Health Insurance Markets & Managed Care Call for Papers New Perspectives on Market Competition & Health Plan Design Implications for Policy, Delivery, or Practice: Policies aimed at controlling prices and spending in the private sector need to recognize the role of market forces. Primary Funding Source: GAO ●The Contribution of Medicaid Managed Care to the Increasing Undercount of Medicaid Beneficiaries in the Current Population Survey Arpita Chattopadhyay, Ph.D., Andrew B. Bindman, M.D. Chair: Dennis Scanlon, Pennsylvania State University Sunday, June 25 • 8:30 am – 10:00 am ●Competition and Other Factors Linked to Wide Variation in Health Care Prices Christine Brudevold, Ph.D., A. Bruce Steinwald, Michael Kendix, Jennifer Rellick, Leslie Gordon, Ann Tynan Presented By: Christine Brudevold, Ph.D., Assitant Director, Health Care, GAO, 441 G Street, NW, Washington, DC 20548; Tel: 202-512-2669; Fax: 202-512-4778; Email: BrudevoldC@GAO.GOV Research Objective: This study examined prices and spending in the Federal Employees Health Benefits Program (FEHBP) Preferred Provider Organizations (PPOs) to determine (1) the extent to which hospital and physician prices varied geographically, (2) which factors were associated with geographic variation in hospital and physician prices, and (3) the extent to which hospital and physician price variation contributed to geographic variation in spending. Study Design: We compared hospital and physician prices among metropolitan areas in the United States using FEHBP health claims data from 2001. We grouped claims by the metropolitan area where care was delivered and adjusted claims for the cost of providing services and differences in case mix. To determine which factors might be associated with geographic differences in price, we examined the relationship between price and indicators of market competition, health maintenance organization price bargaining leverage and cost-shifting pressures for each metropolitan area. To examine how prices affected spending, we calculated measures of utilization and spending for metropolitan areas, and compared the contribution of price and utilization to spending in the metropolitan areas in highest and lowest spending quartiles. Population Studied: Enrollees and their dependents in the FEHBP who were under age 65 and participated in one of the study PPOs. Principal Findings: FEHBP PPOs paid substantially different prices for hospital inpatient physician services across metropolitan areas. The variation in prices appeared to be affected by market characteristics. In general, less competition and less HMO capitation were associated with higher prices. For hospital and physician services, price contributed to about one-third of the variation in spending between metropolitan areas in the highest and lowest spending quartiles. Conclusions: Understanding price variation is essential to understanding why some areas have higher or lower spending in the private sector. Market forces, not just the underlying costs of doing business, help to determine the prices FEHBP PPOs ultimately pay hospitals and physicians. Presented By: Arpita Chattopadhyay, Ph.D., Senior Statistician, Medicine, Division of General Internal Medicine, University of California at San Francisco, Box 1364 Parnasus Avenue, San Francisco, CA 94143; Tel: (415) 206-6166; Fax: (415) 206-5586; Email: achat@medsfgh.ucsf.edu Research Objective: To determine whether the penetration of Medicaid managed care is associated with the magnitude of the underestimate of Medicaid beneficiaries in the CPS. The Current Population Survey (CPS) is an important source of data for comparing beneficiaries across insurance groups. However, the CPS routinely underestimates the Medicaid population, which correspondingly results in an overestimate of one or more of the other insurance categories. Not only is the underreporting substantial, several studies indicate that the extent of Medicaid underreporting in CPS has been rising over time. Study Design: Pooled cross-sectional comparison of survey and administrative databases for the years 1995-1999, corresponding to a period of substantial Medicaid managed care expansion. We compared the CPS estimates of Medicaid person-years in California in 1995-1999 with the gold-standard number derived from the Medicaid eligibility file for the same time period and examined the association between the CPS underestimate and penetration of managed care in the beneficiary’s county, controlling for secular changes in the design and administration of the CPS and Medicaid policy that may impact Medicaid reporting in the CPS. A mixed model methodology, which accounts for clustered observations, was used to estimate the association between CPS undercount and Medicaid managed care. Population Studied: All California Medicaid beneficiaries less than 65 years old. Principal Findings: The CPS underestimated the Medicaid population by approximately a third. At the county level, errors in estimated numbers of Medicaid beneficiaries in the CPS increased in association with the penetration of Medicaid managed care. The unadjusted correlation coefficient between CPS undercount and mediciad managed care penetration is 0.23 (p=0.01). After controlling for secular changes, each percentage point increase in the penetration of managed care was associated with an underestimate in the CPS of 0.4 percentage points (p=0.006). Projecting the results in California to the growth of Medicaid managed care nationally during the same time period suggests that Medicaid managed care might explain more than 80% of the increasing underestimates of Medicaid beneficiaries in the CPS. Conclusions: A substantial portion of the increase in the underestimates of the number of Medicaid beneficiaries in the Current Population Survey can be explained by the growth of Medicaid managed care. Implications for Policy, Delivery, or Practice: Steps must be taken to improve the Current Population Survey if this survey is to remain useful for making accurate estimates of Americans’ health insurance status. Primary Funding Source: No Funding ●What are the Effects of Medicaid and SCHIP Managed Care on Children with Chronic Health Conditions? Amy Davidoff, Ph.D., Ian Hill, M.P.A., M.S.W., Emerald Adams, BA, Brigette Courtot, BA Presented By: Amy Davidoff, Ph.D., Assistant Professor, Public Policy, University of Maryland Baltimore County, 1000 Hilltop Circle, Baltimore, MD 21250; Tel: 410-455-6561; Fax: 410-455-1172; Email: davidoff@umbc.edu Research Objective: Managed care (MC) is an established feature of many Medicaid programs but is relatively new for some enrolled children with chronic health conditions (CHC). The effects of MC may be particularly strong for children with CHC, for whom health plans have strong financial incentive to coordinate care. Alternatively, elevated baseline use by children with CHC may be appropriate given their greater needs; MC may exert its effects by disrupting established provider relationships. The use by capitated plans of behavioral health or specialty “carve-outs,” while intended to direct children to appropriate systems of care, may create fragmentation and access barriers. MC programs designed specifically for children with CHC may ameliorate negative effects. The objective of this study is to examine the effect of different types of mandatory MC programs within Medicaid and SCHIP on children with CHC. Study Design: Pooled data from the National Health Interview Survey (1997-2002) were supplemented with county, year, and population specific data on Medicaid and SCHIP MC program types, assembled from annual CMS Medicaid MC Enrollment Reports and other sources. Children with CHC were identified based on parent report of diagnosed conditions or activity limitations. MC data were linked to children based on state and year specific eligibility for Medicaid or SCHIP. Linear probability models estimated the effects of MC program types relative to fee-for-service (FFS) on access and use for publicly insured children, with and without CHC. Population Studied: Medicaid and SCHIP eligible and enrolled children. Principal Findings: Relative to FFS, mandatory capitated programs without carve-outs are associated with decreased physician visits, reduced likelihood of a specialist visit (-7.2%), ER visits (-7.6%) and hospital stays (-3.1%). When MC programs include carve-outs we also observe reduced probability of mental health specialty visits (-7.8%), vision care visits (-6.4%) and prescription drug use (-9.6%). Special MC programs for children with CHC are associated with increased physician visits. Few significant effects are identified for children without CHC. Conclusions: Our results suggest that the effects of MC in Medicaid and SCHIP operate primarily on children with CHC. Relative to FFS, mandatory MC programs are associated with reduced use of services commonly used by children with CHC. The addition of behavioral health or specialty carve-outs is associated with even greater reductions in use. Reductions in ER and hospital use are suggestive of improved outpatient management; it is not possible to determine whether reductions in other services represent better care management or inadequate care. However, despite the reductions in use, we did not observe a corresponding increase in perceived unmet need, thus, the net change may represent improved care management. Implications for Policy, Delivery, or Practice: Much debate surrounds the issue of whether MC can work for children with CHC. In theory, MC could improve coordination and integration of care, but advocates are concerned that incentives to control costs may lead plans to under-serve these vulnerable children. This study does not resolve this debate, but suggests that MC is associated with improved outpatient management without increases in unmet need, and that specialty managed care models can facilitate access to physician care. Primary Funding Source: HRSA ●Primary Care Physician Response to a Mental Health Carve-Out: An Economic Analysis Ashley Dunham, MSPH, Jennifer L. Troyer, Ph.D., William P. Brandon, Ph.D. Presented By: Ashley Dunham, MSPH, Ph.D. Student, Public Policy, UNC-Charlotte, 528 Maupin Avenue, Salisbury, NC 281441; Tel: 704-239-6039; Email: adunham@uncc.edu Research Objective: The first research objective was to examine changes in depression treatment after the implementation of a Medicaid mental health carve-out that precluded reimbursement of primary care treatment for mental health. A second research objective was to use these results to test, in this particular experiment, two competing economic theories that may explain the way physicians behave: the principal-agent theory and the theory of wealth maximization. Study Design: The creation of a mandatory Medicaid mental health carve-out in Mecklenburg County established an experimental setting that allowed a test of physician behavior when faced with a contradiction between financial incentives created by managed care arrangements (the mental health carve-out) and current "best practice" that supports the treatment of depression in primary care. In 1996, the State of North Carolina received permission to implement mandatory enrollment in risk-contracting HMOs under a 1915(b) waiver in Mecklenburg County, North Carolina, which contains the city of Charlotte. As a result of the waiver, Medicaid recipients were shifted from a fee-for-service system to managed care plans, but mental health and prescripton drugs were "carvedout" of the premium and responsibility of the HMO. Medicaid claims data for 1995 through 1998 (for both the experimental county - Mecklenburg, and a control county - New Hanover) contain prescription drug claims for anti-depressants and claims for mental health visits coded for depression treatment. Claims data from both counties before and after HMO enrollment allowed the construction of two dependent variables that measured any change in depression treatment after the carve-out. Using monthly data, a difference-indifference model tested if the implementation of a carve-out in Mecklenburg county explained any changes in either referrals to mental health providers for depression treatment or antidepressant prescribing activity. Population Studied: A sample of Medicaid recipients in Mecklenburg County, North Carolina (n=3765) who were enrolled in Medicaid from January 1995 to December 1998. In addition, a sample of Medicaid recipients from New Hanover County, North Carolina who were not enrolled in any managed care arrangement (n=1004) served as the control county. Principal Findings: Results showed a significant increase in depression claims from mental health providers as well as a significant decrease in anti-depressant claims after HMO implementation in Mecklenburg County. Conclusions: The results showed that the physicians did respond to the mental health carve-out with more referrals to mental health providers for depression treatment than before the carve-out. With fewer antidepressant prescriptions and more claims from mental health providers, we assume that physicians acted as wealth maximizers, referring out treatment that was not included in their capitated arrangement. Implications for Policy, Delivery, or Practice: If Medicaid programs (as well as other health insurance organizations) want primary care physicians to treat patients for depression, reimbursement arrangements must be tailored to maximize both patient and physician utility, including financial incentives for the primary care physician to implement clinical practice guidelines for depression treatment. Further research should explore the willingness of primary care physicians to treat depression, as well as the effectiveness of depression treatment in the primary care setting. In addition, research should test which reimbursement arrangements maximize both physician and patient utility with regards to depression treatment. Primary Funding Source: No Funding ●Using Williness to Pay to Evaluate Hospital Merger: Results from 16 mergers Richard Lindrooth, Ph.D., David Dranove, Ph.D, Mark Satterthwaite, Ph.D. Presented By: Richard Lindrooth, Ph.D., Associate Professor, Department of Health Administration and Policy, Medical University of South Carolina, 151B Rutledge Ave P.O. Box 250961, Charleston, SC 29425; Tel: 8437922192; Fax: 8437921358; Email: lindrorc@musc.edu Research Objective: To test whether the methodology proposed by Capps, Dranove and Satterthwaite (2003), hereafter CDS, can be used to prospectively identify mergers that will lead to large price increases. Study Design: First, we estimate the CDS measure for hospitals and systems in nine markets. This measure is tied closely to economic theory and is estimated using simulations based on the parameter estimates of a hospital choice model. Next, we regress net inpatient revenues from the Medicare Cost Reports on the CDS measure and control variables and hospital fixed effects. The parameter estimate on the CDS measure reflects the increase in net inpatient revenue as a result of changes in market power associated with a merger. We identify the effect of self-selected mergers, thus our results are predictive rather than causitive. Population Studied: We examine hospitals and systems in the following markets: Bakersfield, CA; Buffalo, NY; Daytona, FL ; Denver, CO ; Jacksonville, FL; and Rochester, NY. The CDS measure is estimate using HCUP-SID data in the year prior to the first merger in the market. Variables from the AHA Annual Survey are included to estimate the patient choice model that underlies the CDS measure. There were 16 involving over 45 hospitals between 1995 and 2000 in these markets. Principal Findings: The average increase in the CDS measures for these mergers was 1162 Units. Based on the regressions results this translates into an average $7,730,183 increase it net inpatient revenue. The total increase in net inpatient revenue associated with the mergers is about $124 million. Note that this increase is solely due to the merger, it is over and above what hospitals would have recieved if they were independent. Conclusions: The CDS measure performed reasonably well in this analysis. About 7 of the 16 mergers led to a greater than 10% increase in market power. The R-squared in the second part of the analysis was much lower than what was observed in the CDS analysis of a single market. Though this is not surprising given we pool 6 years of data over multiple markets. Implications for Policy, Delivery, or Practice: This paper has direct implications for the Federal Trade Commission's merger policy. We have validated an alternative approach to measuring market power that in contrast to current (much critisized) approaches is directly tied to economic theory. This approach shows promise for future merger cases. Primary Funding Source: RWJF Call for Papers Markets, Financial Incentives & Quality of Care Chair: Jon Gabel, Center for Studying Health System Change Tuesday, June 27 • 10:30 am – 12:00 pm ●The Role of Quality in the Formation of Exclusive Networks for Kidney Transplantation David Howard, Ph.D. Presented By: David Howard, Ph.D., Assistant Professor, Department of Health Policy and Management, Emory University, 1518 Clifton Road NE, Atlanta, GA 30322; Tel: 404727-3907; Fax: 404-727-9198; Email: david.howard@emory.edu Research Objective: Almost all large insurers restrict enrollees’ choice of kidney transplant center under the guise of “centers of excellence” programs. We determine the role of quality, as measured by patient outcomes, in the formation of health plans’ exclusive transplant networks. According to plans’ promotional literature, hospitals are selected based on quality. Study Design: From the Scientific Registry of Transplant Recipients we obtained a measure of quality for every kidney transplant hospital in the U.S. (N = 203). The measure is observed kidney graft survival at one year post-transplant minus expected graft survival. Graft survival indicates that the patient is alive and the kidney graft is still functioning. Expected graft survival is computed by the Registry based on a regression model with extensive patient controls. Higher values indicate better performance. We perform two analyses. First, we compare observed-expected graft survival between innetwork and out-of-network hospitals for five large national health plans. Lists of networks were obtained from plans’ websites. Second, using patient level data obtained from the Registry, we estimated a conditional logit model of transplant center choice. We constructed choice sets based on historical registration patterns and distance. Two center attributes, distance from the patients’ home and observed-expected graft survival, were fully interacted with patient characteristics, including insurance type (Medicare versus private). We used the conditional logit model parameters to determine if privately-insured patients are more likely to register at high quality centers. The sample size was 32,652 patients. Principal Findings: Observed-expected graft survival was higher in in-network hospitals by 2-4 percentage points for all five health plans (p<0.05 for each). There was a substantial degree of overlap in the networks, and most networks included the larger, well-established transplant centers (for example, UCLA, University of Pittsburgh). Simulations using the parameters from the patient-level conditional logit model show that for patients with Medicare, a one-standard deviation increase in the graft-failure rate is associated with a 2% decline in patient demand. However, for privately-insured patients, a one-standard deviation increase in the graft-failure rate is associated with a 15% decline in demand. Conclusions: Findings indicate that transplants centers with good outcomes are more likely to be included in plans “centers of excellence” networks, consistent with plans’ claims. Implications for Policy, Delivery, or Practice: Harvard Business School Professor Michael Porter and other thought leaders have recently questioned whether exclusive provider networks provide value to patients. This study, as well as previous studies examining the role of quality in the formation of provider networks for cardiac surgery, indicate that insurers use networks to steer patients to high quality facilities. Primary Funding Source: NIDDK ●Are Insurance Gaps Costly for Diabetes Patients? Who Pays? Hsou Mei Hu, Ph.D., Emily C. Shelton, MAE, David M. Cutler, Ph.D., Allison B. Rosen, M.D., Sc.D. Presented By: Hsou Mei Hu, Ph.D., Research Associate, Internal Medicine, University of Michigan, 300 Ingall, Room 7C27, Ann Arbor, MI 48103; Tel: (734) 615-7975; Email: hsoumeih@med.umich.edu Research Objective: Continuity of care is crucial for diabetes patients in preventing costly complications. As health insurance premiums rise, more chronically ill individuals experience lapses in insurance coverage which, by limiting access, may result in worse health outcomes and require more costly care. The purpose of this paper is to examine the effect of insurance gaps on diabetic patients’ use and costs of inpatient and emergency room (ER) care. Study Design: We used data from three panels (spanning from 2000 to 2003) of the Medical Expenditure Panel Survey (MEPS). Respondents reporting diabetes in the first year of each of the two year panels were dichotomized into those with continuous coverage (n=1901) and those with gaps in coverage (n=369). Those who were uninsured for the entire two year period (n=182) were excluded. We examined the association between coverage gaps and the use and costs of care. Poisson regression was used to assess differences by continuity of coverage in ER visits and hospitalizations, as these utilization events are markers of worse health outcomes. Differences in ER and inpatient spending and total spending (inflated to 2003 dollars) between the two groups were examined using ordinary least squares regression. Spending was further explored by private vs. public payers. All analyses controlled for demographic characteristics, socioeconomic status, and comorbidities, and adjusted for the MEPS complex sample design using STATA 9.1. Population Studied: A weighted population of 6,381,214 civilian noninstitutionalized individuals with diabetes in the US. Principal Findings: Diabetes patients experiencing gaps in insurance coverage had significantly more ER visits (0.99 vs. 0.71) and were more likely to be hospitalized (1.03 vs. 0.71) during the 24-month period than those with continuous coverage. Respondents with insurance gaps incurred higher ER, inpatient, and overall costs than those without insurance gaps (ER & inpatient costs $9,748 vs. $7,129, and total costs $19,187 vs. $17,803, respectively). When examining spending by payer type, public programs (including Medicaid, Medicare and Tricare) paid more for diabetics with insurance gaps than for those without gaps in coverage ($9,085 vs. $7,953, respectively), although private insurers paid comparable amounts for the two groups. Conclusions: Insurance gaps are associated with increased ER visits and hospitalizations, suggesting a possible adverse impact on the health of individuals with diabetes. This increased use is mirrored by concomitant increases in ER and hospital spending. These effects may be more pronounced for public payers. Implications for Policy, Delivery, or Practice: Policies to fill insurance gaps may improve health and potentially even save money if applied in sicker population, such as those with diabetes. Further studies are needed to explore the impact on both health and economic outcomes of policies aimed at addressing these coverage gaps in chronically ill populations. Primary Funding Source: NIA, The Harvard Interfaculty Program for Health Systems Improvement, the Lasker Foundation ●The Effect of Hospital Safety Reports and a Tiered Hospital Network on Inpatient Referrals Dennis Scanlon, Ph.D., MA, BA, Jon B. Christianson, Ph.D., Coleen Lucas, RN, Eric Ford, Ph.D. Presented By: Dennis Scanlon, Ph.D., MA, BA, Associate Professor of Health Policy & Administration, Health Policy & Administration, The Pennsylvania State University, 119B Henderson Building, University Park, PA 16802-6500; Tel: 814865-1925; Fax: 814-863-2905; Email: dpscanlon@psu.edu Research Objective: Recent years have seen a movement towards both ‘consumer directed’ and ‘pay for performance’ programs in health care. Many of these programs utilize ‘tiered networks’ for hospital care, where consumer out-ofpocket co-payments vary based on the hospital chosen. While most of the early tiering efforts were based on hospitals’ charges only, there is an increasing movement towards placing hospitals into tiers based on efficiency and quality/safety indicators. However, little has been published about consumer response to tiered hospital benefits or the impact of tiered networks on hospital admissions and revenues. This paper examines the effect of a tiered hospital benefit in an employed, commercially insured population. Study Design: In conjunction with its major labor unions, the Boeing Company instituted the Hospital Safety Incentive (HSI) for union (i.e., hourly) employees enrolled in Boeing’s Traditional Medical Plan (TMP). The TMP, which took effect in July 2004, is an ERISA, self-funded health plan administered for Boeing by Regence Blue Shield of Washington. The HSI is unique because it gives patients a financial incentive to choose hospitals that meet the Leapfrog Group’s three patient safety "leaps". While the TMP’s standard coverage for hospital care is 95% of allowed hospital charges (up to the annual out-of-pocket maximum), union beneficiaries enrolled in the TMP can achieve a benefit of 100% for hospital care if admitted to a hospital that meets the Leapfrog standards. Boeing’s actuaries have estimated the average value of the 5% payment to be approximately $450 per admission. We estimate the effect of the HSI on patients’ selection of hospital in two of Boeing’s major employment hubs (Seattle, WA and Wichita, KS). We utilize a pre-post study design and take advantage of the fact that the HSI did not apply to nonunion (i.e., salaried) employees. We identify the effect by comparing the change in hospital admissions of hospitalized hourly and salaried beneficiaries after the HSI went into effect. To gauge awareness of the HSI, we also examine differences, pre-post, between hourly and salaried non-hospitalized beneficiaries. We identify enrolled beneficiaries from claims data, and while we examine changes in hospital market shares, our primary outcome variables come from answers collected during a 20 minute telephone survey. The telephone survey was necessary since patients are referred to hospitals by their physicians, and thus it is not clear if admissions decisions are made by patients, physicians, or jointly. Targeted respondents were randomly sampled from four groups in each period (union/non-union, hospitalized/nonhospitalized). The key outcomes include questions regarding the degree to which the patient was involved in the choice of hospital and awareness of the HSI. We completed approximately 1,200 interviews in each period and achieved a 60% survey participation rate. Population Studied: Salaried and hourly employees of the Boeing Company and their beneficiaries in two markets (Seattle, WA and Wichita, KS). Principal Findings: Preliminary results indicate the HSI did not influence the admissions decisions of hourly hospitalized beneficiaries. Answers to other survey questions provide insight about why. In particular, patient-physician relationships are very important, and admission to a Leapfrog compliant hospital would often require switching physicians. In addition, only 20% of hospitals qualified for the PSI during the study period. Conclusions: While benefit plans that include tiered provider networks (e.g., hospital and physicians) aimed at increasing consumer sensitivity to cost, quality and safety have grown in popularity and are theoretically attractive, little is known about how they work operationally. Our study provides important insight regarding the practical limitations of these benefit structures as well as consumer response to them. Implications for Policy, Delivery, or Practice: Our findings are important for informing the emerging health care consumerism and pay for performance movements. Primary Funding Source: AHRQ ●The Effect of Cost-Sharing on the Adherence of Antihypertensive Medications Matthew Solomon, Ph.D., Jose J Escarce, M.D., Ph.D., Dana P Goldman, Ph.D., Geoffrey F Joyce, Ph.D. Presented By: Matthew Solomon, Ph.D., Medical Student, David Geffen School of Medicine at UCLA, 3332 Hamilton Way, Los Angeles, CA 90026; Tel: 323 669 1550; Email: mattsol@ucla.edu Research Objective: To control rapidly rising prescription drug costs, nearly all health plans have increased their costsharing requirements. Although many studies have established a link between cost-sharing and the utilization of prescription drugs, there is little evidence that directly explores the mechanism underlying such cost-mediated utilization changes. Preliminary research has indicated that increased cost-sharing levels are associated with delays in the adoption of antihypertensive therapy. This study aims to understand the effect of cost-sharing on adherence to antihypertensive treatment regimens for those patients with newly diagnosed hypertension who initiate drug therapy. Study Design: This is a retrospective study conducted from 1997 to 2002 examining pharmacy and medical claims data that are matched to the health benefits from 13 large employers and 31 health plans. Outcomes explored are (1) the proportion of days (PDC) covered in the first year and first two years after initiation of antihypertensive therapy, and (2) the number of days until a 60- or 90-day gap in coverage after initiation of antihypertensive therapy. Population Studied: The sample includes 5022 elderly individuals newly diagnosed with hypertension who have employer-based retiree insurance benefits, including insurance for prescription drugs. Principal Findings: The mean PDC for the overall sample was 75.0% (S.D. 29.2%), and the proportion of patients with a PDC greater than 80%, a common measure of "good" adherence, was 60.1% (S.D. 48.9%). In multivariate logistic regression models, cost-sharing levels were not associated with having a PDC greater than 80% in either the first year (P=0.719) or first two years (P=0.765) after the initiation of antihypertensive medication. Sensitivity analyses revealed that cost-sharing was similarly not associated with having a PDC greater than 70%, 60%, 50%, 40%, or 30% in either the first year or two years after initiation. However, nonantihypertensive drug use at the time of initiation was associated with an increased likelihood of good adherence in the first year (O.R.=1.39; P=0.000) and first two years (O.R.=1.45; P=0.000). Similarly, in multivariate Cox models, cost-sharing was not associated with the time to a 60- or 90day gap in coverage (P=0.503 and P=0.521, respectively), and concurrent medication use was associated with a decreased hazard of having a 60- (H.R.=0.816; P=0.000) or 90-day gap (H.R.=0.814; P=0.000). Conclusions: Cost-sharing levels were not associated with adherence to antihypertensive medication regimens among newly diagnosed elderly hypertensive patients. However, concurrent medication use at the time of initiation was a strong predictor of good adherence. The lack of an association between cost-sharing and adherence indicates that adherence may not account for a substantial portion of the observed cost-related utilization reduction of prescription drugs found in the literature. Future research should continue to explore the relationship between prior drug use and adherence. Implications for Policy, Delivery, or Practice: These results will help policy makers understand how drug benefits affect antihypertensive use, and will help physicians address factors that may improve adherence to the treatment regimens they prescribe for their hypertensive patients. Primary Funding Source: AHRQ, NIH Medical Scientist Training Program at UCLA ●Effects of Hospital Price Competition on Quality of Care for 4 High-mortality Conditions Kevin Volpp, M.D., Ph.D., R. Tamara Konetzka, Ph.D., Julie Sochalski, Ph.D., FAAN, RN, Jingsan Zhu, M.B.A. Presented By: Kevin Volpp, MD, Ph.D., CHERP, Philadelphia VA Medical Center, 3900 Woodland Ave., 9th Floor, Philadelphia, PA 19104; Tel: 215-573-9718; Fax: 215-573-8778; Email: volpp70@mail.med.upenn.edu Research Objective: A significant body of health economics research indicates that hospital competition shifted from a quality/amenity basis to a price basis with the growth of managed care in the 1980s and 1990s, lowering the rate of increase in hospital costs. However, in recent years costs have begun to rise at rapid rates and it is less clear that managed care is still effectively controlling costs. In this analysis, we examine effects of price competition and managed care on changes in quality over two time periods: 1991-1997, a period in which price competition and managed care reduced the rate of increase in hospital costs and 19972001, a period in which our previous work shows that price competition still reduced the rate of increase in hospital costs but managed care no longer had these effects. Study Design: Retrospective Observational Study. We assess the effects of price competition and managed care penetration on 30-day mortality for several high-mortality conditions while controlling for patient severity and changes in hospital volume and case-mix. We use hospital-level fixed effects in a longdifferences framework to assess whether the effects on quality of being in a market with high managed care penetration and/or high market competition have changed over time. Population Studied: Patients with principal diagnoses of AMI, stroke, hip fracture, or GI bleed admitted to California hospitals between 1991 through 2001. We use OSHPD data linked with state death certificates and hospital financial data. Principal Findings: Neither the unadjusted nor the adjusted results suggest a shift in the effect of managed care on mortality corresponding to the shift in the effect on hospital cost growth. In both the earlier and later periods, the greatest decline in mortality is found in markets with high competition but low managed care. The smallest decline in mortality (or largest increase) is found in markets with low competition and high managed care. This would suggest that high competition is associated with improvements in quality but that high managed care penetration interferes with rather than enhances that effect. Higher managed care penetration alone is associated with relatively worse mortality rates regardless of competition, but the size of the difference does not differ substantially between the two periods. Conclusions: If the fact that higher MCP ceased to have an effect on costs after the managed care backlash circa 1997 were due to a shift away from competition on price toward competition on quality, we might expect to see relative declines in mortality in high-MCP markets in the post-backlash period. For four common conditions treated in inpatient settings, we find no evidence of this. These results are consistent with qualitative evidence that hospitals have simply gained negotiating power relative to managed care organizations, and that any shift toward quality competition consists of competition on amenities and services that have little influence on clinical outcomes. Implications for Policy, Delivery, or Practice: Provision of data on risk-adjusted outcomes in different hospitals might increase the degree to which hospitals and managed care organizations compete in improving outcomes. Primary Funding Source: Doris Duke Charitable Foundation Related Posters Health Insurance Markets & Managed Care Poster Session B Monday, June 26 • 5:30 pm – 7:00 pm ●Affordability of Health Insurance: Do Assets and Net Wealth Explain the Demand for Health Insurance Better than Income? Didem Bernard, Ph.D. Economics, Jessica Banthin, Ph.D. Economics, Willaim Encinosa, Ph.D. Economics Presented By: Didem Bernard, Ph.D. Economics, Senior Economist, CFACT, AHRQ, 540 Gaither Road, Rockville, MD 20850; Tel: 301-427-1682; Fax: 301-427-1276; Email: dbernard@ahrq.gov Research Objective: Understanding the affordability of coverage is important for evaluating the role of policy in reducing the number of uninsured workers. We study worker health insurance take-up and coverage decisions using data from the Medical Expenditure Panel Survey (MEPS) from 1997 to 2003. Study Design: Unlike previous studies which control only for current income, we include information on the presence and value of family-level assets such as home ownership, vehicles, savings accounts, stocks, bonds, retirement accounts, as well as liabilities, to estimate the effect of net wealth on insurance purchase decisions. Unlike most studies which focus on “worker” take-up, we also take into account the availability of insurance offers through the spouse’s employer in estimating enrollment decisions of workers and their families. We estimate worker demand for employer sponsored health insurance as a function of the premium as well as worker, family, employer and plan characteristics. We use two approaches to deal with the lack of premium data for workers who decline coverage. The first approach uses a sub sample in MEPS from 1997-1999 with linked data from the Household Component (HC) and the Insurance Component (IC) which is a survey of employers. Although the HC-IC link sample is not nationally representative, it contains data on the premiums for takers and decliners as well as the availability of choice of health plans, and types of plans offered. The second approach uses simulated premiums from the Insurance Component List Sample. Using the nationally representative sample of employers in the MEPS-IC, we estimate average plan premiums as a function of predictor variables available on both the employer and household surveys, including location (state, MSA), firm size, industry, and plan types offered. We then use this model to predict premiums for workers in the household survey. Population Studied: We study workers' (aged 18 to 64) health insurance take-up and coverage decisions using data from the Medical Expenditure Panel Survey (MEPS) from 1997 to 2003. Principal Findings: Among adults living in families with health insurance offers in 2001 and 2002, 7.6 percent did not take up private insurance. As expected, probability of take up declined with income: 8.6 percent of adults with middle income, 19.9 percent of adults with low income, and 32.4 percent of poor and near poor adults did not take up private insurance. (Bernard and Selden, 2005) Preliminary work based on this sample, suggests that assets and net wealth play a significant role in insurance coverage decisions. Controlling for income, adults who did not take up health insurance were significantly less likely to have assets. For example, among poor and low income adults, the decliners were less likely to own homes (46% vs. 54%), cars (78% vs. 86%), checking accounts (37% vs. 51%), stocks (1% vs. 5%), and individual retirement accounts (13% vs. 22%). Conclusions: Preliminary results suggest that assets and net wealth play a significant role in insurance take up decisions. Implications for Policy, Delivery, or Practice: Research using affordability thresholds based on income has shown that health insurance was affordable to 25% to 75% of the uninsured in 2000. (Bundorf and Pauly, 2002) Our preliminary results suggest that affordability thresholds based solely on income may be inadequate in explaining health insurance purchase decisions. Primary Funding Source: AHRQ ●Ambulatory Surgery Facility Access and Health Care Cost and Utilization William Cecil, MBA, John Barnes, M.A., M.B.A., Steven L. Coulter, M.D. Presented By: William Cecil, MBA, Director, Health Policy Research, Health Policy Research, BlueCross BlueShield of Tennessee, 801 Pine Street - 1E, Chattanooga, TN 37402; Tel: (423) 763-3372; Fax: (423) 755-5100; Email: bill_cecil@bcbst.com Research Objective: Since January of 2000, the ASF capacity in Tennessee has increased by 70 new facilities at a CON approved cost of $208 million. Since January of 1999, outpatient surgical plan paid costs for BCBST have risen from $7.03 PMPM to $15.19, as of June 2005; a 116% increase. OP surgery utilization has risen from 89.4 per 1,000 members to 144.6 per 1,000; a 61.7 % increase. The paid cost per visit has risen from $943 to $1,228, a 30.2% increase. BCBST TennCare OP surgery paid costs have risen from $4.49 to $12.04, a 168% increase. TennCare OP surgery utilization has risen from 88.2 per 1,000 to 189, a 114% increase. TennCare paid cost per visit has risen from $610 to $785, a 28.7% increase. According to the State of Tennessee Department of Labor, Ambulatory Care Facility employment is projected to grow at about 493 jobs per year. First quarter 2005 average annual wage was $43,732. Listings of employers available at the state web site showed 72% of firms in the ACF category were ambulatory surgery facilities. A principal justification of ambulatory surgery facility use for outpatient surgery compared to hospital based outpatient surgery is lower costs. The objective of this paper is to investigate the relationship between ambulatory surgery capacity, access, cost and utilization and health care costs and utilization. Study Design: An observational study that makes benefit of increases in network ambulatory surgery capacity over a four year period from 2001 through 2004 to test the accessutilization/cost relationships. We examined 357,172 claims records using standard methods to compare plan paid costs and utilization of both hospital based and ambulatory surgery facility based outpatient surgery utilizers. Population Studied: A commercially insured population of 74,959 members from 2001, 89,484 from 2002, 98,219 from 2003 and 94,510 from 2004, all residing within Tennessee. Principal Findings: Both theoretical and empirical models confirm the positive relationship between capacity, access and utilization. Empirical models show higher adjusted plan paid costs for those undergoing outpatient surgery in an ambulatory surgery facility compared to hospital outpatient surgery. Higher use of primary care office visits, specialist office visits and imaging in all service locations were associated with ambulatory surgery facility based outpatient surgery compared to hospital based outpatient surgery. Actual and theoretical marginal access improves by 0.07 and 0.08 miles respectively with the addition of a single ambulatory surgery facility. Marginal ambulatory surgery facility access gains become less than a mile at 18.6 % of 2004 network capacity. Conclusions: The distance to an ambulatory surgery facility is a function of the number of network ambulatory surgery facilities. Ambulatory surgery facility utilization is a function of distance to the facility in addition to out of pocket cost sharing, age, gender and DCG score. We find higher total plan paid costs with the use of ambulatory surgery facility based outpatient surgery. Implications for Policy, Delivery, or Practice: Careful consideration should be given to network composition for outpatient surgery services given the higher cost associated with ASF use. ASF network management may also play an important role in imaging cost management. Primary Funding Source: BlueCross BlueShield of Tennessee ●A Review and Synthesis of the Cost and Benefits of Health Insurance Regulation in the U.S. Christopher Conover, Ph.D., Ilse Wiechers, M.P.P., M.D. Presented By: Christopher Conover, Ph.D., Assistant Research Professor, Terry Sanford Institute of Public Policy, Duke University, Box 90253, Durham, NC 27708; Tel: (919) 613-9369; Fax: (919) 684-6246; Email: conoverc@hpolicy.duke.edu Research Objective: The central objective of this paper is to develop a comprehensive estimate of the costs and benefits of health insurance regulation derived by summing the results of fine-grained cost estimates in 19 different domains of health insurance regulation; a companion objective is to identify domains of health insurance regulation in which regulation currently appears not to be cost-effective. Study Design: A formal literature search was conducted for each major domain of health insurance regulation. For each domain, a synthetic estimate of regulatory costs is obtained using a standardized procedure that systematically assembles evidence from the literature regarding government regulatory costs (such as monitoring and enforcement costs) and compliance costs (including both industry costs and patient time losses), and indirect costs such as costs associated with health losses (morbidity and mortality losses, all monetized using a common value of statistical life year) or increases in uninsured risk (monetized using a standardized estimate of the external cost of being uninsured and the monetized value of the increased mortality risk faced by the uninsured). Corresponding monetized estimates of benefits were developed based on empirical evidence of benefits derived from the literature, including efficiency gains, quality improvements or reductions in uninsured risk. Lower and upper bound estimates were derived using minimum and maximum parameter estimates for the various components used to calculate regulatory costs. Population Studied: The synthesis focused on 19 domains of insurance regulation, including regulations focused on access to care (e.g., high risk pools), cost control (e.g., HIPAA administrative simplification) and quality (e.g., patient bill of rights). Principal Findings: The preliminary base case result for the cost of health insurance regulation in the U.S. is $99.3 billion (2002 $); the minimum estimate is $68.2 billion and the upper bound estimate is $172.7 billion. Estimated benefits are $84.9 billion, but this ranges from a minimum of $49.7 billion and upper bound of $180.9 billion. [final estimates updated to 2004 $ to be available in early February] Conclusions: On balance, the net costs of insurance regulation exceed $14 billion annually. The domains of health insurance regulation having the highest net cost include continuation of coverage requirements, benefit mandates and patient bill of rights. Implications for Policy, Delivery, or Practice: The 12 domains in which costs exceeded benefits are areas warranting closer examination to determine how regulatory objectives might be achieved more cost-effectively. Primary Funding Source: ASPE, DHHS ●Generic Switch Study Melissa Godfrey, MS, William Westerfield, MA, Soyal Momin, MS, M.B.A., Ray Phillippi, Ph.D., Terence Shea, PharmD Presented By: Melissa Godfrey, MS, Research Analyst, Health Services Research, BlueCross BlueShield of Tennessee, 801 Pine Street, 3E, Chattanooga, TN 37402; Tel: (423) 752-6471; Fax: (423) 785-8083; Email: melissa_godfrey@bcbst.com Research Objective: The primary purpose of this study was to estimate the brand-to-generic “switch rate” for a health plan’s members who were offered a generic co-pay waiver from October 18 to December 31, 2004. A secondary objective was to create a statistical model predicting members’ decisions to switch. Study Design: Healthcare claims data were analyzed to identify the health plan’s members who met study inclusion criteria. Those who switched brand prescriptions to generics during this time were assigned to the “switcher” group, and those who did not were assigned to the “non-switcher” group. The “switch rate” was calculated as the ratio of switchers to the study population. The Prescription Drug Survey was created to measure factors – such as perceptions of generic drugs and interaction with healthcare professionals – that previous research has linked to consumers’ decisions to use generic drugs. The survey was mailed to 1,027 switchers and 1,027 non-switchers. A total of 586 surveys were returned (response rate=28.5%). Survey responses were combined with healthcare claims data, and binomial logistic regression was used to create a statistical model to predict switching. Population Studied: Participants were members of the commercially insured population of a large single-state health insurance carrier in the southeastern United States who met the following criteria: (1) received the generic co-pay waiver, (2) purchased the brand version of a multi-source maintenance drug between July 1 and October 17, 2004, (3) purchased either the brand or the generic version of the same multi-source maintenance drug during the co-pay waiver period, (4) were at least 18 years old, (5) were either a subscriber or spouse of a subscriber, (6) were enrolled at the time of survey mailing (June 2005), and (7) qualified for one of the two groups (“switcher” and “non-switcher”). Principal Findings: The brand-to-generic switch rate for the study population during the co-pay waiver period was 11%. The following factors were associated with members’ decisions to switch: brand co-pay (OR=1.024, p=.0164), perceptions of generic drugs (OR=1.264, p<.01) approaching a healthcare professional about switching (OR=2.492, p<.01), taking estrogens (OR=0.207, p<.01), and taking thyroid hormones (OR=0.579, p<.01). Conclusions: Eleven percent of members who were eligible for inclusion in the Generic Switch Study took advantage of the co-pay waiver. Members with high brand co-pays, positive perceptions of generic drugs, and those who approached a healthcare professional about switching were more likely to switch than their counterparts. Members who took estrogens and thyroid hormones were less likely to switch than those who took other drugs. Implications for Policy, Delivery, or Practice: Managed care organizations can encourage members to switch brand prescriptions to generics by offering financial incentives, educating members about the benefits of using generics, and encouraging them to approach healthcare professionals about cost-effective treatments. Directing these tactics to members who take estrogens and thyroid hormones should increase generic adoption. The 11% switch rate can be used as a baseline measure for future analyses. Primary Funding Source: BlueCross BlueShield of Tennessee ●Is Cost-sharing Associated with Seeing Out-of-Network Physicians among Children with Asthma or Diabetes? David Grembowski, Ph.D., William Trejo, BA, Paula Diehr, Ph.D., James Stout, M.D., M.P.H., Frederick Connell, M.D., M.P.H. Presented By: David Grembowski, Ph.D., Professor, Health Services, University of Washington, Box 357660, Seattle, WA 98195-7660; Tel: (206) 616-2921; Fax: (206) 543-3964; Email: grem@u.washington.edu Research Objective: To determine the association between out-of-network and in-network cost-sharing and the likelihood of seeing an out-of-network physician among children with asthma or diabetes in managed care plans. The single, common element across all types of managed care organizations (MCOs) is a provider network. To control their costs, MCOs often build networks composed of physicians with lower fees and/or lower-cost practice styles, and then route enrollees to network physicians through greater out-ofnetwork cost-sharing. Because children with chronic conditions have greater needs and see multiple physicians, it is unclear whether greater out-of-network cost-sharing reduces the likelihood of seeing an out-of-network physician, and whether this matters in terms of expenditures and quality of care. Study Design: Cross-sectional cohort study. Population Studied: 123,645 children/adolescents with 2 years of continuous coverage (July 1999 to June 2001) in a commercial health insurance firm in Washington state. Of these, 301 children met diabetes eligibility criteria (> 1 diagnosis for diabetes or 2+ insulin prescriptions), and 2,242 children met asthma eligibility criteria (> 2 diagnoses for asthma separated by at least a week). Dependent variable was whether a child saw an out-of-network physician in Year 2 (excluding anesthesiologists, radiologists and emergency room physicians). Cost-sharing was measured by out-ofnetwork and in-network cost-sharing indexes for the child’s health plan in Year 2. Data sources include physician network directories, American Medical Association Masterfile, child characteristics from health plan, asthma severity in Year 1, sociodemographic characteristics of child’s Zip code tabulation area, and HMO penetration in child’s county. Principal Findings: Children with asthma saw an average of 2.4 physicians in Year 2 (excluding anesthesiologists, radiologists and emergency room physicians), and children with diabetes saw an average of 2.2 physicians. About 22% of children with asthma and 24% of children with diabetes had emergency room visits, and 5% of children with asthma and 12% of children with diabetes were hospitalized. About 4% of children with asthma and 9% children with diabetes saw an out-of-network physician in Year 2. Over half of the out-ofnetwork physicians had primary care specialties. Logistic regression revealed that greater in-network plan coverage was associated weakly (p=.07) with lower odds of seeing an out-ofnetwork physician for children with diabetes but not for children with asthma. The out-of-network cost-sharing index was not associated with seeing an out-of-network physician. Greater HMO penetration was associated with greater odds of seeing an out-of-network physician (p<.001) for children with asthma. Conclusions: Cost-sharing is not associated generally with seeing an out-of-network physician among children with diabetes or asthma in health plans with provider networks. Greater HMO penetration is associated with seeing an out-ofnetwork physician among children with asthma. Implications for Policy, Delivery, or Practice: For children with asthma or diabetes, cost-sharing does not appear to be a barrier to seeing out-of-network physicians. Greater HMO market penetration may be related to less physician participation in plan provider networks, which may increase utilization of out-of-network physicians. Primary Funding Source: AHRQ ●The Role of State Regulation in Consumer-Driven Health Care Timothy Jost, J.D., Mark Hall, .JD. Presented By: Timothy Jost, J.D., Robert Willett Family Professor, College of Law, Washington and Lee University, Lewis Hall, Lexington, VA 22802; Tel: (540) 458-8510; Fax: (540) 458-8488; Email: jostt@wlu.edu Research Objective: To examine the role that the states are taking in regulating health savings accounts (HSAs) and high deductible health plans (HDHPs) for which federal tax subsidies are available under the Medicare Modernization Act (MMA). Study Design: Interviews with state regulators, insurance company and trade association representatives, independent experts, and HSA advocacy groups, supplemented with a literature review and legal analysis. Population Studied: State regulatory programs and insurance companies. Principal Findings: Most states have responded affirmatively to the incentives offered by the MMA for consumer-driven health care by removing regulatory barriers to qualified HDHPs such as minimum deductible mandates Many states have gone further, adopting or modifying state income tax laws to supplement the federal incentives provided through the MMA for HSAs and HDHPs with state incentives. With a few notable exceptions, the states seem generally supportive of consumer-directed health care and reluctant to impede its progress with regulatory burdens. In embracing consumerdriven health care, however, the states seem to have lost sight of their traditional regulatory concerns about financial accountabilty, on the one hand, and access to insurance, on the other. The states have also not adequately worked through the interface between consumer-driven healthcare and state managed care regulation. Conclusions: The states have generally embraced the federal initiative to encourage consumer-driven health care, but in doing so have largely abdicated their traditional responsibility for insurance regulation. They have not carefully examined the potential problems presented by HSAs and HDHPs and designed an appropriate regulatory response. Implications for Policy, Delivery, or Practice: Public policy with respect to consumer-driven health care seems at this point to be driven exclusively by the federal government. The states have traditionally been responsible for regulating health insurance in our federal system, and need to consider their responsibility for regulating consumer driven health care. This paper raises the issues that the states need to work through. Primary Funding Source: RWJF ●Effects of Health Insurance on Physical Activity of US Adults Eric Keuffel, M.P.H. Presented By: Eric Keuffel, M.P.H., Graduate Student, Health Care Systems Department, Wharton - University of Pennsylvania, 6 Reaney Court, Philadelphia, PA 19103; Tel: (215) 546-9286; Email: ekeuffel@wharton.upenn.edu Research Objective: This study estimates the effects of health insurance status on physical activity decisions. Epidemiological evidence links physical activity with reductions in relative risk for a variety of costly chronic diseases. In theory, the expected effect of insurance coverage on primary prevention effort is ambiguous. Moral hazard may reduce the probability of engaging in prevention behaviors by beneficiaries. However, many insurers now offer benefits and programs that promote physical activity. Prior research found a complementary relationship between coverage and physical activity. This study disaggregates effects by source (public vs. private) and type (HMO vs. Non-HMO) of insurance. Study Design: Physical activity (PA) is defined as “vigorous activity 3 or more times per week for at least 30 minutes per session”. Logit models regressed the bivariate physical activity measure in 2002 (1-yes, 0-no) on insurance variables, demographic characteristics, occupation, income, education, health status measures and medical expenditures. Independent variables either measure 2002 levels or changes (when appropriate) and help extract out variation due to underlying health status, demographic, educational and occupational factors. Variance Inflation Factors (VIF) from linear specifications reject excessively collinear specifications. Marginal effects (ME) are estimated using mean values and reflect the change in probability of being physical active for an “average” individual for each one unit increase in the relevant independent variable. The analysis was conducted using STATA version 9.2. Population Studied: The Individual Component of the Medical Expenditure Panel Survey (MEPS) Panel 6 (20012002) is a weighted representative panel of the civilian noninstitutionalized US population. PA was only measured for adults Principal Findings: Just over half of adults are physically active (.55). Although the primary specification results in a positive, but insignificant, association between insurance coverage and physical activity; the specification which differentiated between public and private coverage shows that private insurance complements physical activity (+.03 marginal effect, p=.06) while public insurance has insignificant effects. Interaction models indicate that, after accounting for type of insurance (public and private), those enrolled in HMO plans within Medicaid have significantly higher probabilities for PA (+.06 ME, p=.03). Much of the effect associated with private insurance depends on employment status. The main private insurance coefficient is positive (+.08 ME, p<.01), but the interaction term between private insurance and employment is negative (-.07 ME, p<.01). Lastly exposure to out-of-pocket costs (measured as a ratio to total health expenditures) is significant but of modest magnitude (+.03 ME, p=.05) Conclusions: To the extent that the covariates eliminate important confounders such health status and account for selection, these regressions suggest that private plans and HMOs increase probability of physical activity for at least portions of the population. Greater exposure to out-of-pocket costs increases the probability of PA, but only modestly. Implications for Policy, Delivery, or Practice: Incentives embedded in private and HMO style plans appear to increase PA behavior modestly. While the precise mechanism is unknown, future identification of these economic fulcrums that promote physical activity and potentially reduce long run health care costs may affect both private and government payers. Further study on the mechanism of action and subgroup analyses are warranted. Primary Funding Source: NRSA ●Cost-Based Physician Profiling: Increasing Efficiency through Accountability Antonio Legorreta, M.D., M.P.H., Yingxu Zhao, Ph.D., William Wright, BA, Elizabeth Lord, BA, Antonio Legorreta, M.D., M.P.H. Presented By: Antonio Legorreta, M.D., M.P.H., Adjunct Professor, Health Services, University of California, Los Angeles School of Public Health, 21650 Oxnard St. Suite 550, Woodland Hills, CA 91637; Tel: 818-676-2872; Fax: 818-7159934; Email: legorreta@ucla.edu Research Objective: To enhance existing methods used by commercial payers to profile physicians based on cost of care delivery in a combined health maintenance organization and preferred provider organization setting. Study Design: Two years of administrative facility, physician, and pharmacy claims ending in first quarter 2005 were employed to identify complete episodes of care using the Symmetry Episode Treatment Groups (ETG)™ grouper software. Episodes that were cost outliers, assigned to members not continuously enrolled during the report period, or did not occur in the most recent year were excluded. Episodes with one physician responsible for a majority of total costs were attributed to that provider. The total episode set was then filtered to include only episodes in the top 50th percentile in terms of both cost and frequency for each specialty as long as 80% of total costs were also included. If this criteria was not met then the top 25th, or 5th percentile was used in stepwise fashion as needed. The traditional ETG risk adjustment approach estimates the expected value for individual episodes by averaging the costs of all episodes within the same type of ETG. The ratio (Observed/Expected) or the standardized cost difference (SCD) between the average actual and expected costs for each physician is then calculated to characterize physician practice efficiency. To improve predictive accuracy, we refined this mean approach by computing the predicted costs from a Gamma regression model based on each “Super ETG” category—a group of similar types of ETGs (e.g., chronic bronchitis with and without complications and comorbidities)with additional adjustment for demographic and disease severity characteristics such as age, gender, physician specialty, complication, level of comorbidity , and medication burden. The predictive accuracy of the risk adjustment model was assessed by the R-squared statistic (percentage of variation explained). Population Studied: A large Midwest commercial health plan including approximately 2 million member lives and 10,000 physician groups was examined. Principal Findings: We identified nearly 1.3 million episodes which corresponded to 1,620 physician groups in 50 specialties. ETG risk adjustment models with the traditional mean approach predicted medical costs with a R-squared of 0.633. The regression modeling approach based on SuperETGs explained more total variance (R-squared=0.882). Conclusions: Regression modeling techniques adjusting for demographics and disease severity improved the predictive accuracy of the existing risk adjustment system. Their application, in combination with the common risk adjustment methodology (ETG), is suggested for use in economic cost profiling of physicians. Implications for Policy, Delivery, or Practice: Measuring providers on both quality and cost is of critical importance in light of the continued interest to curb rising healthcare costs through improved efficiency. Peer-to-peer comparison of the cost of providing care can help increase physician awareness and accountability of how practice patterns drive increased health care spending. This type of profiling may be used by commercial payers to identify physicians who provide improved value in health care. Primary Funding Source: No Funding ●Measuring Managed Care and Its Environment Using National Surveys: a Review and Assessment Su-Ying Liang, Ph.D., Kathryn A Phillips, Ph.D., Jennifer Haas, M.D. Presented By: Su-Ying Liang, Ph.D., Associate Researcher, Clinical Pharmacy, University of California San Francisco, 3333 California Street, Suite 420, San Francisco, CA 94143; Tel: 415514-0457; Fax: 415-502-0792; Email: liangs@pharmacy.ucsf.edu Research Objective: (1) to review empirical measures and analytical examples for examining the impact of managed care, managed care markets, and other characteristics of the area where an individual resides using two widely used national surveys, the Medical Expenditure Panel Survey (MEPS) and the National Health Interview Survey (NHIS); and (2) to discuss practical issues common in the design, analysis, and the interpretation of studies using these surveys. Study Design: We adopt a previously developed framework of health plan factors that guides our review of managed care concepts. This framework is based on a review of literature from 1990-2000 and is updated to incorporate newly developed managed care concepts from the 2001-2005 literature. We map the managed care concepts onto available data sources. We summarize empirical measures of managed care in MEPS and NHIS and measures of area characteristics in other datasets linkable to MEPS/NHIS including the Area Resource File and US census data. We provide empirical applications of these measures and discuss analytical issues that should be considered. Population Studied: Synthesis of dataset documentation, reports, and research articles. Principal Findings: The framework of health plan factors identifies thirteen categories of measures from MEPS and NHIS that can be used to examine health plan characteristics relating to managed care. The framework also identifies the areas in which these national surveys can be improved to advance research on managed care. We discuss the following analytical issues: (1) categorization and interpretation of measures of health plan characteristics; (2) defining markets and relevant areas to measure area characteristics; and (3) analyzing data at multiple levels, including statistical modeling and software options. Conclusions: Despite numerous analytical challenges and the lack of data from the provider perspective, MEPS and NHIS are rich sources of data for examining the impact of health plans and the characteristics of markets or areas on health care expenditures and outcomes. Implications for Policy, Delivery, or Practice: National surveys may consider collecting additional data to allow examination of new innovations of managed care, including consumer-directed health plans, pay for performance, and multi-tiered networks. Primary Funding Source: NCI ●How Does A Specialty Pharmacy Provider Program Impact a Health Insurer? Andrea McAllister, B.S., Andrea DeVries, Ph.D. Presented By: Andrea McAllister, B.S., Decision Support Specialist, Knowledge Discovery, Highmark Inc., 120 Fifth Avenue, Pittsburgh, PA 15222; Tel: 412-544-0664; Fax: 412-5440700; Email: andrea.mcallister@highmark.com Research Objective: To study the impact of the implementation of a specialty pharmacy program for Remicade in Highmark’s Managed Care and PPO products. Study Design: Costs, including administration (admin) fees, and utilization were compared for the twelve months prior and post program which was enacted April 2004. The two main diagnoses of interest were Crohn’s/Ulcerative Colitis and Arthritis. Various measures were calculated (number of claims per patient, Highmark cost per claim, and Highmark cost per patient per month (PPPM)) and were stratified by procedure type (treatment, admin), claim type (outpatient, professional), region of service delivery (Western vs. NonWestern), and Insurance category (Manages Care, PPO, and Traditional) as well as other indicators. Additionally, an analysis was completed examining “total visit” cost. In this case the cost of a “total visit” was equivalent to the treatment cost plus any admin fees occurring during the seven days subsequent to the service date of the treatment claim. Population Studied: All Highmark members receiving Remicade treatment twelve months prior and post program implementation. Principal Findings: Remicade treatment claim costs decreased by 3% among the Western Managed Care Products for both Arthritis and Crohn’s disease. Of the approximately 10,000 Remicade claims during the post program period, about 30% were provided by the specialty pharmacy provider. Major sources of the non-specialty pharmacy provider claims were categories in which the specialty pharmacy provider cannot be used as a provider: Out-of-area (Blue Card), Western region outpatient facility, and Central region (professional and outpatient facility). There was an 11% decrease in outpatient facilities claims coupled with a 4.7% increase in professional office visit claims after the program went into effect. Conclusions: The specialty pharmacy provider program resulted in decreased costs to Highmark as well as a shift in site-of-service as seen by opposing changes in outpatient facility and professional office visit utilization. There was an added incentive to professional providers in the form of increased administration fees. This resulted in increased total professional visit costs but is offset by the decrease in outpatient facility claims which cost more. Compliance to the program appeared to be quite high at 30% considering large categories of claims could not be accessed by the specialty pharmacy provider. The impact of the specialty provider program in Highmark’s Western Region for Managed Care products on site-of-service for Remicade treatments translates to a total savings to Highmark of $535k per year. Implications for Policy, Delivery, or Practice: When taking this total savings result into consideration with the fact that a specialty provider would responsible for multiple drug treatments at an insurer the potential for savings due to shift in site-of-service alone is great. This result can be used as evidence to support the use of a specialty pharmaceutical provider by an insurance plan for all drug treatments of which administration can be performed in either the outpatient or professional office setting. Considering specialty pharmaceuticals are greater than a 20% share of entire pharmacy budgets for an insurer these type of strategies for cost savings are highly desirable. Primary Funding Source: No Funding ●Life Style Coaching: Does the Telephone Work? James Naessens, M.P.H., James Purvis, MS, Rachel Carroll, BA, Barbara Kreinbring, RN, M.B.A., William Litchy, M.D., Holly Van Houten, BA Presented By: James Naessens, M.P.H., Clinical Associate, Health Care Policy & Research, Mayo Clinic, Pavilion 3, Rochester, MN 55905; Tel: (507)284-5592; Fax: (507)284-1731; Email: naessens@mayo.edu Research Objective: To assess the effectiveness of telephonic life style coaching on subjects enrolled in weight control, stress, exercise and nutrition programs. Study Design: A pre-post cohort study comparing baseline and 6 month post-intervention subject-reported status on general health, confidence, healthy practices and outcome variables was performed for each of four lifestyle coaching programs. Primary data was collected with telephone interviews at enrollment, during intervention and at 6 months. The intervention consisted of a life style coaching program with regular telephone contact between subject and coach over a 13 week period. Assessment variables were created by comparing 6 month to baseline values or, where appropriate, analyzing a reported variable reflecting the subject’s assessment of change from baseline. Paired analysis was performed with t-tests, rank-sum tests and Chi-square tests, as appropriate. An intent-to-treat analysis was attempted with inclusion of all participants completing the follow-up interview. Population Studied: Life style coaching enrollees were eligible for employee-sponsored life style coaching benefits and were identified with appropriate risk factors through a health risk appraisal. All enrollees came from across the US from a single multinational company in 2004. Weight control included 1114 participants, exercise included 334 participants, nutrition included 259 participants and stress management included 281 participants. Principal Findings: Significant outcome improvements were seen in weight control (drops in BMI and weight), exercise (improvement in energy levels), stress (reporting fewer stressrelated problems), and nutrition (increased energy and drop in BMI). The average weight loss for the 938 participants with starting and ending weights was 4 pounds. 58% of weight loss participants had at least a 5% or 5 pound weight loss. Significant improvements in confidence dealing with their risk factor were seen among the participants in weight control and stress. Those in the exercise program were inadvertently asked how confident they were at increasing their exercise level. Nutrition program participants had a significant drop in confidence. Significant improvements in healthy practices were seen among participants in all four programs. Significant improvements in self-reported general health were seen among participants in weight control, nutrition and exercise. Conclusions: Telephonic life style coaching appears to be effective in improving outcomes and healthy practices for participants who enter those programs with identified health risks. Participants in weight control, nutrition and exercise programs reported significant improvements in general health. Implications for Policy, Delivery, or Practice: Telephonic delivery of life style coaching would expand the reach of programs to improve health and reduce risk. Further research should be performed to determine whether participants’ improvements were due to program content or frequent attention. Primary Funding Source: No Funding ●Examining Health Reimbursement Account Exhaust in a Consumer Driven Health Plan Ross Owen, M.P.A., Regina Levin, M.P.H. Presented By: Ross Owen, M.P.A., Research Analyst, Research, Definity Health/United Health Group, 1600 Utica Ave. S., St. Louis Park, MN 55416; Tel: 952-277-6013; Fax: 952277-5502; Email: ross_owen@uhc.com Research Objective: The central feature of consumer driven health plans (CDHPs) is the presence of an account from which enrollees receive “first dollar” coverage as well as the opportunity to roll funds over from year to year. Despite the increasing prevalence of such plans, relatively little is known about how the health reimbursement account (HRA) functions differently for members based upon their characteristics and utilization behaviors. This study evaluates the impact of a large commercial group of CDHP enrollees' traits and utilization on their length of time to HRA account exhaust. Study Design: This retrospective cohort analysis uses Cox Regression to examine enrollees over a one year period (calendar 2004) as long as their HRA dollars remained. The number of months to HRA exhaust serves as the dependent variable, and enrollee characteristics and behavior during the HRA period serve as the covariates in the proportional hazards model. Population Studied: The study group is composed of a cohort of commercial (Definity Health) CDHP enrollees grouped at the family level (n=79,050 families, representing ~188,000 lives). The sample contains all continuously enrolled employees from Definity’s 2004 book of business who had plan years that started on January 1st of the study year, as well as any spouses and dependents enrolled with them. Principal Findings: The hazard ratios for each covariate in the Cox Regression model quantify the expected change in the hazard (or instantaneous risk) of a family exhausting their HRA given a unit change in that variable with all other variables held constant. The utilization variables, as expected, increase the hazard of exhaust in proportion to the intensity of utilization. Three additional non-utilization variables have sizable hazard ratios: family size (hazard ratio = 1.295, 95% C.I. = 1.285 – 1.306), chronic disease prevalence (hazard ratio = 1.287, 95% C.I. = 1.270 – 1.303), and the amount of HRA funds rolled over from the previous year (hazard ratio = .899, 95% C.I. = .896 - .902). Conclusions: Controlling for utilization and severity of illness, the model demonstrates that larger families and families with more chronic disease had increased risk of exhausting their HRA. The model also quantifies the impact of rollover dollars in reducing the hazard of exhaust, exploring the impact of members’ ability to roll these funds over from year to year on the HRA’s effectiveness in protecting members from out-ofpocket costs. Implications for Policy, Delivery, or Practice: This study provides a rare look at a large sample of commercial CDHP enrollees’ interaction with their plan’s central feature. Given the increasing prevalence of CDHP and the amount of focus given account-based coverage, results like these do much to evaluate how this plan design functions “on the ground”. These results will be of interest to practitioners and policymakers as they relate to benefit design and the suitability of CDHP for enrollees with widely varying health needs. With better knowledge about how the HRA will work differently for individuals and families based on their composition and behavior, CDHPs can more precisely target financing mechanisms, communication strategies, and incentives to encourage clinically beneficial and cost-effective care. Primary Funding Source: Definity Health/United Health Group managed care and government constraints, which are restraining growth about one percent per year. Summing up, we estimate that the current stimulus to real per capita personal health care spending to be in the range of 4.5-5 percent per year. Conclusions: Because the growth in real per capita GDP is only expected to be two percent per year or a little more, our results indicate that the percentage of GDP devoted to health care will grow from 16 percent in 2004 to about 21 percent in 2014. Serious attempts to slowdown health care spending growth might consider consumer-friendly ways to reduce insurance coverage. Implications for Policy, Delivery, or Practice: High deductible insurance coupled with health savings accounts may be a way of lowering coverage in a consumer-friendly way for the working age population. However, because government health care spending (46 percent of the total in 2003) is quickly approaching half of all health spending as the baby-boomers retire, consumer options leading to lower coverage in this portion of the market might be considered as well. Primary Funding Source: American Enterprise Institute ●Health Insurance and the Rising Cost of Health Care Edgar Peden, Ph.D., Mark S. Freeland, Ph.D. ●Deductible-based Cost-sharing: Informed Consumers or Uninformed Confusion? Mary Reed, Dr.P.H., Richard Brand, Ph.D., Joe Newhouse, Ph.D., Joe Selby, M.D., M.P.H., John Hsu, M.D., M.B.A. Presented By: Edgar Peden, Ph.D., Author, American Enterprise Institute, 7509 Oakmont Dr, Frederick, MD 21702; Tel: (301) 473-8553; Email: rhohat@adelphia.net Research Objective: An empirical analysis of the factors driving medical spending growth from 1960 through 2003. Study Design: Using data from the National Health Accounts, and other national data, our study is based on a time-series (regression) analysis of the health spending growth and its causal factors. Population Studied: U.S. population and health care market. Principal Findings: Our analysis of the post World War II health market experience shows that government policies have fostered ever greater insurance coverage both indirectly (employer tax deductibility) and directly (Medicare and Medicaid). In the late 1940s consumers paid 70 cents on the dollar out-of-pocket for health care, but in the ensuing 50 plus years it dropped almost continuously so that in the early 2000s it had fallen to less than 15 cents. This inter-temporal change has exacerbated the familiar ‘moral hazard’ problem of over consumption of health care. But more importantly it has also resulted in ready markets for high cost medical technology, as consumers have leveraged their out-of-pocket spending to pay for ever more sophisticated, ever more costly medical services. The marginal benefits from this higher spending may be less than the marginal cost. The econometric model we estimated for 1960-2002, implies that insurance coverage (the percentage of medical spending paid by insurers) and its growth are currently driving real per capita personal health care spending higher at the rate of 3.5 percent per year. This stimulus is being abetted by growth in real permanent income, non-commercial medical research spending, provider monopoly power, and the changing age/gender population mix. Together these account for another 2-2.5 percent per year. Offsetting these factors are Presented By: Mary Reed, DrPH, Research Associate, Division of Research, Kaiser Permanente, 2000 Broadway, Oakland, CA 94612; Tel: 510.891.3808; Fax: 510.891.3606; Email: mer@dor.kaiser.org Research Objective: Deductible-based plans with high levels of cost-sharing could encourage patients to participate more actively in their health care. Effective incentives, however, require that consumers are well-informed. We examined patient knowledge of their deductible plans in a population with a new deductible plan. Study Design: We conducted a cross-sectional telephoneinterview study in a random sample of adult health system members, with over-sampling of members with chronic diseases (asthma, coronary artery disease, heart failure, hypertension, or diabetes). Participants reported whether they faced any deductible, the amount of their deductible, and which medical services applied to their deductible. We compared these patient reports with each respondent’s actual deductible structures and amounts using the health-plan’s automated databases. Population Studied: All 479 study participants had a deductible: for 56.5%, the deductibles applied only to hospital care; for the remaining 43.5%, the deductible applied to all care except office visits. Participants had a mean age of 44.8 years, 52.4% were female, and 76.0% reported being of white race/ethnicity; 49.7% had “excellent” or “very good” health status; 49.7% had less than a college-graduate education; and 17.1% had an annual household income of less than $35,000. Principal Findings: While all respondents actually faced a deductible, 44.6% of participants were not aware that they faced any deductible. Among those who reported having a deductible, 54.7% did not know that they had to pay full price for hospital care before reaching their deductible. Similarly, less than half knew that their actual deductible applied to other services: 34.8% knew that emergency department care was included in their deductible and 33.9% knew that medical tests were included. Although no one’s deductible applied to office visits, 39.3% believed that office visits were included toward their deductible. In multivariate analyses, respondents whose deductibles applied to all care (except office visits) were significantly more likely to know that they faced a deductible than those whose deductibles applied only to hospital care (OR: 1.89, 95%CI: 1.23, 2.94). In addition, those with nonwhite race/ethnicity (OR: 0.57, 95%CI: 0.36, 0.89) and who had “excellent” or “very good” self-reported health (OR: 0.59, 95%CI: 0.38, 0.91) were significantly less likely to be aware of their deductible. Conclusions: Almost half of patients did not know that they faced a deductible at all. Among patients who did know about their deductible, most were unaware of the specific medical services that were included or excluded from their deductible. Implications for Policy, Delivery, or Practice: New deductible forms of cost sharing have the promise to involve patients more directly as consumers of healthcare, but effective patient participation requires detailed knowledge of their benefits. We found that patients had limited understanding of their plans. This confusion could inhibit the ability of these plans to engage patients in their care. Additionally, these inaccurate perceptions could even lead some patients to avoid care unnecessarily. More research is needed to understand how these deductibles affect patient care-seeking behavior, especially as benefits grow increasingly complex. Primary Funding Source: Kaiser Foundation Research Institute ●More Than Half of Californians in HMOs are Overweight or Obese Dylan Roby, Ph.D., Gerald Kominski, Ph.D., Natalya Mostovskaya, BA Presented By: Dylan Roby, Ph.D., Senior Research Associate, Center for Health Policy Research, UCLA, 10911 Weyburn Ave, Suite 300, Los Angeles, CA 90024; Tel: 310-794-3953; Fax: 310794-2686; Email: droby@ucla.edu Research Objective: This project examines overweight and obesity among HMO enrollees in the state of California. Study Design: Using data from the 2003 California Health Interview Survey (CHIS), it examines the proportion of private HMO enrollees who are overweight or obese (based on BMI and age) in each of the seven major health plans in the state. The study also calculated rates by different age and ethnic categories. The UCLA Center for Health Policy Research collected information on height, weight, and health plan name during the most recent California Health Interview Survey. Population Studied: Private HMO Enrollees, ages 12-64. Principal Findings: More than five million Californians enrolled in private HMOs - over half of all private HMO enrollees aged 12-64 - are overweight or obese. Members of Aetna (59%), Kaiser (54%), and Health Net (53%) reported higher combined prevalence of overweight and obesity relative to the statewide rate; Cigna (52%) was comparable to the state average; while members of Blue Cross (51%), Blue Shield (49%) and Pacificare (49%) had combined overweight and obesity rates lower than the statewide average. Aetna had the highest percentage of obese enrollees (23%), while Blue Cross had the lowest (17%). Conclusions: Our estimates provide a baseline measure of overweight and obesity prevalence rates in California's HMOs, and thus provide a starting point for assessing whether plans are making progress in dealing with this important threat to public health in the future. Since HMOs provide insurance to almost half of California's population, it is essential that they take further action to improve the current baseline of overweight and obesity rates in HMOs. Implications for Policy, Delivery, or Practice: HMOs face a serious challenge in addressing the problem of overweight and obesity among their members, and should actively design appropriate interventions to address this problem. Emphasizing diet, exercise, and regular healthy activity are important steps in confronthing this nationwide epidemic. Health plans can also take steps to help the process, such as physician incentives to discuss weight management and providing educational materials to enrollees. Several health plans, including LA Care and Kaiser Permanente, are actively involved in weight management strategies for their members. Primary Funding Source: California Office of the Patient Advocate ●The Early Disease Detection Study Chris Stehno, MBA (magna cum laude), University of Notre Dame; Graduate Studies in Bio-Statistics, University of Iowa; BS in Biology, University of Kansas Presented By: Chris Stehno, MBA (magna cum laude), University of Notre Dame; Graduate Studies in Bio-Statistics, University of Iowa; BS in Biology, University of Kansas, Healthcare Management Consultant, Denver Health Practice, Milliman, Inc., 1099 18th Street, Suite 3100, Denver, CO 80202-1931; Tel: (303) 299-9400; Fax: (303) 299-9018; Email: chris.stehno@milliman.com Research Objective: Disease Management (DM) programs have been plagued by faulty and judgmental ROI estimations and conclusions. The fact is that current predictive models used in DM rarely can predict an event in the early or predisease states wherein it is well known that intervention can make a significant difference. The problem lies not in the models themselves, but rather in the data being used to power these models. This paper will discuss the statistical significance of lifestyle-based data in relationship to early disease detection. Study Design: The use of historical medical data as a predictor of future events has been the mainstay for the insurance and healthcare industry for over 50 years. This study looks at the statistical significance of historical medical data, lifestyle data, and the combination of the two when defining the at-risk population for four different events: aortic aneurysms, atherosclerosis, osteoporosis, and stroke. Population Studied: Our population involved over 100,000 individuals who participated in a series of medical screenings. These screenings included: ultrasound imaging measurement of the abdominal aorta, Ankle Brachial Index (ABI index) test, Bone Mineral Density (BMD) measurement, and Doppler ultrasound imaging of the carotid arteries for blood flow velocity. In addition, participants provided medical history and self-reported lifestyle-based data elements. Principal Findings: The lifestyle elements provided as good as or better predictive capabilities than the medical history elements for at-risk status in all four of the tests studied. The combination of the two datasets, historical medical and lifestyle-based, provided the optimal results in all four of the cases. Conclusions: Current predictive models and data mining techniques based solely on historical medical inputs seriously impede our ability to predict diseases in the early onset or predisease state. The need to look ‘out of the box’ at new datasets, such as lifestyle-based datasets, is critical to fully developing disease management and wellness initiatives. Implications for Policy, Delivery, or Practice: ROI questions surrounding disease management initiatives will continue to plague the industry as long as DM applications focus on late stage disease identification and management techniques. However, a move to incorporate the early and pre-disease states will provide significant improvements to both costs and patient health. In order to achieve these goals, a new way of thinking and new datasets are necessary to move our antiquated predictive models into the next generation. Primary Funding Source: No Funding ●Severe Medical Conditions and Loss of Health Insurance Coverage: Evidence from the Health and Retirement Study Hsin-yu Tseng, Ph.D. (expected in June 2006) Presented By: Hsin-yu Tseng, Ph.D. (expected in June 2006), Ph.D. candidate, Department of Economics, University of Chicago, 5550 S. Dorchester Ave. Apt. 607, Chicago, IL 60637; Tel: 773-256-1392; Email: htseng@uchicago.edu Research Objective: The diagnosis of severe medical conditions has competing effects on health insurance coverage. On the one hand, it increases the demand for coverage because individuals diagnosed with severe medical conditions (high-risk individuals) not only face a substantially higher risk of future health problems, but also need the coverage to pay for regular checkups helping to manage their medical conditions. On the other hand, this diagnosis may place individuals at risk of losing coverage due to declines in resources and due to risk discrimination in private health insurance markets. Individuals would want to avoid this risk because loss of coverage hinders their consumption smoothing and may exacerbate their medical conditions (e.g., Institute of Medicine, 2003; Levy and Meltzer, 2004). Public insurance and price restrictions in private markets aim to reduce this risk. However, little is known about whether severe medical conditions affect the likelihood of losing coverage and whether these price restrictions affect this likelihood. To address these issues, this paper conducts two tests. First, I test whether high-risk individuals, specifically high-risk policyholders who purchase coverage from the individual (non-group) health insurance market, continue or lose coverage. In the individual market, risk discrimination is prevalent in most states because whether coverage is offered and at what premium rate depends on risk status. In the employment-based (group) market, however, federal laws prohibit individual risk discrimination. Second, I test whether price restrictions in insurance markets result in an adverse selection effect leading low-risk policyholders to drop out of the market. I conduct this test by exploiting state variations in the degree of risk discrimination in the individual market. Some states implement price restrictions, while the other states do not. These restrictions limit the range of variation in premium rates due to risk status. Thus premium rates charged to high-risk policyholders are lowered, but rates charged to low-risk policyholders are increased. In a voluntary market, these restrictions may result in an adverse selection effect in which high-risk policyholders retain coverage, but lowrisk policyholders drop out of the market. Study Design: I use individual data on 50-to-64-year-old Americans from the Health and Retirement Study (HRS). Conditional on current insurance status, I focus on changes in insurance coverage at the extensive margin for two reasons. First, the HRS statistics indicate that approximately 98% of the policyholders have comprehensive plans. These statistics imply that coverage changes at the extensive margin result in larger financial risk changes than changes at the intensive margin. Second, HRS provides limited, qualitative measures on intensive-margin changes. Additionally, the coverage analysis is augmented by an analysis of premium payment level because HRS is one of the few public data sets that contain information on premium payments. I divide high-risk individuals into two groups: (1) individuals who “just become high-risk;” (2) individuals who “have been high-risk for two or more years.” I compare changes in insurance coverage between each of these two groups and a low-risk reference group. This division is justified because the diagnosis of severe medical conditions resembles a permanent health shock, and the degree of risk discrimination may vary across the duration of experiencing medical conditions. Population Studied: 50-to-64-year-old Americans Principal Findings: The findings indicate that severe medical conditions have a lagged effect on the probability of becoming uninsured. Policyholders who become high-risk do not lose coverage immediately. This finding may be explained by prohibition of selective increases in premium rates and by short duration of experiencing medical conditions. In contrast, policyholders who have been high-risk for two or more years are 5.1 percentage points (57%) more likely to become uninsured. Three factors can explain this likelihood, including risk discrimination, an imperfect safety net, and an income effect. Four factors, however, cannot, including the crowding out effect of Medicaid coverage or of uncompensated care, better health conditions, and fatalism. Findings from premium payment analysis also support this lagged effect. The findings further indicate that state price restrictions result in an adverse selection effect. Price restrictions are associated with a 4.7 percentage points (45%) increase in low-risk policyholders’ probability of becoming uninsured, and a 5.5 percentage points (24%) decline in high-risk policyholders’ probability of becoming uninsured. The overall coverage rate declines by 2.6 percentage points (20%). Robustness checks indicate that this adverse selection effect is not driven by declines in income. The findings indicate that insurance coverage responds to changes in risk status. This response implies an ex-ante welfare loss. I calibrate this loss and find that two-thirds of the respondents are willing to pay 14% of their income to avoid coverage fluctuation. This welfare loss arises because before diagnosis, risk-averse individuals would want to insure against the risk of losing coverage if they become ill. However, this insurance is imperfect in spite of public insurance and price restrictions in private markets. Conclusions: High-risk individuals have a higher demand for health insurance coverage, but they may be at risk of losing coverage due to risk discrimination in private health insurance markets and due to declines in resources. This paper examines whether high-risk individuals, specifically high-risk policyholders who purchase coverage from the individual market, continue or lose coverage. The findings suggest that policyholders who have been high-risk for two or more years are more likely to become uninsured than their low-risk counterparts. This likelihood implies imperfect “smoothing” in coverage so it indicates an ex-ante welfare loss. I calibrate this loss and find that two-thirds of the policyholders are willing to pay 14% of their income to avoid coverage fluctuation. States have implemented price restrictions to narrow the range of premium rate variation due to risk status. Yet, these restrictions are prone to market failures in the individual health insurance market. This paper confirms the premise of market failure because the findings show that price restrictions have helped high-risk policyholders retain coverage, but result in a negative externality leading low-risk policyholders to drop out of the market. This dropout provides evidence of adverse selection and implies a welfare tradeoff between regimes with and without price restrictions. Future research will employ the joint distribution of risk status and risk aversion to figure out which regime is associated with a larger overall welfare loss. Implications for Policy, Delivery, or Practice: Among policies that address the aforementioned welfare loss, price restrictions with compulsory purchase of catastrophic plans impose the greatest degree of regulation on the health insurance market. Such policies may reduce welfare since individuals cannot choose to be uninsured. However, they do eliminate the risk of losing coverage so welfare may be improved from an ex-ante point of view. Although an attempt to implement universal health insurance coverage failed in 1994, coverage expansion remains a key policy issue. This paper examines why policyholders become uninsured. So, on the flip side, it analyzes how to help the uninsured obtain coverage. In addition to the risk discrimination discussed earlier, the findings imply that the success of coverage expansion hinges on whether the society has a consensus on substantial income redistribution. Income effects play a vital role in dropping insurance coverage; for policyholders who have been high-risk, low income is associated with an 11.8 percentage points (109%) higher probability of becoming uninsured. Medical progress has made genetic testing possible, and it is expected that this testing will become part of the routine physical examination of the twenty-first century (Pyenson, 2003). My findings can be applied to genetic testing because government faces the same challenge when it decides whether the insurer can use genetic information for risk discrimination. Information from genetic testing may facilitate preventive treatments; however, if its use is unregulated, this information may deprive individuals of insurance possibilities much before they actually contract the genetic disease (e.g., Alzheimer’s disease). If the insurer is prohibited from using genetic information, but individuals have access to it, this prohibition may result in an adverse selection effect leading low-risk individuals to drop out of the market. The extreme case of this dropout is the collapse of insurance markets. Primary Funding Source: Chicago Center for Excellence in Health Promotion Economics ●Why Small Are Firms Less Likely than Big Firms to Offer Health Insurance? Chapin White, M.P.P., Ph.D., David Auerbach, Ph.D., Stuart Hagen, Ph.D. Presented By: Chapin White, MPP, Ph.D., Analyst, Health and Human Resources Division, Congressional Budget Office, Ford House Office Building, Room 424c, Washington, DC 20515; Tel: 202-226-4931; Fax: 202-225-3149; Email: chapin_white@post.harvard.edu Research Objective: Small firms are substantially less likely than large firms to offer health insurance. Three explanations are commonly offered for this phenomenon are: higher loading and administrative costs for small firms, differences in the workforces, and adverse selection. Our first research objective is to propose an additional explanation for the low offer rates at small firms, which we term the "dispersion" effect. Dispersion refers to the variation across firms in the firm-level aggregate preferences for health insurance. Small firms will tend to exhibit more dispersion than large firms, even if there is no sorting of workers into firms based on worker preferences, due simply to the law of large numbers. Our second research objective is to gauge the relative importance of those four explanations for the disparity in offer rates between small and large firms. Study Design: We construct a microsimulation model based on the 2001 Survey of Income and Program Participation (SIPP). That microsimulation model is used to measure the magnitude of each of the explanations for the large firm-small firm disparity in offer rates. Each individual in the SIPP is assigned a reservation price (based on income, age, sex, health status, education, Medicaid eligibility and other factors) and a health insurance premium (based on expected medical spending multiplied by an appropriate load factor). Workers are grouped into synthetic firms of various sizes. Those synthetic firms choose, based on the aggregated preferences of their workforces, whether to offer health insurance and, if they offer, what level of employee premium contribution to impose. Workers, if they are offered health insurance, choose whether to enroll in an employer-sponsored family policy, an employer-sponsored single policy, a non-group policy or no policy. The reservation prices and premiums are calibrated so that the microsimulation produces firm-level and individuallevel behavior that roughly matches the observed offer and enrollment behavior reported in the SIPP. Population Studied: The U.S. non-Medicare population. Principal Findings: In the microsimulation runs in which all workers are grouped into synthetic large firms, firm-level offer rates are substantially higher than in the microsimulation runs in which all workers are grouped into small firms. The effects of large-firm versus small-firm loading are minor, as are the effects of adverse selection. Conclusions: The two major factors that account for the discrepancy in offer rates between large and small firms are, first, the composition of the workforce and, second, the dispersion effect. The differences in loading, and the effects of adverse selection account for little of the observed discrepancy. Primary Funding Source: No Funding