design: Thirteen focus groups were conducted for the Nebraska State Planning Grant to complement its household survey. We used the stages of preparation, contact with key contacts to negotiate and confirm participants, and follow-up to lay groundwork for future contact. Incentives for recruitment included a monetary stipend, transportation, a light meal, and child care. The three Hispanic groups were conducted in Spanish. For all groups conducted with minority populations, a researcher co-moderated with the key contact. Group proceedings were audio-recorded and transcribed to allow coding and analysis using NVivo software. We identified themes and subthemes unique and similar across all groups. Population Studied: The household survey identified populations likely to be uninsured. Thirteen focus groups totaling 133 participants were conducted across Nebraska’s six health planning regions. Nine groups included populations at risk of being uninsured: urban African-Americans, urban Native Americans, rural Hispanics (two groups), urban Hispanics, young adults attending community college, lowincome adults (two groups), and refugees. Four employer groups included urban employers with fewer than 50 employees, rural employers with fewer than 50 employees (two groups), and rural self-employed or micro-employers with 5 or fewer employees. Principal Findings: All groups wanted access to affordable, comprehensive health care for themselves and their families. Reasons for not obtaining insurance were high cost and a perceived lack of value due to high deductibles—paying for basic necessities took precedence. High out-of-pocket costs limited utilization of primary care by the uninsured and the underinsured—those who purchased catastrophic coverage to protect assets. The uninsured and underinsured described fragmented care due to gaps in the safety net. The high cost of premiums prevents small businesses from offering insurance, which discourages community development and results in the uninsured and underinsured incurring debt when obtaining needed care. Immigrants, refugees, and young adults do not understand the role insurance plays in access to care. Conclusions: The uninsured and underinsured value the security provided by insurance but use limited resources to pay for basic necessities. The lack of affordable health insurance affects quality of life: people are unable to fulfill moral obligations to care for family and employees; they delay or forgo needed care, allocate limited resources to their children, and worry about debt. The cost of health insurance impedes rural development as out-of-pocket costs prevent spending in other economic sectors and lack of employersponsored insurance contributes to depopulation. Quantitative studies of the uninsured do not capture the difficulty those with catastrophic coverage experience in paying for primary care. Implications for Policy, Delivery, or Practice: Rural states with diverse at-risk populations (immigrants, minorities, lowincome adults, the self-employed) and a high proportion of small employers should identify the unique barriers to coverage for each group so that limited resources are used to improve coverage for those most likely to forego needed care or incur debt paying for it. Additional research is needed to understand the extent and effect of underinsurance in Nebraska. Primary Funding Source: HRSA ●Health Care Coverage and Access to Care: The Status of Minnesota’s Veterans Yvonne Jonk, Ph.D., Kathleen Thiede Call, Ph.D., Andrea H. Cutting, MA, Heidi O’Connor, MS, Vishakha Bansiya, M.D., Kathleen Harrison, MBA, MPH Presented By: Yvonne Jonk, Ph.D., Health Economist, Center for Chronic Disease Outcomes Research, Minneapolis VA Medical Center, One Veterans Drive 1110, Minneapolis, MN 55417; Tel: (612)467-3882; Fax: (612)725-2118; Email: yjonk@umn.edu Research Objective: The primary objective of this study is to examine veterans’ reliance on health care services provided by the Veterans Health Administration (VHA) within Minnesota and estimate the potential effect on uninsurance rates if all eligible veterans relied on VHA coverage. Secondary objectives are to compare veterans and non-veterans’ by geographic location, demographic characteristics, health status, and health insurance coverage, and to compare insured and uninsured veterans especially with regard to access to care. Study Design: Data are from the 2001 Minnesota Health Access Survey of a stratified random sample of over 27,000 respondents of whom 3,500 were self-identified veterans. Though all veterans were eligible to obtain health care services from the VHA in 2001, veterans not reporting VHA coverage and having no other source of insurance coverage were considered uninsured. Differences in weighted population characteristics are reported. Logistic regression analysis is used to identify factors associated with veterans' reliance on VHA coverage. Population Studied: A random sample of over 27,000 Minnesota residents of whom 3,500 were self-identified veterans in 2001. Principal Findings: Veterans represented 13.4% of the state's adult population and 9.3% of the state’s uninsured non-elderly adult population in 2001. Uninsured veterans were more likely to be single, unemployed, living in rural areas, and in poorer health than insured veterans. They also report constrained access to services. Insured veterans were less reliant on VHA coverage. Conclusions: The state’s uninsurance rate would significantly decrease if veterans relied on the VHA safety net. Implications for Policy, Delivery, or Practice: If veterans’ insurance status matters in states with low uninsurance rates, VHA coverage has broader implications for states with higher veteran concentrations and higher uninsurance rates. Primary Funding Source: HRSA ●Emergency Department Overcrowding in Large Metropolitan Areas: From Bad to Worse Phillip R. Kletke, Ph.D., Jon Gabel Presented By: Phillip R. Kletke, Ph.D., Director, Health System Studies, Health Research and Educational Trust, 1 North Franklin, Chicago, IL 60616; Tel: (312)422-2641; Fax: (312)4224568; Email: pkletke@aha.org Research Objective: To monitor the increasing prevalence of overcrowding in the emergency departments (EDs) of hospitals in large metropolitan areas; and to determine how these EDs are responding to the growing demand for emergency services. Study Design: We conducted a survey of ED administrators in the 729 ED hospitals in the 20 largest metropolitan areas of the US. The field period was between February and June of 2004. The survey questionnaire collected data on: trends in ED visits and ED capacity; the prevalence of ED overcrowding and ambulance diversions; the perceived causes of overcrowding; the methods used to minimize overcrowding and ambulance diversions; and strategies used by EDs to minimize excess demand for their services. EDs could respond to the survey by mail, by fax, or on-line. We received responses from 191 hospitals for a response rate of 29%. The data are weighted to control for potential response bias. Population Studied: Hospitals with EDs in large metropolitan areas. Principal Findings: ED overcrowding in large metropolitan areas is worsening. Over 90% of the EDs indicated that the demand for their services had increased in the past 3 years. Average waiting time for emergency services was long: 20 minutes for emergent patients; 52 minutes for urgent patients; and 76 minutes for non-urgent patients. Almost three-fourths (73%) of the EDs indicated that the boarding of patients (i.e., housing patients in the ED due to the lack of beds or staff in inpatient areas) had increased during the past three years. More than half (52%) indicated that the incidence of ambulance diversions (re-routing ambulances to more distant hospitals because of overcrowding at the nearest ED) had increased during the past 3 years. During 2003, hospitals were on full or partial diversion for an average of 694 hours— the equivalent of almost a full month. The most frequent causes of diversions were bottlenecks in the processing of patients—the lack of inpatient beds, the lack of available ICUs, or the lack of ED capacity. Conclusions: EDs are attempting to cope with overcrowding in a variety of ways: coordinating the discharge of ED patients with hospital admissions (77%); increasing ED staffing (73%); increasing the use of contract workers (59%); increasing the physical capacity of the ED (57%). Nevertheless, some EDs are clearly losing the struggle. Because of overcrowding and long waiting times, approximately 3% of ED patients who complete triage leave the ED before being evaluated, perhaps forgoing needed care. Implications for Policy, Delivery, or Practice: Overcrowding in hospital EDs has increased for at least the past 25 years, and our findings indicate that this trend shows no signs of abating. The prospect that EDs may become even more overcrowded should generate considerable concern within the health policy community. Overcrowding in the ED delays the treatment of patients with emergent conditions and thus prolongs their pain and suffering and increases the likelihood of poor clinical outcomes. Furthermore, ED overcrowding limits access to care to the most vulnerable of patients by interfering with the ED’s ability to function as a safety net provider to the poor, the uninsured, and Medicaid patients. Primary Funding Source: RWJF ●Effects of the Missing Medigap Guarantee-Offer for Disabled Beneficiaries Jill Klingner, MS, RN Presented By: Jill Klingner, MS, RN, Rural Health Research Center, University of Minnesota, 1240 Juliet Avenue, St. Paul, MN 55105; Tel: (651)698-8045; Email: klin0089@umn.edu Research Objective: This study assessed the scope of protection available to the Medicare-eligible under-age 65 disabled through supplemental insurance plans, including Medicaid, Veterans Administration care, and employersponsored (ESSI) or individually purchased policies(IPSI). It also estimated the effect of state Medigap guarantee-offer and Medicaid eligibility policies on supplemental insurance acquisition. Study Design: This study is a longitudinal, natural experiment comparing the supplemental insurance rates of disabled Medicare beneficiaries across states with differing and changing Medigap and Medicaid policies over the 5-year period of 1997-2001. The Medicare Current Beneficiary Survey (MCBS) provided information on personal characteristics, health insurance, health status, functional status, chronic conditions, and living arrangements. Additional information on state Medicaid and Medigap policies were available from CMS and state offices. Population Studied: The study subjects are disabled, under65 Medicare beneficiaries in the MCBS sample for at least two years between 1997-2001. Those who died before their second complete year were excluded from the sample because their information was incomplete. Those with eligibility due to End Stage Renal Disease (ESRD) were excluded because their Medicare benefits and experiences are different than those of other Medicare beneficiaries under-65. Principal Findings: The Medicare disabled are less likely than the elderly to have Medicare supplemental health insurance coverage. Reflecting their lower incomes, they are more likely to have Medicaid. The largest difference in access to additional coverage appears in the Medigap category. Only 8% of the disabled purchase these policies. State policies mandating Medigap-offering at eligibility strongly affects the likelihood of enrollment. Conclusions: Due to the increased barriers to supplemental insurance and the higher health-related needs with fewer resources the disabled person’s ability to maintain a health regimen, to improve his or her health, or maintain his or her health, or prevent deterioration of health, may be compromised. Implications for Policy, Delivery, or Practice: Access to Medigap policies is greater for the elderly because of a 1990 federal law requiring Medigap policies to be available for all age-eligible during their initial Medicare eligibility. Some states mandate a guarantee-offer of this type (Medigap or IPSI) of policy for the disabled at their initial eligibility. As of 1998, disabled Medicare beneficiaries in only nine states had guarantee access to Medigap policies without underwriting. As of 2003, 22 states had guaranteed access to Medigap policies for the disabled without underwriting. An expansion of the federal or state policies requiring Medigap offerings at initial eligibility for the disabled would increase the proportion with Medicare supplemental insurance and may have a positive effect on the health- and functioning status of disabled under-65 beneficiaries. Primary Funding Source: No Funding Source ●The Role of Technical Assistance in the Success of Rural Health Networks: The ‘Right’ TA at the ‘Right’ Time Patricia Kota, MSPL, RN, Dora Ward Kyabu, MPH, BA, Lindsey Lonergan, MSH, BS Presented By: Patricia Kota, MSPL, RN, Director, Community Health Systems Development, Georgia Health Policy Center, Andrew Young School of Policy Studies, 14 Marietta Street, Atlanta, GA 30303; Tel: (404)463-9564; Fax: (404)651-3147; Email: pjkota@gsu.edu Research Objective: The development of rural health networks is an inherently complex process involving an array of rural health network grantees (RHNs). On the presumption that standardized technical assistance processes and tools improve efficiency and predictability, rural health network grantees are provided technical assistance with health information technology, business planning, and care management implementation, among other topics. The Georgia Health Policy Center has developed a proactive technical approach grounded in the assertion that systematic but flexible technical assistance, provided by experts in relationships with networks, is the most effective and efficient means to improve access to care for rural residents. The goal of this analysis is to describe the impact of meaningful technical assistance contact provided early in a network’s development versus the impact of technical assistance provided later. Study Design: This qualitative study is based on case studies of the Health Resources and Services Administration’s Office of Rural Health Policy Rural Health Network grantees receiving technical assistance from the Georgia Health Policy Center. Case studies of three vertically integrated RHNs and three horizontally integrated RHNs receiving technical assistance visits in the first year of development are compared with case studies of three vertically integrated RHNs and three horizontally integrated RHNs receiving technical assistance visits in the second year of development or later. All RHNs are selected for homogeneity of budgets, overall goals, and demographic factors in order to control for extraneous factors influencing network success. Data sources include monthly and quarterly reports, interviews with network directors and technical assistance providers, and responses to quality assurance surveys. Measures of success include degree of governing board involvement, strategic and business planning process and commitment, community involvement as determined by formal involvement of community leaders and citizens, and the level to which networks adopt an integrated, holistic health approach. Population Studied: Rural Health Networks recieving grants from the Health REsources and Servies Administration. Principal Findings: Analysis is presently underway. Expected completion is May, 2005 Conclusions: Technical assistance is presumed to be of benefit to rural health networks, but its effectiveness has not been documented. By contrasting the impact of technical assistance delivered early and later in the network development process, we expect this study will demonstrate the importance of providing technical assistance as early as possible in the network development process. Implications for Policy, Delivery, or Practice: The power of technical assistance tailored to a community’s needs and provided early and consistently throughout the process of development or reformation of any community health system may be immense. Demonstrating the impact of early technical assistance to community health systems may support the provision of this support to underserved communities throughout the country, resulting in improved health among the nation’s most needy populations and increased return on investment for grantors. Primary Funding Source: HRSA ●Expanding Health Care Access with DSH: Georgia's Indigent Care Trust Fund Glenn Landers, MBA, MHA, Presented By: Glenn Landers, MBA, MHA, Senior Research Associate, Georgia Health Policy Center, 14 Marietta Street, Atlanta, GA 30303; Tel: (404)463-9562; Fax: (404)651-3147; Email: glanders@gsu.edu Research Objective: Evaluate the effectiveness of Georgia's Indigent Care Trust Fund (ICTF) - the program created to distribute federal disproportionate share hospital (DSH) funds. Georgia is the only state that requires hospitals to spend a portion their DSH awards on community-based primary care. Study Design: 1.) Systematic review of hospital primary care plans. 2.) Written questionnaire 3.) Qualitative interviews with hospital, public health, service agency, and state personnel. Population Studied: 92 hospitals participating in Georgia's Indigent Care Trust Fund Principal Findings: 1.) About half of the hospitals studied attempted some measurement of the impact of their primary care programs on the community; slightly more than 20% extensively measured impact; about 25% did not attempt to measure program impact. 2.) More than 50% of the time, programs reached the intended population - Medicaid clients and the medically indigent; about 40% of the time, programs reached unintended populations; less than 10% of the time, programs did not reach the intended population at all. 3.) Almost 60% of the time, hospitals engaged District Public Health Directors in health planning to some degree; about 22% of the time, hospitals extensively engaged District Health Directors; less than 20% of the time, hospitals did not engage District Health Directors at all. 4.) Large hospitals were able to leverage additional funds more effectively than smaller hospitals. 5.) Hospitals were particularly successful in leveraging their primary care dollars with indigent drug programs. Conclusions: The Indigent Care Trust Fund Primary Care Plan requirement plays a vital role in Georgia's health care safety net. Without it or comparable funding, fewer Georgians would have access to primary health care services, and uncompensated hospital expenses would most likely increase due to individuals forgoing routine care and seeking hospital emergency care once illness progresses. Implications for Policy, Delivery, or Practice: Georgia's primary care requirement is an innovative model for other states that wish to expand access to uninsured individuals. DSH funding targeted to community-based primary care has the potential to reduce high cost inpatient admissions. Primary Funding Source: Georgia Department of Community Health ●Effect of Health Insurance on Workplace Absenteeism in the US Workforce Jennifer Lofland, PharM.D., MPH, Ph.D., Kevin D. Frick, Ph.D. Presented By: Jennifer Lofland, PharM.D., MPH, Ph.D., Research Assistant Professor, Health Policy, Thomas Jefferson University, 1015 Walnut Street, Suite 115, Philadelphia, PA 19107; Tel: (215)955-7348; Fax: (215)923-7583; Email: jennifer.lofland@jefferson.edu Research Objective: 1) To examine the relationship between health insurance and the annual number of missed workdays, 2) to determine how this relationship is modified by one’s level of access to care and healthcare use. Study Design: This was a retrospective, pooled, crosssectional study using 1996 to 1999 data of the Medical Expenditure Panel Survey. An access to care index was developed using Rasch Partial Credit Analysis. A two-part model was used to estimate the annual number of missed workdays. The first part of the model estimated the probability of incurring any missed workdays. Since the missed workdays data had a skewed distribution, the number of missed workdays was transformed using a logarithmic scale. For the second part of the model, the number of log transformed missed workdays, conditional on missing at least one workday, was estimated. Population Studied: Non-institutionalized, civilian, employed individuals within the US. Principal Findings: Of the 25,676 employed individuals, 13,957 (54%) reported missing time from work. Of those who missed time from work, the mean (SE) annual number of missed workdays was 4.3 (0.092). Having health insurance was significantly associated with a decreased likelihood of missing work (p=0.038). When access to care was included in the model examining the relationship between having health insurance and missed workdays, a higher level of access to care was significantly (p=0.000) associated with missing a greater number of workdays and the coefficient associated with the health insurance coefficient decreased slightly from 0.017 to -0.046. However, health insurance was still not associated with the number of missed workdays. When healthcare use was included in the same model, 1) healthcare use was significantly (p=0.000) associated with an increased likelihood of missing work and missing a greater number of workdays, 2) having health insurance was significantly associated with missing a smaller number of workdays (p=0.000), and 3) the health insurance coefficient decreased from -0.017 to -0.081. When access to care and healthcare use were simultaneously included in the model, the results were the same as above. Conclusions: Having health insurance at any time point over the course of a year is significantly associated with a decreased likelihood of missing work as well as missing a smaller number of workdays. Access to care and healthcare utilization are important explanatory variables. Healthcare utilization appears to be a confounder between health insurance and workplace absenteeism. Implications for Policy, Delivery, or Practice: These results suggest that the potential financial benefits of employers ensuring that all employees have health coverage should to be explored. Primary Funding Source: AHRQ ●Changing Paterns of Care by Veterans Judith Long, M.D., Daniel Polsky, Ph.D., Joshua Metlay, M.D. Presented By: Judith A. Long, M.D., Physician Researcher, Philadelphia VA CHERP, 1201 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104; Tel: (215)898-4311; Fax: (215)5738778; Email: jalong@mail.med.upenn.edu Research Objective: We describe how characteristics of veterans who use the VHA have changed from 1992 to 2000. Study Design: We merged two waves of the National Survey of Veterans to determine changing patterns of out-patient care use. We evaluated, conditioning on receiving out-patient care, the choice of veterans to receive at least some of their care from the Veterans Health Administration (VHA). Population Studied: Veterans Principal Findings: Compared to veterans who receive all their care outside of the VHA, there was an increase in the relative number of VHA users who were 65 or older, less educated, poor and had no additional health insurance. Although the odds that a VHA user compared to a non-VHA user had no additional health insurance in 1992 was 2.12 (95% CI: 1.60-2.81), by 2000 the odds was 4.35 (95% CI 3.71-5.10). The increase in use by the poor and less educated appears to be driven by veterans who receive their care both within and without the VHA (dual users). Conclusions: Among users of out-patient care there was an increase in the proportion of veterans using the VHA who are older, without additional medical insurance and low-income. Implications for Policy, Delivery, or Practice: An important mandate of the VHA is to serve vulnerable veterans. Since the VHA re-engineering of the mid 1990s the VHA seems to have made significant strides toward fulfilling this mandate by increasing the amount of care it provides to older, sicker, low income veterans, as well as those veterans without additional health insurance. Although there is still room for improvement the VHA plays a vital role in the U.S. health care safety net. Primary Funding Source: VA ●Constructing an Index for Predicting Health Access Deprivation in the U.S. Jessica McCann, MA, Robert L. Phillips, M.D., MSPH, Martey S. Dodoo, Ph.D. Presented By: Jessica McCann, M.A., Analytic Geographer, The Robert Graham Center, 1350 Connecticut Avenue, N W Suite 201, Washington, DC 20036; Tel: (202)331-3360; Fax: (202)331-3374; Email: jmccann@aafp.org Research Objective: To use logistic modeling techniques in constructing a health access deprivation index that will allow us to determine the variables that best predict reported health access problems in this country Study Design: We reviewed the relevant literature and identified variables that have been associated in some way with health access and health deprivation in this country. We modeled two measures of health care access deprivation derived from the 2002 National Health Interview Survey (NHIS): These were (1) was care delayed at anytime in the past 12 months, and (2) a combination of was care delayed or did respondent not get care at anytime in the past 12 months. Using techniques and software for analyzing surveys with complex designs like the NHIS, we identified independent variables that had significant associations with the health care access variables. We used those independent variables in stepwise logistic regressions to determine the extent to which these variables predict whether a person had care delayed or did not get care, and the odds associated with these. Population Studied: We used merged national data from the person, household, and adult components of the 2002 NHIS. The sample size was 31,044 persons 18 or more years old. Principal Findings: Out of the 32 independent variables identified in the literature as potential predictors of health access deprivation, 23 were available in the NHIS. After further deliberation, we tested 18 of these for significance and 16 were found to be significantly associated with the delay of care or not receiving care. In the sample analyzed, 15.6 percent had care delayed and 19.4 percent delayed or did not get care. The key results are that someone is most likely to have care deprived, or to have care delayed, if they are female, do not own a home, live in a single adult household, have a functional limitation, smoke, have no usual source of care or one place where they receive care, have no health insurance or are less than 65 years old. Conclusions: The results were as expected - that health access is highly associated with a number of socioeconomic variables. Variables such as being female, not owning a home, living in a single adult household, having a functional limitation, being a smoker, having no usual source of care or one place where they receive care, having no health insurance, and being less than 65 years old can be good predictors of potential health access problems. Implications for Policy, Delivery, or Practice: This study has provided a model that may be used as a basis for local community assessments and health planning. As a next step the results of this study may be used as a foundation for merging and using Census data to identify specific communities and/or populations at risk of having health access deprivation problems or issues. Primary Funding Source: No Funding Source ●Regional Variation in ED Utilization by Oregon Health Plan Enrollees K. John McConnell, Ph.D., Robert A. Lowe, M.D., MPH, Rochelle Fu, Ph.D., Annette A. Adams, MPH, Cody Weathers, BS, Charles A. Gallia, MS, Judy Mohr-Peterson, Ph.D. Presented By: K. John McConnell, Ph.D., Assistant Professor, Center for Policy and Research in Emergency Medicine, Oregon Health & Science University, 3181 SW Sam Jakcson Park Road Mail Code CR-114, Portland, OR 97211; Tel: (503) 494-1989; Fax: (503)494-4640; Email: mcconnjo@ohsu.edu Research Objective: Although high utilization of emergency departments (EDs) has been well documented as a marker of poor access to medical care, less is known about the regional variations in access that occur among different medical service areas. The objective of this study was to determine the extent to which case-mix adjusted ED visits varied among 130 small market areas. Study Design: This historical cohort study used Oregon Health Plan administrative data to determine number of ED visits, duration of enrollment, age, gender, race, ethnicity, eligibility category, managed care/fee-for-service coverage, pregnancy, and presence of 33 chronic medical conditions. The enrollee's zip code was used to place each patient in one of 130 primary care service areas (PCSAs). PCSAs were developed by the Oregon Office of Rural Health and are based on topography, social/political boundaries, and travel patterns We used a zero-inflated negative binomial model to estimate ED visits, adjusting for patient characteristics and chronic medical conditions, and fixed effects for PCSAs. We then examined the contribution to ED visits that could be attributed to each PCSA. This approach was designed to identify the variation in ED visits that could not be attributed to patient characteristics. Population Studied: 676,345 Oregon Health Plan beneficiaries aged <65, enrolled in 2001 and 2002. Principal Findings: Unadjusted rates of ED utilization in different PCSAs varied from 0.15 visits/member/year (95% CI 0.11-0.19) to 1.5 visits/member/year (95% CI 1.4-1.6). Case-mix adjusted rates showed a similar range, with a coefficient of variation (CoV) of 0.4. Variations in ED use by region were larger among older patients and patients enrolled in managed care plans. Conclusions: Rates of ED use vary substantially, by a factor of 10, among different medical service areas. These variations remained even after adjusting for patient demographics and comorbidities. Implications for Policy, Delivery, or Practice: Variations in ED use among these medical service areas suggest that some communities have successfully provided access to care to the underinsured, while others exhibit substantial barriers to care. Understanding the factors that drive disparities in access in different regions may lead to improvements in primary care for these vulnerable populations. Improving primary care has the potential to improve health outcomes and also lead to more efficient allocation of health resources. Primary Funding Source: Collins Foundation and Emergency Medicine Foundation Riggs Family Grant ●Making Coverage for the Uninsured Karen Minyard, Ph.D., Deborah Chollet, Ph.D., Laurie E. Felland, MS, Lindsey Lonergan, MHS, Chris Parker, MBBS, MPH Presented By: Karen Minyard, Ph.D., Director, Georgia Health Policy Center, Andrew Young School of Policy Studies, 14 Marietta Street, Atlanta, GA 30303; Tel: (404)651-3104; Fax: (404)651-3147; Email: kminyard@gsu.edu Research Objective: As health care costs and the number of uninsured Americans continue to increase, community initiatives across the country are steadfast in their efforts to bridge the growing gap between public and private health insurance coverage. The object of this research was to understand how community initiatives have developed and sustained financing to serve the uninsured and more specifically to: a. Describe efforts to increase coverage or access and impact on target populations. b. Examine the cost effectiveness and efficiency of operations. c. Identify factors that enhance or challenge sustainability and expansion of the initiatives. d. Understand the lessons for replication. e. Examine how states and communities cooperate to close gaps in funding and access. Study Design: Five case studies were conducted over an 18 month period. Key study methods included: feedback on research design from public and private experts; development of site selection criteria: site selection; assembly of data regarding the case-study community initiatives; development of interview protocols; site visits with 10-25 key informants from each site (a combined total of 84); follow-up integration of information; coding and uploading interview notes to a qualitative database (Atlas-ti); and team analysis of each site and comparative analyses across sites. Analyses frameworks included: intent; effectiveness; sustainability; replication; state/community interface; and leadership. Population Studied: Five case study sites (Wichita, Kansas; Paris, Arkansas; Milwaukee, Wisconsin; Olympia, Washington; and Forsyth, Georgia) were selected because of their geographic and operational diversity. Each of these initiatives attempts to provide coverage and/or access to care to individuals who have difficulty finding or navigating conventional insurance arrangements and public programs. Principal Findings: All five local community initiatives attempt to provide access and coverage to a specific population: non-elderly adults. In each community, there is evidence that providing more appropriate health care services is cost effective for providers and the community at large. Program leaders have fought hard to maintain momentum and remain relevant despite diminishing grant funding. The difficulty of replication can be attributed to the complexity of the innovations and to differences in context between alpha and beta sites. The context of community programs – the presence of supportive public programs and/or strong private insurance capacity is an essential success factor. Conclusions: Encouraged by private foundations and government efforts to strengthen local health care safety nets, some communities have developed programs to piece together access and financing for individuals and families who live between the worlds of private insurance and public coverage. These efforts attempt to weave the threads of provider discounts, free care, and public financing into a more coherent system of coverage and care. Implications for Policy, Delivery, or Practice: Public policy that supports provider participation and state-level leaders who believe in local innovation are important to programs that entail provider volunteerism or acceptance of reduced compensation. Local programs can support or complement state public and private insurance programs, but they are unlikely to thrive independent of them. Primary Funding Source: CWF, Healthcare Georgia Foundation, Washington Health Foundation ●The Role of Targeted External Facilitation and Technical Assistance in Improving the Clinical Relevance and Financial Viability of Rural Health Systems Karen Minyard, Ph.D., Tina Anderson Smith, MPH, Lindsey Lonergan, MHS Presented By: Karen Minyard, Ph.D., Director, Georgia Health Policy Center, Andrew Young School of Policy Studies, 14 Marietta Street, Atlanta, GA 30303; Tel: (404)651-3104; Fax: (404)651-3147; Email: kminyard@gsu.edu Research Objective: The health care environments in many rural communities in the United States have been in crisis for more than a decade, yet community leaders are often not organized to strengthen health and human services that contribute to improved access and health status. Rural/urban disparities exist in that rural Americans are generally older, poorer, and sicker than their urban counterparts. This research examines the role of targeted external facilitation and technical assistance in improving the collaboration, clinical relevance, and financial viability of rural health systems. Study Design: Review of the literature, a survey of national best practices, and pilot interventions in three rural communities suggested a framework for building and sustaining successful rural health systems. Case studies to test the framework were conducted in 12 rural communities. Each community experienced a fragile, failing health care system. Each community was provided with external facilitation and technical assistance during the study period. Community specific information, technical assistance resources, and outcomes were documented for each community in 1999 and 2000. The data were examined vertically (within each site) and horizontally (across sites) for themes and outcomes. The framework was revised based on learnings from the case studies. Population Studied: Twelve rural communities in Georgia were studied. Principal Findings: The key modifications to the framework were the time sequencing of the financial and clinical outcomes and the introduction of contextual modifiers. When provided with technical assistance, all 12 rural communities were able to maintain or improve local collaboration. The local collaboration was followed by financial gains in 9 of the 12 communities and clinical gains in 3 of the 12 communities. Several things were found to moderate the theorized relationships and process: Longer duration technical assistance; an environment of crisis was not associated with clinical or financial gains; and local strong leadership seemed to be associated with improved outcomes. Conclusions: Community leaders in areas with fragile, failing health and health systems are often not organized to strengthen health and human services that contribute to improved access and health status. External neutral facilitation can help local leaders representing hospitals, physicians, public health, clinics, other health and human services, government, business, schools, churches, and philanthropies organize along a continuum from fragmented to integrated local systems. Once the local collaboration reaches a certain level and the ability to learn, plan, and implement exists, local leaders are able to begin to strengthen the local system. Local health system financial viability and clinical relevance are theoretical precursors to access and health status improvement. Implications for Policy, Delivery, or Practice: Health and health care are local. Local health systems provide the foundation on which the overall health care system is built. The overall system consists of national, state, and local public and private programs, policies, and resources. Investment in strong local health care systems can make a difference and is an important investment in the health care system as a whole. Primary Funding Source: Georgia Department of Community Health ●The Effect of Medicare Managed Care Plan Benefit Levels on Use of the VA Medical Care System Robert Morgan, Ph.D., Dolly John, MPH, Jessica Davila Ph.D., Margaret Byrne, Ph.D., Raji Sundaravaradan, BS, Iris Wei, Dr. PH, , Debora Paterniti, Ph.D., Nora Osemene, Pharm.D., Laura Petersen, M.D., MPH Presented By: Robert Morgan, Ph.D., Senior Research Scientist and Associate Professor, Houston Center for Quality of Care and Utilization Studies, Michael E. DeBakey VA Medical Center and Baylor College of Medicine, 2002 Holcombe Boulevard (152), Houston, TX 77030; Tel: (713) 7948635; Fax: (713) 748-7359; Email: rmorgan@bcm.tmc.edu Research Objective: Changes in benefits provided by Medicare managed care plans (Medicare HMOs) may affect VA use among dually eligible veterans, possibly affecting the continuity and quality of their health care. The purpose of this study was to examine the relationship of Medicare HMO plan benefit levels with VA outpatient, inpatient, and ER service use by veterans enrolled in Medicare HMOs. Study Design: We used national VA inpatient and outpatient data for calendar year (CY) 2000 merged with CY 2000 national Medicare enrollment and plan benefit data. Plan benefit packages were categorized as low, medium, and high based on a combination of the physician and hospital co-pays, as well as pharmacy and supplemental benefits (vision, dental, hearing and podiatry). Logistic regression analyses were used to examine any use of the VA during CY 2000, followed by negative binomial regressions to examine magnitude of use among VA users. Analyses were conducted separately for outpatient, inpatient, and ER use, and were adjusted for age, sex, race/ethnicity, VA priority level, income, Medicare state buy-in status and metropolitan versus non-metropolitan location. Population Studied: Analyses were performed on data for elderly (aged 65+) veterans who had any use of the VA health system within the past five years and who were enrolled in an Medicare HMO plan for some part of CY 2000 (N = 282,939). Principal Findings: The level of Medicare HMO plan benefits was significantly and monotonically related to the likelihood that veterans would use the VA system for any of their health care. Veterans enrolled in plans with high benefit and medium benefit packages were less likely to use any VA outpatient services than those enrolled in a plan with low benefit levels (ORs = .68 [.65-.70] and .82 [.79-85], respectively). Interestingly, among veterans who did use outpatient care, those in high and medium benefit plans used slightly more outpatient care than veterans in low benefit plans (ORs = 1.08 [1.06-1.09] and 1.07 [1.05-1.08], respectively). Similarly, veterans enrolled in high benefit and medium benefit plans were significantly less likely to have a VA ER visit than were those in low benefit plans, with enrollees in high benefit plans being the least likely to have a VA ER visit (ORs = .52 [.46-.58] and .83 [.73-.95] for high benefit and medium benefit versus low benefit enrollees, respectively). There was no association between plan benefit levels and VA inpatient use. Conclusions: This is the first demonstration that VA use among Medicare HMO enrolled veterans varies directly with the extent of benefits offered by their plans, particularly for outpatient and ER care. Implications for Policy, Delivery, or Practice: Our findings strongly suggest that changes in Medicare HMO plan benefits are likely to influence whether veterans turn to the VA for care as an alternative or additional provider of care, ultimately affecting resource availability at individual VA medical centers. Primary Funding Source: VA visit. Probit analysis will be conducted to analyze the effect of coverage status and gains or losses of coverage on the probability of a visit. Covariates include coverage status at time of visit, whether an individual gained or lost coverage over the study period, presence of chronic conditions, selfperceived health status, and demographic characteristics such as age, gender, race/ethnicity, and income. Population Studied: A two-year panel of nonelderly adults (age 19-64) constructed from the 2000 and 2001 Medical Expenditure Panel Survey (MEPS), a nationally representative survey of the U.S. civilian, non-institutionalized population. The unweighted sample of 5,737 represents 165,500,334 adults. Principal Findings: Over the two year period, 68% of the panel is continuously insured, 11% is continuously uninsured, and 21% experience one or more transitions in health insurance status. Health status of the three populations does not differ significantly. Adults with a change in insurance status account for 27% of all ED visits, while the continuously insured and uninsured represent 18%. Preliminary results suggest that people with a transition have twice as many visits by the continuously insured or uninsured. Further analysis will stratify ED visits by insurance status at the month of visit, to see if the probability of an ED visit is greater when insured or uninsured, and whether insurance gainers or losers are more likely to have a visit. Conclusions: Preliminary results suggest ED use might not be driven by insurance status as much as it is driven by changes in coverage status, highlighting the importance of stable coverage. Implications for Policy, Delivery, or Practice: Policymakers need to take into account the dynamic aspect of insurance coverage when designing programs aimed at improving the care received by the uninsured, or relieving overcrowding of EDs. Gaining a better understanding of the transitionally uninsured allows us to see if stability of coverage is a more important determinant of ED use than coverage itself. Primary Funding Source: RWJF ●Health Insurance Transitions and Emergency Department Utilization Karoline Mortensen, MA, Catherine McLaughlin, Ph.D. ●Managing Care for Uninsured Patients: Five Success Stories from America’s Public Hospitals and Health Systems Christina Moylan, MHS, Lynne Fagnani, MBA Presented By: Karoline Mortensen, MA, Health Services Organization & Policy, University of Michigan, 4000 Purdue #149, Houston, TX 77005; Tel: (713)661 1526; Email: karoline@umich.edu Research Objective: Recent findings in the academic literature suggest that, contrary to conventional wisdom, the uninsured do not use the emergency department (ED) more than the insured. Previous studies focus on the utilization of the continuously insured and continuously uninsured, yet a gap exists in studying people who transition into or out of coverage, their coverage status at the time of ED visit, and how ED utilization may vary depending on their insurance status. The goal of this project is to examine ED utilization for nonelderly adults who experience a transition (gain or loss of insurance) in health insurance status, relative to the ED use of the continuously insured and continuously uninsured. Study Design: ED visits of adults followed over a two year period are analyzed. Monthly health insurance coverage status is linked to each episode of care. The unit of analysis is the ED Presented By: Christina Moylan, MHS, Policy Advisor, National Association of Public Hospitals and Health Systems, 1301 Pennsylvania Avenue, NW, Suite 950, Washington, DC 20004; Tel: (202) 585-0100; Fax: (202) 585-0101; Email: cmoylan@naph.org Research Objective: Collect descriptive information on innovative public hospital and health systems programs that use managed care principles in delivering health care services to uninsured patients. Identify common programmatic strategies and policy recommendations based upon the research. Study Design: Key informant telephone and on-site interviews with five public hospitals and health systems with programs to manage care for the uninsured. Hospitals were selected based upon the age of the program, use of managed care principles with uninsured patients, and demographic and geographic diversity. The interview protocol included general background, eligibility criteria, enrollment, enrollee demographics, services and utilization, network structure and provider payment, care management strategies, and program costs, funding, and administration. Population Studied: Boston Medical Center in Boston, MA (Care Net Plan); Health and Hospital Corporation of Marion County in Indianapolis, IN (Health Advantage); Jackson Memorial Hospital in Miami, FL (Trust Care); Parkland Health & Hospital System in Dallas, TX (Parkland HEALTHplus); and VCU Health System Authority in Richmond, VA (Virginia Coordinated Care for the Uninsured). Principal Findings: While these models apply managed care principles to caring for uninsured patients, they differ from Medicaid and commercial managed care plans in that they are not insurance products – there is no separate payment source and no defined benefit package. The upper income eligibility level ranges from a low of 150 percent of the FPL to a high of 200 percent of the FPL. All offer enrollees access to a comprehensive set of services, including primary, specialty, and inpatient care and operate using existing provider networks. Care management strategies are targeted at primary care, inpatient care management, discharge planning, and chronic disease management. Costs are difficult to quantify given the lack of a dedicated funding stream. Conclusions: These models appear to be having a positive impact on the continuity and quality of care for uninsured patients. They offer interesting lessons learned for other safety net providers and policymakers. Additional research is needed to fully understand the impact of these programs on continuity, quality of care, and health outcomes for uninsured patients. Research is also needed to better understand the costs and benefits of this approach. Implications for Policy, Delivery, or Practice: For safety net providers, these models offer valuable lessons learned about how to create and administer these programs. For policy makers, these models demonstrate that public hospitals and health systems continue to fill the gap left by existing public insurance programs and that it is essential to support these efforts. There is also an opportunity to explore the benefits of using health information technology (HIT), such as electronic medical records, to manage care for uninsured patients. HIT may be particularly beneficial with this population given the high prevalence of chronic disease and the needed coordination among providers. Primary Funding Source: National Association of Public Hospitals and Health Systems ●Unworried Parents of Well Children: a Look at Uninsured Children who Reportedly do not Need Health Insurance Kathleen S. O'Connor, MPH, Stephen Blumberg, Ph.D., Genevieve Kenney, Ph.D. Presented By: Kathleen S. O'Connor, MPH, Survey Statistician, Division of Health Interview Statistics, Centers for Disease Control and Prevention, National Center for Health Statistics, 3311 Toledo Road Room 2114, Hyattsville, MD 20782-2003; Tel: (301)458-4181; Fax: (301)458-4035; Email: kdo7@cdc.gov Research Objective: To examine the characteristics of lowincome uninsured children whose parents reported that health insurance coverage was not needed. Study Design: This study uses data from the 2001 National Survey of Children with Special Health Care Needs, conducted by the National Center for Health Statistics, Centers for Disease Control and Prevention, on behalf of the Maternal and Child Health Bureau of the Health Resources and Services Administration. This random-digit-dial telephone survey screened 373,055 children under 18 years of age for special health care needs. In each household, up to one child with special needs and one child without special needs were randomly sampled for an interview that included questions about health insurance coverage, health status, health care access and utilization, and awareness of, experience with, and intention to enroll in Medicaid or the State Children’s Health Insurance Program (SCHIP). The survey respondent was the parent or guardian who was most knowledgeable about the child’s health. Population Studied: Uninsured children from low-income households whose parent or guardian said that public or private health insurance coverage was not needed. Principal Findings: When asked why their children did not have health insurance, parents of 6.8% of the low-income uninsured children gave either primary or secondary answers that were categorized as “don’t need insurance/don’t get sick”. Non-Hispanic children of American Indian or Alaska Native (AIAN) race were more likely than low-income uninsured children from any other racial or ethnic group to reportedly not need insurance (15.2%). Hispanic children were also more likely than non-Hispanic white children to reportedly not need insurance (8.9% and 3.2% respectively). After adjusting for the language of interview and sociodemographic characteristics, the relationship of Hispanic ethnicity to reported need for insurance was not statistically significant. Instead, with both language of interview and ethnicity in the regression model, the model revealed that language of interview had a stronger relationship to reports of not needing insurance. Many children who reportedly did not need insurance were not receiving the preventive care that is recommended by pediatricians and did not have a medical home. Roughly half of the children reportedly not needing insurance had parents who had heard of Medicaid or SCHIP and said they would enroll their child in Medicaid or SCHIP if told they were eligible. Conclusions: Parents’ perceptions of the need for insurance are related to children’s health status and health care use. Lack of awareness of preventive care needs for healthy children, of the availability of low-cost public health insurance programs, and of the eligibility of most children from lowincome households were related to sociocultural characteristics. Implications for Policy, Delivery, or Practice: The majority of children reportedly not needing insurance coverage had parents who said they would enroll them in Medicaid or SCHIP if told their child was eligible. However, for parents of the small fraction of uninsured children who were generally reported to be healthy and not need insurance, learning about eligibility might not be sufficient to motivate participation in Medicaid, SCHIP, and other insurance programs. Primary Funding Source: HRSA, DHHS/ASPE ●Does Unauthorized Residency Status Determine Mexican Immigrants' Access to Employer-Based Health Insurance? Victoria D. Ojeda, Ph.D., MPH, Enrico Marcelli, Ph.D. Presented By: Victoria D. Ojeda, Ph.D., MPH, NIMH PostDoctoral Research Fellow, Health Care Policy, Harvard Medical School, 180 Longwood Avenue, Boston, MA 02115; Tel: (617)432-0819; Email: ojeda@hcp.med.harvard.edu Research Objective: To determine whether unauthorized residency status explains low rates of employment-based health coverage among unauthorized immigrants or whether this outcome is predicted by other sociodemographic, employment, or community characteristics. Study Design: This study is based on data obtained from the 1994 and 2001 Los Angeles County Mexican Immigrant Residency Status Surveys (LAC-MIRSS) and the March 19942003 Current Population Surveys (CPS). The LAC-MIRSS data include information on adult respondents' residency status, health insurance coverage, and other sociodemographic and employment data. We first apply demographic predictors of unauthorized residency status obtained from the 1994 and 2001 LAC-MIRSS (e.g., age, sex, years residing in the United States, educational attainment) to the noncitizen foreign-born Mexican adult population in the 1994-2003 March CPS to estimate the proportion of unauthorized Mexican immigrants who had employer-sponsored heath insurance. We next analyze how unauthorized residency status influenced the probability of having had job-based coverage using logistic regression models and controlling for other sociodemographic, employment, and neighborhood characteristics. Population Studied: The 1994 LAC-MIRSS sample includes 558 foreign-born Mexican adults residing in one of 275 randomly selected households within one of nine census tracts in Los Angeles County. The 2001 LAC-MIRSS sample includes 780 foreign-born Mexican adults residing in one of 456 randomly selected households within one of 125 census blocks in Los Angeles County. Principal Findings: Bivariate analyses demonstrate significant differences in health insurance coverage rates by residency status. Overall, more than two-thirds of unauthorized immigrants are uninsured and about one-third have some form of private or public health insurance. When data are disaggregated by source of coverage, we find that unauthorized immigrants have the lowest rates of employerbased coverage of the groups analyzed, with only about 1 of every 10 covered through their workplace. However, results from logistic regressions suggest that unauthorized status was not a significant predictor of employer-based coverage. Rather, unauthorized Mexican immigrants' lower coverage rates through the workplace appear to have been a function of time residing in the United States, educational attainment, occupation, participation in local community groups, and neighborhood homeownership. Conclusions: This study provides new insight into patterns of coverage among one subgroup of Latinos, unauthorized Mexican immigrants. Results provide evidence of vulnerability among working unauthorized immigrants residing in the U.S. Contrary to expectations it is not unauthorized status per se that determines immigrants' access to employer-based coverage, but rather how immigrants are distributed in the labor market, which is likely also due to human capital as measured by their educational attainment, as well as social factors such as civic participation and neighborhood homeownership rates. In addition, length of time in the U.S. may reflect level of integration into mainstream society, ability to navigate workplace bureaucracies, and increased level of experience in the workforce. Implications for Policy, Delivery, or Practice: Continued efforts to reduce the uninsured population and to expand private health insurance coverage to low-wage sectors of the labor market where unauthorized immigrants are likely to be concentrated are needed. Making private coverage more affordable is critical to safeguarding the health and access to care of those individuals and their families as they are likely to include children who are U.S citizens by birth. Primary Funding Source: No Funding Source ●National Data on the Characteristics of Rural Residents Hospitalized in Rural as Compared with Urban Hospitals Maria Owings, Ph.D., Margaret Hall, Ph.D., Jill Marsteller, Ph.D., Maria Owings, Ph.D. Presented By: Maria Owings, Ph.D., Health statistician, Hospital Care Statistics Branch, National Center for Health Statistics, 3311 Toledo Road, Hyattsville, MD 20782; Tel: (301)458-4409; Fax: (301)458-4032; Email: mowings@cdc.gov Research Objective: To compare demographic and clinical characteristics of 5.1 million rural residents discharged from rural hospitals to 2.2 million rural residents who received inpatient care in urban hospitals; and to examine the availability of health services in their counties of residence. Study Design: Data were from the 2001 National Hospital Discharge Survey (NHDS). The survey gathered data on diagnoses and procedures, length of stay, expected payment source, and discharge disposition. Based on diagnosis codes, the Charlson Index was calculated for each patient as a measure of comorbidity burden. In addition, county FIPS codes were used to link NHDS data with the Area Resource File in order to obtain socioeconomic and health service supply data on the counties where rural patients lived. The June 2003 OMB definitions were employed to designate patients' and hospitals' counties as urban (i.e. metropolitan) or rural (all others, including micropolitan and noncore). Population Studied: A nationally representative sample of inpatients in short-stay, nonfederal hospitals in the United States. Principal Findings: Thirty percent of hospitalized rural residents crossed over to urban areas to receive their inpatient hospital care. These patients were younger, had more procedures and longer stays than rural residents hospitalized in rural hospitals. A smaller proportion of rural patients in urban hospitals expected to pay with Medicare. Rural residents hospitalized in urban hospitals were more likely than their counterparts in rural hospitals to have cancer, circulatory (including heart) diseases, nervous system or musculoskeletal disorders, and injuries, and they were less likely to have diseases of the endocrine, respiratory, digestive or genitourinary systems. Rural patients in urban hospitals averaged twice as many procedures performed during their hospitalization as those in rural hospitals. Residential counties of rural patients who crossed over to urban hospitals had fewer active MDs and fewer short-term hospital beds per capita than those of rural patients hospitalized in rural hospitals. Counties of crossover patients also had fewer hospitals with a cardiac intensive care unit. Consistent with the limited availability of certain types of health services within their counties, a relatively large share of rural crossover patients received complex surgical procedures including invasive cardiac procedures during hospitalization. Conclusions: Many rural hospitals provide basic medical and surgical care, but their ability to provide specialized surgery and care for complex medical conditions may be limited. Geographic disparities in access to technologically advanced medical and surgical care may present barriers to the receipt of timely and appropriate care for rural patients, especially the elderly and the poor. Implications for Policy, Delivery, or Practice: Research on existing patterns of inpatient care for rural residents can be helpful to both urban and rural hospitals as they develop collaborative arrangements to ensure efficient use of resources and the provision of high quality care that meets the needs of all rural patients requiring hospitalization. Primary Funding Source: CDC ●Stresses to the Safety Net: The Public Hospital Perspective Marsha Regenstein, Ph.D., Jennifer Huang, MS Presented By: Marsha Regenstein, Ph.D., Director, National Public Health and Hospital Institute, 1301 Pennsylvania Avenue, NW, Suite 950, Washington, DC 20004; Tel: (202) 585-0100; Fax: (202) 585-0101; Email: mregenstein@naph.org Research Objective: Safety net hospitals care for large numbers of uninsured, low-income and vulnerable populations. Given their mission, these hospital systems are extremely dependent upon public sources of financing to deliver high-quality care. The downturn in the economy over the past several years has placed increasing stress on these systems, with limited opportunities for new sources of revenue. This study describes trends affecting the financial viability of safety net hospitals across the country in light of increasing demand for services. The study examines specific challenges facing large safety net hospital systems in eight states. Study Design: A review was conducted of available national surveys documenting utilization and financing of large public hospitals in the United States. Data was drawn primarily from the annual surveys of members of the National Association of Public Hospitals and Health Systems (NAPH) and the American Hospital Association. In-depth interviews were conducted with chief executive officers, chief financial officers, chief medical officers, emergency department directors and other key leaders in safety net hospitals in eight states: California, Florida, Georgia, Minnesota, New Mexico, New York, Ohio and Texas. States were selected to provide variation in terms of geographic location, diversity and size of the uninsured population in the state, and characteristics of the Medicaid program. Population Studied: Approximately 120 public and not-forprofit safety net hospitals that are members of NAPH were included in the study. These hospital systems collectively accounted for nearly 1.5 million discharges, 30 million outpatient visits, and over 5 million emergency department visits in 2002. About 42% of outpatient visits are to patients who are uninsured and another 27% are covered by Medicaid. Principal Findings: Safety net hospitals have implemented numerous programs over the past five years to improve efficiencies and leverage scarce resources on behalf of their patient populations. Increasingly, public hospitals are dependent on federal Medicaid revenues for their financial health and therefore are extremely sensitive to Medicaid changes both at the federal and state level. With increasing demand for services due to the growing numbers of uninsured and underinsured individuals, CEOs and other hospital leaders indicate that they have few options for increasing revenue from other payer sources. Most juggle the need for direct care with the need for capital improvements, including advances in information technology. Conclusions: Federal sources of financing, primarily from Medicaid, are the single most important source of revenue for safety net hospitals. Opportunities for improvements in safety and efficiency rest with IT enhancements, although investments in IT are difficult given heightened financial stresses. With growing demand from uninsured, underinsured, undocumented, and otherwise vulnerable populations, efforts to curtail federal support will have profound and immediate effects on safety net hospitals’ ability to maintain current services within their communities. Implications for Policy, Delivery, or Practice: Communities have always turned to public hospitals for their health care needs. However, without sustained support from public sector financing, public hospitals will be forced to limit service availability, limit eligibility for free or reduced care, or reduce quality of care. Primary Funding Source: Kaiser Family Foundation ●Estimating the Cost of Caring for the Uninsured in California Dylan Roby, M.Phil, Gerald Kominski, Ph.D. 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; Email: droby@ucla.edu Research Objective: This study documents the relative disparities in spending for direct personal health care services between California’s uninsured and insured population (ages 0-64) and provides an estimate of the direct costs that would be incurred if the uninsured were provided with health insurance, while controlling for demographic differences and health status. Study Design: The 1998-2000 Medical Expenditure Panel Survey (MEPS) was used in conjunction with the 2001 California Health Interview Survey (CHIS) in order to model direct medical expenditures in the state of California based on patient characteristics, health status, and various other measures. Based on these direct expenditure models, we were then able to project how much in direct expenditures would be needed to cover the uninsured in the state. Population Studied: Californians aged 0 to 64. Principal Findings: The majority of personal health care expenditures in the state come from full-year insured adults, with $43.1 billion (69%) spent from private insurance, public insurance, out-of-pocket costs, and other sources. Insured children account for 18% of expenditures, or almost $11.3 billion. In comparison, spending by uninsured children and adults constitutes a combined 12% of expenditures, a total of $7.4 billion. However, if provided with health insurance, the uninsured would spend about $14.8 billion in total on direct health care if fully insured, compared to the $7.4 billion they already spend out-of-pocket or that is paid on their behalf from other public and private sources. This $7.4 billion increase in expenditures represents a 12% increase in total direct personal health care expenditures in the state. The additional expense would be about $1,180 per uninsured person in the state. Conclusions: Covering the uninsured is a significant step in improving the health status and quality of life for all Californians by guaranteeing that no Californian will face access barriers as a result of losing their health insurance. The projections in this policy brief show that an expansion of health insurance would increase direct expenditures. However, it would also provide much needed care to the 6.3 million people who lack health insurance, as well as provide equal access to health care in the state. Implications for Policy, Delivery, or Practice: Shifting resources could enhance the ability of the state to insure the uninsured, improving access to care and the health of Californians. This increase in direct expenditures does not reflect the billions of dollars spent to care for the uninsured through funding from DSH, health center grants, or other indirect subsidies. It will be important to consider the amount of indirect sources of expenditures made on behalf of the uninsured, and to develop options on how all these direct and indirect sources of expenditures might be combined to provide affordable insurance for California’s uninsured population. In light of our estimates, the costs of an insurance expansion may be more affordable if the state can find mechanisms for redirecting current sources of public and private expenditures on behalf of the uninsured. Primary Funding Source: California Endowment ●Do Employers Respond to Public Hospitals Crowding Out of Private Insurance? Eric Seiber, Ph.D. Presented By: Eric Seiber, Ph.D., Assistant Professor, Public Health Sciences, Clemson University, 527 Edwards Hall, Clemson, SC 29634; Tel: (864) 656-6206; Fax: (864) 656-6227; Email: seiber@clemson.edu Research Objective: Multiple studies have found public health care services crowding out private insurance in the United States. A separate literature reports increasing numbers of employers deciding not to offer a group health insurance plan to their employees. However, no study links these two themes in the literature to examine the extent of public insurance crowding out on the supply of employer sponsored insurance plans. The objective of this study is to decompose the public hospital crowd out effect into changes in take-up and changes in the supply of employer sponsored insurance. Study Design: The study uses data from the 1995 Contingent Workers Supplement in the Current Population Survey and data from the 1995 American Hospital Association’s Annual Hospital Survey to estimate a nested multinomial logit model of workers’ insurance choice. A two tier choice set captures both whether the employer offers a group plan and the employee’s choice to participate, purchase on the individual market, or to remain uninsured. Simulations decompose the effect of public hospitals on the offer of employer sponsored plans and the take-up of those plans. Separate simulations are conducted for workers under the poverty line, 100-199%, 200-299%, and those over 300% of poverty. Population Studied: Contingent workers in the 100 largest Metropolitan Statistical Areas. Principal Findings: The study finds public hospitals having a significant crowding-out effect on demand and a modest effect on the supply of employer sponsored insurance. The simulations reveal the most substantial effects for the poor and near-poor. A simulated doubling of public hospitals has the uninsured rate increasing by five percent amongst the poor and a three percent decrease in the offer of employer sponsored plans (3% and 1% for the near poor). Conclusions: The results indicate that employers do respond to the availability of public substitutes, with each one percent of private coverage crowded out by public hospitals being matched by a 0.5-0.66% reduction in the availability of employer sponsored insurance. Implications for Policy, Delivery, or Practice: Faced with a public hospital undergoing an ownership conversion, employer sponsored insurance markets will catch workers that previously used public hospitals, and policy makers should focus their efforts on the unemployed and others with no access to the employer sponsored insurance market. Primary Funding Source: No Funding Source ●Economic Stress and the Safety Net Peter Shin, Ph.D., MPH, Sara Rosenbaum, JD, Julie Darnell, MHSA Presented By: Peter Shin, Ph.D., MPH, Assistant Research Professor, Center for Health Services Research and Policy, George Washington University, 2021 K Street, NW Suite 800, Washington, DC 20006; Tel: (202)530-2313; Fax: (202)2960025; Email: pshin@gwu.edu Research Objective: This paper examines the impact of the recent economic downturn on health centers in selected communities, exploring the effect of declining private health insurance coverage and the elevated unemployment levels among lower wage workers, and assesses how recent increases in federal appropriations relate to growing demand for health center services as the number of low-income, uninsured persons continues to rise, and challenges health centers face as they experience widespread state cutbacks in Medicaid – the single most important source of health center financing. Study Design: The study uses both a quantitative and qualitative approach. The Uniform Data System is used to provide new information on the demographic profile of health center patients and the revenue sources available for financing their care, including recent increases in federal discretionary funding. The 2002 Uniform Data System (UDS), which managed by the federal government, includes data on 843 health centers and 11 million users at more than 4,600 urban and rural sites. The UDS data was also used to estimate the proportion of each state's low-income and uninsured population (along with 2001-2002 Kaiser Commission on Medicaid and 2002-2003 new users of federally-funded health centers from the National Association of Community Health Centers). During 2002 and 2003, interviews were conducted with health center staff in 8 urban and rural communities and representatives from state and regional primary care associations regarding the effects of the downturn on health centers and their patients. Researchers selected urban and rural communities located in states with elevated unemployment rates related to the downturn using data from the U.S. Department of Labor and based on Medicaid generosity levels. Respondents were asked about the impact of economic conditions on their financial support, patient caseload and health status, and overall operations; whether and how their experiences differed from health centers in less affected communities around the state; and what actions if any their state or local governments had taken to help them. Respondents also provided additional data regarding their patients, revenues, and staffing. Population Studied: Given their location and their service missions, health centers operate in the communities most likely to bear the worst brunt of an economic downturn such as the one that experts say began in March 2001. The slowdown has been particularly difficult for lower wage workers with limited job skills, and the very low level of job growth in the current economic recovery suggests that employment hardships for these Americans may continue for some time. In addition to facing a rising number of patients without jobs and who are at increased risk for lack of health insurance, health centers have also had to weather state Medicaid cutbacks and the loss of state and local operating funds, as state budgets have constricted in response to declining revenues. Even where Medicaid-enrolled health center patients do not lose their coverage entirely, coverage for important benefits and services may have been lost. Principal Findings: The roughly 7% increase in federal funding between 2002 and 2003 occurred at the same time as the number of uninsured health center patients is estimated to have risen by an even faster rate of 11.4%. From a national perspective, for every one uninsured, low-income patient that a health center is able to treat, there are an average of four additional low-income, uninsured persons. Recent data suggest that Medicaid is paying significantly more per patient than are grant funds. Medicaid contributed on $450 per year for each Medicaid patient served in 2002. In contrast, grant payments provided only $299 per year for each uninsured patient served in a health center. Therefore, Medicaid coverage remains the largest revenue source for health centers and is the engine that supports health center capacity during an economic downturn. Health center respondents reported that their centers felt the impact of the downturn in two ways. First, respondents found that many patients lost coverage, thereby complicating the provision of their health care and driving down revenue expectations for the centers. Second, respondents experienced a surge in new patients who lost coverage as a result of either job cutbacks or unemployment and were left without affordable health care. Respondents further noted that their new, or newly uninsured, patients tended to experience serious physical or mental conditions, which they attributed either to the ramifications of job loss or to underlying health problems that might have contributed to job loss as the economy softened. Respondents noted that while Medicaid payments were more generous than payment levels from private insurers, revenues realized from newly enrolled patients were inadequate to offset the surge in costs that health centers experienced from having to rapidly expand treatment for uninsured patients. Only in the District of Columbia did the proportion of low-income health center patients exceed 30% of the low-income population. In 20 states, health center penetration into the low-income population was less than 10% of the low-income population. Furthermore, in only 8 states did health centers reach at least 30 percent of the low-income uninsured population. Conclusions: Findings underscore the fact that as essential as health center grants are for supporting care for the uninsured, establishing new health center sites and supporting and improving existing services, the reach of these funds alone is modest in relation to need. Even a program as favorably regarded as the health center program has received federal discretionary appropriations increases that pale in comparison to the level of need, particularly at times when the economy slows, the number of uninsured patients rises, and the severity of problems seen at health centers increases. These findings suggest that for health centers, just as for non-safety net providers, health insurance remains the engine that drives health care capacity and responsiveness. Its impact can be measured both by the comprehensive benefit package it provides health center patients and the cost-based payment rates that it offers specifically to health centers. These rates help ensure that federal discretionary grant funds tied to the uninsured patient population are not spent to offset public program payment shortfalls. Medicaid’s importance is such that even in states with more limited programs and in communities with high immigrant populations, health centers report investing in Medicaid outreach and enrollment assistance, not only to help patients secure coverage but also because of Medicaid’s impact on the centers’ operating capacity. Implications for Policy, Delivery, or Practice: Federal discretionary spending increases are, however, a vital means of expanding the health center program. Such increases help stabilize centers and support care for uninsured patients, allow new health centers to be established and expand existing programs, and support highly beneficial quality improvement activities. But even in the case of a program as popular as health centers, these discretionary funding investments are modest in relation to the number of low-income persons without health insurance. At the same time, the survival of health centers is especially sensitive to federal and state Medicaid policies in all areas: eligibility, enrollment, benefits and coverage, and provider payment rates. Our discussions with local health centers suggest that when economic hardship hits families, even modest increases in the cost of health care will deter them from seeking care for fear of incurring expenses they absolutely cannot afford. Therefore, reforms aimed at improving Medicaid’s reach into the newly unemployed and uninsured lower wage workers are essential to health centers. Primary Funding Source: Kaiser Family Foundation ●Providers’ Lack of Knowledge as a Potential Barrier to Clinical Trial Participation for Patients with Melanoma Karyn Stitzenberg, M.D., MPH, Nancy E. Thomas, M.D., Ph.D., David W. Ollila, M.D. Presented By: Karyn Stitzenberg, M.D., MPH, Fellow, Cecil G. Sheps Center for Health Services Research and Department of Surgery, University of North Carolina, 725 Airport Road, CB# 7590, Chapel Hill, NC 27599-7590; Tel: (919)966-9801; Fax: (919)966-8806; Email: karyn_stitzenberg@unc.edu Research Objective: Proven adjuvant therapies for patients with metastatic melanoma are limited. Consequently, clinical trials are often the best therapeutic option for patients with metastatic melanoma. The purpose of this study is to assess different providers’ awareness of the availability of clinical trials for their patients with metastatic melanoma. Study Design: A self-administered questionnaire was developed to examine the practice and referral patterns of melanoma providers in NC. Several questions specifically addressed use of adjuvant therapy and clinical trials. The survey was pilot tested then administered in three mailings with phone follow-up after the second mailing to maximize response. Pearson’s chi square was used to test associations between provider characteristics and reported access to clinical trials for metastatic melanoma. Provider characteristics examined included: age, gender, race, years in practice, practice type, melanoma patient volume, and affiliation with an academic institution or cancer care network. Population Studied: All licensed NC dermatologists and a subset of licensed surgeons were selected for inclusion. Surgeons were included if, according to a statewide inpatient and ambulatory surgery discharge database, they performed at least three melanoma procedures between 1998 and 2001. Principal Findings: Overall response rate was 60%; 147/232 dermatologists and 116/206 surgeons. 3% of dermatologists reported they enroll patients in trials, 66% reported access to another physician who enrolls patients, and 28% reported no access to clinical trials for metastatic melanoma. 11% of surgeons reported they enroll patients in trials, 56% reported access to another physician who enrolls patients, and 31% reported no access to clinical trials for metastatic melanoma. Even among surgeons who reported that they perform surgical nodal staging for patients with melanoma, 29% reported no access to trials. Surgeons who were affiliated with an academic center and those who had seen >2 melanoma patients in the past year were more likely to report access to clinical trials than those without academic affiliations and those who saw <3 melanoma patients in the past year (87% vs. 65%, p=0.04 and 77% vs. 53%, p=0.009, respectively). In contrast, for dermatologists, none of the factors examined were associated with reported access to clinical trials. Conclusions: For many patients with metastatic melanoma, clinical trials may be the only adjuvant therapy option. Despite this, roughly one third of melanoma providers surveyed reported they had no access to clinical trials for patients with metastatic melanoma. Providers’ lack of knowledge of available options may present a significant barrier to clinical trial participation for patients with melanoma. Implications for Policy, Delivery, or Practice: Educational efforts directed at providers should be undertaken to improve access to clinical trials for patients with metastatic melanoma. Primary Funding Source: AHRQ ●Safety-Net Hospitals: Serving the Uninsured Donald Stull, Ph.D., Roxanne Andrews, Ph.D., Irene Fraser, Ph.D., Bernard Friedman, Ph.D. Presented By: Donald Stull, Ph.D., Research Scientist, MEDTAP, 7101 Wisconsin Ave, Suite 600, Bethesda, MD 20814; Tel: (301)654-9729; Email: donald.stull@unitedbiosource.com Research Objective: In 2000, the Institute of Medicine called for improved monitoring of the structure, capacity, and financial stability of the hospital safety net to meet the health care needs of the uninsured and other vulnerable populations. This analysis uses national data to describe how the burden of caring for the uninsured is spread across the U.S. hospital system and focuses on the characteristics of hospitals that care for a higher proportion of the uninsured. Study Design: The analysis uses discharge-level data from the 2001 Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample, supplemented with hospitallevel data from the American Hospital Association Annual Survey of Hospitals and information from Medicare cost reports. This combination of data yields a unique and comprehensive picture of safety-net hospitals in the U.S. Population Studied: This study provides national estimates for all US community hospitals and their patients based on a weighted 20 percent sample of community hospitals (nearly 1,000 hospitals and 7 million discharges). Principal Findings: The study found that hospital care for the uninsured is distributed unequally across U.S. hospitals. Ten percent of hospitals accounted for approximately one third of uninsured discharges: we defined these as “safety-net” hospitals. Another 20% of hospitals accounted for an additional one third of uninsured discharges. The remaining 70% of hospitals provided care for the other one third of uninsured discharges. The analysis found that safety-net hospitals are more likely to be small (less than 100 beds), non-teaching, and to treat a higher share of Medicaid patients and fewer privately insured patients. They also treat a mix of patients with different conditions. This patient mix makes it more difficult for them to cross-subsidize care for the uninsured. As a result, a larger proportion of safety-net hospitals experience financial losses. Losses were more pronounced among non-teaching safety-net hospitals where over 40% experienced a loss compared to 25% of the teaching safety-net hospitals. Conclusions: Safety-net hospitals provide critical access to uninsured patients and other vulnerable populations. Because of their typically small size, small profit margins, and patient mix, they are among the most vulnerable of U.S. hospitals. Implications for Policy, Delivery, or Practice: As the number of uninsured remains worrisome and persistent, the availability of a strong safety net is vital. Given the financial vulnerability of the safety-net hospitals, communities may want to examine how they are paying for the care of the uninsured. They also may want to consider developing rewards for hospitals that take on the role of safety-net provider or risk the possibility that some of these providers will not be able to provide services with continuing losses. Primary Funding Source: AHRQ ●National Underutilization of Cardiac Rehabilitation Postacute Myocardial Infarction Among Elderly Medicare Patients Jose A. Suaya, M.D., MBA, MPH, Donald S. Shepard, Ph.D., William B. Stason, M.D., MS, Grant Ritter, Ph.D., Sarita Bhalotra, M.D., Ph.D., Gail K. Strickler, MA, MS Presented By: Jose A. Suaya, M.D., MBA, MPH, The Schneider Institute for Health Policy, Heller School, Brandeis University, 415 South Street, MS 035, Waltham, MA 024549110; Tel: (781)736-3904; Fax: (781)736-3928; Email: jsuaya@brandeis.edu Research Objective: Cardiac rehabilitation (CR) as part of the management of coronary heart disease is recommended by the U.S. Department of Health and Human Services, and many medical professional associations. Medicare reimburses for outpatient CR up to three sessions per week for 12 weeks (36 sessions) following acute myocardial infarction (AMI), coronary artery bypass graft (CABG), or stable angina pectoris. This study analyzes national utilization of CR within 1 year after hospital discharge for AMI in elderly Medicare patients, and describes unadjusted and adjusted variations of CR use according to patients’ characteristics. Study Design: Using Medicare claims, we identified a retrospective cohort of patients with AMI in 1997. The individual patient was the unit of analysis. Outcome variables were use, onset, amount, and completion of CR within one year after discharge. Independent variables included sex, age group, race/ethnicity, socio-economic and education status, severity of the hospitalization, management of the AMI, type of discharge, major Census region, state of residence, and type of patient’s residence (urban, rural). Univariate, bivariate and multivariate analyses (binary logistic regression) were used to determine unadjusted and adjusted rates of CR use among candidates. Population Studied: Patients included fee-for-service Medicare beneficiaries aged 65 and above enrolled in both Parts A and B, who had a first hospital admission in an U.S. non-federal acute hospital in 1997 for AMI, a stay shorter than 32 days, and who survived at least 30 days after discharge. We identified a cohort of 203,685 such individuals, who were considered candidates for CR for preliminary analyses. Extensions to 400,000 additional candidates are planned for CR from other indications. Principal Findings: Preliminary findings follow. Of the cohort, 27,537 (13.5%) received some outpatient CR within the first year after discharge. Seventy one percent of users started CR within two months after discharge; the average time from discharge to start was 56 days; the average user received 24 sessions. Thirty-six percent of the CR users completed fewer than 19 sessions and 20.5% completed 36 sessions. Utilization rates were 16.0% among males, 14.4% among whites, and 19.4% among patients younger than 75. Those rates differed from those observed among females (10.8%), blacks (4.9%), and patients older than 84 (2.5%). Mean days from discharge to initiation of CR services also differed by sex (males 54 vs. females 61 days), race (whites 56 vs. blacks 68 days) and age (56 days among patients younger than 75 years vs. 53 days among patients aged 85 and above). Conclusions: Despite national guidelines recommending the use of CR, our analysis suggests that CR post-AMI is underused among elderly Medicare beneficiaries. On average one out of 8 candidates received CR in 1997 and recent rates remain low. CR use varies markedly by patients’ demographic and socio-economic characteristics and follows utilization patterns of other preventive services. Implications for Policy, Delivery, or Practice: Convincing evidence supports the effectiveness of CR services in reducing mortality and morbidity in patients with heart disease. Reasons for this apparent underutilization of CR among post AMI patients should be explored, including assessment of physician opinions about CR, capacity in CR programs, and the adequacy of Medicare reimbursement. Primary Funding Source: CMS ●"It's Broke. Can We Fix It?": a Community Forum Theatre Dialogue on Disparities in Acces to Health Care John Sullivan, MA, Mathew Stanford, BA Presented By: John Sullivan, MA, Co-Director: Public Forum & Toxics Assistance, Preventive Medicine, Community Health, National Institute of Environmental Health Sciences, Sealy center for Environmental Health & Medicine, 301 University Boulevard, Galveston, TX 77555; Tel: 409-747-1246; Fax: 409772-1790; Email: josulliv@utmb.edu Research Objective: Purposes of the Forum included: 1) using images and performative tools of TO to analyze social, economic and cultural implications of health care disparities, 2) framing issue experientially to illustrate impacts of service denial, and encourage audience empathy, 3) using interactive facets of TO to encourage participatory dialogue and problemsolving among audience members and performers. (Dialogue was expressed as actions by audience volunteers who entered scene(s), replaced the oppressed protagonist or allied with protagonist, and proposed solutions through stage actions.) Study Design: A workshop convened for training in Theatre of the Oppressed dramaturgy. Participants learned: 1) how image tableaus may represent affective and conceptual elements of the issue, 2) uses of dramatic structure to sequence and focus personal stories, 3) development of scenarios through improvisation, 4) uses of special structures to deconstruct significance of scenic actions, and 5) using audience interaction to invite dialogue, novel actions and processing of spectator interventions. The show was performed for community members in an informal space, using an issue-specific sociometric exercise as a warm-up. Audience selected one scene for the Forum process, and spectators initiated interventions. Interventions were processed, and closure scene was performed by original protagonist. Population Studied: 15 Workshop participants varied by SES / Race / Ethnicity but all had direct experience with denial of access to health care. Audience (64) varied across categories, but medical students and health care professionals composed >50%. Principal Findings: The following perceptions, biases, attitudes and differentiating characteristics emerged: 1) <12% of audience had direct experience with denial of care, 2) audience interventions express judgement that UTMB’s Demand & Access Management Program (DAMP) for rationing care poses a problem for medical ethics, 3) health care access disparities compromise the social fabric of communities, 4) strong ameliorative action is necessary to “transform the system,” 5) none of the proposed interventions gave the scenes satisfying closure. Conclusions: This Theatre of the Oppressed Forum provided a dynamic structure for dialogue and predisposed audiences toward direct/authentic engagement with real life aspects of a difficult and frustrating issue. Though no clear solutions emerged, the audience demonstrated active empathy for the uninsured, and political will to engage public policy-makers. All participants, including primary care physicians, voiced an urgent need for more detailed understanding of DAMP. Audience members generally agreed that DAMP should be simplified and infused with social justice. Implications for Policy, Delivery, or Practice: Care consumers and providers need basic information on mechanics of DAMP – and similar systems – as well as the fiscal parameters behind health care resource allocation. The health care community also needs to engage in further dialogue to develop ethical guidelines for navigating the health care crisis that transcend free market pragmatism, the designation of health care as solely a commodity, and moves the dialogue into the realm of disparities in social justice (eg. John Rawls’s difference principle, per audience member suggestion). Primary Funding Source: N/A, Institute for the Medical Humantities @ UTMB / Health Care for All Texas (HCFAT) ●Participation in Breast Cancer Screening Through the National Breast and Cervical Cancer Early Detection Program, 2001-2002. Florence Tangka, MS, Ph.D., Sajal C. Chattopadhyay, Ph.D., Joseph Dalaker, MMP, James G. Gardner, MSPH, Janet Royalty, MS, Donald K. Blackman, Ph.D. Presented By: Florence Tangka, MS, Ph.D., Economist, Division of Cancer Prevention and Control, Centers for Disease Control and Prevention, 4770 Buford Highway NE, MS K-55, Atlanta, GA 30341; Tel: (770)488-1183; Fax: (770)4884639; Email: fbt9@cdc.gov Research Objective: To estimate the number of women eligible for NBCCEDP-funded breast cancer screening and the percentage of them screened during 2001 and 2002. Study Design: NBCCEDP data were used to obtain the number of women who received NBCCEDP-funded breast cancer screening. Data from the Annual Social and Economic Supplement (ASEC) to the Current Population Survey (CPS) were used to estimate the number of women eligible for breast cancer screening through the NBCCEDP. NBCCEDP data and CPS estimates were used to calculate the percentages of eligible women, aged 40-64 and 50-64, at both the national and state levels, who received an NBCCEDPfunded breast cancer screening. The rates are reported for overlapping age groups because some states screen women who are between the ages of 40 to 64 years, while other state screen women who are between the ages of 50 to 64 years. Population Studied: Women eligible(Low-income uninsured women, aged 40-64 years old)for breast cancer screening through the National Breast and Cervical Cancer Early Detection Program. Principal Findings: An estimated 7.3% (3.3 million) of all U.S. women aged 40-64 were eligible for breast cancer screening through the NBCCEDP. Approximately 12.6% of programeligible women aged 40-64 years and 20.9% of program- eligible women aged 50-64 years received a breast cancer screening via mammography through the NBCCEDP. Screening rates varied substantially by state but not by racial and ethnic group. For the 40-64 age group, the lowest state level screening rate was 1.1%, and the highest screening rate was 52.8%. Conclusions: With current congressional appropriations and screening mechanisms, rates of NBCCEDP-funded mammography screening were 12.6% among program-eligible women aged 40-64 and 20.9% among those aged 50-64. These findings underscore the need for additional resources and new strategies for recruitment and service delivery. Implications for Policy, Delivery, or Practice: The estimates reported in this article provide a valuable resource for the CDC and its state partners for program assessment and planning strategies to increase mammography screening among lowincome, uninsured women. Primary Funding Source: No Funding Source ●An Evaluation of Contextual Factors in Breast and Cervical Cancer Screening Anne Tomolo, M.D., MPH, David Litaker, M.D., Ph.D., David Aron, M.D., MS Presented By: Anne Tomolo, M.D., MPH, Chief of Admitting, Medicine, Louis Stokes Cleveland DVAMC, 10701 East Boulevard, 170A (W), Cleveland, OH 44106; Tel: (216)2313292; Fax: (216)421-3008; Email: anne.tomolo@med.va.gov Research Objective: Screening tests for cervical and breast cancers are valuable in maintaining women’s health because of their association with lower mortality through early detection and treatment. Although having a usual source of care and health insurance are characteristics of the individual associated with higher screening rates, the extent to which characteristics of the local health care system (HCS) and the social and economic context in which women live also influence receipt of these services has not been fully explored. We examined the independent association between contextual factors and breast and cervical cancer screening in women. Study Design: The Ohio Family Health Survey, a crosssectional survey conducted in 1998 by telephone with a 53.1% response rate assessed demographics, insurance status, socioeconomic status, and health care needs of Ohio residents. These data were linked at the county-level using data from the 1998 Area Resource File and the 1990 U.S. Census. Guided by the Behavioral Model of Health Care Utilization, we identified operational measures for individuallevel and contextual characteristics potentially associated with our outcome measure, the receipt of either breast or cervical cancer screening in age-eligible women. Characteristics reflecting the social and economic context included urbanization, residence in an Appalachian county, and the proportion of 1) residents living in poverty, 2) residents completing high school, and 3) households headed by a woman. HCS attributes included the supply of primary care physicians (PCPs), the proportion of all physicians engaged in primary care, the presence of a residency-training program, designation as a medically underserved county, and the level of managed care activity (MCA). Nested hierarchical logistic regression models tested associations between contextual factors and receipt of either breast or cervical cancer screening tests, adjusting for individual-level characteristics. Population Studied: A state-representative sample of 8473 women between the ages of 18-75 years old, living in urban, suburban, and rural settings throughout Ohio. Principal Findings: The weighted proportion of women screened by either test was 60.1% (N=5094). There were significant bivariate associations between receipt of screening and individual characteristics: having a high school diploma, continuous insurance, higher family income, a usual source of care, being married/coupled, current employment, and better rating of general health; and county characteristics: greater supply of PCPs, presence of a physician-training program, greater MCA, and increased urbanization (all p <0.001). After accounting for both individual and contextual characteristics, the following county characteristics were independent predictors of receipt of either screening procedure: MCA (adjusted Odds Ratio [AOR], 1.01 (1.00-1.01); p<0.01) and poverty (AOR, 0.97 (0.95-0.99); p<0.01). Significant associations remained with an individual’s current health status, insurance status, education, and having a usual source of care. Conclusions: Characteristics of the HCS and those reflecting economic context are associated with the likelihood a woman will receive cancer screening services. Implications for Policy, Delivery, or Practice: Although we cannot infer the mechanisms through which this effect occurs, our findings underscore the importance of context on the health of the individual in developing more effective future interventions that improve delivery of preventive services to women. Primary Funding Source: VA ●The Association of Distance to Care with Severity of Disease, Length of Stay and Charges for Persons Hospitalized with HIV/AIDS Roberto Vargas, M.D. MPH, Paul Robinson, Ph.D., Susan Ettner, Ph.D., William Cunningham, M.D., Ph.D. Presented By: Roberto Vargas, M.D. MPH, Assistant Professor, General Internal Medicine and Health Services Research, UCLA School of Medicine, 911 Broxton Avenue, Los Angeles, CA 90024; Tel: (310) 794 3703; Fax: (310) 794-0372; Email: RBVargas@mednet.ucla.edu Research Objective: Our study examines the association of patient residential distance to the nearest HIV/AIDS medical provider and nearest ancillary services site with severity of disease, length of stay and hospital charges for patients with HIV/AIDS admitted to acute care hospitals Study Design: We combined data from the California Office of Statewide Health Planning and Development 2000 discharge files, the AIDS Project Los Angeles list of HIV services, and the 2000 Census. Our main predictors were distances from the patient’s population-based zip code centroid to the nearest HIV/AIDS medical provider site and nearest ancillary care service site. Our outcomes were dichotomized severity of disease indexes for conditions present at the time of hospitalization: 1) Deyo Charlson 2) Dartmouth Manitoba (DM) Charlson scores and 3) the Severity Classification for AIDS Hospitalizations (SCAH); length of stay and hospital charges were additional outcomes. We used multiple logistic and linear regression to control for differences in patient age, gender, race, and insurance status; and levels of poverty, English language proficiency, racial and ethnic segregation, education level and number of hospital beds in the patient’s neighborhood of residence. Generalized Estimating Equations were used to adjust the variance for clustering by zip-code Population Studied: We identified 3543 hospitalizations for persons with HIV/AIDS in Los Angeles County in 2000. Patients were 85% male, 56% white, 29% black, and 37% Hispanic irrespective of race. Insurance coverage was 26% Medicare, and 47% Medicaid and other indigent care programs. Patients came from communities where, on average, 35% of people had less than a high school education and the ratio of households making less than $20,000 to those making over $60,000 was 2:1 Principal Findings: Mean patient’s distance from care was 0.68 miles (range 0.03-13.3; s.d.0.82) from ancillary care services and 1.81 miles (range 0.05-15.8; s.d.1.5) from medical care services. Multiple logistic regressions suggested that a greater distance to ancillary services was associated with a significantly lower disease severity index among patients admitted to acute care hospitals for both Charlson indexes, Deyo: Odds Ratio (OR) 0.84, 95% Wald Confidence Interval (CI) (0.73, 0.97) and DM: OR 0.86 CI (0.75,0.99) but not for the SCAH index: OR 1.04 CI (0.93,1.16). Multiple linear regressions suggested that greater distance from ancillary care services was associated with a significantly greater length of stay (estimate 0.078; p=0.002) but not with hospital charges (estimate 0.044; p=0.120). Regression models did not show a significant association between distance to medical care provider and length of stay, charges or any severity indexes Conclusions: Our results suggest that greater distance to ancillary, but not medical, HIV services is associated with a lower disease severity at admission and a longer length of hospital stay during admission, implying a lower clinical threshold for admission and a greater threshold for discharge for patients from communities that are further away from ancillary HIV care services Implications for Policy, Delivery, or Practice: Improving geographic access to ancillary services for persons with HIV/AIDS may reduce hospitalizations and length of stay for these patients Primary Funding Source: National Center on Minority Health and Health Disparities ●Surgical Release of Work-Related Carpal Tunnel Syndrome: Geographical Variation in Utilization and Associated Post-Surgical Outcomes Radoslaw Wasiak, Ph.D., Glenn Pransky, M.D., MOccH Presented By: Radoslaw Wasiak, Ph.D., Research Scientist, Center for Disability Research, Liberty Mutual Research Institute for Safety, 71 Frankland Road, Hopkinton, MA 01748; Tel: (508)497-0242; Fax: (508)435-8136; Email: radoslaw.wasiak@libertymutual.com Research Objective: Medical treatment of compensated work-related conditions has two objectives—improve injured workers’ health status and allow safe and sustained return to work (RTW). Theoretically, the choice of treatment method should be based primarily on these objectives. Surgical treatment of work-related carpal tunnel syndrome (CTS) provides an opportunity to evaluate whether this occurs. The traditional method of open release has been complemented by an endoscopic procedure, particularly useful in work-related cases due to the anticipated benefit of earlier RTW. The objective of this study was to investigate the differences in surgical treatment for work-related CTS across eight U.S. states, and the associated post-surgical work disability outcomes. Study Design: From all workers’ compensation claims (WC) reported to a single insurer during the 1995-1999 period, we identified individuals with a one or two surgical procedures for work-related CTS. Work disability duration was divided into pre- and post-surgical duration according to the date of surgery. Work disability duration included compensated lost work days and compensated days at work in partial capacity. Population Studied: Individuals with surgical release of compensated work-related carpal tunnel syndrome. Principal Findings: Among selected individuals (n=4,421), about 20% were treated using the endoscopic procedure; this percentage had a ten-fold variation across the eight jurisdictions. However, utilization of endoscopic release did not increase during the study period, despite reports of better RTW outcomes. We found no statistically significant differences in total work disability between alternative surgical procedures; however, in all jurisdictions, those with endoscopic releases had less lost work days (p < .01) and more days at work in partial capacity (p < .01). Conclusions: The highly jurisdictional nature of the U.S. WC system, with significant differences in reimbursement levels for endoscopic procedures, and geographical differences in medical training were among the potential contributors to the observed variation in utilization. The benefits of the endoscopic technique show in redistribution of work disability from total work loss to light duty arrangements. Implications for Policy, Delivery, or Practice: Earlier RTW, even in partial capacity, has been shown to increase patients’ likelihood for sustained long-term RTW. Given the potential to increase the rate of early RTW in patients with a surgical release of carpal tunnel, substantial short- and long-term gains can be achieved by increasing the rate of endoscopic releases. Primary Funding Source: No Funding Source