Health Insurance Markets & Managed Care Call for Papers The Market for Health Insurance Chair: Jon Christianson, University of Minnesota Sunday, June 26 • 3:30 pm – 5:00 pm ●What Happens When Workers Fail to Take up Employment-Related Health Insurance? Evidence from 1996 and 2001 Didem Bernard, Ph.D., Tom Selden, Ph.D. Presented By: Didem Bernard, Ph.D., Economist, CFACT, AHRQ, 540 Gaither Road, Rockville, MD 20850; Tel: (301) 4271682; Email: dbernard@ahrq.gov Research Objective: A growing percentage of workers do not take up employment-related health insurance, and persons in families with incomplete take-up represent a growing share of the uninsured and publicly-insured population in the U.S. Yet little is known about what happens to families when workers do not take up offered coverage. Do these families face reduced access or financial obstacles to needed care? Do they bear significant financial burdens and/or risks? To what extend do they rely on publicly-funded and uncompensated care? How large are these subsidies compared to the taxrelated subsidies they would have received had they taken up employment-related coverage? The objective of this paper is to answer these questions using data from the Medical Expenditure Panel Survey for 1996 and 2001. Study Design: We examine access to health care, utilization, expenditures, and risk of high expenditure burdens among families of workers who do not take up employment-related health insurance. Population Studied: Families of workers who do not take up employment-related health insurance. Principal Findings: We find that families turning down coverage do not have higher out-of-pocket expenditures on care. Nor do they have increased risk of facing high expenditure burdens as a percentage of income. Rather, our analysis indicates that families turning down offered coverage avoid paying high out-of-pocket burdens by relying on good health, reduced consumption of needed health care, and, when care is consumed, safety net financing through public insurance and uncompensated care. We compare the tax subsidy available to these families for health insurance premiums to the subsidies for health care they receive “through the safety net” when they do not take up coverage. We find that the net flow of subsidies favors not taking up employer offers of coverage. Implications for Policy, Delivery, or Practice: A subject for public policy discussion is that the public and private transfers required to fund these families’ safety net expenditures appear to be substantially larger than the tax-related subsidies for taking up employment-related coverage. Primary Funding Source: AHRQ ●Employer Offers of Health Insurance and Worker Enrollment Decisions: The Role of Health Risk M. Kate Bundorf, Ph.D., MBA, MPH, Bradley Herring, Ph.D. Presented By: M. Kate Bundorf, Ph.D., MBA, MPH, Assistant Professor, Health Research and Policy, Stanford University School of Medicine, HRP Redwood Building, Stanford, CA 94305; Tel: (650) 725-0067; Fax: (550) 723-2586; Email: bundorf@stanford.edu Research Objective: Individuals who decline health insurance offered by an employer represent a growing proportion of the uninsured. The implications of this trend for both the stability of the employer-sponsored market and the likely effects on health for the uninsured depend in part on the health status of those who decline employer-sponsored coverage. In this paper, we investigate the role of health risk in a worker’s decision to enroll in the coverage offered by an employer. Study Design: Using data from the 2-year panel component of the 1996-2001 Medical Expenditure Panel Survey, we develop estimates of an individual’s expected health expenditures based on the presence of medical conditions. We use these estimates to develop measures of health risk which distinguish between demographic characteristics and medical conditions. Using these measures, we compare the health risk of workers accepting and declining health insurance offered by their employer as well as workers not offered health insurance by their employer both overall and by family income. In addition, we use multivariate models to obtain estimates of the relationship between health risk and coverage choices controlling for other demographic and socioeconomic characteristics of workers that affect demand for health insurance. Population Studied: A nationally representative sample of workers 25-64 who were not covered by public insurance at any point during the year. Principal Findings: We find that workers who decline the health insurance offered by their employer are characterized by lower expected health care expenditures than those who choose to enroll, particularly among workers in low and medium income households. In addition, the expected health expenditures of workers in low and medium income households who decline coverage from an employer are similar to workers who are not offered coverage from an employer. Among workers in high income households, we find that workers not offered health insurance from an employer have lower expected medical expenditures than those offered coverage, whether or not they enroll. Conclusions: Selection into coverage based on health risk in the employer-sponsored market indicates that employmentbased coverage does not necessarily effectively pool health risks. Among workers in low and medium income families, this selection occurs both when workers choose jobs and when they make enrollment decisions. For workers in high income families, selection into coverage based on health risk occurs primarily through the choice of a job. Implications for Policy, Delivery, or Practice: Both employers offering health insurance to workers and policymakers developing policies targeted at the employersponsored market to reduce the number of uninsured should consider how benefits may be made more attractive to low risk workers. Primary Funding Source: RWJF, ERIU ●Hospital Market Effects on Uptake and Utilization of Innovative Healthcare Technologies: 1983-2001 Peter Groeneveld, M.D., MS, Gregory Kruse, MSc, Ellen K. Brookstein, BA, Jingsan Zhu, MBA, Zhen Chen, Ph.D., Kevin G.M. Volpp, M.D., Ph.D. Presented By: Peter Groeneveld, M.D., MS, Assistant Professor of Medicine, Center for Health Equity Research and Promotion, Philadelphia VA Medical Center, 9-East, 3900 Woodland Avenue, Philadelphia, PA 19104; Tel: (215)898-2569; Fax: (215)573-8779; Email: peter.groeneveld@med.va.gov Research Objective: Prior studies of market effects such as managed care market penetration on hospital acquisition of healthcare technology have examined static measures of hospital technical capacity, but few studies have investigated actual procedure utilization. The preponderance of prior investigations have been confined to a single type or class of technology. We hypothesized that increases in managed care market share in California during the 1980s and 1990s would have blunted the uptake of a wide array of innovative medical technologies, and that this effect would be most clearly evident in records of actual procedure use. Study Design: We examined hospital discharge data provided by California's Office of State Health Planning and Development, comprising care delivered at all non-federal acute-care California hospitals. Assuming that rapid-growth procedures would be the most likely to demonstrate variation in use, we selected eleven procedures that were generally performed in inpatient settings and that rapidly increased in volume from 1983-2001. Patient-level logistic regression models were used to generate propensity scores for each procedure based on the presence of indicator diagnoses (e.g. cardiac valve disease for valve replacement surgery), comorbidities, age, race, sex, and year of admission. The annual sums of hospitalized patients' propensity scores for each hospital were included in a series of eleven hospital-level, negative binomial regression models predicting use of the eleven procedures. Covariates included each hospital's bedsize, previous year's net income, ownership status, and academic affiliation. We also included health service area (HSA) level measures including managed care penetration, hospital market concentration, physicians-per-capita, urban/rural status, and the percentage of white-collar workers in the population. Model parameters were estimated using generalized estimating equations. Population Studied: Adult patients admitted to non-federal, acute-care California hospitals from 1983-2001. Principal Findings: Eleven new medical procedures were performed in 472 California hospitals located in 28 different HSAs. A 5% increase in managed care market share was associated with higher utilization rates of inferior vena cava filters (rate ratio [RR] 1.11, 95% confidence interval [CI] 1.081.15), cardiac electrophysiology studies (RR 1.25, 95% CI 1.081.45), endoscopic treatment of gastrointestinal bleeding (RR 1.11, 95% CI 1.08-1.14), endoscopic biliary procedures (RR 1.19, 95% CI 1.15-1.24), internal mammary artery coronary bypass grafting (RR 1.14, 95% CI 1.04-1.25), dual-chamber pacemaker implantation (RR 1.43, 95% CI 1.40-1.48), and transpleural thoracoscopy (RR 1.21, 95% CI 1.12-1.30). There were no significant differences in bioprosthetic aortic or mitral valve replacement surgery (p=0.24 and 0.26), cardioverter- defibrillator implantation (p=0.87), or vascular access device implantation (p=0.11). Conclusions: Increasing managed-care market share was associated with greater use of innovative medical procedures by hospitals. This association persisted despite control for variation in the hospitals' case-mix, physician workforce, urban/rural location, socioeconomic status of the local population, and financial stresses from decreasing revenue or local competition. Implications for Policy, Delivery, or Practice: This surprising finding was possibly due to these procedures being sources of increased revenues for hospitals, or because new procedures were more cost-efficient to implement compared to older technologies. Primary Funding Source: Thomas B. and Jeanette E. Laws McCabe Fund ●Why Employer Coverage Changed: 1997-2003 James Reschovsky, Ph.D., Bradley Strunk, BA Presented By: James Reschovsky, Ph.D., Senior Health Researcher, Center for Studying Health System Change, 600 Maryland Avenue SW, Suite 550, Washington, DC 20850; Tel: (202)484-4233; Fax: (202)484-9258; Email: jreschovsky@hschange.org Research Objective: To explore reasons behind changes in employer sponsored insurance (ESI) coverage, 1997-2003. Study Design: This study uses nationally representative survey data on individuals and families from the 1997, 1999, 2001, and 2003 Community Tracking Study Household Survey (n˜60,000 each year) to decompose factors driving changes in ESI coverage rates among the nonelderly. The annual samples were divided into 122 groups, of which 120 were defined by 3 family work status categories (2 full-time workers in the family, one full-time worker, part-time workers only), 8 job quality categories, and (among those with offers) 5 income groups. Job quality was based on a pooled model estimating the likelihood jobs will offer access to ESI, using industry, firm size, and time-adjusted wages as regressors. The highest quality job in each family was assigned to all members. The two additional groups were for those in non-working families and families with only self-employed workers. We calculated offer, eligibility rates and take-up rates for all but the last two groups, using family-level concepts to define offer and eligibility. These data were then used in a shift-share analysis to isolate the change in ESI coverage over time attributable to changes in labor force participation, job quality, offer rates, eligibility rates, income levels among those with access to ESI, and take-up rates. The analysis was replicated for people in 3 socio-economic status (SES) groups, based on whether predicted family income from a human capital regression equation was <200%, 200-399%, or =400% of poverty. Population Studied: Nonelderly persons in 1997,1999,2001, and 2003. Principal Findings: The percent of nonelderly Americans with ESI dropped from 67 to 63.4% between 2001 and 2003. After accounting for population growth, this amounts to 8.9 million fewer people with ESI. The decline came after a populationadjusted gain of 4.5 million people with ESI from 1997 to 2001. Out of that 8.9 million, 5 million (53%) is attributable to lower workforce participation; 2.7 million (30%) to a drop in job quality, only .5 million (5%) to changes in employer offer rates, and 1.5 (17%) million to lower take-up. This was somewhat offset by gains in incomes of those with access to ESI and in eligibility rates. The relative importance of these factors differed in the 1997-1999 and 1999-2001 periods, though work status is consistently important. There is evidence of secular declines in take-up. We classified 19%, 39%, and 42% of the population into low, middle, or high SES categories. These groups accounted for 27%, 57%, and 17% of those losing ESI coverage between 2001 and 2003, respectively. Changes in family work status had its greatest proportional effect on lower SES families, while changes in job quality affected higher SES families the most. Surprisingly, declining take-up affected higher income persons the most in proportional terms Conclusions: Reflecting macroeconomic trends, family work status is a primary factor governing ESI coverage trends. Changes in job quality, a reason often cited as contributing to the long-term decline in ESI coverage rates. Trends in job quality appear to be influenced by macroeconomic trends, explaining improved job quality in the late 90’s. Rising health care costs appear contribute to lower ESI rates by discouraging people from accepting ESI offers. These factors have different effects on different SES groups. Finally, the analysis illustrates the importance of thinking about insurance coverage trends in a family context. Implications for Policy, Delivery, or Practice: Apart from explicit government subsidies, there are two primary avenues for bolstering ESI coverage rates. The first is through macroeconomic and trade policy, which affects work force participation and job quality. Apart from these policies, efforts need to be directed at controlling health care costs. Successful policies to lower costs will increase take-up and, to a lesser extent, offer and eligibility rates. Moreover, lower health care costs will benefit the macro economy, further improving ESI coverage rates. Primary Funding Source: RWJF ●Health Insurance Enrollment Decisions: Understanding the Role of Preferences for Coverage Jessica Vistnes, Ph.D., Alan Monheit, Ph.D. Presented By: Jessica Vistnes, Ph.D., Senior Economist, CFACT, AHRQ, 540 Gaither Road, Rockville, MD 20852; Tel: (301)427-1671; Email: jvistnes@ahrq.gov Research Objective: Policy initiatives to expand access to health insurance in the US remain prominent on the public agenda. However, the weak responsiveness of the uninsured to such efforts has perplexed researchers and policymakers and raised the issue of whether the uninsured value health insurance. Individual valuation of health insurance also has important implications regarding the choice between voluntary and mandatory approaches to expanding coverage. Our purpose is to explicitly examine the association between health insurance preferences and coverage status. We consider whether uninsured adults are more likely to exhibit weak preferences for coverage than are insured adults. We also examine whether preferences for coverage are associated with worker decisions to sort among jobs that offer/fail to offer coverage and to enroll in offered coverage. We compute the subsidy necessary to overcome weak preferences and induce such workers to enroll in offered coverage. We consider whether demographic variables are appropriate proxies for reported preferences. Study Design: We use data from the 2000 MEPS on attitudes toward health insurance, risk, and medical care to characterize individuals and families as having strong or weak preferences for coverage. These responses are obtained independent of insurance status thus minimizing their role as rationales for lack of coverage. We describe the prevalence of weak insurance preferences among the non-elderly US population, its association with health insurance status, demographic, health and economic characteristics. We apply econometric models to examine whether preferences for coverage affect worker sorting between jobs with/without coverage and enrollment decisions. We use MEPS data on health insurance premiums to estimate the subsidy required to offset the utility loss borne by persons with weak preferences were they to enroll in health insurance. Population Studied: The non-elderly U.S. civilian noninstitutionalized population. Principal Findings: Uninsured adults are more likely to report weak preferences for health insurance than are insured adults. Econometric analyses reveal that single workers and married couples with weak insurance preferences are less likely to sort into jobs that offer coverage and less likely to take up offered coverage. We find the premium subsidy necessary to overcome the disutility associated with weak preferences for coverage is equivalent to the out-of-pocket premium costs faced by such workers. Conclusions: We find striking differences in attitudes toward health insurance between insured and uninsured adults. Preferences for health insurance play an independent role in determining whether single workers and married couples obtain jobs with health insurance and whether they take-up offered coverage. Implications for Policy, Delivery, or Practice: There may be a considerable gap between perceptions of policymakers and some uninsured persons regarding the social and private value of health insurance. Mandatory provision of health insurance is likely to lead to welfare losses for some uninsured but voluntary efforts are likely to require subsidies as well as educational efforts to inform targeted groups among the uninsured about the value of health insurance. Primary Funding Source: Robert Wood Johnson Foundation/University of Michigan Economic Research Initiative on the Uninsured Call for Papers Insurance & Pharmaceuticals Chair: Timothy McBride, Saint Louis University Sunday June 26 • 5:30 pm – 7:00 pm ●The Relationship Between Prescription Drug Benefits and Use and Physician Office Benefits and Use William Cecil, MBA, John Barnes, MBA, MA, Terence K. Shea, PharmD, 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: The prescription drug benefit is frequently designed and employed as if it is a stand-alone product without consideration for influence from aspects of the medical benefits design. An increasingly common tactic is to carve out the prescription drug benefit so that it is vended by a completely separate source from the medical benefits. There is a need to study the relationship between medical and pharmacy benefit designs to determine the level and nature of any relationship. The purpose of this study is to assess the relationship between the physician office visit cost sharing and utilization and prescription drug cost sharing and utilization in a commercially insured population. Study Design: Analysis of the claims records of 448,180 commercially insured Tennesseans for the incurred service year of 2003. We used standard tabulations and regression analysis to characterize and control for demographics and estimate elasticities for cost sharing and other factors that could influence the decision to buy prescription drugs or physician office visits and the number of units bought. Population Studied: A commercially insured population of 448,180 Tennesseans with medical and prescription drug benefit coverage. Principal Findings: Higher prescription drug cost sharing was associated with decreased prescription drug use. Higher office visit cost sharing was associated with increased use of generic drugs, reduced use of preferred brand drugs and average use of non-preferred brand drugs. The decision to buy primary care physician office visits was associated with reduced generic drug use, increased preferred brand use and reduced non-preferred brand use. The decision to buy specialist office visits was associated with reduced generic prescription drug use, increased preferred and non-preferred brand use. Increased prescription drug use was associated with female gender, age and DCG score. Decreased generic drug use was associated with increased distance to physician office. Reduced primary care physician office visits were associated with age, office visit cost sharing, generic drug copay, female gender, distance to physician’s office and the number of specialist visits. Reduced specialist visits were associated with physician office visit cost sharing, generic and preferred brand copay and the decision to buy primary care physician office visits. Conclusions: The physician office visit cost sharing may influence consumer preference for prescription drug use and prescription drug cost sharing may influence consumer preference for physician office visits. Prescription drug use varies with the physician office visit copay level. Consumers, preferred physician office visits and prescription drug use in combinations that allow for generic drugs to be a complement and preferred brand drugs to be a substitute for office visits. Preferred brand drugs are a substitute for generic drugs. Plan costs for prescription drug and physician office visits could be influenced significantly by how physician office benefits are distributed across plan membership. The cost-optimizing bundle occurs at an office visit copay of $20. Implications for Policy, Delivery, or Practice: Benefits managers should not assume that higher cost sharing always means lower costs, as higher cost sharing in medical benefits may cause higher prescription drug costs. There is significant value in considering both medical and pharmacy benefits in the same management construct. Primary Funding Source: BlueCross BlueShield of Tennessee ●Prescription Drug Demand for Psychotropics: the Impact of Out-of-Pocket Payment Jieling Chen, Ph.D. Presented By: Jieling Chen, Ph.D., Health Economist, Health Economic Statistics, Merck Research Laboratories, 10 Sentry Parkway, Blue Bell, PA 19422; Tel: (610) 650-0714; Email: whjane@hotmail.com Research Objective: This paper examines the effect of out-ofpocket payment on the purchase of psychotropic drugs for treating depressive disorders. Study Design: Using the Medical Expenditure Panel Survey (MEPS) data from 1996 to 2000, the numbers of psychotropic purchases among patients with depressive disorder in a time period are modeled using truncated count data models, with covariates that include the out-of-pocket payment and other measurements related to demographics, socioeconomic characteristics, and health status. Potential endogeneity of the out-of-pocket payment variable is handled with an instrumental variable approach. Population Studied: Medical Expenditure Panel Survey (MEPS) data from 1996 to 2000 Principal Findings: We find an overall out-of-pocket price elasticity of -0.225. But the demand elasticity for elderly is quite inelastic (-0.046). We also find evidence of plan manipulation. Conclusions: The demand for psychotropics is responsive to changes in out-of-pocket price, but the degree of responsiveness varies dramatically between depressed elderly and depressed non-elderly. Some degree of moral hazard is likely to be present among the young and the middle age, but not among the elderly. Moreover, insurers have been successful in manipulating prescription drug benefits for depressive patients, particularly among young and middle age patients. Implications for Policy, Delivery, or Practice: In an era with increasing emphasis on high out-of-pocket payments to control moral hazard, highly responsive psychotropic demand, compounded by significant plan manipulation, could aggrevate the undertreatment problem among patients with depressive disorder. For depressed elderly, high copayments will simply increase the financial burden on already financially stressed elderly, thereby reducing their welfare without the benefit of significantly reducing moral hazard. For depressed non-elderly, the moral hazard profile of this disease suggests that imposing a high rate of copayment may actually be socially inefficient. Primary Funding Source: No Funding ●Prescription Drug Copayments, Mail Order Substitution and Stockpiling of Maintenance Medications for Early Retirees with Chronic Conditions Teresa B. Gibson, Ph.D., Catherine G. McLaughlin, Ph.D., Dean G. Smith, Ph.D. Presented By: Teresa B. Gibson, Ph.D., Director, Research Division, Medstat, 777 East Eisenhower Parkway, Ann Arbor, MI 48108; Tel: (734)913.3481; Fax: (734)913.3200; Email: teresa.gibson@thomson.com Research Objective: Many employers provide financial incentives for enrollees to switch to mail order pharmacies from retail pharmacies and to lower-cost generic drugs from more expensive brand name medications. We assess the effects of the introduction of an incentive-based formulary on maintenance prescription drug utilization for early retirees with chronic conditions enrolled in an employer-sponsored plan. Study Design: We take advantage of a natural experiment where one employer introduced an incentive-based formulary and a comparison employer that did not change copayments. A monthly panel data set was created from the 1996-1997 MarketScan database representing the maintenance prescription drug experience. Negative binomial difference-indifference models were estimated to measure the effects of the benefit change. Distributed lag models were estimated to assess the extent of pre-implementation stockpiling of prescription drugs. Population Studied: Continuously-enrolled early retirees and spouses of two large firms with at least one chronic condition (n=6,165). Principal Findings: Utilization of maintenance prescription drugs declined 5.5% (p<.01) after the introduction of an incentive-based prescription drug benefit, a decline that was limited to retail pharmacy prescriptions. In contrast, mail order utilization rose 23% (p<.001), dominated by an increase in mail order brand name prescriptions (31%, p<.001). There was little evidence of generic substitution as the utilization of generic drugs filled in a retail pharmacy dropped 8.9% (p=.001) and mail order generic prescriptions were largely unchanged. No evidence of pre-implementation stockpiling of brand name retail drugs was discovered, although enrollees may have anticipated the change and increased utilization of mail order brand name drugs 15.7% (p=.059) in the month prior to the new benefit and retail generic drugs in the two months prior to the new benefit (5.6% two months before (p=.078) and 6.6% one month before (p=.054)). Conclusions: The introduction of an incentive-based formulary had significant effects on maintenance drug utilization patterns for early retirees with chronic conditions, a subgroup that typically depends upon prescription drugs to maintain and improve health. Contrary to expectations, generic drug use did not increase. However, many enrollees switched to mail order pharmacy for brand name drugs. Implications for Policy, Delivery, or Practice: An increase in prescription drug copayments can produce intended and unintended effects on the maintenance drug consumption patterns of near-elderly retirees with chronic conditions who were enrolled in a relatively generous employer-based health plan. The copayment increase may be associated with unintended effects such as a reduction in consumption of maintenance medications. However, retirees responded to financial incentives and increased utilization of the mail order pharmacy for maintenance medications. In contrast, the generic incentive did not appear to induce retirees to switch to generic maintenance drugs. Primary Funding Source: No Funding ●Prescription Drug Benefit Caps and Drug Adherence: Quitting When the Money Runs Out Mary Reed, MPH Presented By: Mary Reed, MPH, Research Associate, Division of Research, Kaiser Permanente, 2000 Broadway, Oakland, CA 94610; Tel: (510)891.3808; Fax: (510)891.3606; Email: mer@dor.kaiser.org Research Objective: We investigated patient self-reported responses to a prescription drug benefit cap. Study Design: We conducted a cross-sectional telephoneinterview study in a random sample of integrated delivery system (IDS) adult members, with over-sampling of members age 65+ years. Participants reported whether they faced a drug benefit cap (after exceeding a benefit cap, subjects pay the full price of their prescription drugs), the amount of their cap, and whether they had exceeded a cap in the previous calendar year. Participants then reported whether their cost-sharing affected their drug adherence against their physician’s recommendation (took less than prescribed, did not fill a new prescription, or did not refill a prescription). We linked the survey data with information from the IDS’s automated databases. Population Studied: Overall, 298 study participants (32% of all respondents) faced a drug benefit cap. Those with a cap had a mean age of 74 years, and predominantly were female (64%), and all had Medicare insurance. In addition, 22% reported being of White race; 43% had “excellent” or “very good” health status; 75% had less than a college-graduate education; and 62% had an annual household income of less than $35,000. Principal Findings: Overall, 80% of those with a cap accurately reported having this benefit structure. Actual cap amounts were either $500 or $1000; 72% of subjects accurately stated their benefit cap amount, 21% overestimated, and 6% underestimated. Among respondents with a cap, 16% were not adherent to their prescription drug treatment because of the cost ; specifically, 9% took less medication than prescribed, 9% did not fill a new prescription, and 8% did not refill a prescription because of the cost. Additionally, 10% had exceeded their benefit cap in the previous year. In multivariate logistic models, subjects’ who had exceeded their cap in the previous calendar year (OR:4.1, 95%CI:1.4-12.2) were more likely to report being non-adherent Those who had incomes under $35,000 (OR:3.4, 95% CI:1.48.4), who had a college degree (OR:2.5, 95%CI:1.1-5.8) and those with lower age (10 year increment, OR:2.7, 95% CI:1.45.0) were also more likely to report being non-adherent. Conclusions: Nearly one in six patients with a drug benefit cap reported being non-adherent to their drug therapy because of costs. Patients who had exceeded their cap in the past year were more vulnerable to these cost-induced behaviors. Implications for Policy, Delivery, or Practice: Some forms of cost sharing such as benefit caps appear to result in poor drug adherence. More research is needed, especially as partial coverage health plans (e.g. caps, deductibles, donuts) increase in use. Primary Funding Source: Kaiser Foundation Research Institute ●The Effect of Pharmacy Benefits on Initiation of Antihypertensive Therapy Matthew D. Solomon, MA, M.Phil., Jose J Escarce, M.D., Ph.D., Dana P Goldman, Ph.D., Geoffrey F Joyce, Ph.D., Robert H Brook, M.D., Sc.D. Presented By: Matthew D. Solomon, MA, M.Phil., Graduate Fellow, Health, RAND, 1776 Main Street, Santa Monica, CA 90407; Tel: (310)393-0411 x7403; Email: msolomon@rand.org Research Objective: To control rapidly rising prescription drug costs, nearly all health plans have increased their costsharing requirements. This study aims to understand the effect of cost-sharing on the initiation of treatment for hypertension, the most commonly treated chronic condition. Study Design: This is a retrospective study conducted from 1997 to 2000 examining pharmacy and medical claims data that are matched to the health benefits from 7 large employers and 18 health plans. Population Studied: The sample includes 4010 elderly individuals newly diagnosed with hypertension who have employer-based retiree insurance benefits, including insurance for prescription drugs. Principal Findings: Mean time to initiation of antihypertensive therapy was 103.5 days (S.D. 192.6) from the date of initial diagnosis to the date of first prescription among enrollees who purchased drugs in our observation window. In multivariate Cox models, doubling copayments was associated with a decrease in the proportion of patients initiating antihypertensive therapy by 100 days post-diagnosis from 45% to 24%. Concurrent use of non-antihypertensive medications, age, and female gender were positively associated with initiation of antihypertensive treatment. Conclusions: The initiation of antihypertensive medication was negatively associated with increased cost-sharing. Since delays in antihypertensive treatment may be responsible for adverse health outcomes, these results raise concern about the potential health effects of high levels of cost-sharing. Implications for Policy, Delivery, or Practice: These results will help policy makers understand how drug benefits affect antihypertensive adoption and use, and will help physicians address factors that may interfere with treatment regimens they prescribe for their hypertensive patients. CMS should carefully monitor cost-sharing arrangements in the new Medicare Part D benefit to ensure that beneficiaries with chronic diseases adopt needed drugs in a timely manner. Primary Funding Source: AHRQ Related Posters Poster Session A Sunday, June 26 • 2:00 pm – 3:15 pm ●Implications of Incentives Increasing Mail-Service Pharmacy Utilization Bartholomew Clark, Ph.D., Mark Siracuse, Pharm.D., Ph.D., Robert Garis, Ph.D. Presented By: Bartholomew Clark, Ph.D., Assistant Professor, Creighton Health Services Research Program, Creighton University School of Pharmacy & Health Professions, 2500 California Plaza (Boyne 141), Omaha, NE 68178; Tel: (402)280-3724; Fax: (402)-280-4809; Email: bclark@creighton.edu Research Objective: Examine implications of pharmacy benefit plan design incentives that encourage patients to utilize mail-service pharmacies in lieu of community-based retail pharmacies. Study Design: Study design utilized the untreated control group with pretest-posttest quasi-experimental design with non-random assignment. Prescription transactions for identical drug products and dates of service were compared at the per-unit level based on whether they were dispensed by a community-based retail pharmacy or by a mail-service pharmacy. Prescriptions dispensed by a retail pharmacy were considered the control group. The intervention group was comprised of prescriptions dispensed by mail-service pharmacies. Differences in means between retail pharmacies and mail-service pharmacies were examined for the following variables in pharmacy benefit expenditures: drug ingredient costs, pharmacist dispensing fees (i.e., fill fees), patient copayments, amounts owed by plan sponsors and total amounts. “Total amount” is “drug ingredient cost” plus “dispensing fee”. “Amount owed by plan sponsor” is “total amount” less “patient copayment”. The Wilcoxon rank-sum test was utilized for all comparisons. Separate comparisons were performed for all claims, for generic drug claims, for brand name drug claims, and for non-drug supply claims (e.g. diabetic testing strips). Population Studied: Prescription transaction records from five employer groups totaling 19,300 prescription transactions (a total of 9,650 exact matches for drug product and date of service) were examined. Each match consisted of one prescription from a retail pharmacy and one prescription from a mail-service pharmacy on identical dates of service for identical drug products within the same employer group. Prescription claims data were from publicly held and private employers involved in enterprises as diverse as food processing, transportation, and higher education. Principal Findings: Analyses of retail-to-mail comparisons for all transactions generated the following significant results (p < 0.001). Plan sponsors paid 4.0% more for prescriptions filled at mail-service pharmacies. Drug ingredient costs were 9.8% lower for mail-service pharmacy prescriptions. Pharmacist dispensing fees were 98.6% lower for mail-service pharmacy prescriptions. Patient copayments were 46.7% lower for mailservice pharmacy prescriptions. The total amounts were 13.6% lower for mail-service pharmacy prescriptions. Separate analyses of retail-to-mail comparisons for brand name drugs, generic drugs and non-drug supplies revealed that plans sponsors paid 2.1% more for brand name mail-service prescriptions (p < 0.01), 25.6% more for generic mail-service prescriptions (p < 0.001), and 36.5% more for mail-service non-drug supplies (p < 0.001). Conclusions: Despite the fact that mail-service drug ingredient costs, dispensing fees and total amounts were lower, the incentive of discounts in patient copayments for these transactions resulted in higher amounts owed by plan sponsors. Implications for Policy, Delivery, or Practice: Recent actions of large employer groups (e.g. the “Big Three” U.S auto manufacturers) that require their employees to utilize mailservice pharmacies have created a conventional wisdom that conversion of retail prescriptions to mail-service prescriptions will result in cost savings to plan sponsors (i.e., employers). Policies encouraging or requiring such conversions may not achieve the desired result. Public policy that requires full disclosure of pharmacy benefit manager revenue streams and true costs to plan sponsors may be required. Primary Funding Source: Creighton School of Pharmacy & Health Professions Internal Grant ●What Drives Health Insurance Options at the Workplace? Evidence from 1997-2002 MEPS Employer Data Philip Cooper, Ph.D., Kosali Simon, Ph.D., Jessica Vistnes, Ph.D. Presented By: Philip Cooper, Ph.D., Senior Economist, CFACT, AHRQ, 540 Gaither Road, Rockville, MD 20850; Tel: (301)427-1675; Email: pcooper@ahrq.gov Research Objective: 1)To examine trends in the availability of different plan types from 1997 to 2002 using a large, nationally representative survey of roughly 25,000 private sector employers. 2)To assess whether the observed trends were related to changes in the characteristics of employers, workers and labor market characteristics or changes in the relationship between those characteristics and the plan types offered. Study Design: We analyze data from repeated cross-sections of the 1997-2002 Medical Expenditure Panel Survey -Insurance Component List Sample, which ascertains whether an employer offers health insurance and if so, collects detailed information on up to four health plans. The survey also contains information on employer and employee characteristics. For the first research objective, we present trends in indicator measures of whether establishments offer any Fee-for-Service (FFS), HMO, Preferred Provider Organization (PPO) or Point-of-Service (POS) plans by firm size. For establishments offering two or more plans, we also present trends in the combination of plans offered to employees. We then estimate econometric models and perform simulations to assess whether the observed trends in the availability of different plan types were related to changes in the characteristics of employers, workers and labor market characteristics or changes in the relationship between those characteristics and the plan types offered. Population Studied: Private sector establishments in the US. Principal Findings: Employers offered workers a very different mix of health insurance plans in 2002 compared to 1997. Specifically, there was a steady decline in the availability of FFS plans and an increase in the availability of PPOs. Additionally, we find different trends for small and large firms. Lastly, we find that characteristics of employers, employees and local labor markets were fairly stable over this period and therefore do not predict observed trends in plan type availability. Conclusions: We find that characteristics of the employer, labor force and local markets were fairly stable over this period and therefore do not predict much of the observed trends in plan type availability. Instead, relationships between certain employer, worker, and other characteristics have changed over this period. Implications for Policy, Delivery, or Practice: Understanding the reasons for the change in prevalence of certain types of plans offered by employers, particularly managed care, is important for policymakers’ assessment of whether the current employer provided health insurance system functions in a manner that provides employees with adequate choice. Primary Funding Source: No Funding Source ●Public Sector Retiree Health Benefit Challenges Under GASB 43 and 45 Julia Costich, JD, Ph.D. Presented By: Julia Costich, JD, Ph.D., Assistant Professor, Health Services Management, University of Kentucky College of Public Health, 121 Washington Avenue, Lexington, KY 40536; Tel: (859)257-6712; Fax: (859)257-3909; Email: julia.costich@uky.edu Research Objective: Government Accounting Standards Board (GASB) statements 43 and 45 require public sector employers or multi-employer plans to report net present liability for future retiree benefits on an accrual basis, in contrast with the current prevailing practice of accounting for such benefits on a pay-as-you-go basis. The challenges to public sector retiree health benefit programs presented by the GASB statements are assessed along with an analysis of employers' potential responses. Study Design: We analyze available public sector employer responses to GASB 43 and 45 in light of simultaneous pressure on retiree health plan financing from rising costs, volume of impending retirees, competing entitlements such as public education and Medicaid, and state budget constraints. Public sector considerations are compared with those taken by the corporate sector in response to FASB 106 in the early 1990s. Population Studied: Public sector employers and retiree health benefit plans covering state, school system, municipal, public university, and other public sector employees. Principal Findings: The largest public sector retiree health benefit plans will be required to comply with the new reporting standards beginning in 2006; employer compliance begins in 2007. Since FASB implementation, some 46% of private sector firms have capped retiree health benefit contributions to avoid open-ended liability. Other strategies under consideration in the public sector include longer service requirements, proportionally lower employer contribution rates to monthly premiums, and disaggregation of active and (under-65) retiree rate computations. The availability of Medicare prescription drug coverage may mitigate the fiscal impact of the new standards, but the magnitude of aggregate retiree health benefit liability for public sector organizations is already a source of serious concern. Conclusions: The scope of health coverage for public sector retirees is threatened by GASB 43/45 implementation. The high risk of overreaction to the size of accrued liability makes informed, principled assessment of alternatives imperative. Implications for Policy, Delivery, or Practice: Retiree health benefit cutbacks in reaction to GASB 43/45 implementation put public sector employees at risk for inadequate postretirement coverage and have the potential to shift significant cost to Medicare and Medicaid. Primary Funding Source: Institutional ●National Trends in Length of Stay and Associated Hospital Charges Before and After State and Federal Postpartum Stay Legislation Matthew Davis, M.D., MAPP, Devesh K. Tiwari, MPP Presented By: Matthew Davis, M.D., MAPP, Assistant Professor, Pediatrics, Internal Medicine, & Public Policy, University of Michigan, 300 NIB, Ann Arbor, MI 48109-0456; Tel: (734) 615-3508; Fax: (734) 764-2599; Email: mattdav@med.umich.edu Research Objective: In response to declining length of stay (LOS) for postpartum discharges during the early 1990s attributed to the implementation of managed care, several states enacted postpartum LOS legislation beginning in 1996. State laws mandated that insurance companies cover a postpartum stay of =48 hours for uncomplicated vaginal births and =96 hours for cesarean sections, but such regulations did not involve all states and did not apply to ERISA-exempt health plans. A federal postpartum LOS law (implemented in 1998) overrode ERISA and applied across all states. We examined national trends in LOS and related expenditures in the pre- and post-legislation periods. Study Design: We analyzed data from the Nationwide Inpatient Sample (NIS), an annual nationally representative sample of non-federal hospital discharges, to characterize trends in postpartum LOS and associated hospital charges for the years 1993 through 2002. Delivery types were assigned to 4 categories: vaginal deliveries with and without complications and cesarean sections with and without complications. Trends in LOS and associated charges (adjusted for inflation to 2002 US$) were analyzed with respect to patient age and expected primary payer. All analyses were weighted to reflect the complex stratified sampling of the dataset and permit inferences at the national level. Analyses reported below emphasize results for uncomplicated cases that were the focus of the legislation. Population Studied: Cases eligible for analyses in each of the NIS years were identified by diagnosis related group (DRG) codes 370 to 375. Principal Findings: From 1993 through 1995, postpartum LOS for all delivery types fell from a mean of 2.38 days (95% CI: 2.33-2.43) to 2.11 days (2.07-2.15). With the subsequent enactment of postpartum stay legislation in selected states and then nationally, all-delivery LOS increased to a high of 2.56 days (2.52-2.60) by 2002. LOS for uncomplicated vaginal deliveries (61.4%-65.0% of deliveries across all years) mirrored the overall trend, with a decline from 1.93 days (1.89-1.97) in 1993 to 1.71 days (1.68-1.74) in 1995 followed by an increase to 2.14 days (2.11-2.17) by 2002. In contrast, for uncomplicated cesarean sections (14.3%-20.3% of deliveries) there was a slight decline in mean LOS from 1993 (4.00 days; 3.91-4.09) through 1995 (3.61 days; 3.54-3.69), but mean LOS did not subsequently change significantly through 2002 (3.71 days; 3.64-3.78). Notably, inflation-adjusted mean hospital charges for uncomplicated vaginal deliveries and uncomplicated cesarean section deliveries remained unchanged from 1993 through 1995. However, from 1995 through 2002 mean hospital charges for uncomplicated vaginal deliveries increased annually by $263; for uncomplicated cesarean deliveries, charges increased annually by $577. There were no significant time trends in LOS or hospital charges with respect to age or payer. Conclusions: This study, the first to characterize longitudinal trends in LOS and associated charges related to postpartum stay legislation, indicates differential effects of such legislation for vaginal versus cesarean births. While postpartum hospital charges remained steady for uncomplicated births while LOS declined in the pre-legislation period, post-legislation charges rose steadily for vaginal deliveries along with LOS, compared to rising charges but stable LOS for cesarean deliveries. Implications for Policy, Delivery, or Practice: Legislative insurance mandates have the potential to alter practice fundamentally, with economic consequences. Primary Funding Source: No Funding Source ●How Significant are “Mismatches” Between Workers’ Preferences and Employers’ Health Insurance Provision in Explaining Aggregate Patterns of Coverage? Richard Hirth, Ph.D., Reagan Baughman, Ph.D., Michael E. Chernew, Ph.D., Emily C. Shelton, BS Presented By: Richard Hirth, Ph.D., Associate Professor, Health Management and Policy, University of Michigan SPH, 109 South Observatory, Ann Arbor, MI 48109-2029; Tel: (734)936-1306; Fax: (734)764-4338; Email: rhirth@umich.edu Research Objective: The majority of working-age Americans and their families obtain health insurance through employers. To assess the performance of this employer-sponsored insurance (ESI) system, it is necessary to understand how well workers sort into jobs that offer their desired mix of cash wages relative to benefits. However, few studies directly measure the extent of sorting. We attempt to quantify the prevalence of mismatches between workers’ preferences and firms’ insurance offerings, considering two types of mismatch, 1) workers who would desire coverage through their employer, but do not receive an offer from their employer, and 2) workers who do not desire coverage through their employer, but do receive an offer. Study Design: To help assess the quality of matching between workers’ demand for ESI and their employers’ decisions to offer coverage, we divide workers into 8 groups based on ESI offers and takeup, insurance obtained through alternative sources and type of source (e.g., spouse's ESI, nongroup coverage, public program), and a multivariate model estimating the worker's demand for insurance coverage. For workers in each of these 8 groups, we then infer the quality of the labor market match. We considered individuals that received an offer of coverage from their employer as mismatched if they declined the offer. This designation stems from the literature indicating that wage offsets do not operate perfectly at the individual level. Thus, workers declining an offer are unlikely to receive enough additional cash wages to recoup the entire value of the foregone offer. We considered individuals that did not receive an ESI offer mismatched if they were not insured through a public program or another source of private group coverage, and they appear to have a high demand for ESI. High demand was inferred either from the purchase of non-group coverage (which is likely to be undesirable relative to a typical offer of ESI due to higher cost, greater intertemporal premium risk, and/or more limited coverage), or from having a high estimated demand for insurance based on observable characteristics. These "high demanders" were workers we believed to be likely to have desired and taken up an offer of ESI. Our classification scheme is, of course, an approximation, subject to both Type I and Type II errors. For example, we classify workers who were not offered ESI by their own employers and hold group coverage from another source as well-matched. Certainly, some of these workers may have prefered an offer of ESI and are took up the alternative coverage as means of compensating for a poor labor market match. Similarly, some workers declining coverage may be willing to forego some cash wages in exchange for the "option value" or access to a redundant source of coverage. Population Studied: The study population includes workers in the 1996-1997 Community Tracking Household and Followback Survey (CTS). This survey includes households in 51 metropolitan statistical areas (MSA) and 9 rural sites. Survey sites were chosen to yield a nationally representative survey sample. Our analysis sample includes 24,454 working individuals between 18 and 64 years old. The unemployed, the self-employed, and non-workers were excluded. Principal Findings: Based on the classification algorithm described above, most workers (78.1 percent) enjoy labor market matches that appear consistent with their preferences, either receiving an offer that they accept, or not receiving an offer that we believe they would not have valued highly (either because of the availability of alternative group or public coverage, or because they were low demanders of insurance). The remaining 21.9 percent of workers appear to be mismatched. The most common types of mismatches involved workers who declined an offer of ESI (12.9% of the sample declined and held coverage from another source; 2.9% declined and were uninsured). In addition, 2.6% purchased non-group coverage, and 3.4% were uninsured, had no offer of ESI, and were classified as high insurance demanders by virtue of having a predicted probability of coverage in excess of 70%. The prevalence of different types of mismatches across the 60 labor markets represented in the CTS is correlated with several labor market characteristics. Conclusions: Using our classification scheme, it appears that the employment-based health insurance system works well for most workers, at least in terms of whether or not the worker’s desire for ESI matches their employer’s offer decision. We do not explore more subtle manifestations of mismatching such as workers who have insurance but would prefer a plan not offered by their current employer. Nonetheless, more than one worker out of five appears to be mismatched. Implications for Policy, Delivery, or Practice: We identify different types of missorting that may have quite different policy implications (e.g., uninsured workers who would desire insurance, vs. dual income families accepting one member’s offer of ESI and rejecting the other’s offer). For most of the identified mismatches, the primary consequence is lower cash earnings or higher insurance premiums than they would face if they were better matched in the labor market. However, some of the identified mismatches (representing the 3.4% of the total sample) were uninsured, lacked an ESI offer and had characteristics indicating high demand for insurance. These individuals appear to be “involuntarily uninsured” workers who would gain insurance if they were to find a better match or if their existing employer began offering coverage to them. Extrapolating from the analysis sample, these involuntarily uninsured workers and their uninsured dependents may represent more than one in six uninsured individuals in the United States. Therefore, policies either facilitating labor mobility or encouraging offers of ESI by employers of these involuntarily uninsured workers could alleviate a significant source of uninsurance in the United States. Primary Funding Source: RWJF ●The Value of Quality in Healthcare Russell Mardon, Ph.D., L. Gregory Pawlson, M.D., Frank Lichtenberg, Ph.D., Steven Teutsch, M.D., MPH, Pamela Landsman, DrPH, Claire Sharda, RN, MBA Presented By: Russell Mardon, Ph.D., Director of Analysis, Research and Analysis, NCQA, 2000 L Street NW, suite 500, Washington, DC 20036; Tel: (202)955-3567; Fax: (202)9553599; Email: mardon@ncqa.org Research Objective: To quantify the effect of higher quality (as measured by HEDIS indicators) on utilization and cost in disease or demographically homogeneous populations, while controlling for demographic and case-mix differences. Study Design: A longitudinal cohort design with multiple regression estimates based on claims-level hospital, ambulatory, and laboratory data from the Medstat Marketscan database. Sample sizes ranged from 42,000 to 318,000 depending on the population. The quality measurement year was 1999. Then, through the use of individual pseudoidentifiers, subsequent utilization was tracked for each person in the eligible population through 2002 for as long as they remained continuously enrolled in the same health plan. For the expenditure variable, we used the actual paid cost (i.e. the sum of insurance reimbursements, patient copayments, and deductibles) for each inpatient or outpatient claim. Comorbidities were measured using flag variables for 99 diagnostic categories produced by Medstat. Population Studied: The study populations were commercial managed care plan members with asthma, diabetes, or major depression, as well as women in the managed care population in general. The conditions chosen have a relatively high prevalence, a high impact on costs, and at least one established Health Plan Employer Data Information Set (HEDIS) quality measure with an accepted definition and methodology for defining the population based on claims data. Each subgroup was modeled separately, taking into account co-morbidities to better identify the clinical areas in which health plan services can have the greatest economic impact. Principal Findings: For three of the four quality areas examined (breast cancer screening for women age 50-69, the use of inhaled steroids for patients with persistent asthma, and a set of tests related to diabetes monitoring, there were substantial and significant negative correlations between patients' quality of care and total costs of care. For example, the estimated average inpatient and outpatient expenditure of women who had a mammogram was 13% -24% lower than the average expenditure of women who did not have a mammogram in the subsequent three years. For antidepressant medication management there were no significant associations found. Conclusions: Using patient-level data from commercial health plans, we found substantial and significant negative correlations between effectiveness of care as measured by selected HEDIS measures, and cost of care based on paid claims in three areas, breast cancer screening, asthma and diabetes. These findings are in agreement with those of Wennberg et al in examining the relationship of resource use and cost in the Medicare program at the hospital-physician practice level. Implications for Policy, Delivery, or Practice: Higher quality care can result in cost savings in the years immediately following, at least for some disease groups and types of services. This reinforces the importance of quality measurement and evaluation in the health care system. Primary Funding Source: Merck, Inc. ●Tiered Hospital Plans: Are They Tiered Drug Plans Writ Large? James Maxwell, Ph.D., Peter Temin, Ph.D., Tanaz Petigara Presented By: James Maxwell, Ph.D., Director of Health Policy and Management Research, Health Services, JSI Research & Training, 44 Farnsworth Street, Boston, MA 02210; Tel: (617) 482-9485; Fax: (617) 482-0617; Email: maxwell@jsi.com Research Objective: Based on data from a survey of large California employers, we compare the implications of tiered hospital plans where consumers pay more for high-cost hospitals to already existing tiered drug plans. Study Design: During 2002-2003, we conducted qualitative interviews with employers and major health plans in California, asking them about new health insurance products and trends. Population Studied: Health benefits managers at 318 of California's largest public and private employers. We also interviewed all the major health plans, provider groups, and employer purchasing coalitions in California. Principal Findings: Tiered hospital plans resemble tiered drug plans in name only. While drug tiers reduce costs by affecting consumer behavior, hospital tiers reduce costs by affecting provider behavior. Conclusions: Tiered strategies can be applied across a variety of health services including drugs, hospitalizations, and specialty care. However, the application of these strategies to different services leads to very different effects. Infrequent hospitalizations, difficulty in evaluating hospital services, and the subsequent provision of adequate information to consumers may limit the ability of tiered hospital plans to affect consumer behavior. Exclusive networks—the ultimate tier—demonstrate the point. Like hospital tiers, they work through suppliers, since the great gap between the cost of included and excluded services precludes almost all choices by consumers. Implications for Policy, Delivery, or Practice: Tiered hospital plans are not part of the new wave of consumerism in health benefits purchasing. New trends such as tiering must be examined carefully to determine who they affect most. Primary Funding Source: California HealthCare Foundation ●Are Large Employers Moving to Catastrophic Health Insurance Coverage?: Results from a Study of Large California Employers James Maxwell, Ph.D., Peter Temin, Ph.D., Tanaz Petigara Saminaz Zaman Presented By: James Maxwell, Ph.D., Director of Health Policy and Management, Health Services, JSI Research & Training, 44 Farnsworth Street, Boston, MA 02210; Tel: (617) 482-9485; Fax: (617)482-0617; Email: maxwell@jsi.com Research Objective: To document the adoption and penetration rates of new insurance products such as highdeductible PPOs and consumer-driven health products within the California marketplace. Study Design: To document these new products, in 20022003 we administered a 30 minute telephone survey to health benefits managers at California’s largest employers. We surveyed 318 out of 374 employers for a response rate of 85 percent. We asked about offer rates, penetration rates, and plans to implement high-deductible PPOs and consumerdriven health plans in the future. Population Studied: 318 of the largest public and private employers in California. Principal Findings: Although the California market has traditionally been dominated by HMOs, we found that a significant number of employers now offer a PPO option to their employees (85% of private employers and 67% of public employers). High-deductible PPOs, however, are not a widespread offering. We found that only a modest number (12% of private employers and 5% of public employers) of employers offer these plans and that they do not offer adequate incentives for their employees to move to them. There appears an equal interest in high deductible PPOs and consumer-driven health plans. Despite the publicity, only 5 percent of private employers offered consumer driven health plans, and an even smaller fraction of large public employers have moved to these new products. Only about one-third of large employers offer either a high-deductible PPO or a consumer-driven health plan. Conclusions: We found no evidence of a swift and linear movement towards catastrophic health plans. Instead, employers have moved incrementally to add new product options, hoping that they will yield cost savings. Implications for Policy, Delivery, or Practice: While media, the President, and many policy analysts have touted consumer-driven health products as the answer to rising health care costs, employers have only responded cautiously to these developments. Employers are not moving all their employees to high-deductible PPOs or consumer-driven health plans. Neither are they providing strong financial incentives for their employees to chose these plans. Primary Funding Source: California HealthCare Foundation ●Health Purchasing Among Large Private and Public Employers in California James Maxwell, Ph.D., Tanaz Petigara, Saminaz Zaman Presented By: James Maxwell, Ph.D., Director of Health Policy and Management Research, Health Services, JSI Research and Training, 44 Farnsworth Street, Boston, MA ; Email: maxwell@jsi.com Research Objective: Past research has shown that public employers are, in general, unable to be as aggressive as their private counterparts in rate negotiations. California has been an exception because of the presence of CalPERS, the University of California system, and SISC, a coalition of state and community colleges. These public agencies more closely resemble large private firms, with the financial and administrative resources needed to aggressively negotiate with carriers. Study Design: In 2002-2003, we conducted a study of the purchasing practices of over 300 private and public employers in California, using a 30-minute telephone interview that targeted the official with the most responsibility and detailed knowledge of health benefits. In-depth in-person interviews were also conducted with employers, consultants and all the major California health insurance carriers. Population Studied: Principal Findings: Both large public and private employers in California reported losing their purchasing clout in the last few years due to consolidation among health carriers, hospitals, and provider groups. We found that both private and public employers in California now showed a greater reliance on financial incentives (e.g., copayments and deductibles) aimed at influencing consumer decisions. However, the purchasing practices of large private and public employers in California have started to diverge as private employers have greater leeway to implement these changes. Private employers often have younger workforces and less unionization. In contrast, heavy unionization and entrenched public sector culture of rich benefits discourages public employers from rapidly adopting cost-cutting measures aimed at the consumer. Conclusions: The disparities between the private and public sector workforces, coupled with differences in management culture, suggest that a higher portion of health care costs will continue to beset public employers in California. ●A State by State Comparison of U.S. Small Group Market Rating Methods Andrea McAllister, BS, Andrea DeVries, Ph.D. Presented By: Andrea McAllister, B.S., Decision Support Specialist, Health Services Research, Highmark, Inc., 120 Fifth Avenue, Pittsburgh, PA 15222; Tel: (412)544-0664; Fax: (412)544-0700; Email: andrea.mcallister@highmark.com Research Objective: To (1) identify small group insurance rating structures among the 50 United States during 2003. (2) Determine disparities, if any, between regulations for for-profit and non-profit health insurers and HMO’s within each state. (3) Define categories and identify patterns of rating structures between states. Study Design: For the purpose of health insurance, a “small group” is defined as an employer group containing between two and fifty employees. Multiple sources of insurance rating structure information are researched for 2003. Individual state insurance departments are contacted when the information is ambiguous or sources provide conflicting information. At that point, the actuary or other insurance analyst responsible for the small group market in that state is interviewed. Subsequent to gathering states’ regulation information, categories are created to define the various types of rating structures among the states. Population Studied: The population consists of the 50 states comprising the United States. Principal Findings: Almost all of the states, 48 of 50, do not differentiate between for-profit and non-profit health insurers and HMO’s. Michigan has passed reform legislation to change their status. This will leave Pennsylvania as the only state to have small group rating structures that distinguish between these types of insurers. Additionally, Pennsylvania is only one of two states, Hawaii being the other, which does not have specified premium limits. Over time, the trend has been for states to move toward equitable regulation of for-profit and non-profit health insurers and HMO’s. Additionally, more states have adopted the use of health status as a determinant of premium level for small groups. Conclusions: There is little available information regarding the direct comparison of small group insurance market rating structures among the United States. It became quite evident that Pennsylvania holds a unique status in the U.S. as a state with a small group market rating structure different from any other. It has a rare combination of differentiating between forprofit and non-profit health insurance plans and HMO’s while having premium limits only for the non-profit insurers and HMO’s and no limits specified for the for-profit insurers. Implications for Policy, Delivery, or Practice: In Pennsylvania, for-profit insurers have much more flexibility in determining premium rates for small groups than the nonprofits and HMO’s. As a consequence, the for-profit insurers leave the non-profit insurers and HMO’s to insure the small groups that cannot afford the high rates offered to them. This results in the non-profit insurers and HMO’s keeping the high-risk small groups while the for-profit insurers keep the low-risk small groups. Potentially, the costs for these groups could increase to the point that the non-profits and HMO’s may not be able to continue to insure the small group market. This would unfairly affect high-risk small businesses, leaving them with no alternative to for-profit insurers and could potentially hamper the retention and growth of small businesses in the state. Primary Funding Source: No Funding Source ●Does Relying on the VA Health System Affect Enrollment in Medicare Managed Care? Robert Morgan, Ph.D., Jessica Davila, Ph.D., Margaret Byrne, Ph.D., Debora Paterniti, Ph.D., Dolly John, MPH, Jennifer Hasche, MS, Laura Petersen, M.D., MPH Presented By: Robert Morgan, Ph.D., Senior Research Scientist and Associate Professor, Houston Center for Quality of Care and Utilization Studies, Michael E. DeBakey VA Medical Center and Baylor College of Medicine, 2002 Holcombe Boulevard (152), Houston, TX 77030; Tel: (713) 7948635; Fax: (713) 748-7359; Email: rmorgan@bcm.tmc.edu Research Objective: Compared to traditional Medicare, Medicare managed care plans (‘Medicare HMOs’) often provide additional services and/or reduced costs. Since many veterans enroll in Medicare HMOs, changes in Medicare HMO policies can have a significant impact on VA system use. However, we have little information on why veterans decide to enroll in Medicare HMOs. This study examines reasons for enrolling in Medicare HMOs and attitudes about the VA system among Medicare-enrolled veterans who do and do not use the VA for care. Study Design: Prospective mailed survey of VA-using and VA non-using (no VA use in prior 3 years), elderly, male veterans in six large metropolitan areas. Analyses were weighted to reflect the sampling proportions used in each area. Population Studied: We used survey data from elderly, Medicare enrolled male VA-users and elderly male veterans who did not use the VA system. VA users were sub-classified based on whether they reported receiving most of their medical care from the VA (‘primary’ VA users) or from non-VA sources (‘secondary’ VA users). Principal Findings: The overall survey response rate was 52%. Our final analysis sample included 1,258 VA-users and 828 VA non-users. 31.3% of VA users and 32.9% of VA non-users in our sampling areas were enrolled in Medicare HMOs. Among VA users, 29.3% of HMO enrollees were primary VA users, compared to 38.2% of non-HMO enrollees (p <= .0001). Among HMO enrollees, secondary VA users and VA nonusers were more likely to cite cost factors (e.g., reducing copayments), coverage factors (e.g., availability of specialty care), or physician or hospital availability as reasons why they joined their HMO plan (p <= .01 for all comparisons). In contrast, primary VA users enrolled in HMOs were more likely to regard the quality of care at the VA as both affordable and better (p <= .001 for both). Conclusions: Primary and secondary VA users view Medicare HMOs and the VA system quite differently, with secondary VA users resembling VA non-users in terms of their reasons for joining Medicare HMOs. Reductions in service and/or price increases by Medicare HMOs are likely to mitigate the reasons secondary VA users and VA non-using veterans cite for joining HMOs. In contrast, increases in services, or price reductions by HMOs do not appear as likely to influence primary VA users to seek care outside of the VA. Implications for Policy, Delivery, or Practice: 70% of HMO enrolled VA users report using the VA as a secondary source of care. Even in the absence of increased enrollment in the VA system, reductions in service provided by Medicare HMOs, or increases in costs, could put the VA system at substantial risk for increased demand. Primary Funding Source: VA ●Extent of Selection Bias in Choice of Level of Patient Cost Sharing James Naessens, MPH, Holly VanHouten, BS, Sidna Scheitel, M.D., Rebecca Pautz, BA, Prathiba Varkey, M.D., Karen Ytterberg, M.D. Presented By: James Naessens, MPH, Clinical Associate, Health Care Policy & Research, Mayo Clinic, Pavilion 3, Rochester, MN 55905; Tel: (507)284-5592; Fax: (507)285-1731; Email: naessens@mayo.edu Research Objective: Explore the extent of selection bias among eligible employees choosing health insurance with different levels of cost sharing. Study Design: Assessment of the selection bias in a natural experiment, within the self-funded medical care plan of a large medical group practice. In 2004, the medical care coverage of the main medical benefit option was changed. 82% of eligibles in 2003 were enrolled in the Comprehensive plan which covered full first dollar coverage of services provided within the group practice and 17% were in the Select plan with copays and deductibles. The new options were between a plan which included relatively small copayments for visits to specialty care physicians and emergency rooms, and relatively low (10%) coinsurance for all other services(Universal); or a plan with the same level of copayments, but a deductible and 20%coinsurance(Choice). In addition the Universal plan had higher premiums and a lower out-of-pocket maximum than the Choice plan. Selection bias was determined by comparing the characteristics of the enrollees in the Universal versus Choice plans in January 2004. Health status was measured by assigning diagnoses from medical care claims for each enrollee for the previous calendar year to the Johns Hopkins Adjusted Clinical Groups (ACGs) and the AHRQ comorbidities developed by Elixhauser. The demographic, health status, previous plan and 2003 incurred medical expenses were compared using Chi-square tests for nominal factors and Wilcoxon rank-sum tests for quantitative factors. A logistic regression model was used to assess the simultaneous effects of the independent variables. Population Studied: Employees and their dependents of the Mayo Clinic in Rochester, Minnesota (MCR) who were under 65 years of age in January 2004. MCR employees included the medical, administrative, educational and research staff of the clinic, hospitals and schools located at the Rochester campus. N=59,374. Principal Findings: 95% of those choosing Universal had been in the Comprehensive option in 2003, while almost 42% of those choosing Choice had been in the Select option. Sicker patients were much more likely to choose the Universal plan. 18.2% of Universal versus 9% of Choice had ACGs based on 2003 diagnoses which indicated they sought medical care for at least 6 different ambulatory diagnosis groups. 22% of Choice versus 15% of Universal had either only preventive or no diagnoses in 2003, while almost 12% of Choice versus 9% of Universal had only an acute condition in 2003. Conclusions: Employees with either no health care utilization or utilization for only preventive or acute care in 2003 tended to select the health plan with a low premium, 20% coinsurance and deductible. Enrollees with multiple problems the previous year were more likely to select the plan with a higher premium, but to lower personal financial risk. Past medical experience clearly dominates the extent of cost sharing eligible employees are willing to undertake. Implications for Policy, Delivery, or Practice: Recent studies have discounted the impact of selection bias based on level of employee cost sharing. Selection bias is alive and well in 2004. Primary Funding Source: No Funding Source ●The Association of Switching Primary Care Physicians with Utilization and Disease Burden Chih-Wen Pai, Ph.D. Presented By: Chih-Wen Pai, Ph.D., Senior Research Associate, Office of the Dean, University of Michigan Medical School, M7325B Medical Science Building I, Box 0624, Ann Arbor, MI 48109; Tel: (734) 615-7041; Fax: (734) 615-5154; Email: cwpai@umich.edu Research Objective: Little is known about how primary care physician (PCP) switching affects utilization and disease burden of academic health systems (AHSs). Anecdotal evidence suggests that patients switch to PCPs at the AHS in order to receive specialty care. In this study we examined if managed care members switching to an AHS had higher utilization and disease burden compared to others. Study Design: We applied a standardized fee schedule to estimate all health care use in one common monetary unit, including inpatient and outpatient services as well as pharmacy and ancillary uses. We classified members based on age and sex, Adjusted Clinical Groups (ACGs), and Aggregated Diagnosis Groups (ADGs). We compared ACG distributions to examine different patterns of disease burden. Population Studied: Commercial members enrolled in a health maintenance organization for the entire 12 months of 2001. Principal Findings: About 3% of the study population switched PCPs across the academic and community boundary. Members switching to the AHS only had higher utilization than those staying with community providers. There was no dominating pattern of different disease burden for these two groups. Compared to members staying with the AHS, those switching out of the AHS had lower utilization but significantly higher disease burden, characterized by more acute minor conditions. Conclusions: There was no direct evidence that members switching into the AHS had medical need discernibly different from those staying with community providers. Members switched their care from the AHS to community providers probably due to access and capacity constraint at the AHS. Implications for Policy, Delivery, or Practice: Although members switching to the AHS decreased burden of community providers, they did not add to the AHS’s “absolute” disease burden that was already high compared to the rest of the population. However, PCP switching is not necessarily beneficial or harmful to the AHS, depending on capitation payment level and cost estimate. Decision making on whether to carve out switchers from capitation pool is complicate due to sensitivity in payment negotiation and accuracy in cost estimate. Nevertheless, a small size of switcher population is not likely to have much impact on adjusting the financial status. Primary Funding Source: No Funding Source ●Why Do Fewer Californians Have Job-Based Health Insurance? Ninez Ponce, MPP, Ph.D., E.Richard Brown, Ph.D., Thomas Rice, Ph.D., Shana Alex Lavarreda, MPP Presented By: Ninez Ponce, MPP, Ph.D., Assistant Professor, Health Services, UCLA School of Public Health, 31-254B CHS, Los Angeles, CA 90095; Tel: (310)206-4021; Email: nponce@ucla.edu Research Objective: California’s lower rates of employmentbased health insurance (EBHI) compared to other parts of the United States has been attributed to compositional differences in its population—primarily that California has more Latinos and noncitizens in its labor force. We examine if given similar employee/employment characteristics and other area-level factors as the rest of the US, whether California employers would be more likely to offer EBHI, and whether California employees would increase its already higher take-up rates of EBHI, than their counterparts in the rest of the country. Study Design: Algebraic decompositions distinguished whether California’s offer and take-up rates of EBHI were different than the rest of the nation because of the sociodemographic composition of its workforce and its communities, or, because of structural differences in its labor market. Population Studied: Linked individual-level data on employees, ages 19-64 from the February 1999 Current Population Survey, with metropolitan-statistical area (MSA)level data from the March 2000 Current Population Survey and InterStudy 1997/98 HMO Data. Principal Findings: Compared to the rest of the US, California has a lower offer rate (80.96% vs. 86.31%), but has a higher take-up rate (87.05% vs, 84.41%). Large differences due to California’s population composition (-8.50 percentage-points) is not fully offset by the state’s positive structural factors (3.15 percentage-points). For take-up, both structural and compositional factors are favorable in California. The structural environment advantage (1.90 percentage-points) is the key driver of California’s higher take-up rates. Greater concentration of noncitizens had the largest compositional effect on take-up. And unlike its negative effect on offer-rates, its effect on take-up was positive. Conclusions: With a large vulnerable workforce that is disadvantaged when it comes to getting EBHI, California’s structural environment, though favorable relative to the rest of the nation, still needs to be ameliorated in such a way to encourage employers to offer health insurance. For take-up, strong contextual effects suggest that California’s large noncitizen communities as a whole attribute high value to health insurance. Implications for Policy, Delivery, or Practice: (1) Offer rates: Nationally, the findings on the structural effects on offer rates suggest that employers in the rest of the US would be less likely to offer EBHI if they had the same types of workers who lived in California. This is troubling since the growth in the immigrant, principally Latino and Asian workforce in California, foretells of the spread of the growing foreign-born population in the rest of the country. (2) Take-up rates: With the same types of California workers and communities, nationally, workers would be less likely to take-up. Increasing immigration also plays a role: unlike California, workers in the rest of the US would be less likely to take up EBHI with higher concentrations of noncitizens in communities. Future studies that focus on why California is different in this respect are needed. In the absence of policies and incentives that increase the propensity of employers to offer, and employees to take-up, then EBHI coverage rates nationwide would be expected to decline even more. Primary Funding Source: California Wellness Foundation ●Does Managed Care in California Close the Equity Gap? For Whom and For What Services? Ninez Ponce, MPP, Ph.D., Robert Nordyke, Ph.D., Ellen Wu, MPH Presented By: Ninez Ponce, MPP, Ph.D., Assistant Professor, Department of Health Services, UCLA School of Public Health, 31-254B CHS, Los Angeles, CA 90095; Tel: (310)206-4021; Email: nponce@ucla.edu Research Objective: Managed care (MC) may encourage use of a usual source of care (USOC) via requirements to choose or be assigned a primary care provider. However, it is unclear whether this translates into greater access to preventive or diagnostic care and whether it reduces or exacerbates race/ethnic disparities in access to these services. Our study examined whether utilization by different racial and ethnic groups (African American, Asian, Latino, non-Hispanic White, and American Indian/Alaskan Native (AIAN) or Other ) is higher in MC than FFS. Study Design: Weighted multivariate logit models examined factors associated with the likelihood of: 1) having a USOC; 2) receiving appropriate cancer screening; and, 3) receiving appropriate care for chronic diseases. Separate estimates were obtained for those with insurance coverage in: 1) public Medicaid and Children’s Health Insurance (CHIP) programs; and 2) employment-based or privately purchased health plans (EBI). We evaluated MC’s potential differential effect by race/ethnicity by specifying MC interaction terms for each racial and ethnic group. Population Studied: Insured adults ages 18-64 covered by Medicaid/CHIP program or by EBHI from the 2001 California Health Interview Survey (CHIS 2001) random-digit dial and Asian oversample files ( n =38,130 ). Cancer screening use was evaluated on a subset of individuals meeting the ageappropriate guidelines. Principal Findings: Overall, MC is associated with greater chance of having a USOC (OR=2.86 public, =2.54 private; p<0.001 for both) and appropriate cancer screening (OR=2.25 public, =1.22 private; p<0.001 for both). There was no overall or differential race/ethnicity impact of MC on chronic disease management. We found no race/ethnic difference in having a USOC among the publicly insured. However, among adults covered by EBI, Latinos, Asians, and African Americans tend to have higher USOC in EBI than Whites. Latinos and AIANs also have higher rates of cancer screenings in public programs than Whites. However, consistent with other studies, APIs have substantially lower rates of cancer screening in both public and EBI plans. Conclusions: The evidence on the differential racial/ethnic effect of MC is mixed. For USOC, MC appears to lower the likelihood of having a USOC among AIANs with EBI coverage, but we saw no additional MC effect for other groups. For cancer screening, MC appears favorable for APIs and African Americans covered by EBI, but not favorable for Latinos and AIANs in public programs. Implications for Policy, Delivery, or Practice: MC in California may result in better utilization than FFS depending on the service and source of coverage. However, MC may also have negative effects on utilization – the difference is determined by demographics and service sought. Additional research and public policies to ensure that minorities are aided by the implementation of MC is essential. These efforts will be supported by: 1) investigation of health plan characteristics that influence utilization differences among California’s racial and ethnic population groups; 2) Identifying best practices that may reduce disparities that exist; collection and analysis of race/ethnicity data by all health insurance purchasers. Primary Funding Source: California Program on Access to Care / California Policy Research Center ●Cost Savings Associated with an Integrated Workers’ Compensation/ Managed Care Pharmacy Benefit Program Shadi Saleh, Ph.D., MPH Presented By: Shadi Saleh, Ph.D., MPH, Assistant Professor, Health Policy, Management and Behavior, State University of New York at Albany, School of Public Health, One University Place, Rensselaer, NY 12144; Tel: (518) 402-0299; Fax: (518) 402-0414; Email: ssaleh@albany.edu Research Objective: The study was aimed at examining the cost savings’ potential of an integrated workers’ compensation managed care prescription drug program in New York State. The program, called ONECARD Rx, allowed New York State employees to use their general health insurance card to buy WC-related prescription drugs. Study Design: A retrospective cost analysis design conducted through the use of a prescription drug claims’ dataset. Population Studied: The study population consists of New York State government employees who filed a workers’ compensation-related prescription drug claim through ONECARD Rx in 2002. A total number of claims processed through the use of the ONECARD Rx system during 2002 was 6705. Principal Findings: : The results of this study showed that the use of ONECARD Rx produced total cost savings of $322,067 ($48.04/claim) in 2002. Projections of the cost savings of ONECARD Rx were even higher for subsequent years. Findings revealed that the combination of the increase in use of ONECARD Rx coupled with the buffered effect of generic and brand-name price increases results in increased savings each year. Conclusions: Integrated WC/managed care prescription drug program has the potential to produce significant cost savings to employers. Implications for Policy, Delivery, or Practice: The study results may have an impact on how employers organize, deliver and finance their WC systems, specifically the prescription drugs' benefit package. Merging WC prescription drug benefits with the employees' general health insurance may be one way to control the skyrocketing prescription drug costs. In addition, this study may be a stepping stone towards evaluating the effects of combining/integrating all WC services (medical and prescription drug benefit packages) into the general health insurance plan of their employees for potential cost reductions. Primary Funding Source: RWJF ●Prescription Drug Insurance: Effects on Drug Use, Expenditure and Out-of-Pocket Spending Among the NearElderly Jaeun Shin, Ph.D., Sangho Moon, Ph.D. Presented By: Jaeun Shin, Ph.D., Assistant Professor, , KDI School of Public Policy and Management, 207-43 Cheongnyangri2-dong, Dongdaemoon-gu, Seoul, 130-868; Tel: (822) 3299-1037; Fax: (822) 3299-1240; Email: jshin@kdischool.ac.kr Research Objective: As the Baby Boom generation begins to reach at 55, the near-elderly population between ages 55-64 receives special attention recently. Despite the growing concern on the potential consequences of a Medicare prescription drug benefit, relatively little is known about how prescription drug insurance relates to the patterns of drug use and expenditures among the prospective beneficiaries, the near-elderly. The purpose of this study is to understand differences in prescription drug use and expenditures in the near-elderly population by drug insurance status; to identify the factors associated with prescription drug insurance coverage; to examine the effect of drug insurance on the use and spending on prescription drug; and to assess the health policy implications of the findings for a Medicare prescription drug benefit and its impact on the beneficiaries. Study Design: Prescription drug insurance is identified for respondents to the 2001 Medical Expenditure Panel Survey (MEPS). The primary outcome variables are total number of prescription refills per year and three categories of prescription drug expenditures (total expenditures, total outof-pocket spending, total payment by insurance). Statistical significance of the differences in means is calculated using ttests. Race, gender, marital status, region of residence, educational attainment, employment status, poverty level, family size, and health conditions are included in the logistic regression as predictors of whether or not an individual has prescription drug insurance coverage. We use a zero-inflated negative binomial model (logistic regression to separate those with nonzero outcomes from those with zero outcomes, followed by negative binomial regression) to understand the relationship between drug insurance coverage and prescription drug use and expenditures among the nearelderly. Population Studied: The population studied includes the near-elderly, individuals ages 55-64 (N=2,606) from the 2001 MEPS. The respondents are divided into two groups, those with prescription drug insurance versus those without prescription drug insurance for further analysis. Principal Findings: Respondents with drug insurance refill smaller number of prescriptions on average (18 prescriptions) compared to 25 prescriptions for those without insurance. The mean of total prescription drug expenditure (RXEXP) for respondents with drug insurance is significantly lower compared to those without drug insurance ($1032.2 versus $1293.2). Substantial difference in total out-of-pocket spending for prescription drug (RXSLF) is also found between respondents with drug insurance ($340.6 or 46% of total outof-pocket spending for all health care use (TOTSLF), 40% of RXEXP) and those without drug insurance ($687.3 or 64% of TOTSLF, 69% of RXEXP). These differences are all statistically significant at 1% level (p<0.01). Being Hispanic or nonHispanic Afro-American, being unemployed and living in a larger family significantly lower the probability of having prescription drug insurance, while being married, living in urban area, having higher levels of education and family income increase the probability. The family income effects (200% or higher than the Federal Poverty Level) and high education effect (college education or more) are stronger than the effect of race/ethnicity. Prescription drug insurance coverage has no significant relationship with level of utilization and total prescription drug expenditure. However, it shows a positive and significant (p<0.01) association with outof-pocket spending (RXSLF) and a negative significant (p<0.01) association with payment by insurance for prescription drug (RXPRV). Being Hispanic/non-Hispanic Afro-American and being unemployed are negatively and significantly (p<0.05) correlated with drug use and RXSLF, while being unemployed, being in poor self-rated health status, having at least one priority disease, having larger number of co-morbidities, having limitations in (instrumental) activities of daily living, and being diagnosed with certain chronic conditions show significantly (p<0.05) positive associations. Conclusions: This study provides preliminary evidence that the near-elderly without drug insurance demand more use and higher costs for prescription drugs compared to those with drug insurance. Without drug insurance, prescription drug is a financially heavy medical item to afford, requiring high consumer cost sharing of out-of-pocket spending. Income, education, race/ethnicity are major predictors for prescription drug insurance status, implying insurance coverage disparity by these characteristics may lead to disparity in access to prescription drugs and, subsequently, in health outcomes between those with insurance and those without insurance. Providing drug insurance does not increase level of utilization and total expenditures for drug. It shifts burden of payment from consumers to insurers in a significant way. Medicare prescription drug benefit would help remove insurance disparity by income, education and race/ethnicity and reduce relatively high financial burden for those without drug insurance under private insurance system. It may enhance accessibility to prescription drugs among the elderly and promote their health, while special attention is needed for the potential fiscal impact of additional prescription drug coverage in Medicare to support the current near-elderly population. Implications for Policy, Delivery, or Practice: Policy makers can use results from this study in designing and modifying Medicare prescription drug benefit. Further research should pay attention to quantitatively approximate the fiscal effect on Medicare program when the current near-elderly becomes eligible for Medicare prescription benefit. Primary Funding Source: KDI School of Public Policy and Management; Sungkyunkwan University, Korea (South) ●The Effect of Area HMO Market Share on Colorectal Cancer Screening Within the VA Health Care System Lynn Soban, Ph.D., MPH, RN, Elizabeth Yano, Ph.D., MPH, Patricia Parkerton, Ph.D., Lisa Rubenstein, M.D., Susan Ettner, Ph.D. Presented By: Lynn Soban, Ph.D., MPH, RN, Post-doctoral fellow, VA Greater Los Angeles HSR&D Center of Excellence, 16111 Plummer Street (152), Sepulveda, CA 91343; Tel: (818) 891-7711 x9954; Email: lsoban@ucla.edu Research Objective: Managed care “spillover” effects are reflected by the degree to which local managed care activity influences the structure and functioning of other health care systems in the same environment. These effects may be important as they may explain local variations in the delivery of care. Recent research has demonstrated that the VA’s fundamental reorganization towards managed care principles in the mid-1990’s has resulted in substantial quality gains, outstripping Medicare performance. We aimed to evaluate the extent to which managed care spillover might have contributed to VA quality reports, using colorectal cancer (CRC) screening as a case example. Study Design: This cross-sectional study uses facility and patient-level data. Data on the level of managed care activity in the local VA facility environment (defined as percent HMO enrollment per total area population) were obtained from Interstudy. This was linked to other area characteristics (e.g., rural/urban location, area physician capacity) from the Area Resource File; and with individual VA health care facility characteristics (e.g., facility size, academic affiliation) from the National VA Survey of Primary Care Practices. CRC screening data among veteran users were obtained from the VA’s External Peer Review Program (EPRP). These data are generated from chart review conducted by trained nurse abstractors using standardized abstraction protocols. CRC screening is “achieved” when a guideline-eligible patient receives screening from a VA or non-VA provider. Using generalized estimating equations, we then examined the effect of HMO penetration in each VA’s locale on the probability of a veteran receiving CRC screening in general (by a VA or community provider), and, among screened veterans, the probability that the screening was performed by a VA provider (as opposed to by a community provider, but credited to the VA). Population Studied: 52,213 randomly sampled outpatient clinic patients from 214 VA facilities across the U.S. Principal Findings: The level of HMO activity in a VA’s local market is not a significant predictor of the receipt of CRC screening. However, among veterans who were screened, those who received care in areas with higher levels of HMO market share were less likely to have received screening from a VA provider (RR: 0.98; 95% CI: 0.96, 0.99). Patient characteristics predictive of receipt of screening from a nonVA provider included: female gender (p<.05), having private insurance (p<.001), or Medicare coverage (p<.001). Conclusions: In areas with higher levels of HMO penetration, VA patients are substituting non-VA screening for VA screening; this effect is more pronounced among women veterans and those with outside insurance coverage. Implications for Policy, Delivery, or Practice: Most screening occurred within the VA system, thus, the magnitude of the shift to non-VA providers in areas with higher levels of HMO penetration is not large enough to inflate VA performance. However, it appears that VA facilities located in areas with higher levels of HMO penetration may benefit from their location by relying more on community providers to perform screening. Primary Funding Source: AHRQ ●Impact of a Medication Copayment Increase on Medication Acquisition in the VA Kevin Stroupe, Ph.D., Bridget Smith, Ph.D., Todd Lee, PharmD, Ph.D., Ramon Durazo-Arvizu, Ph.D., Elizabeth Tarlov, Ph.D., Lishan Cao, MS Presented By: Kevin Stroupe, Ph.D., Research Scientist, Department of Veterans Affairs, Midwest Center for Health Services and Policy Research, PO Box 5000 (151H), Hines, IL 60141; Tel: (708)202-3557; Fax: (708)202-2316; Email: kevin.stroupe@med.va.gov Research Objective: In February 2002, the Department of Veteran’s Affairs, VA, raised medication copayments from $2 to $7 per 30-day prescription. Veterans are subject to medication copayments based their service-connected health conditions and income. Veterans fall into three groups where they face no copayments, copayments for non-service connected medications only, or copayments for all medications. We examined the number of 30-day equivalent prescriptions veterans obtained from VA during 12-months before and after the increase. We evaluated the impact of the copayment increase across several medication categories: all chronic medications, defined as medications with at least one 30-day supply; higher and lower-cost medications, defined as medications with retail cost more or less than the copayment; over-the-counter and prescription only medications; generic and brand medications; and more or less essential medications, defined as medications preventing deterioration in health or relieving symptoms only. Study Design: This was a retrospective cohort study using data from VA national administrative databases. We used zero-inflated negative binomial count models to examine the effect of the copayment increase on the number of 30-day prescriptions, controlling for age, race, comorbidities, and insurance status. We employed a difference-in-differences approach to estimate the change in number of 30-day prescriptions, i.e., the marginal effects, after the increase for veterans with copayments relative to veterans without copayments, across the medication categories. We used these marginal effects to calculate price elasticities, which give the percentage change in 30-day prescriptions per 1 percent increase in the copayment. Population Studied: We took a 5.5 percent random sample of male VA users in fiscal year 2001, excluding veterans with no VA healthcare in the 12 months before the study or who died during the study period, resulting in 150,983 veterans: 19,620 with no copayments, 102,643 with some copayments, and 28,720 with copayments for all medications. Principal Findings: Following the copayment increase, the number of 30-day prescriptions for all chronic medications fell by 6 to 7 percent among veterans subject to copayments for some or all medications relative to veterans with no copayments, implying an elasticity of -0.05 to -0.06. The number of lower-cost medications fell by 16 to 25 percent, while the number of higher-cost medications fell by 5 percent. For over-the-counter medications, the number of 30-day supplies fell by 25 to 44 percent, while for prescription-only medications the number fell by only 3 percent. For both brand and generic medications, the number of 30-day supplies fell by 5 to 8 percent. The number of essential medications fell by 4 percent; however, there was no significant difference among non-essential medications. Conclusions: Following the copayment increase, the number of medications veterans obtained from VA fell among veterans subject to the copayment. The copayment increase had relatively larger impacts on lower-cost and over-the-counter medications. Implications for Policy, Delivery, or Practice: As a managed care organization, VA may wish to monitor medication acquisition by its chronically ill veterans. However, charging veterans copayments larger than the price they might pay elsewhere for certain lower-cost medications might lead veterans to obtain those medications outside of the VA, if at all, limiting VA’s ability to monitor medication acquisition. Primary Funding Source: VA ●The Effect of Co-Payment Change on the Discontinuation of VA Pharmacy Use for Chronic Prescription and Overthe-Counter Medication Kevin Stroupe, Ph.D., Todd Lee, Ph.D., Ramon Darazo-Arvizu, Ph.D., Elizabeth Tarlov, Ph.D., Lishan Cao, M.S. Presented By: Kevin Stroupe, Ph.D., Social Science Analyst, Midwest Center for Health Services and Policy Research, Hines VA Hospital, P.O. Box 5000, 151-H, Hines, IL 60141; Tel: (708)202-4870; Fax: (708)202-2499; Email: Bridget.Smith@med.va.gov Research Objective: The co-payment for a 30-day supply of medication obtained from Veterans Affairs’ pharmacies was increased from $2 to $7 in February 2002. Based on the relationship of their condition to their military service and income, veterans may be subject to co-payments for none, some, or all of their medications. The objective of this study was to determine if veterans affected by co-payments were more likely than those not affected to discontinue use of VA pharmacy for chronic prescription and over-the-counter medications following the co-pay increase. Study Design: This was a retrospective observational study using VA pharmacy and utilization data. We examined the effect of the co-payment change on the discontinuation of VA pharmacy use for chronic prescription and over-the-counter medications and the intensity of outpatient use with bivariate probit models. A medication was defined as chronic if a subject received more than a 30-day supply. Other variables in the models included race, age, a co-morbidities score, insurance coverage, and distance to the nearest VA hospital. Population Studied: The study population was an approximately 5 percent sample of male veterans who had pharmacy utilization 12 months before the increase and had some VA healthcare utilization 12 months after the increase. A total of 140,232 veterans had complete data for all the characteristics and were included in the model for prescription medications and 96,093 were included in the over-the-counter model. Principal Findings: After the change, 2.9 percent of veterans with no co-payments, 4.46 percent of veterans with some copayments and 3.5 percent of veterans always subject to the co- payment discontinued their VA pharmacy use for chronic prescription medications. The percentages of veterans who discontinued their chronic over-the-counter medications were much higher: 26.6 percent of veterans with no co-payments, 40.5 percent of those with some co-payments and 61.0 percent of veterans who always paid a co-payment. Based on the results of the multivariate models, the probability of discontinuing VA pharmacy use was 0.99 percentage points higher for veterans with some co-payments than for prescription use and 12.4 points higher for over-the-counter medications than for veterans not subject to the co-pay. For veterans who had to pay the co-payment for all medications, the probability of discontinuing use was 0.79 percentage points higher for prescription and 24.9 points higher for overthe-counter medications than for unaffected veterans. Conclusions: Veterans that were required to pay co-payments for some or all of their chronic medications were significantly more likely to stop filling both their prescription and over-thecounter medications through the VA after the co-payment change. The marginal effects for veterans that paid some or all of the co-payment were substantially larger for over-thecounter medications than for prescription medications. Implications for Policy, Delivery, or Practice: The fact that Veterans that were required to pay a co-payment for some or all of their chronic medications were more likely to discontinue their VA pharmacy use than unaffected veterans has the potential to result in some unintended health consequences. It will be important to determine if veterans stopped using their medications or obtained them elsewhere, and to examine the impact on health outcomes. Primary Funding Source: VA ●Prescription Drug Management: Are Health Maintenance Organizations Different From Traditional Insurance Plans? Cindy Thomas, Ph.D., Stanley S. Wallack, Ph.D., Timothy Martin, MS, Grant A Ritter, Ph.D. Presented By: Cindy Thomas, Ph.D., Senior Scientist, Schneider Institute for Health Policy, Brandeis University, 415 South Street MS 035, Waltham, MA 02454; Tel: (781)736-3921; Fax: (781)736-3905; Email: cthomas@brandeis.edu Research Objective: This research determines whether the different management approaches used by health maintenance organizations (HMOs) versus other types of insurers result in different patterns of prescription drug use and expenditures. We assess differences in prescription drug expenditures and use between two different types of insurance programs: HMOs; and self insured employer groups whose health care benefits are not provided by HMOs. We examine whether these insurers differ on several aspects of drug management, and how these translate to cost and utilization outcomes. Study Design: We examined differences between HMOs and other insurers to determine unadjusted and adjusted differences in the types of benefit design features implemented, such as average prescription drug use and copayments for three tier plans, and average cost sharing. We also compared utilization and expenditure outcomes between the two types of insurers. We decomposed the difference between HMOs and non-HMO plans in overall expenditures, the average price of a prescription, and average number of prescriptions, into the contribution of generic versus brand use, and into each therapeutic class of drugs. Multivariate regression was also used to estimate the importance of demographic and prescription program design features on total drug expenditures per person, and its two components: price per prescription and number of prescriptions per member. Population Studied: The sample used in the analyses included pharmacy claims from a total of 734 specific insurance plans, representing nearly 2 million working age individuals and their dependents, all under age 65, in a national sample. Among the 332 three-tier programs, 218 were HMOs and 114 were non-HMO (employer groups that were self-insured and did not provide health care benefits through an HMO). All study plans were managed by the same pharmacy benefits manager. Principal Findings: The enrollees in both health maintenance organization plans and non-HMO plans responded significantly to higher cost sharing. The results also indicate that there is clearly an “HMO effect” in the cost and utilization of prescription drugs, managed by the same PBM. The price per prescription was significantly lower in HMOs than in other types of health plans (p<.0001), due to greater use of generics and less expensive brand drugs. After controlling for demographics and cost sharing, HMOs experienced lower per capita expenditures (p<.01), number of prescriptions (p<.05) and average price per prescription (p<.01). Conclusions: Prescription drug utilization and spending patterns differ between HMOs and non-HMO health plans administered by the same PBM. Per capita prescription drug expenditures and use are affected both by consumer cost sharing and other management strategies. Implications for Policy, Delivery, or Practice: Measuring differences in prescription drug management between HMOs and other insurers will be critical for evaluating the impact and estimating costs of the new Medicare prescription drug benefit, and prescription drug management in other insured groups. Studies that do not take into account the effect HMO management strategies are likely to overestimate the effect of cost sharing alone and underestimate the long-run impact of additional prescription drug management efforts. As well, future studies on prescription use and cost should take into consideration the fact that health plans use a number of concurrent strategies, with the plan’s culture, physician organizational arrangement and market dictating the strategy and its impact. Primary Funding Source: California Health Care Foundation ●Employer-Sponsored Retiree Health Benefits in 2004 Heidi Whitmore, MPP, Jon Gabel, MS, Jeremy Pickreign, MS Presented By: Heidi Whitmore, MPP, Deputy Director, Health Research and Educational Trust, 17215 49th Avenue North Unit H, Plymouth, MN 55446; Tel: (763)478-6725; Fax: (763)478-6725; Email: hwhitmore@aha.org Research Objective: This study examines the current state of employer-based retiree health benefits for Medicare retirees, recent changes in their coverage, and possible employer responses to the Medicare Modernization Act (MMA) of 2003. Study Design: The Kaiser Family Foundation and HRET conduct this annual survey of employer-sponsored health benefits, with MedPAC supporting the portion of the survey on retiree health benefits in 2004. The survey collected data from 1,925 firms on the largest Medicare retiree plan and included questions on offer rates, eligibility and coverage, premiums, retiree cost-sharing, prescription drug benefits, and possible responses to the MMA. HRET retained National Research to conduct telephone interviews with benefits managers from January to May 2004, with a response rate of 50%. Population Studied: Interviews were conducted with 1,925 U.S. employers with a minimum of three employees. The survey asked benefit managers about health benefits offered to their Medicare retirees. Principal Findings: About three-fourths of employers that offer retiree health benefits extend them to retirees who have reached Medicare age. Some firms currently providing retiree coverage have already terminated retiree coverage for active workers. About 2/3 of workers from firms currently offering coverage will be eligible for retiree benefits when they retiree. Medicare retirees pay a higher percentage of the monthly premium than do active workers – 25% vs 16%. For those firms currently offering Medicare retiree coverage, about onefourth reported they are very or somewhat likely to eliminate retiree health benefits for new hires in the next two years, and 13% indicated they might do so for active employees. Most firms reported they are very or somewhat likely to require Medicare retirees to pay a larger proportion of the premium – 73 percent of retirees will likely pay more. Most (78%) retirees who now have benefits would continue to have some supplement to the new Medicare drug benefit from their former employer. The plan option that qualifies employers to receive a 28 percent subsidy was reported as “most likely” by more firms than other options, encompassing almost half of covered retirees. About one fourth of retirees are in firms that were considering wrapping around the Medicare benefit. Conclusions: We are witnessing a continuation of past trends – erosion of retiree health benefits. We will not see wholesale dropping of retiree health plans, but rather dropping of coverage for current and new hires. Firms will have tougher eligibility requirements, and a greater financial burden for retirees. The survey does not support the view that enacting a drug benefit would induce employers to drop retiree coverage altogether, or even retiree drug benefits, in response to the MMA. Implications for Policy, Delivery, or Practice: While the data do not support the concern that the MMA would crowd out employer-based retiree coverage, erosion of health benefits in general remains a growing problem for retirees. A smaller share of today’s workers being eligible for employer-based retiree coverage, coupled with less generous benefits, could place a large financial strain on future retirees. Primary Funding Source: MedPAC ●Patterns of Antibiotic Utilization among Managed Care Enrollees That Potentially Contribute to Drug Resistance Lok Wong, MHS, Jennifer Lis, MBA, MHSA, Barbara Souder, PhD, MPH, RN Presented By: Lok Wong, MHS, Senior Health Care Analyst, Quality Measurement, National Committee for Quality Assurance, 2000 L Street NW, Suite 500, Washington, FL 20036; Tel: (202)955-1784; Fax: (202)955-3599; Email: wong@ncqa.org Research Objective: To identify utilization patterns of antibiotic use in managed care populations, including specific antibiotics that potentially contribute to antibiotic drug resistance which diminishes antibiotic drug efficacy. Study Design: Retrospective pharmacy claims data analysis. Number of antibiotic prescriptions dispensed by drug class, plan, age cohort, member months enrolled, and drug class; average days supplied per prescription. Population Studied: Five million commercial members of all ages 0-85+ enrolled in seven managed care plans across the U.S. in 2002 and 2003 for which 4.6 million antibiotics were prescribed in 2003. Principal Findings: Overall antibiotic prescribing rates across all field-test plans increased by 1% from 2002 to 2003; however change varied by plan (range -3% to 5%). Antibiotic use in plans ranged from 738 to 1,040 antibiotic prescriptions per 1,000 members. Despite increase in total number of prescriptions per member, average days per prescription supplied declined (-1% to -3% change) between 2002 and 2003. Despite national efforts to reduce prescribing of antibiotics that potentially contribute to drug resistance, results show increased use of antibiotics of concern. Prescribing of antibiotics of concern made up approximately half of all antibiotic prescribing across plans (range 43% – 58%). Between 2002 and 2003, the percentage of these antibiotics generally increased as a proportion of all antibiotics prescribed (-0.2% to 9%) but with high levels of regional variation. By drug class, prescribing antibiotics of concern increased in most plans: clindamycin (plan range: 1% to 21% change), amoxicillin/clavunate (-2% to 7%), azithromycin and clarithromycin (2% to 17%), quinolones (-4% to 5%). Only later-generation cephalosporins generally showed a decline in prescribing (-2% to -13%, except one plan increased 2%). Prescribing for vancomycin was low at 0.06 to 0.22 per 1000 members, with variable change (-40% to 27%). In contrast, prescribing of antibiotics not known to contribute drug resistance generally declined (range -6% to 1%) from 2002 to 2003. Prescribing rates varied widely by class of antibiotic and patient age. Across all classes of antibiotics, females had a higher rate of antibiotic utilization than males. Conclusions: Study results show that despite national concern about increased antibiotic drug resistance due to overuse and misuse, there is an overall increase in antibiotic use. Of particular concern is increased prescribing in managed care of antibiotics of concern, compared to declining prescribing of antibiotics not related to drug resistance. Regional differences in prescribing antibiotics of concern suggest some regions could be experiencing or contributing to higher levels of drug-resistant organisms Implications for Policy, Delivery, or Practice: This national overview of antibiotic utilization trends in managed care using a standardized HEDIS measure will help to identify opportunities for quality improvement and inform national efforts to educate providers and patients about the importance of reducing use antibiotics that contribute to drug resistance. Further research is needed to understand for what conditions and patient populations antibiotics of concern are prescribed to further target quality improvement and reduce antibiotic drug resistance. Primary Funding Source: CDC, Council for Affordable Quality Healthcare, National Committee for Quality Assurance