Medicare & Medicare Prescription Drugs Call for Papers Medicare Part D: Successes & Challenges Chair: Scott Young, Agency for Healthcare Research and Quality Monday, June 26 • 8:30 am – 10:00 am ●Factors Associated with Regional Variation in Medicare Part D Prescription Drug Plan Participation and Beneficiary Premiums Leslie Greenwald, Ph.D., Judith Shinogle, Ph.D., John Kautter, Ph.D., Greg Pope, Ph.D., Boyd Gilman, Ph.D. Presented By: Leslie Greenwald, Ph.D., Senior Scientist, , RTI, International, 5104 Wetheredsville Rd, Baltimore, MD 21207; Tel: 410-448-2611; Fax: 410-448-2611; Email: lgreenwald@rti.org Research Objective: To understand the factors that influence regional variation in Medicare Part D prescription drug plan (PDP) participation and beneficiary premiums. Study Design: This study investigated the relationship between PDP regional characteristics and the number of Medicare Part D stand alone PDP plans and their average 2006 premiums. Data on PDP regional characteristics were developed from a comprehensive database of U.S. Census, Medicare administrative files, the Area Resource File, Interstudy, AARP, and other sources. This database was originally created for the Centers for Medicare & Medicaid Services (CMS) in their development of the 34 PDP regions, and has been updated for this study The total number of PDP plan options, and average premium charged, were correlated with regional characteristics available in this database to detect factors related to PDP plan participation and beneficiary premiums. Population Studied: Medicare PDP plans and regions. Principal Findings: Several regional characteristics are correlated with the number of PDPs offered in each region and the premiums charged. The percent rural population in regions has a small, negative relationship with the number of PDP plans, but a strong, positive relationship with increased premiums. A greater penetration of PPO enrollment in commercial markets was negatively correlated to the number of PDP options, and positively correlated to higher premiums. Conversely, both commercial and Medicare HMO penetration was positively correlated to the number of PDP options, and negatively correlated to premiums. These findings suggest that PDP plans may be more attracted, and offer lower premiums, in regions with higher HMO penetration; the opposite is the case in regions with a more dominant PPO presence. Other factors that were positively correlated to the number of PDP options included the number of physicians per capita, whether a discount drug program was offered in that region in 2003, and the diagnosis-based risk score; negatively correlated to the number of PDP options was the combined market share of the 3 largest private insurers in the commercial market. Average premiums for PDPs in the region was positively correlated to the percent of elderly in poverty, and negatively associated with the percent of the work force with union sponsored health coverage and the number of general practice physicians. Conclusions: PDP plans are widely available to Medicare beneficiaries in all regions, though the number and cost of these plans vary and are related to key PDP regional characteristics. Our findings are generally consistent with views expressed by PDP organizations, including that rural regions would have fewer PDPs and face higher premiums; and that the structure of health insurance markets in the PDP regions would affect the viability of the new Part D products. Implications for Policy, Delivery, or Practice: Regional characteristics such as the rurality and the managed care market may influence the number and costs of PDP options for Medicare beneficiaries. Currently options are widely available, however, attention to these characteristics may help policy makers understand and predict where PDPs may be successful in future years as the market for this benefit matures. Primary Funding Source: CMS ●Dollars to Doughnuts: Predicting Prescription Drug Costs of Beneficiaries and the Medicare Program Under Part D M. Christopher Roebuck, Dominick Esposito, Meredith Lewis Presented By: M. Christopher Roebuck, Health Economist, R&D, Caremark, 11311 McCormick Road, Suite 230, Hunt Valley, MD 21031; Tel: 410-785-2136; Email: chris.roebuck@caremark.com Research Objective: To examine drug utilization and out-ofpocket costs of Medicare beneficiaries using a Medicare prescription drug discount card, including beneficiaries who qualified for the Transitional Assistance Program (TAP). Study Design: Data included eligibility and prescription claims for enrollees in 34 separate Medicare drug discount card programs managed by Caremark. We used claims data to calculate annualized utilization and costs for beneficiaries and, in turn, simulated Medicare beneficiaries’ out-of-pocket costs (excluding premiums) and costs to Medicare under the Part D benefit. We estimated a generalized linear model (GLM; gamma distribution with log link function) for both beneficiary costs and Medicare payments under Part D to identify factors associated with drug expenditures. A probit model for the likelihood of falling into the doughnut hole was also specified. Explanatory variables in the models included demographic characteristics (age, gender, region, and TAP status), the generic dispensing rate, and 62 disease indicators derived using a pharmacy-based classification system. Population Studied: Beneficiaries enrolled for a minimum of six months with at least one claim between June 2004 and November 2005 (n=37,425). Participants were largely female (67%), between the ages of 65 and 80 (70%), and had an average of 2.2 medical conditions, with hypertension (52%), hypercholesterolemia (27%), and diabetes (16%) being among the most prevalent. Principal Findings: On average, beneficiaries in the sample filled 19 prescriptions at an annual cost of $538. Under the standard Part D benefit, mean total drug expenditures for these seniors would be $849 annually with $412 paid by the beneficiary and $437 paid by Medicare. About 6% of these beneficiaries have annual spending greater than $2,250 (the benefit’s “doughnut hole”). TAP beneficiaries (46%) would have higher out-of-pocket costs under Part D than the drug discount card ($429 versus $256; p<0.001). Multivariate analyses suggest a 10-percentage point increase in the generic dispensing rate is associated with a drop in beneficiary costs by $23 (p<0.001), a 0.5-percentage point reduction in the probability of falling into the doughnut hole (p<0.001), and a $42 decrease in Medicare payments (p<0.001). Conclusions: TAP beneficiaries who do not qualify for subsidized coverage under Part D will face higher out-ofpocket costs than under the discount drug card program, assuming fixed drug utilization. Increased use of generic drugs in proportion to brand name drugs would benefit the Medicare program more than beneficiaries, on average, due to the standard benefit’s structure. Implications for Policy, Delivery, or Practice: In choosing whether or not to enroll in Medicare Part D, seniors will compare their annual premium with the expected payout of the Medicare program. These results suggest the average, risk-neutral beneficiary would only enroll at monthly premiums below $36 ($437 divided by 12). If faced with higher out-ofpocket costs, low-income beneficiaries who do not qualify for subsidies may reduce their prescription drug utilization potentially resulting in adverse health effects. Finally, to reduce costs to both beneficiaries and taxpayers, Medicare should promote the substitution of generic medications whenever possible. Assuming 29 million Medicare Part D enrollees, the Medicare program could save more than $1.2 billion annually by increasing the generic dispensing rate 10%. Primary Funding Source: Internal- Caremark ●Disease Burden and the Intensity of Medication Therapy for Medicare Beneficiaries with Diabetes: Will Part D Make a Difference? Bruce Stuart, Ph.D., Thomas Shaffer, M.H.S., Linda SimoniWastila, RPh, Ph.D., Ilene Zuckerman, PharmD, Ph.D. Presented By: Bruce Stuart, PhD, Professor, Peter Lamy Center, University of Maryland Baltimore, 515 W. Lombard St., room 157, Baltimore, MD 21201; Tel: (410) 706-5389; Fax: (410) 706-1488; Email: bstuart@rx.umaryland.edu Research Objective: Guidelines for treating diabetes emphasize the need to maintain glycemic control and to take preventive measure to avoid hypertension, hyperlipidemia, and common infections like influenza and pneumonia. Medicare already covers flu shots and pneumonia vaccinations, but coverage for blood glucose regulators, antihypertensive agents, and lipid lowering agents has not been universally available until this year under the MMA. This paper is intended to assess the potential impact of Part D coverage on Medicare beneficiaries with diabetes taking into account the difficulties of managing diabetes for beneficiaries with significant disease burden. Study Design: This study categorized a national sample of Medicare beneficiaries with diabetes into 10 mutuallyexclusive, equal-sized groups based on their cumulative medical spending in 2001 (a proxy for disease burden). We then computed the prevalence and annual utilization rates within each group for 6 drug categories recommended for treatment of diabetes (blood glucose regulators, dyslipidemics, ACE-inhibitors, other antihypertensive agents, annual flu shot, and pneumonia vaccination in the past 5 years). Group-specific robust regression models were estimated for each drug class. Explanatory variables included domains for demographic characteristics, income, drug coverage, health status, and medical encounters. Chow tests for significant differences in coefficient values across the 10 groups were conducted to identify the differential impact of disease burden on factors predictive of the prevalence and intensity of drug use. Population Studied: 1,746 elderly and disabled Medicare beneficiaries with 2 or more paid claims indicating diabetes (ICD-9=250.xx, 357.2, 362.00-362.02, 366.41) In 2001 drawn from community-dwelling respondents in the 2001 Medicare Current Beneficiary Survey weighted to be nationally representative. Principal Findings: Annual population prevalence rates for the 6 drug categories ranged from a low of 42% for dyslipidemics to a high of 86% for antihypertensive agents. The lowest prevalence rates were found in the groups with the least disease burden. For 4 of the 6 drugs (blood glucose regulators, dyslipidemics, flu shots and pneumonia vaccinations) prevalence rates increased with disease burden but then sharply declined in the top two groups. A similar inverted “U” shape pattern was observed in the annual utilization rates for the 4 chronic care drug categories. Preliminary analysis indicates that these patterns are not significantly reduced by adjustments for demographics, income, drug coverage, and health status, but are highly sensitive to medical encounters, particularly the number of different physicians seen in the year. Conclusions: Medicare beneficiaries with diabetes receive suboptimal treatment for recommended drug therapies particularly among those with minimal or significant disease burdens (the highest prevalence rates and intensity of treatment levels were found in the middle of the range of disease burden). Implications for Policy, Delivery, or Practice: Our analysis indicates that expanded prescription coverage under the Medicare Part D drug benefit is unlikely to significantly improve these drug treatment patterns for Medicare beneficiaries with diabetes. However, changes in disease management programs that may be associated with the new benefit may prove beneficial. Primary Funding Source: CWF ●Prescription Drug Use and Expenditures Among Dually Eligible Medicare and Medicaid Beneficiaries in 2001 Implications for the Medicare Part D Drug Benefit James Verdier, J.D., Myoung Kim, Ph.D., Deo Bencio Presented By: James Verdier, J.D., Senior Fellow, Research, Mathematica Policy Research, Inc., 600 Maryland Ave., SW, Suite 550, Washington, DC 20024-2512; Tel: (202) 484-4520; Fax: (202) 863-1763; Email: jverdier@mathematica-mpr.com Research Objective: To illuminate issues likely to arise in the new Medicare Part D prescription drug benefit by examining, both at the national and state levels, patterns of Medicaid prescription drug use and expenditures in 2001 among dual eligibles by beneficiary characteristics (age, gender, race, and use of nursing facilities) and drug characteristics (brand status, drug type, and statutory exclusion from Part D). Study Design: Using Medicaid Analytic eXtract (MAX) 2001 files for 50 states and the District of Columbia released by the Centers for Medicare and Medicaid Services (CMS) in 2005, we created analysis files that enabled us to tabulate various measures of dual eligible drug use and expenditure for each state and the nation. We draw state-by-state comparisons and examine national patterns of use and associated Medicaid expenditures among dual eligibles in 2001, with comparisons to MAX 1999 data. Population Studied: All fee-for-service (FFS) Medicaid beneficiaries who were eligible for Medicare and received full Medicaid benefits in 2001. A limited number of dual eligibles enrolled in capitated managed care arrangements were excluded because of lack of data. Principal Findings: While the results for 2001 cannot be released until they been cleared by CMS, we expect clearance in time for the meeting. We will be able to report comparative state-by-state measures such as Medicaid expenditures on prescription drugs for dual eligibles in 2001 as a share of total Medicaid prescription drug expenditures, average annual perbeneficiary drug expenditures for dual eligibles under age 65 compared to those age 65 and over, drug expenditures for dual eligibles residing in nursing facilities as a share of total drug expenditures, the number of prescriptions filled per benefit month among duals in nursing facilities compared to those in the community, use of antidepressants and antipsychotics by duals living in the community and in nursing facilities, and use by dual eligibles of drugs statutorily excluded from Part D coverage, such as benzodiazepines and barbiturates. Conclusions: There is wide variation among states in dual eligible drug use and expenditures. Expenditures grew sharply between 1999 and 2001. Dual eligible drug use by those under age 65 and those in nursing facilities is substantially higher than for those age 65 and over and those living in the community. Use by dual eligibles of drugs statutorily excluded from Part D is widespread, but varies significantly from state to state. Implications for Policy, Delivery, or Practice: With the start of the new Medicare Part D prescription drug benefit in 2006, drug coverage for over 6 million dual eligibles has shifted from Medicaid to Medicare. Most of the health plans to which duals have been assigned for Part D have little experience with this population. The 2001 MAX data are the most detailed and comprehensive comparative state-by-state data on drug use by dual eligibles that will be publicly available until and unless the Part D drug data for 2006 are released by CMS, so they represent a major resource for both Part D plans and those monitoring and analyzing Part D who seek to anticipate and respond to the prescription drug needs of this important population. Primary Funding Source: CMS Call for Papers Medicare: Impact of Different Payment & Coverage Policies on Utilization & Outcomes Chair: Karen Milgate, Medicare Payment Advisory Commission Monday, June 26 • 10:30 am – 12:00 pm ●Medicare Advantage Plans: Different Payments, Different Benefits Brian Biles, M.D., M.P.H., Nora Super, Lauren Nicholas, M.P.P. Presented By: Brian Biles, MD, MPH, Professor, Department of Health Policy, George Washington University, 2021 K St NW, #800, Washington, DC 20006; Tel: (202) 416-0066; Fax: (202) 543-5311; Email: bbiles@gwu.edu Research Objective: The Medicare Moderization Act (MMA) of 2003 sought to greatly increase the role of private health plans in providing Medicare benefits to the elderly and disabled. MMA policies included substantial increases in payments to MA plans with the generaly understanding that plans would use the additional funds to provide additional benefits and attract additional plan enrollees. This study analyzes the pattern and trends in Medicare payments to Medicare Advantage plans and in the benefits provided by MA plans in order to understand the value of Medicare Advantage to the overall Medicare program and to MA plan enrollees. Study Design: This study first analyzes the amounts paid to MA plans both as the actual annual amount and relative to average Medicare costs in the traditional fee-for-service (FFS) system. This analysis is conducted at the county level. The study next analyzes the benefits provided to MA plan enrollees by comparing estimated out-of-pocket (OOP) costs for enrollees in individual MA plans at the county level. OOP costs are estimated for enrollees in good, fair and poor health. The study finally analyzes the benefits provided to MA plan enrollees, as indicated by estimated OOP costs, in different regions, states and cities across the nation. Population Studied: The study analyzes key factors in the Medicare Advantage program including Medicare payment policies, private plan decisions on the design of benefit packages, and beneficary choices regarding enrollment in MA plans. The analysis is conducted using data at the county level aggregated to state and national levels. Principal Findings: Study findings indicate that Medicare payments to MA plans will exceed costs in the FFS system by as much as $30 b over the next 5 years and will vary greatly across the nation at the county level. Policies that pay MA plans in rural areas as much as 20% more than FFS contribute relatively little to total extra payments to plans, primarily because MA plans have not attacted rural residents. The major policy that leads to extra payments is the urban floor that pays more to plans in metro areas with low FFS costs. Findings indicate that OOP costs of MA plan enrollees in poor health exceed total OOP costs of beneficaries with Medigap coverage in more than one-third of cities studied. A number of MA plans now have hospital deductables or copayments that can exceed $1,000 per stay. While monthly Medigap premiums are substantially greater than MA premiums, Medigap covers virtually all Medicare OOP costs for acute care services. Study findings also indicate that there is great variation across the nation in the extent to which MA plans provide additional benefits to enrollees. Plans with substantial additional benefits are concentrated in approximately 20 cities, generally in the West and other areas with a long-time history of Medicare managed care. Less additional benefits are provided to beneficaries in rural areas and many urban areas. Conclusions: This study sheds light on the overall value of the MA program as currently structured. Findings regarding Medicare costs detail how the MA program costs more than FFS due to a set of specific payment policies. Findings regarding benefits suggest that the policies that allow individual MA plans to design benefit packages have led plans in a number of cities to set high OOP costs for the use of health services. Additional benefits of MA plans are not widely distributed across the nation. Implications for Policy, Delivery, or Practice: The Medicare Advantage program with its reliance on private plans to control costs has been proposed as the foundation for a solution to the long-range fiscal problems of the Medicare program. To increase the role of MA plans in Medicare, the MMA provided substantial increases in payment to plans to the extent that MA plans are now paid more the average Medicare FFS costs in every county in the nation. In 2005, Congressional efforts to reduce the Federal budget deficit began with a focus on reducing health spending by the Medicaid program. Difficulty reaching agreement on Medicaid reductions led to bipartisan proposals for Medicare reductions, including reductions in extra payments to MA plans. The final legislation includes a provision to reduce MA plan budget neutral risk adjustment payments over the next 5 years by over $6 b. In an era of Federal deficits of $300 b a year, the MMA policies that increase payments to MA plans may well remain a major area of interest to policy makers. Analysis of the specific policies that increase payments to MA plans, as well as patterns in the amount of additional benefits provided to MA plan enrollees and the extent of variations of MA plan benefits across the nation, will be useful in these deliberations. Primary Funding Source: CWF ●Do Medicare Coverage Policies Matter?: Evaluating the Impact of Coverage Policies on Utilization of Services Susan Bartlett Foote, J.D., MA, Robert J. Town, Ph.D., Beth A. Virnig, Ph.D. Presented By: Susan Bartlett Foote, JD, MA, Associate Professor, Health Services Research and Policy, University of Minnesota School of Public Health, 420 Delaware Ave., S.E., MMC 729, Minneapolis, MN 55455; Tel: (612) 626-2851; Fax: (612) 624-2196; Email: foote003@umn.edu Research Objective: The Medicare program has invested significant resources in the development of local and national coverage policies. Coverage policies determine whether a procedure or technology will be provided in the Medicare program, and include details about the conditions under which that coverage will be permitted. There is a growing body of research on the development of coverage policies, but little is known about their impact on practice patterns. The principal objective of this project was to determine whether and to what extent Medicare coverage policies matter. Do policies affect utilization of services in Medicare? Are there differences based on whether the policy affects new technology or widely disseminated technologies? What is the link to regional variation in the Medicare program? Study Design: We analyzed Medicare claims data focused on 10 case studies pre- and post- issuance of a local or national coverage policy to examine how trends in their use are influenced by Medicare coverage decisions. We used Medicare data from 2000-2002 in three Medicare claims files: the MedPAR hospitalization file, the Outpatient file, and the Carrier file. To analyze the data, we used a difference-indifference approach to identifying the impact of Medicare policies on utilization. That is, we used a carrier fixed effects framework that allows for autocorrelated errors. Population Studied: Medicare beneficiaries receiving services in our 10 case studies over a three year period. Principal Findings: We found that in only one procedure, deep brain stimulation, did the promulgation of a policy impact utilization. In that case study, the issuance of a policy significantly reduced utilization. For the other 9 cases, we found no measurable effect before and after issuance of a policy on utilization patterns. Conclusions: We conclude that while Medicare’s coverage policies may provide important information assessing the value of a technology or procedure, Medicare’s policies don’t appear to change provider behavior. Potential explanations include the lack of detailed information on claims forms to allow evaluation of compliance with policy provisions, limitations on the use of computerized edits to review claims, and the lack of resources or incentives to manually review specific medical records which is the only way to assure compliance with many parameters in coverage policies. Implications for Policy, Delivery, or Practice: There are significant policy and practice implications of these findings. Medicare has invested major resources in the development of local and national coverage policies. Medicare, however, has not provided either the incentives or the tools for the contractors who process the claims to monitor and enforce compliance with coverage policies. These findings point to the need for reconsideration of the way that claims are reported and processed in Medicare. As the Medicare program moves in the direction of quality improvement and pay for performance, issues related to compliance with carefully crafted, evidence-based coverage policies gain greater salience for providers and policymakers alike. Primary Funding Source: RWJF ●Race and Non-Adherence to Prescription Medications Among Seniors: Results of a National Survey Walid Gellad, M.D., Jennifer S. Haas, M.D., MSPH, Dana Gelb Safran, Sc.D. Presented By: Walid Gellad, M.D., General Medicine Fellow, Division of General Medicine, Brigham and Women's Hospital, 1620 Tremont Street, 3rd Floor, Boston, MA 02120; Tel: (617)732-5500 x33513; Email: wgellad@partners.org Research Objective: Prescription drug coverage is currently at the forefront of public discussion in healthcare. Patient nonadherence to prescription regimens results in poorer control of chronic health conditions. While non-adherence due to financial factors (“cost-related non-adherence”) might be addressed by drug coverage, other types of non-adherence will not. Because of significant racial/ethnic disparities in the prevalence of many chronic health conditions, we examined whether reasons for non-adherence (cost, self-assessed need, personal experience) varied by race and ethnicity among seniors. Study Design: Cross-sectional analysis of data from a 2003 national survey. Respondents were asked about nonadherence to prescriptions during the last 12 months, specifically asking about cost-related non-adherence, nonadherence due to experiences (e.g. “ skipped doses because it was making you feel worse”), and non-adherence due to selfassessed need (e.g.“ skipped doses because you felt you were taking too many medicines).” We compared each of these causes of non-adherence among races using Chi-square tests. Multivariable logistic regression was used to calculate odds ratios for each type of non-adherence, controlling for age, insurance status, number of chronic conditions, and income. Population Studied: Medicare beneficiaries greater than 65 years of age, who reported their race/ethnicity as white, black or Hispanic and who reported taking at least one medication (n=14,829). Principal Findings: 41.5% of these seniors reported some form of non-adherence. Blacks and Hispanics were more likely to report cost-related non-adherence than whites (35.1%, 36.5%, 26.7% respectively, p=.0001). When asked if they had “spent less on food, heat, or other basic needs to afford prescriptions,” 11.3% of whites said yes, 26.6% of blacks, and 23.9% of Hispanics (p=<.0001). There were no racial/ ethnic differences in non-adherence due to experiences or nonadherence due to self-assessed need (p all >0.39). Among respondents who reported any non-adherence, there was a significant relationship between race and the types of nonadherence reported (p<.0001): 38.9% of blacks and 41.4% of Hispanics reported only cost-related non-adherence, versus 28.4% of whites. Blacks had 1.48 times the odds of being non-adherent due to cost (95%CI 1.19,1.86) compared to whites, and Hispanics had an odds ratio of 1.58 (95%CI 1.22,2.04), and these values did not change when controlling for age, number of chronic conditions, and the presence of drug coverage. When income was added to the model, race was no longer a significant predictor for non-adherence due to cost. Conclusions: This 2003 national survey of seniors suggests that racial/ ethnic disparities in non-adherence are largely explained by cost concerns and not by self-assessed need or differences in experience. The disparity in cost-related nonadherence persists even when accounting for the presence of drug coverage, but the relationship is substantially attenuated when income is taken into account. Implications for Policy, Delivery, or Practice: Reducing racial disparities in chronic diseases requires that prescription cost issues be explicitly and aggressively addressed, both in the clinic and in health policy, especially for the poor. This is particularly important as outreach efforts are underway to encourage low-income Medicare beneficiaries to sign up for subsidies for the new drug benefit. Primary Funding Source: CWF, Henry J. Kaiser Family Foundation ●Effect of Physician Referral Patterns on Medicare Utilization of Diagnostic Imaging Rebecca Lewis, Mythreyi Bhargavan Presented By: Rebecca Lewis, American College of Radiology; Email: rebeccal@acr.org Research Objective: Diagnostic imaging constitutes the fastest growing component of Medicare costs, growing at a rate of 10% per year between 1999 and 2003, and 18% between 2003 and 2004 as reported by the Medicare Payment Advisory Commission in its June 2005 data book. The objective of this study is to examine the effect of physician referral patterns on Medicare imaging use, while controlling for beneficiaries’ demographic and clinical characteristics. Study Design: Medicare physicians are classified into four categories based on referral patterns for diagnostic imaging: (i) self-referrers if the referring physician performs imaging procedures on any of their patients, (ii) radiologist-referrers if the referring physician always refers imaging to a radiologist, (iii) same-specialty if the physician is not a self-referrer but always refers patients to another physician of the same specialty as himself or herself, and (iv) others. Physicians are classified into these categories for each imaging technique (for example, MRI, CT). For a set of diagnoses, such as heart disease, knee pain, and back pain, we compare the rate of imaging utilization ordering of physicians in each of these categories, while controlling for patient characteristics such as age, gender, race, whether the person has state buy-in for Medicare, and characteristics of the state and county of residence. Outcomes of interest are (a) whether a patient receives an imaging procedure, and (b) imaging dollars per patient. We use the Medicare 5% Limited Data Set claims files for these analyses. All statistical analyses will use standard techniques, such as logistic regression for outcome (a) and linear or log-linear regression for outcome (b). Population Studied: 5% sample of Medicare beneficiaries for the period 2002-2004. Principal Findings: Analysis of summarized Medicare claims data reveal that imaging utilization by non-radiologists in nonhospital settings grew substantially faster than imaging by radiologists. During the period 2000-2003, the costs of inoffice CT by non-radiologists grew, on average, at 26.5% per year and the costs of in-office MRI at 27.9% per year. The corresponding growth rates for imaging by radiologists are 14.4% for CT and 12.7% for MRI. This suggests that the rate of imaging ordered by self-referrers and same-specialty referrers may differ significantly from imaging ordered by radiologist referrers. This study will measure those differences. Conclusions: We do not have any specific conclusions at this point. Analysis results will reveal whether referral patterns explain the recent growth in imaging utilization among Medicare beneficiaries. Implications for Policy, Delivery, or Practice: Much of the recent growth in healthcare expenditures may be explained by technological progress in medicine, i.e., the fact that medicine can do more for each patient now than it could several years ago. However, this growth has been found to be highly variable across geographical areas, sub-populations, and physicians. Identifying the part of variability that can be attributed to physician referral patterns will assist payers and policy makers design optimal educational initiatives to help ensure physicians make appropriate utilization of diagnostic imaging. Primary Funding Source: No Funding ●Drug Use in Severely Mentally Ill Medicare Beneficiaries: Impact of Discontinuities in Drug Coverage Ilene Zuckerman, Ph.D., PharmD, Linda Simoni-Wastila, RPH, Ph.D., Christopher Blanchette, MS, Bruce Stuart, Ph.D. Presented By: Ilene Zuckerman, PhD, PharmD, Associate Professor, Lamy Center on Drug Therapy and Aging, University of Maryland Baltimore School of Pharmacy, 515 W. Lombard Street, Room 163, Baltimore, MD 21201; Tel: 410706-3266; Fax: 410-706-1488; Email: izuckerm@rx.umaryland.edu Research Objective: The purpose of this study is twofold: To 1) describe the extent of drug benefits possessed by severely mentally ill Medicare beneficiaries and 2) determine how gaps in drug benefits influence use of prescription medications used to treat mental illness. Study Design: This study uses an observational cohort analysis using ordinary least squares regression, adjusting for sociodemographics, health status, comorbidities, and death. The primary outcome measures are total mental health drug, antidepressant and antipsychotic utilization, assessed as total counts of drugs used over 3 years. Gaps in drug benefits are measured as proportion of months not covered over three years, and reported categorically (0 gaps (full drug coverage), 1-24% gaps, 25-49% gaps, 50-74% gaps, 75-99% gaps, and 100% gaps (no coverage). Outcome and drug coverage gap measures are annualized, and account for time lost to death and loss-to-follow-up. Population Studied: The population studied is a pooled sample of three, 3-year cohorts of community-dwelling, severely mentally ill disabled and aged Medicare beneficiaries derived from the 1997 - 2001 Medicare Current Beneficiary Survey. Severe mental illness is assessed based on the existence of 2 or more relevant ICD-9 codes extracted from Part A and B claims. Principal Findings: Among our sample representing over 2.5 million SMI beneficiaries, over half (54.4%) had full drug coverage and 18.9% had no coverage. Controlling for sociodemographic and clinical confounders, beneficiaries with drug coverage gaps received 2.2 to 3.8 fewer mental health drugs annually than their fully covered peers, with those experiencing a 25-49%, 75-99% and 100% drug coverage gaps impacted the most severely (-3.7, -3.8, and -3.7 prescriptions, respectively). Drug coverage gaps involving the antipsychotics followed a similar pattern, while antidepressant use was not as dramatically impacted. Among antidepressant users, only those experiencing 50-74% and 100% coverage gaps receiving significantly fewer antidepressants annually (-1.3 and -1.7, respectively). Conclusions: This study is the first to demonstrate that any discontinuity in drug benefits can adversely impact use of mental health drugs in severely mentally ill Medicare beneficiaries. Because such coverage gaps in drug coverage are possible under the Medicare Modernization Act's "doughnut hole" design, this study provides an important first glimpse at what might happen to severely mentally ill individuals who enroll in the Part D plan. Such reductions in utilization of these important medications for this frail population may lead to adverse consequences, including symptom breakthrough, impaired cognitive function, and even mortality. Implications for Policy, Delivery, or Practice: As state and Federal policy-makers implement and evaluate the Part D component of the Medicare Modernization Act, further scrutiny of its impact on this medically vulnerable group is warranted. Further research is needed to examine outcomes associated with disruptions in pharmacologic treatment. Primary Funding Source: RWJF Related Posters Medicare & Medicare Prescription Drugs Poster Session A Sunday, June 25 • 2:00 pm – 3:30 pm ●The Heart of the Matter: Cardiovascular Health Issues Among Adult Patients Using Stimulant Medications for ADHD and Narcolepsy Karyn Kai Anderson, Ph.D., M.P.H. Presented By: Karyn Kai Anderson, Ph.D., M.P.H., social science research analyst, Office of Research, Development and Information, Centers for Medicare and Medicaid Services, 7500 Security Blvd., C3-20-17, Baltimore, MD 21244-1850; Tel: (410)786-6696; Fax: (410)786-5534; Email: karyn.anderson@cms.hhs.gov Research Objective: With reports of sudden death and other cardiovascular events occurring in individuals taking medications for ADHD and narcolepsy, the FDA announced in the first week of the new year, 2006, that it will launch a new probe into the safety of these drugs. This newfound alarm also brings focus to the patient care guidelines ill advising against prescribing stimulant medications to those diagnosed with a heart-related condition. While most attention is given to pediatric use of stimulant medication, it is plausible that these complications may be especially pronounced among older adults. Given the importance of these issues coupled with the dearth of extant research, we used CMS data to investigate the extent to which stimulant medications are prescribed to adult dual eligible beneficiaries (i.e., those eligible for both Medicaid and Medicare) with preexisting heart conditions. Study Design: The sample was a 5% extraction of Medicare beneficiaries that was then matched to Medicaid claims files (N=285,448). Medicare eligibility data from 1999 was matched to prescription drug claims data from 2000. A key variable under study included cardiovascular diagnosis as indicated by ICD-9 codes for hypertensive disease (401-405), ischemic heart disease (410-414), diseases of pulmonary circulation (415-417), other forms of heart disease (420-429), and cerebrovascular disease (430-438). Use of stimulant therapy was denoted by NDC codes for the following ADHD and narcolepsy medications on the market between 1999 and 2000: Adderall, Ritalin, Dexedrine, dextroamphetamine sulfate AND amphetamine sulfate, Cylert, and Methylin ER. Population Studied: The population studied was the adult "dual eligible" population (i.e., those eligble for both Medicare and Medicaid). Principal Findings: As expected, only a small portion of this mostly elderly dual eligible population had taken medication for ADHD or narcolepsy (0.05%). Overall 47.89% of this sample had a diagnosis for a cardiovascular condition. Of the 154 that had a prescription for a stimulant medication, 33.77% (52) had been diagnosed with some type of cardiac diagnosis prior to or while being prescribed stimulants. Specifically, among stimulant users, 16.23% had a prior or concurrent diagnosis of hypertension, 5.84% had ischemic heart disease, 8.44% had other forms of heart disease including dysrythmia and heart failure and 3.25% had a cardiovascular disease or stroke. Conclusions: The findings from this study suggest unequivocally that minimum precautions are not being adhered to in terms of screening older adults for heart-related conditions prior to and while treating ADHD and narcolepsy with stimulants. Implications for Policy, Delivery, or Practice: In the current debate regarding the safety of stimulant medications, very little attention has been given to physicians’ adherence to counterindication guidelines. Clearly, medical screening and ongoing health monitoring should have a strong role in treatment plans for patients with ADHD and narcolepsy. Primary Funding Source: CMS ●Recent Trends in Use of Generic Drugs among Medicare Beneficiaries and Their Potential Impact on Out of Pocket Costs under Medicare Part D Jessica Banthin, Ph.D., G. Edward Miller, Ph.D. Presented By: Jessica Banthin, Ph.D., Division Director, Center for Financing, Access and Cost Trends, AHRQ, 540 Gaither Rd., Rockville, MD 20850; Tel: 301-427-1678; Fax: 301-427-1276; Email: Jbanthin@AHRQ.GOV Research Objective: With the start of Medicare Part D in 2006, there may be more pressure for Medicare beneficiaries to switch from brand name to generic drugs. Many of the Medicare Part D plans provide better coverage for generics than for brand name drugs in the same class. In addition, some widely used drugs will lose patent protection in the near future. We document the role of generic drugs over time in some of the most expensive categories of drugs used by Medicare beneficiaries. We then simulate the potential impact on out of pocket costs of increased substitution of generic for brand name drugs under several specific Medicare Part D plans. Study Design: Until Medicare Part D claims data are available, the Medical Expenditure Panel Survey provides some of the most accurate and nationally representative data on Medicare beneficiaries’ use of and expenditures on prescription drugs. The MEPS verifies household reports with claims data collected from pharmacies. Using the National Drug Code, these data are linked to secondary data. The first part of our analysis provides a comprehensive view of recent trends from 1997 to 2003 in total drug expenditures for the Medicare population and for key subpopulations such as dual eligibles. We identify the fastest growing clinical classes of drugs and decompose aggregate trends into changes in population with use and changes in expenditures per user for each class. We then estimate changes in the percent of prescriptions that are generic versus brand and identify classes where generic drugs have played a large role or may in the future. The second part of the analysis simulates the changes in total and out of pocket costs for beneficiaries under a sample of popular Part D plans when generics are substituted for brand name drugs. We focus on beneficiaries with common chronic conditions where generic drugs are or will be available for treatment, including statins, anti-diabetic drugs, anti-depressants, pain relief and gastro-intestinal drugs. We examine the impact from the perspective of individuals with chronic conditions and take account of their current coverage, if any, compared to the coverage offered under Medicare Part D. Population Studied: Total Mediare population, with a focus on dual eligibles, persons with chronic conditions, and persons without coverage. Principal Findings: Using data from 2001, earlier work showed that the fastest growing categories of drugs among Medicare beneficiaries included statins, proton pump inhibitors, anti-diabetic agents, antipsychotics and antidepressants. There is substantial variation in the availability of generic drugs within these categories, ranging from a low of 5 percent of prescriptions for statins to a high of 46 percent for all types of psychotherapeutic drugs. Final results will update these findings and simulate future scenarios. Implications for Policy, Delivery, or Practice: The true costs of Medicare Part D are difficult to predict given the uncertain enrollment but nationally representative data on beneficiaries’ current use of generic drugs provides a baseline from which further savings are then estimated. This analysis indicates the degree to which Part D plans may protect beneficiaries from high out of pocket costs for some common chronic conditions. Primary Funding Source: AHRQ ●Medicare Prescription Drug Program: Report of Findings from Cognitive Interviws to Develop Measures for the CAHPS-PDP Survey Module Julie Brown, Pam Dardess, Ph.D., Amy Heller, Ph.D., Kristin Carman, Ph.D. Presented By: Julie Brown, Director, Survey Research Group, RAND Corporation, 1776 Main Street, PO Box 2138, Santa Monica, CA 90407-2138; Tel: (310) 393-0411 ext. 6212; Fax: ; Email: Julie_Brown@rand.org Research Objective: The Medicare Part D prescription drug program (PDP) presents beneficiaries with many choices. Providing senior citizens with information about the quality of prescription drug plans allows them to choose the best plan for their needs. The Consumer Assessment of Healthcare Providers and Systems (CAHPS®) program is a public-private initiative to develop standardized surveys of patient experiences. These survey results are reported to consumers to use as a decision aid. The objective of the current research effort is to use cognitive interviews to inform the development of a CAHPS-PDP survey module that will allow beneficiaries to report their experiences and satisfaction with their Medicare Prescription Drug Plans. Study Design: This paper presents the findings from 30 cognitive interviews conducted in late winter, 2006, with ten interviews in each of three heterogenous regions of the country. Cognitive interviews provide a unique opportunity to learn how well matched are researchers’ and respondents’ understandings of the meaning of candidate survey items. Cognitive testing involves intensive one-on-one interviews with members of the target survey population. During these 2hour interviews, candidate survey questions are administered and the cognitive interview subjects are concurrently debriefed with regard to what they thought the questions were asking, how they formulated their responses, and whether or not their answers fit into the offered response categories. Population Studied: These cognitive interviews are conducted with Medicare beneficiaries, aged 65 and older, in three distinct regions of the US: in New England, the MidAtlantic, and in California. Subjects are recruited to represent as diverse a group as possible with regard to race, ethnicity, and level of education. Principal Findings: Based on the results of the 12 focus groups we conducted at an earlier stage of instrument development and subsequent refinement from cognitive testing, items to measure satisfaction with prescription cost structure, ease of filling and refilling prescriptions, pharmacy access, PDP services, etc. were developed. Conclusions: These cognitive interviews will inform both item reduction and the modification of individual question wording for the item set that will comprise the CAHPS-Medicare Prescription Drug Program survey module. Implications for Policy, Delivery, or Practice: The Medicare PDP is a complex and significant undertaking for CMS. Currently, many products and offerings have been developed, creating an overload of information for some seniors. Both CMS and beneficiaries need information on the quality and performance of plans to make informed decisions about PDPs. The CAHPS-PDP instrument will provide such information from the consumer perspective, allowing CMS and others to assess consumers’ experience with the prescription drug benefit. The data may be used for quality improvement as well as for patient decision making. Primary Funding Source: CMS, AHRQ ●Prescription Drug Utilization Among the CommunityDwelling Elderly Population from the 2003 Medical Expenditure Panel Survey Joey Crosby, Ph.D, R.Ph, Rod McAdams, Ph.D. Presented By: Joey Crosby, Ph.D, R.Ph, Associate Professor of Health Sciences, Health Sciences, Armstrong Atlantic State University, 11935 Abercorn St., Savannah, GA 31419; Tel: (912) 921-7316; Fax: (912) 921-7350; Email: crosbyjo@mail.armstrong.edu Research Objective: To estimate patterns of prescription drug utilization among the non-institutionalized elderly population in 2003 using data from the Medical Expenditure Panel Survey. Study Design: Descriptive, cross sectional Population Studied: Non-institutionalized, Medicare-eligible, elderly population living during 2003. Principal Findings: Compared with earlier studies, the economic burden of prescription drug utilization among the community-dwelling elderly population (greater than or equal to 65 years of age in 2003) continued to increase, with the average elderly person consuming approximately 27 prescriptions, expending over $1600 total and $875 out-ofpocket, translating into an out-of-pocket burden of 53% of total spending. This economic burden varies significantly across subgroups of the non-institutionalized elderly population stratified by demographic, income, and health status categories, with females, the poor, and those reporting lower levels of physical health facing a larger total burden. Conclusions: The economic burden of prescription drug utilization continued to increase during 2003 for the noninstutionalized elderly population compared with previous years, with more socially vulnerable subgroups facing the largest economic burden overall. Implications for Policy, Delivery, or Practice: The Medicare Modernization Act of 2003, which added a prescription drug benefit to the Medicare program after 40 years of existence, was enacted to address the perceived growing economic burden with respect to prescription drug use within the elderly population. As the future cost of such a benefit grows beyond initial projections, it may be necessary to adjust future benefits as some point so as to target those subgroups that are most in need of subsidization. This study, and others like it, assists with the identification of such groups. Primary Funding Source: No Funding ●Medicare Chronic Condition Data Warehouse Debbie Dean, MS, Brian O'Donnell, Ph.D., Kathy Schneider, Ph.D., Debbie Dean, MS Presented By: Debbie Dean, MS, Manager, Health Informatics, Information Systems, Health Informatics, Iowa Foundation for Medical Care, 6000 Westown Parkway, West Des Moines, IA 50266; Tel: (515) 440-8656; Fax: (515) 457-3793; Email: ddean@ifmc.org Research Objective: As part of the Medicare Modernization Act of 2003 (MMA), the Chronic Condition Warehouse (CCW), a research database, has been launched by the Centers for Medicare & Medicaid Services (CMS). Section 723 of the MMA outlined a plan to “improve the quality of care and reduce the cost of care for chronically ill Medicare beneficiaries.” An essential component of this plan was to establish a data warehouse that contained various types of Medicare data from multiple settings and providers. The CCW provides the research community with beneficiary enrollment, health services, and assessment data. These sources are linked across the continuum of care by a unique, unidentifiable beneficiary key to construct a patient-centric view of enrollment and service utilization. Goals for the CCW include improving the quality of care for chronically ill Medicare beneficiaries, reducing program spending, and making current Medicare program data more readily available to researchers studying chronic illness in the Medicare population. Study Design: The Iowa Foundation for Medical Care (IFMC) contracted with CMS to design, implement, and support this extensive patient-level continuum of care database. For a 5% random sample of Medicare beneficiaries, data for care received in inpatient, outpatient, home health, skilled nursing, hospice, and nursing home settings, as well as durable medical equipment services, are available from 1999 forward. Future plans for the CCW may include the addition of Medicare prescription drug event and Medicaid data. Population Studied: A 5% random sample of Medicare beneficiaries from 1999 forward. Principal Findings: The CCW links beneficiary enrollment, health services, and assessment data across the continuum of care. Conclusions: The CCW is a new and valuable resource for the research community to study health services, including quality of care, cost, utilization, outcomes and other areas of interest for Medicare beneficiaries, across care settings. Implications for Policy, Delivery, or Practice: While data for all beneficiaries in the sample are available, researchers may request data for a specific, predefined cohort based on a set of twenty-one common chronic condition categories. These conditions include diabetes, chronic obstructive pulmonary disease, congestive heart failure, osteoporosis, depression, five types of cancer (i.e., breast, colorectal, prostate, lung and endometrial), and other prevalent chronic diseases. The preset cohort definitions will assist in simplifying and expediting the fulfillment of research requests, and allow for more cost efficient methods of delivering data files to researchers. Additionally, researchers may request data based on diagnosis and procedure categorization schemes or assigned comorbidities as defined by the Agency for Healthcare Research and Quality (AHRQ) Clinical Classification Software or Comorbidity Software. These tools allow for grouping of AHRQ-defined conditions and procedures, or assigning variables that identify comorbidities included in hospital discharge records, based on the International Classification of Disease, Ninth Edition, Clinical Modifications. Lastly, researchers may customize research requests for unique beneficiary populations specific to their research focus. For example, a researcher may be interested in a chronic condition not already predefined by the CCW project. A review board will work closely with researchers to refine all data requests and ensure requests incorporate all relevant data files, fields, and time frames. The various data types available to researchers, as well as the data fields, selection, output and delivery options will be discussed during this presentation. Primary Funding Source: CMS ●Mood Disorder Costs During Inpatient Rehabilitation Deborah Dobrez, Ph.D., Allen Heinemann, Ph.D., Anne Deutsch, Ph.D. Presented By: Deborah Dobrez, Ph.D., Assistant Professor, Health Policy and Administration, University of Illinois at CHicago, 1603 West Taylor Street, Chicago, IL 60618; Tel: 312413-8854; Fax: 312-996-5356; Email: ddobrez@uic.edu Research Objective: Mood disorder diagnoses may complicate service delivery during inpatient rehabilitation care, increasing the intensity of resource needs and length of stay. Although payment is adjusted for specific patient characteristics and the presence of many comorbidities, under Medicare’s prospective payment system, no adjustment in reimbursement is made for mental disorders, raising concern that facilities may be burdened with additional, uncompensated cost for providing care to patients with mental disorders, and that access or quality of care may be negatively affected. The objective of this study was to estimate how the presence of a mood disorder affects the cost of rehabilitative care, the extent to which reimbursement is indirectly affected. Study Design: Episode payment and length of stay were compared between patients with and without mental disorders. The relationship between case mix group assignment, comorbidity-related adjustments in payment, and the presence of mental disorders was estimated. Multiple linear regression was used to estimate the effect of mental disorders on length of stay, controlling for patient characteristics. Population Studied: Data were drawn for Medicare stroke from the American Medical Rehabilitation Providers Association subscription database, eRehabData, for 20022004. Exclusion criteria included death during the rehabilitative stay or transfer to an institutional setting within three days. Patients with mood disorders were identified by ICD-9-CM codes recorded in the Inpatient Rehabilitation Facility – Patient Assessment Inventory. Principal Findings: 80,764 patients were identified during the three year study period, 14% of whom had a mood disorder. Mood disorders were associated with a nearly one day increase in length of stay. Although payment is not directly adjusted for the presence of mood disorders, prospective payment reimbursements were estimated to be $750 more for patients with mood disorders. Patients with mood disorders were more likely to be younger and have lower admission function (p < 0.01), resulting in a more highly reimbursed case-mix group. However, patients with mood disorders were less likely to have a comorbidity that would result in an increase in reimbursement (p < 0.05). Overall, patients with mood disorders had one-third of a day longer stays relative to patients without mood disorders, after adjustment for factors that indirectly increase payment. Conclusions: Patients with mood disorders have longer length of stays that are indirectly reimbursed in part by related patient characteristics and comorbidities in Medicare’s Prospective Payment System. However, mood disorders are likely to affect patients’ needs during the stay, so that actual costs may be much higher. Implications for Policy, Delivery, or Practice: An adjustment in reimbursement for the presence of a mood disorder would bring Medicare reimbursement in to line with facility treatment costs. The failure to directly compensate facilities for providing care to patients with mood disorders may result in reduced access to care for these patients, or unmet need during the rehabilitative episode. New or continuing episodes of mood disorders should be diagnosed and treated early in the acute care or rehabilitative stay to directly address the mood disorder, and to improve the ability of the patient to participate in the rehabilitative process. Primary Funding Source: National Institute for Disability and Rehabilitation Research ●Transitioning Dually Eligible Individuals with Mental Disorders to Medicare Part D: A Simulation of Effects on Access to Care Julie Donohue, Ph.D., Richard Frank, Ph.D., Arnold Epstein, M.D., MA Presented By: Julie Donohue, Ph.D., assistant professor, Health Policy and Management, University of Pittsburgh, 130 DeSoto Street Crabtree Hall A613, Pittsburgh, PA 15261; Tel: 412-624-4562; Fax: 412-624-3146; Email: jdonohue@pitt.edu Research Objective: Over 6 million Medicare beneficiaries receive drug benefits through Medicaid (dually-eligible). Mental disorders are highly prevalent among the duallyeligible and most Medicaid programs provide generous coverage with few restrictions on psychotropic drugs. Duallyeligible beneficiaries suffer a higher prevalence of chronic conditions than non dually-eligible beneficiaries and higher drug expenditures. In 2006, dually-eligible beneficiaries will transition from Medicaid drug coverage to a Medicare Part D prescription drug plan. Medicare drug plans vary widely in terms of cost sharing and formulary inclusion. Further, drug plans bear financial risk for management of drug benefits which introduces an incentive to restrict use of costly medications like psychotropic drugs through the use of formularies. In this work, we analyze the potential impact of transitioning to Medicare Part D drug plans on dually-eligible Medicare beneficiaries with mental disorders. Study Design: We profiled prescription drug use and spending patterns among dually-eligible Medicare beneficiaries with mental disorders using utilization information from the Medicare Current Beneficiary Survey. We obtained estimates of the most commonly used psychotropic medications among this population. Second, we obtained a sample of 14 Medicare Part D drug plans and extracted information on formulary coverage and pharmacy management tools for these commonly used medications. We used the results of those two inquiries and estimated the number of dually eligible beneficiaries with mental disorders who would be affected by restrictions on specific medications. Population Studied: Disabled and aged Medicare beneficiaries who have mental disorders and are duallyeligible. Principal Findings: Fifty-nine percent of dually-eligible disabled and 20 percent of dually-eligible aged beneficiaries have been diagnosed with a mental disorder. The most commonly used psychotherapeutic classes among disabled beneficiaries with mental disorders are antidepressants, antipsychotics and mood stabilizers. Antidepressants, benzodiazepines and antipsychotics are the most commonly used psychotropic drugs among dually-eligible elderly with mental disorders. Eighty six percent of Medicare drug plans covered the 10 most commonly used antidepressants and 74 percent covered all 10 of the most commonly used antipsychotics. The application of pharmacy management tools to psychotropic drug categories is prevalent among Medicaid drug plans. Among the plans analyzed, 57 percent imposed quantity limits on commonly used psychotropic drugs, 36 percent required prior authorization, and 14 percent required step therapy. Depending on their particular condition, we estimate that approximately one-quarter to twofifths of dually-eligible beneficiaries will be subject to these utilization management policies. Conclusions: Mental disorders are highly prevalent among dually-eligible Medicare beneficiaries. Substantial numbers of the dually eligible will be automatically enrolled in plans that place restrictions on commonly used psychotropic medications. Depending on how these utilization management programs are implemented, the transition to Medicare drug plans could result in high out-of-pocket costs and lead to reductions in medication adherence for this vulnerable population. Implications for Policy, Delivery, or Practice: Reductions in use of psychotropic medications among the dually eligible could lead to negative health outcomes and increased service utilization. It will be important for the Medicare program to monitor the use of pharmacy management tools for ensuring access to medications. Primary Funding Source: NIMH ●Structural Implications of the Medicare Modernization Act for the Broader Health Care System Robert Field, J.D., M.P.H., Ph.D. Presented By: Robert Field, J.D., M.P.H., Ph.D., Associate Professor of Health Policy and Director, Health Policy Program, Health Policy, University of the Sciences in Philadelphia, 600 South 43rd Street, Philadelphia, PA 19066; Tel: 215-596-7618; Fax: 215-596-7614; Email: r.field@usip.edu Research Objective: Identify features of the Medicare Modernization Act of 2003 (MMA), which implemented the outpatient prescription drug benefit, that could profoundly affect the larger health care system. Study Design: Policy analysis of provisions of the MMA and consideration of their political context. Principal Findings: In its present form, the MMA could have the following long-term structural effects on American health care: Move Medicare overall closer to a managed care voucher model. Revitalize managed care in the private market. Shift the focus of pharmaceutical research toward those cateogories of illness favored by the law (those with more required formulary categories). Encourage greater price sensitivity in pharmaceutical marketing and promote greater use of pharmacoeconomics. Reduce the attractiveness of Medicaid as a funding mechanism for long-term care and increase the federal role in its operation. Reduce the effectiveness of state pharmacy assistance programs. Conclusions: If the MMA remains in its present form, its influence will extend well beyond prescription drug availability to include profound structural effects throughout the health care system. It is important that policy debates place greater emphasis on these issues, which are largely neglected in most public discussions of the law. Implications for Policy, Delivery, or Practice: Whatever the outcome of the 2006 Congressional elections, Congress will have to consider at least some revisions of the MMA. Public appreciation of broader implications of the law could affect Congresssional debates. Primary Funding Source: No Funding ●Loss of Brand-Name Drug Coverage: Inhaled Steroid Use by Medicare Beneficiaries with Asthma Vicki Fung, Richard Brand, Ph.D., John Hsu, M.D., M.B.A., M.S.C.E. Presented By: Vicki Fung, Consulting Data Analyst, Division of Research, Kaiser Permanente, 2000 Broadway, 3rd Floor, Oakland, CA 94612; Tel: 510-891-3527; Email: vicki.fung@kp.org Research Objective: Eliminating brand-name drug coverage is increasingly common, especially for Medicare Advantage beneficiaries. Little is known about the effects of losing brandcoverage, and in particular, for patients prescribed drugs without generic alternatives, such as inhaled steroids. In 2004, Medicare beneficiaries in a large integrated delivery system (IDS) lost coverage for brand-name drugs. We examined the effect of this change on inhaled and oral steroid expenditures and use in 2003-2004. Study Design: We used automated data from Kaiser Permanente–Northern California (KPNC). Using annual difference-in-difference linear regression models, with a concurrent Medicare control group, we examined coveragerelated changes in total and patient out-of-pocket costs for inhaled steroids; and use of inhaled and oral steroids (proportion of days covered, PDC). We used similar logistic regression models to examine changes in adherence (PDC>=80%). We adjusted for sex, age, race/ethnicity, neighborhood socioeconomic status, comorbidity level (DxCG), chronic disease status, baseline inhaled steroid dose, prior oral steroid use, and a high-risk asthma registry flag. Population Studied: All 2,211 subjects were IDS asthma registry members, continuously enrolled (2003-2004), age 65+, who had inhaled steroid use in 2002-2003, and no chronic obstructive pulmonary disease diagnosis or ipatropium bromide drug use. Starting in 2004, 73.3% of these subjects switched from paying a $30 copayment for brand drugs, with a $1,000 annual benefit cap, to paying full price, with no other limits; 26.7% paid $15-25 copayments for brand drugs in both years. Principal Findings: Loss of brand-coverage was associated with increased mean annual patient out-of-pocket costs (difference-in-difference=+$57 dollars, p-value<.0001) and decreased total inhaled steroid expenditures (difference-indifference=-$115, p-value<.0001) in 2004. Overall, the coverage change was not associated with significant decreases in inhaled steroid use or adherence; however, adherence levels were low, and lower for subjects with no brand- versus brandcoverage (21.9% and 29.3%, respectively). On average, out-ofpocket price increases after the policy change were smaller for the IDS’s preferred ($30-copayment to $45 for the median prescription) versus non-preferred brands ($30-copayment to $238 for the most commonly prescribed drug). Among subjects receiving non-preferred brands in 2003, use of these drugs decreased in 2004 (difference-in-difference=-14.9% PDC, p-value=<.0001), without a concomitant increase in use of less expensive substitutes (difference-in-difference=-0.4%, p-value=0.90). Oral steroid use did not significantly increase after the coverage change. Conclusions: Loss of brand-coverage was associated with higher out-of-pocket and lower total expenditures for inhaled steroids. Adherence levels for inhaled steroids were low overall, and lower among patients with less generous benefits, but did not significantly decrease with the loss of brandcoverage. Patients taking non-preferred inhaled steroids decreased use, but did not switch to less expensive substitutes after losing brand-coverage. Implications for Policy, Delivery, or Practice: Within this IDS setting, loss of brand-coverage was not associated with lower inhaled steroid adherence; however, the IDS offered a relatively inexpensive brand option, and might monitor drug use more closely compared with other settings. Additional efforts are needed to encourage efficient drug choices and improve adherence. Future research will examine adherence levels, the behaviors of non-preferred brand users, and the clinical effects associated with this policy change. Primary Funding Source: AHRQ ●Prescription Drug Plan Choices of Dual Eligibles in New Hampshire Thomas Grannemann, Ph.D., Stefanie Johnson, M.B.A., Donald Hunter, M.P.A. Presented By: Thomas Grannemann, Ph.D., Senior Health Policy Analyst, Office of Medicaid Business and Policy, State of New Hampshire, 129 Pleasant Street, Concord, NH 03301; Tel: 603-271-4404; Fax: 603-271-8431; Email: tgrannemann@dhhs.state.nh.us Research Objective: This study measures the extent to which dual eligibles took action to enroll in the Prescription Drug Plan of their choice and effectively made choices that resulted in in less restrictive coverage of the drugs they used or were likely to use. Study Design: This study uses PDP enrollment data provided to the state by CMS, information on drug coverage provided by the PDPs, and state Medicaid enrollment and claims data on recipient characteristics, diagnoses, and prior use of drugs and services. It considers each of the major dual subgroups: the elderly, physically disabled, mentally disabled, and working poor. It focuses on high-cost drugs commonly used by many elderly and disabled Medicaid recipients. It distinguishes individual PDPs by their coverage restrictions for these drugs (tier assignment, required copayments, utilization restrictions). We use multiple regression analysis to identify factors associated with choosing a plan other than that assigned and predictive factors affecting the specific plan chosen. Population Studied: 19,000 dual eligibles in New Hampshire at the start of 2006 Principal Findings: For each of the groups of dual eligibles, the paper shows (1) the extent to which the dual eligibles took advantage of the option to switch from a plan selected by auto-enrollment and (2) the extent to which such switches resulted in less restrictive coverage of drugs currently used or commonly used by duals. Conclusions: The propensity of duals to select or change PDPs in the first months of the Medicare drug benefit differed among the elderly, physically disabled, mentally disabled and depended on presences of chronic conditions and prior use of prescription drugs. Implications for Policy, Delivery, or Practice: With their limited resources and high medical needs, the appropriate choice of a PDP is particularly important for dual eligibles. Any evidence of inaction or poor choices of plans by duals may suggest the need for better coverage options, or additional information or assistance for duals in making plan choices. Primary Funding Source: State of New Hampshire ●Cost and Cost-Effectiveness of Adding a New Monoclonal Antibody to Standard Chemotherapy in Advanced Non-Small Cell Lung Cancer Patrick Grusenmeyer, ScD (cand) Presented By: Patrick Grusenmeyer, ScD (cand), MPA, Vice President, Cancer Services, Helen F. Graham Cancer Center, Christiana Care Health Services, 4701 Ogletown Stanton Rd., Newark, DE 19713; Tel: (302) 623-4550; Fax: (302) 623-4554; Email: pgrusenmeyer@christianacare.org Research Objective: To determine the cost and cost effectiveness of adding a new targeted monoclonal antibody to standard chemotherapy for advanced non-small cell lung cancer. The Medicare Modernization Act changed the manner in which physicians offices are reimbursed for chemotherapy, while new treatments increase the cost of care. Lung cancer is the deadliest form of cancer, causing an estimated 163,510 deaths in the United States in 2005; more than the next four highest cancers combined. Lung cancer has the second highest incidence of any cancer, an estimated 172,570 cases. Study Design: This study is a cost and cost effectiveness analysis based on clinical findings from a national randomized clinical trial. Costs of the chemotherapy regimens were developed from the Medicare national reimbursement rates (from the payer’s perspective.) Two-drug platinum-containing regimens have been considered the standard of care in advanced non-small cell lung cancer (NSCLC), with several regimens FDA approved and other commonly used regimens felt to have similar outcomes. The Eastern Cooperative Oncology Group, a National Cancer Institute funded research base completed a multi-center, randomized, placebo controlled clinical trial (ECOG 4599) comparing the effectiveness of one standard chemotherapy regimen (carboplatin and paclitaxel) with or without a new targeted agent, the monoclonal antibody (bevacizumab). The trial demonstrated significantly prolonged median overall survival with bevacizumab (12.5 months) compared to chemotherapy alone (10.2 months.) The addition of bevacizumab adds considerable cost to the regimen. Population Studied: The 2005 Medicare reimbursement was calculated using the Centers for Medicare and Medicaid Services (CMS) Drug Payment Table, effective January 1, 2005 with drugs reimbursed at Average Sales Price (ASP) + 6%. Chemotherapy infusion reimbursement was calculated based on the CY 2005 revisions to the Medicare Physician Fee Schedule (PFS). Principal Findings: The carboplatin and paclitaxel regimen costs $14,073. The addition of bevacizumab increases cost by $66,271 to $80,343 and provides a statistically significant increase of 2.3 months in median overall survival than chemotherapy alone, (hazard ratio = 0.77 (0.65, 0.93), p = 0.007.) The addition of bevacizumab to chemotherapy costs $345,762 per year of life gained. Conclusions: The addition of bevacizumab to a chemotherapy regimen containing carboplatin and paclitaxel costs an additional $66,271 to gain an additional 2.3month survival in advanced NSCLC. Bevacizumab costs $345,762 for 1-year additional survival. Medical interventions are generally considered cost-effective at $50,000 - $100,000 per Year of Life Gained (YLG). In order to be cost effective at the $50,000/YLG level bevacizumab reimbursement must be reimbursed at $7.35/10 mg. ($882/cycle) or 13% of 2005 Medicare reimbursement of $57.08/10 mg. ($6,849/cycle). Implications for Policy, Delivery, or Practice: Currently, Medicare reimburses physicians Average Sales Price (ASP) plus 6% for Part B drugs administered in the office. This essentially allows the pharmaceutical company to set the reimbursement rate for Part B drugs, whereas Medicare establishes the rate it pays for other Part B services, particularly for physician services through the sustainable growth rate formula. Medicare should reassess how it pays for chemotherapy drugs under Part B and should set the rate it pays based in part on cost effectiveness of the treatment. Additional analysis should be conducted to determine if reimbursement changes lead to different utilization. Primary Funding Source: No Funding ●Predictors of Physically and Mentally Unhealthy Days for Medicare Managed Care Beneficiaries Mary Anne Hope, MS, Beth Hartman Ellis, Ph.D., David Drachman, Ph.D., Laura Giordano, RN, M.B.A. Presented By: Mary Anne Hope, MS, Senior Analyst, Surveys Research & Analysis, Health Services Advisory Group, 1600 E. Northern Ave., Phoenix, AZ 85020; Tel: 602-745-6312; Fax: 602-241-0757; Email: mhope@azqio.sdps.org Research Objective: The Medicare Health Outcomes Survey (HOS) assesses the physical and mental health outcomes of the Medicare elderly enrolled in managed care in the United States and is the first health outcomes assessment for the Medicare population. This research examines a new measure in the HOS, the number of unhealthy days reported per month, as it relates to profit and non-profit status of managed care health plans. Study Design: The Medicare HOS includes the SF-36, which is a widely used multi-purpose, short-form health survey. In addition to the SF-36, questions regarding Activities of Daily Living (ADLs), chronic conditions, negative symptoms, and demographic information are included in the survey. Population Studied: The respondents in this study were 40,211 beneficiaries in Cohort VIII baseline. The final sample consisted of beneficiaries representing 166 managed care plans. Clustering among health plans was tested with the intraclass correlation coefficient and found to be not significant. Multiple regression was then used to analyze physically and mentally unhealthy days. Predictors included demographics, comorbidity, impaired ADLs, a depression screen, and negative physical health symptoms. Because of the large sample size available, an effect size criterion was used to assess statistical significance. The effect size criterion was set at 0.5% for the contribution of variance for each predictor. Principal Findings: The findings indicate that plan profit status is not related to the number of physically or mentally unhealthy days. However, proxy status, urinary incontinence, impaired ADLs, depression, bodily pain, and shortness of breath all met the effect size criterion for physically and mentally unhealthy days. Comorbidity, race, education, and income did not meet the effect size criterion. Conclusions: Demographic characteristics were not related to unhealthy days for these beneficiaries, nor was plan profit status. However, a number of treatable conditions did contribute to mentally and physically unhealthy days. Implications for Policy, Delivery, or Practice: Beneficiaries with depression and bodily pain should be strongly considered as a focus for quality improvement programs. Primary Funding Source: CMS ●Effect of Benefit Package Design on Demand for Medicare Advantage Plans Chunchih Jim Huang, Ph.D., Mahmud Khan, Ph.D. Presented By: Chunchih Jim Huang, Ph.D., Senior Outcomes Research Associate, Outcomes Measurement Programs, Medstar Research Institute, 6495 New Hampshire Avenue, Suite 305, Hyattsville, MD 20783; Tel: (301)560-2920; Fax: (301)560-2974; Email: jim.huang@medstar.net Research Objective: The research was to examine the effect of various benefit-specific attributes of Medicare plus Choice (M+C, now Medicare Advantage) plans on proportion of Medicare beneficiaries selecting the plans. Specifically, the assumptions tested in the study were as follows: (1) plans with lower out-of-pocket payments and more generous and comprehensive coverage would attract more Medicare beneficiaries and therefore, would experience higher penetration rates at the county level; (2) plans with lower barrier to access to care would encourage Medicare beneficiaries to enroll in the plans; (3) plans provided by the managed care organizations (MCOs) with higher years of Medicare experience are likely to attract more enrollees. Study Design: The prospective study was designed based on the fact that the M+C benefit packages are offered about half year before Medicare beneficiaries choose them. Simultaneous equations with random effect model have been used. The study first considered the relationship between the monthly premium and the benefit components of the M+C plans, and then analyzed the relationship of the M+C benefit options and the enrollment rate for each plan at each service county. The response variable was defined as the percentage of the M+C plan enrollees among total Medicare beneficiaries in a county. In addition to considering different categories of plan benefits and out-of-pocket expenses, factors such as organizational attributes, healthcare market structure, and geographic and demographic characteristics were also included in the model. Population Studied: This study was based on all of the M+C plans and their service counties during the years 2001 and 2002. Data was obtained from various resources. Analyses were carried out at county-plan level. Special programs like PACE (Program for All-Inclusive Care for the Elderly) and HCPP (Health Care Prepayment Plans) were not included. Counties in Alaska and Puerto Rico were not included, either, due to insufficient information. The sample size in the study was 7,250. Principal Findings: The results show that plans with lower out-of-pocket premium, lower co-pay for outpatient visits, and zero co-pay for inpatient stay have higher enrollment rates. Plans with $1 more monthly premium decrease the penetration rate by 0.03%. Plans with $1 more for office visit co-pay decrease the penetration rate by 0.05%. We also find that the coverage of brand-name drugs or the provision of unlimited generic drugs increases the enrollment probability of Medicare beneficiaries in the plan. Dental care coverage provided by MCOs, however, does not affect the plan’s attractiveness. On the other hand, MCOs with one additional year of experience in Medicare business increase their penetration rate by 0.08%. An 1% increase in the MCO enrollees having access to preventive or ambulatory services increase their M+C plans’ penetration rate by 0.05%. Conclusions: (1) The inverse relationship between out-ofpocket cost and the enrollment rate in M+C plans is consistent with our hypothesis. (2) Beneficiaries appear to prefer plans with brand name drug or unlimited generic drug coverage. (3) Coverage of preventive care services plays an important role in the beneficiaries’ decision making on plan choice. (4) MCOs’ experience and their performance are important factors affecting enrollment in their Medicare Advantage plans. Implications for Policy, Delivery, or Practice: This research provides not only the insight into the Medicare beneficiaries’ preferences for benefit package components but also the guidelines for improving the market share of MCOs among Medicare beneficiaries. Primary Funding Source: No Funding ●The Medicare Drug Coverage Gap: Damoclean Swords? Jie Huang, Ph.D., Richard Brand, Ph.D., Rita Hui, Pharm.D, Joe Newhouse, Ph.D., John Hsu, M.D, M.B.A., MSCE Presented By: Jie Huang, Ph.D., Statistical Demographer, Division of Research, Kaiser Permanente, 2000 Broadway, Oakland, CA 94612; Tel: (510) 891-3571; Fax: (510)-891-3606; Email: jyh@dor.kaiser.org Research Objective: Starting in 2006, many Medicare beneficiaries will have drug costs that vary with their overall drug expenditure level. After $2,250, some beneficiaries will have no prescription drug coverage, until they reach an out-ofpocket spending maximum. For some beneficiaries, drug expenditures could remain high in subsequent years, thus affecting the likelihood of entering this coverage gap. To examine these issues, we investigated prescription drug expenditures during consecutive years among Medicare beneficiaries in a prepaid, integrated delivery system. Study Design: We examined the percentage of subjects who had total drug expenditures greater than $2,250 in 2002 and 2003 respectively, and the percentage of subjects who did so in both years. Using logistic regression models, we examined the association between individual characteristics and exceeding the expenditure threshold. Characteristics included age, gender, neighborhood (census block group) median annual household income and mean education level, race/ethnicity, comorbidity (DxCG score), chronic disease status, having a primary care physician (PCP), and medical center. We used 2002-2003 automated and 2000 US Census data. Population Studied: All 41,904 subjects were Medicare+Choice beneficiaries in Kaiser Permanente-Northern California, who had prescription drug coverage without any coverage limits in 2002-2003. The beneficiaries had a mean age of 74.7 years (SD=6.7); 53.5% were female; and 71.9% were white. Principal Findings: In 2002, 10.2% of the study subjects exceeded the $2,250 expenditure threshold; and 11.5% exceeded the $2,250 threshold in 2003. Among subjects with $2,250+ expenditures in 2002, 71.3% also exceeded the threshold in 2003. In multivariate analyses, beneficiaries who were younger (OR=1.80, 95%CI: 1.59-2.03 for 65-74 vs. 85+; OR=1.30, 95% CI: 1.16-1.47 for 75-84 vs. 85+), female (OR=1.53, 95%CI: 1.43-1.64), of white race/ethnicity (OR=1.53, 95%CI: 1.40-1.68), had greater comorbidity (OR=4.70, 95%CI: 4.405.01), or had more chronic diseases (OR=1.22, 95%CI: 1.171.26) were significantly more likely to exceed the $2,250 threshold in 2002. The associations are comparable for exceeding the threshold in 2003 and exceeding in both years. Subjects with high drug expenditures ($2,250+) in 2002 were substantially more likely to have similarly high expenditures in 2003 (OR=32.5, 95%CI: 29.8-35.4). Conclusions: One in ten Medicare+Choice beneficiaries is at risk for entering the coverage gap under the new drug benefits in Medicare Part D. Once beneficiaries have $2,250+ in drug expenditures, they are at risk for entering the coverage gap in subsequent years. As expected, subjects with multiple chronic diseases or a high comorbidity level are at high risk for entering the coverage gap. Implications for Policy, Delivery, or Practice: Medicare Part D drug benefits involve significant amounts of cost-sharing, particularly for beneficiaries with high levels of expenditures. Not surprisingly, beneficiaries with greater clinical need and who have had high drug expenditures in the past are at great risk for having high expenditures in future years, thus would have high out-of-pocket costs under Part D. These costs could affect drug consumption behavior, even below the coverage gap. Additional research is needed to assess the clinical impact of the various drug cost-sharing structures under the new Part D plans. Primary Funding Source: AHRQ ●Does Medicare Related Pharmacy Coverage Affect VA Pharmacy Use? Michael Johnson, Ph.D., Raji Sundaravaradan, BS, Laura Petersen, M.D., M.P.H., Nora Osemene, Pharm D, Iris Wei, DrPH, Robert Morgan, Ph.D. Presented By: Michael Johnson, Ph.D., Assistant Professor, Houston HSR&D, Baylor College of Medicine, 2002 Holcombe Blvd., Houston, TX 77030; Tel: (713) 794-8608; Fax: (713) 748-7359; Email: mjohnson@bcm.tmc.edu Research Objective: Veterans with Medicare managed care plans have access to pharmacy benefits outside the VA, but this coverage affects VA pharmacy use is unclear. We examined patterns of pharmacotherapy among Medicareenrolled veterans with diabetes (DM), ischemic heart disease (IHD) and chronic heart failure (CHF), and how those patterns differ among patients enrolled in fee for service (FFS) or managed care (HMO) plans. Study Design: A retrospective cohort of 702,542 veteran enrollees was identified from ICD-9 diagnosis codes in calendar year 2000 (CY00) and 881,290 in CY02 with DM, IHD, or CHF from the VA National Patient Care Databases. Population Studied: Veteran enrollees were classified as having only one of the three conditions or all 3. Use of 25 categories of selected cardiovascular and anti-diabetic agents in the VA was tabulated from pharmacy prescription records obtained from the Pharmacy Benefits Management Strategic Healthcare Group (PBM) and stratified by Medicare FFS or HMO enrollment status. The proportion of users of conditionrelated medications within disease groups was compared. We hypothesized that patients enrolled in HMO plans would utilize the VA less often than FFS enrollees, and that differences would be due to the presence of a pharmacy benefit within the HMO plan. Principal Findings: Differences between FFS and HMO enrollees were generally minimal (+/- 2%). The most notable exceptions were statin use, which was on average 4.8% less frequent among FFS enrollees in CY00 and 5.7% less frequent in CY02 compared to HMO enrollees across all disease condition groups. For example, in CY02, of the 293,213 FFS enrollees with DM, 45.8% received statins from the VA compared to 50.8% of HMO enrollees with DM (n=36,707). Among the HMO enrollees, 51.4% of enrollees in plans with pharmacy coverage received statins compared to 49.6% without pharmacy coverage. Other notable differences were ACE inhibitor use among patients with CHF or DM (about 4% less frequent among FFS enrollees), and increased use (about 4%) of glucose monitoring supplies among FFS enrollees with DM. Differences between HMO enrollees with drug coverage available compared to those without were uniformly very low (+/- 2%). Conclusions: Differences in VA pharmacy use by patients enrolled in Medicare FFS or HMO plans were minimal in these patients with isolated or multiple chronic conditions. Contrary to our expectation, consistent increased use of statins and ACE inhibitors was found among HMO enrolled veterans. However, differences did not appear to be driven by the presence of drug coverage within HMO plan, suggesting that case-mix differences among enrollees such as income levels, or quality of care differences in FFS or managed care plans may be influencing the choice to use VA pharmacy services. Implications for Policy, Delivery, or Practice: Veterans use of VA pharmacy services is largely unaffected by veterans’ access to Medicare-related pharmacy benefits. Future work will examine the policy implications under the new Medicare Modernization Act. Primary Funding Source: VA ●High Cost, Low Coverage Drugs: Expenditures and Medicare Formulary Coverage For Biotechnology Drugs Su-Ying Liang, Ph.D., Kathryn Phillips, Ph.D., Jennifer Haas, M.D. Presented By: Su-Ying Liang, Ph.D., Associate Researcher, Clinical Pharmacy, University of California San Francisco, 3333 California Steet, Suite 420, San Francisco, CA 94143; Tel: 415514-0457; Fax: 415-502-0792; Email: liangs@pharmacy.ucsf.edu Research Objective: To examine: (1) recent trends in expenditures for the top 10 biotechnology drugs and (2) formulary coverage for these drugs under the Medicare Prescription Drug Plan. Study Design: Cross-sectional analysis of US sales of the top 10 biotechnology drugs from 2002 to 2004. Pharmaceutical sales data were extracted from the Knowledge Express Data Systems (UTEK Corporation). Formulary coverage was obtained from the Centers for Medicare and Medicaid Services. We calculate the percentages of prescription drug plans that include these biotechnology drugs in their formularies. We examine whether formulary coverage varies by drug, state (California vs Massachusetts), and plan type (stand-alone prescription drug plans vs Medicare Advantage drug plans). Population Studied: Medicare prescription drug plans in California (n=471) and Massachusetts (n=293). Principal Findings: Sales of the top 10 drugs have increased 47% from 2002 to 2004, with the biggest increase in sales of Aranesp (495%). A minority of these Medicare prescription drug plans included these drugs in their formularies, ranging from 12% for Rituxan to 35% for Remicade and Enbrel. Coverage for these drugs varies between California and Massachusetts. Stand-alone prescription drug plans are more likely to cover these biotechnology drugs than Medicare Advantage plans (77% vs 19%). Conclusions: Despite increasing sales of the top biotechnology drugs, a minority of Medicare plans in two states cover these products. Coverage varies by drug, state, and plan type. Stand-alone prescription drug plans may have better coverage for biotechnology drugs compared to Medicare Advantage drug plans. Implications for Policy, Delivery, or Practice: Careful evaluation of the value and formulary coverage of biotechnology drugs is needed. Primary Funding Source: NCI ●Effect of Sample Size on Models’ Performances in Predicting Medical Expenditures Hangsheng Liu, MS, Peter Veazie, Ph.D., Andrew Dick, Ph.D., Katia Noyes, Ph.D., M.P.H. Presented By: Hangsheng Liu, MS, Ph.D. Student, Community & Preventive Medicine, University of Rochester, 601 Elmwood Ave, Box 644, Rochester, NY 14642; Tel: 585275-7380; Email: Hangsheng_Liu@urmc.rochester.edu Research Objective: To investigate the association between sample size and various regression models’ performances in predicting the medical expenditures. Study Design: 1992-2000 Medicare Current Beneficiary Survey (MCBS) data is used in this study. Considering the MCBS data as the population, we draw 500 random samples from that population, run various models on the annual total medical expenditures, and compare the models’ performances as the sample size changes from 200 to 20,000. Each sample is randomly split into two parts for the purpose of crossvalidation. The models include simple ordinary linear squares (OLS), log transformed OLS with homoscedastic retransformation (LogOLS-hom) or heteroscedastic retransformation (LogOLS-het), square-root transformed OLS with heteroscedastic retransformation (SqrOLS-het), generalized linear models (GLM) with log link and Poisson family (Poisson GLM) or Gamma family (Gamma GLM). We exclude the institutionalized beneficiaries, those with End Stage Renal Diseases, and those under 65 years old, because their medical expenditure distributions are quite different from those of others. In order to simplify the analysis, we also exclude those beneficiaries without any medical expenditure in a calendar year and do not use sampling weights. The model performance measurements include: predicted mean bias (PMB, the difference between the population mean and the mean of predicted values), absolute prediction error (APE) and root mean square error (RMSE) for both in-sample and out-of-sample predictions. Population Studied: Non-institutionalized Elderly over 65 years old in the US Principal Findings: (1) In-sample prediction. As the sample size increases, PMB is decreasing for all models, where SqrOLS-het, OLS and Poisson GLM perform best and have a faster decreasing rate, followed by Gamma GLM, LogOLS-het and LogOLS-hom. APE does not show a strong pattern of decrease as the sample size increases, and Poisson GLM has the smallest APE, followed by SqrOLS-het, Gamma GLM, LogOLS-het, OLS, LogOLS-hom. The same pattern is observed for RMSE. (2) Out-of-sample prediction. PMB shows the same pattern as that of in-sample prediction, where SqrOLS-het, OLS, and Poisson GLM perform much better than other models and have a faster declining rate as the sample size increases. When sample size is small, simple OLS generates the smallest PMB, which is about half of SqrOLShet and one ninth of Poisson GLM. In contrast to the insample predictions, APE and RMSE do demonstrate the declining trend as the sample size increases. In terms of these two measures, SqrOLS-het, OLS, and Poisson GLM perform better than other models. Conclusions: The LogOLS models do not out-perform other models in predicting the medical expenditures for the noninstitutionalized elderly, while simple OLS shows its advantages. The best models include SqrOLS-het, Poisson GLM and simple OLS. This pattern persists even in small sample sizes. Implications for Policy, Delivery, or Practice: Since simple OLS can out-perform standard alternatives, such as logtransformed models, it is necessary to compare them before one can find a suitable model for a specific dataset. Primary Funding Source: NIA ●Familiarity with Medicare and Self-Reported Access to Care Robert Morgan, Ph.D., Cayla Teal, Ph.D., Jennifer Hasche, MS, Laura Petersen, M.D., M.P.H., Margaret Byrne, Ph.D., Debora Paterniti, Ph.D., Beth Virnig, Ph.D., M.P.H. Presented By: Robert Morgan, Ph.D., Associate Professor and Research Scientist, Medicine (Health Services Research), Baylor College of Medicine and Houston VA Medical Center, 2002 Holcombe Blvd (152), Houston, TX 77030; Tel: (713) 7948635; Fax: (713) 748-7359; Email: rmorgan@bcm.tmc.edu Research Objective: There are substantial concerns regarding Medicare beneficiaries’ abilities to navigate an increasingly complex array of program choices under Medicare, and ultimately, whether the expansion of the Medicare managed care program and the implementation of the Medicare Prescription Drug Benefit, or Part D, will actually improve beneficiaries’ ability to access care. While Medicare beneficiary knowledge has increasingly been studied, there have been few studies of the relationship between beneficiary knowledge and access to health care. In this paper, we examine the relationship between a global self-reported measure of Medicare program familiarity, reported health care use, and perceived access to care. Study Design: Mailed survey. Our overall response rate was 53%, or 2,997 of 5,697 mailed surveys. Population Studied: White, black, and Hispanic Medicare beneficiaries living in nine metropolitan and non-metropolitan areas distributed across the U.S. We stratified our random sample by race-ethnicity, sex, Medicare+Choice or M+C plan enrollment, and geographic area. Principal Findings: All analyses were weighted to reflect our survey sampling proportions. In our bivariate analyses, respondents who reported being unfamiliar or very unfamiliar with Medicare were less likely to report having routine physician visits, p < .05, but reported more ER visits, p < .02. They were also more likely to report cost-related delays in physician visits, p < .02, ER visits, p < .0005, inpatient admissions, p < .02, and prescription fills, p < .02. They reported poorer perceived access to medical care in general, p < .0004, to life-saving/urgent care, p < .0002, and to nonurgent/routine care, p < .0001, and perceived having less access to affordable medical care, p < .0008. In multivariate analyses controlling for race-ethnicity, sex, income, M+C enrollment, possession of supplemental insurance, and geographic area, lower familiarity remained associated with increased numbers of ER visits, increased risk of delays in physician visits, ER care, inpatient admissions, and prescription fills, and to perceptions of poorer access to lifesaving/urgent care, to non-urgent/routine care, and to affordable medical care. Conclusions: Beneficiaries who report being unfamiliar with the Medicare system are more likely to delay care due to cost, and are more likely to use services, e.g., ER visits, which have been previously associated with poor health care access. Further, these beneficiaries consistently perceive themselves as having poorer access to care. Implications for Policy, Delivery, or Practice: Researchers and policy makers are increasingly aware of the integral role that beneficiary knowledge plays in health system enrollment decisions. Medicare Part D is expected to significantly increase the complexity of health care decisions under Medicare. Much of the commentary to date has focused on the difficulties that beneficiaries may have in choosing among the available prescription drug plans. Our findings suggest that regardless of their choice, beneficiaries’ familiarity with the Medicare program will significantly mediate their ability to access care. Primary Funding Source: NIA ●Did the MMA Payment Rate Increases to Medicare Managed Care Plans Increase Enrollment in 2004? Lauren Nicholas, M.P.P. Presented By: Lauren Nicholas, M.P.P., Doctoral Student, School of Social Work, Columbia University, 509 W. 112th St. #3W, New York, NY 10025; Tel: 202-246-8020; Email: ln2123@columbia.edu Research Objective: To determine whether increasing payment rates to Medicare Advantage plans in 2004 caused an increase in Medicare managed care enrollment. Study Design: Multivariate regression models with state and county fixed effects were used to test for a causal effect of increases in Medicare Advantage payment rates on county enrollment rates. This methodology exploits a natural experiment that was created by an unexpected increase in annual payment rates to plans that occurred in early 2004 and was instituted retroactively for the year. Population Studied: County-level Medicare Advantage data for 901 counties with at least 2% of the population enrolled in Medicare managed care at the end of 2003. Principal Findings: Enrollment in Medicare managed care increased following the Medicare Modernization Act increased payment rates to all plans nationwide. However, there is no statistical evidence that these rate increases caused the increases in enrollment. These findings are robust across model specifications which consider absolute dollar value of payment rate increases and percentage increase in payment rate. Enrollment changes were considered for the 3 month and 6 month periods following the payment rate increases. Conclusions: These findings suggest that increasing payment rates to Medicare Advantage plans may not be an effective way to increase enrollment in the program at the county level. Implications for Policy, Delivery, or Practice: Managed care companies were required to use the increased payments to improve benefits or reduce cost sharing for Medicare managed care enrollees. Economic theory would suggest that enrollment should increase in response to the improved offering, although we do not observe this in practice. This suggests that other incentives may be more appropriate and more cost effective if we want to increase managed care enrollment. Primary Funding Source: CWF ●Estimating Out-of-Pocket Pharmaceutical Expenditrures Under the New Medicare Drug Law for Patients with Mental Disorders – An Analysis of Claims Data from Retiree Medical Plans Ronald J. Ozminkowski, Ph.D., Teresa B. Gibson, Ph.D., Tami L. Mark, Ph.D., M.B.A., Laurie A. Costa, M.P.H. Presented By: Ronald J. Ozminkowski, Ph.D., Director, Health & Productivity Research, HPR, Thomson Medstat, 777 East Eisenhower Parkway, 903R, Ann Arbor, MI 48108; Tel: (734) 913-3255; Fax: (734) 913-3850; Email: ron.ozminkowski@thomson.com Research Objective: Estimate out-of-pocket pharmacy expenditures for elderly retirees with employer-sponsored coverage, under the standard Medicare drug plan that began on January 1, 2006. Study Design: Data came from the Medstat MarketScan Medicare Database, for 2004. Payments covered by the Medicare program and by private insurance were included. Analyses focused on 30,681 patients with diagnoses of, and treatment for, depression, anxiety disorder, or schizophrenia in 2004. We estimated the out-of-pocket payments that would have been incurred by these beneficiaries, if the Medicare drug law had been in effect during 2004. We also estimated how long it would have taken to enter the “donut hole” where coverage stops (i.e., that period during which total pharmacy expenditures range from $2,251 to $5,100). Population Studied: Data included all medical and pharmacy claims for 1,220,902 seniors enrolled in retiree plans offered by large employers. (About 30% of non-institutionalized Medicare enrollees have retiree coverage.) Principal Findings: As expected for people with these conditions, more than half were female (i.e., 56% of those with schizophrenia were female, as were 64% of the depression patients, and 68% of the anxiety patients). 83% to 88% lived in urban areas. Several thousand sample members came from each U.S. census region. Average ages ranged from 66 years (for schizophrenia patients) to 73 years (for anxiety and depression patients). Mean total pharmacy payments for these mental health patients were higher than payments for other patients with retiree coverage ($4,491 for depression patients, $3,364 for anxiety patients, and $5,321 for schizophrenia patients, versus $2,587 for the average Medicare beneficiary). The average mental health patient filled over 50 prescriptions for all of their physical and mental disorders. About 61% of the anxiety patients, 69% of depression patients, and 73% of the schizophrenia patients would have fallen into the donut hole if the law had been in effect in 2004, assuming no changes in utilization behavior would be induced by the law’s deductible and copayment features. Schizophrenia patients would have entered after about 4 months (in April). Depression patients would fall into the donut hole after about 6 months, while anxiety patients would enter after about 9 months, on average. Mean out-ofpocket pharmacy payments under the new law would have ranged from $1,858 for anxiety patients to $2,407 for schizophrenia patients. Conclusions: The results to date suggest that most patients with these psychiatric disorders will face high out of pocket costs, and many will be faced with a donut hole of no coverage. Adjusting for any changes in demand due to different copayment features under the new law may modify the results somewhat. Implications for Policy, Delivery, or Practice: As beneficiaries move from one form of insurance into the new Medicare plans, differences in out-of-pocket requirements may have implications for adherence to pharmacotherapy and health status, especially as people enter the donut hole. These should be considered by doctors and policy makers, as patients are treated and research agendas are set. Primary Funding Source: NIMH ●Trends in Characteristics of Medicare Enrollees in Risk HMOs Gerald Riley, M.S.P.H. Presented By: Gerald Riley, M.S.P.H., Social Science Research Analyst, Office of Research, Development, and Information, Centers for Medicare and Medicaid Services, 7500 Security Blvd., Room c3-20-17, Baltimore, MD 21244; Tel: (410) 7866699; Fax: (410) 786-5534; Email: gerald.riley@cms.hhs.gov Research Objective: To examine trends in demographic, health, and functional status measures in the Medicare risk HMO and fee-for-service (FFS) sectors between 1991 and 2004. Study Design: The study used data from the Medicare Current Beneficiary Survey (MCBS) Access to Care Files. The MCBS is a multipurpose survey of a nationally representative sample of the Medicare population, including institutionalized individuals. It has been conducted every year since 1991. Respondents are asked about health status, difficulties with six Activities of Daily Living (ADLs) and six Instrumental Activities of Daily Living (IADLs), and whether a doctor has ever told them they have twenty specific conditions. Information on demographics, income, and supplemental health insurance is also collected. Medicare administrative records are routinely linked to survey information. For each study year, administrative records were used to identify which respondents were enrolled in risk-based HMOs and in FFS at the time of the Fall round interview. Descriptive statistics on individual demographic, income, health, and functional status measures were computed separately for the HMO and FFS samples for each year and were adjusted to the age and sex distribution of the total Medicare population in 1997 (the midpoint of the study period). In addition, a summary proxy health status measure was developed for each respondent, consisting of predicted Medicare costs as a function of individual health and functional status measures, age, and sex. Population Studied: Medicare beneficiaries in risk-based HMOs and FFS. Some measures were available for the noninstitutionalized population only. Principal Findings: Preliminary findings show that enrollees in risk-based HMOs grew from 3.5% of the Medicare population in 1991 to 18.3% in 2000, and decreased thereafter to 13.5% in 2004. During that time, the demographic mix of the HMO population changed (proportionately more women aged 85 and over, fewer women under age 65, more residents in the northeast region). Compared to FFS, fewer HMO enrollees were institutionalized or eligible for Medicaid each year, but the differences between the HMO and FFS sectors lessened over the study period. HMO enrollees generally reported fewer numbers of chronic conditions, fewer problems with ADLs and IADLs, and fewer instances of fair or poor health, but the differences attenuated somewhat over time. The ratio of average predicted costs among HMO enrollees to average predicted costs among enrollees in FFS increased, but remained below 1.0. Conclusions: Health status differences between the Medicare HMO and FFS populations have gradually lessened over time, but HMO enrollees remain somewhat healthier. Implications for Policy, Delivery, or Practice: Increasing HMO enrollment of older and sicker beneficiaries suggests that more vulnerable subpopulations are taking advantage of enhanced choices available under Medicare. However, some prior studies have identified concerns with access to and quality of care for chronically ill populations in HMOs. If the HMO program, recently reformed as Medicare Advantage, continues to enroll more chronically ill beneficiaries, its impact will increase in the areas of care efficiency, access, and quality. Primary Funding Source: CMS ●Medicare Part D Can Reduce Out-of-Pocket Expenditures for Some Low-Income Veterans Randall Rupper, M.D., M.P.H., Byron D. Bair, M.D. Presented By: Randall Rupper, M.D., M.P.H., Physician Investigator, GRECC, VA Salt Lake Medical Center, Bldg. 2 (182), 500 Foothill Drive, Salt Lake City, UT 84148; Tel: (801)582-1565; Fax: (801)584-5640; Email: randall.rupper@hsc.utah.edu Research Objective: Because CMS considers the VA drug benefit to be creditable coverage, many advocacy groups suggest that veterans need not enroll in Part D. The VA and Medicare Part D offer individuals living near poverty waived or reduced co-payments, but the qualifying thresholds and the benefit structures differ. We sought to determine how these different thresholds and benefits would affect pharmacy outof-pocket expenses for low-income veterans. Study Design: Using VA income and asset means testing and pharmacy data from 2005, we estimated the projected cost differences in out-of-pocket expenditures during 2006 for elderly veterans eligible for assistance through each program. Population Studied: We studied users of the VA Salt Lake Medical Center who were age 65 or older, and who had completed means and asset testing during fiscal year 2005. Principal Findings: The threshold for receiving assistance through one or both programs on the basis of lowincome/assets is met by 4,127 veterans. Of these, less than half, 1,810 (49%), would be eligible for complete waiver of copayments through the VA. Of the remaining veterans, 1,367 would be eligible for full premium waivers under Medicare Part D, with lower scheduled co-payments than they would have at the VA. The remaining 950 veterans would pay a sliding scale premium under Medicare Part D, but have lower scheduled co-payments than they would have at the VA. Of the 1,367 veterans who would have clearly lower out-of-pocket expenditures under Medicare Part D, 799 used the VA to fill an average of 70 prescriptions each in FY2005. Depending on refill intervals and use of generic medications, this translates into an average difference of $210 to $525 dollars per year in out-of-pocket expenses. Conclusions: Many veterans living near poverty could lower their drug co-payments by enrolling in Medicare Part D. Under current policies, optimizing individual coverage requires individual analysis that can be supported using available electronic data. Implications for Policy, Delivery, or Practice: The different thresholds and pharmacy benefit structures of the VA and Medicare Part D create a complex decision for many lowincome veterans seeking to reduce out-of-pocket expenditures. Groups or individuals who provide advice regarding the Medicare Part D benefit should not assume that veterans will fare best using their VA pharmacy benefit alone. Low income veterans may benefit from decision support that is adapted to their individual utilization and priorities. Alternatively, policy makers should consider reducing this complexity by making the assistance thresholds interoperable. The future effects on VA enrollment and pharmacy costs, patient medication adherence, and the overall cost to the Federal government are important topics for additional research. Primary Funding Source: VA ●Impact of Medicare Part D on Low-Income Populations: A Case Study Randy Scheid, M.P.A., Thomas Cleare, M.B.A., Jennifer Houlihan, Michael Greene Presented By: Randy Scheid, M.P.A., Planner, Research and Planning, Health Care District of Palm Beach County, 324 Datura Street Suite 401, west palm beach, FL 33401; Tel: 561659-1270; Fax: 561-659-4620; Email: rscheid@hcdpbc.org Research Objective: Establish a monthly and annual cost for an individual under Medicare Part D's low-income subsidy program. Study Design: Case study involving historical utilization patterns of low-income Medicare enrollees. Population Studied: Low-income Medicare enrollees in the provider's prescription drug program. Principal Findings: Using historical utilization rates derived from the sample, the study establishes the monthly and annual projected financial costs a low-income individual might incur under Medicare Part D's low-income benefit. In addition, the study identifies a percentage of the low-income population that will incur significant costs in the doughnut hole. Conclusions: The monthly and annual cost to benefiaries of full prescription benefits under Medicare Part D is established. The savings compared to out-of-pocket costs are substantial under Part D, but the cost incurred may still represent a significant cost for a low-income individual to bear. Implications for Policy, Delivery, or Practice: This study has implications for presecription drug providers under Medicare Part D, state drug programs which also provide prescription drugs to a low-income population, and low-income individuals. Primary Funding Source: No Funding Source ●Modeling the Impact in Drug Cost and Utilization of the Implementation of Mail Order Pharmacy in a Dual-Eligible Population Enrique Seoane-Vazquez, Ph.D., Satish Valluri, MS, M.P.H., Rosa Rodriguez-Monguio, Ph.D., Sheryl L. Szeinbach, Ph.D., R.Ph. Presented By: Enrique Seoane-Vazquez, Ph.D., Assistant Professor, College of Pharmacy and School of Public Health, Ohio State University, 500W 12th Avenue, Columbus, OH 43210; Tel: (614) 292-3907; Fax: (614) 292-1335; Email: pharmacoeconomics@osu.edu Research Objective: Although cost containment is the main reason behind the recent development of mail order pharmacy, there is no empirical evidence that substitution of community pharmacy with mail order services reduces overall drug expenditures. The study of the impact of mail order pharmacy is especially important today as the new Medicare prescription drug benefit could increase the use of this dispensing channel. The objective of this study is to compare drug utilization and drug product cost for a dual-eligible patient population (Medicare/Medicaid patients) using community pharmacy services with the results of a model simulating the effect of mail order pharmacy services for the same population. Study Design: This study is a retrospective cohort study comparing utilization and cost patterns in community vs. mail order pharmacy. A simulation model was employed to assess drug utilization and cost in mail order pharmacy using community pharmacy claim data. Drug claims were aggregated to obtain a set of that courses of drug therapy (CDTs) representing unique patient IDs and unique drug products. The model assumed that CDT in mail order pharmacy would have utilization patterns similar to those found in community pharmacy. Drug product cost estimates excluded dispensing fees. A 95% confidence interval surrounding changes in average utilization and average cost were estimated using bootstrap analysis. A sensitivity analysis was performed by varying fill number where mail order was initiated and compliance. Population Studied: Dual-eligible patients included in Ohio Health Plans (program managing Ohio Medicaid) pharmacy claim data for the period January 2000-September 2004. Principal Findings: The use of mail order by dual-eligible patients was estimated to result in an average day supply increase of 35.4 days, representing a 4.7% increase in utilization and a 4.7% increase in drug product cost. The use of mail order by dual-eligible patients consuming drugs for chronic use was estimated to result in an average day supply increase of 34.8 days, representing a 4.3% increase in utilization and a 4.4% increase in product cost. Conclusions: The results of the simulation model indicate that mail order pharmacy increases drug utilization and can also increase drug product cost if the cost per unit is not reduced accordingly. Implications for Policy, Delivery, or Practice: Decision makers should carefully evaluate the utilization and cost effects that result from the implementation of mail order pharmacy programs for dual-eligible population. In addition, any program that increases day supply should carefully target specific populations and therapies whose potential savings and/or health outcomes overcome the potential extra costs that result from increased utilization and wastage. Primary Funding Source: Other Government ●The Rate of Discharge to a Skilled Nursing Facility (SNF) Differs Among Medicare Beneficiaries who Undergo Elective Open Versus Endovascular Repair of Abdominal Aortic Aneurysms (AAAs) Kara Suter, MS, Marla R. Kugel, M.P.H. Presented By: Kara Suter, MS, Partner, The Moran Company, 1655 N. Fort Myer Drive, Suite 1250, Arlington, VA 22209; Tel: (703) 465-9970; Fax: (703) 465-9969; Email: klsuter@themorancompany.com Research Objective: The aim of this study is to determine whether the rate of discharge to a SNF differs among Medicare beneficiaries who undergo open versus endovascular repair of AAAs. Study Design: This study uses data from the LDS versions of the 2001-2004 inpatient, SNF, and denominator 5% Medicare Standard Analytic files (SAFs). Selected beneficiaries who received either open or endovascular repair were compared. The odds of SNF discharge following open versus endovascular repair of AAA were calculated using multivariate logistic regression. The model controlled for the following independent factors: age, race, gender, co-morbidity, hospital volume, and previous SNF admissions. Population Studied: The population consisted of Medicare beneficiaries who underwent and survived elective AAA repair without rupture in 2002-2004. Open versus endovascular repair was determined by the presence of a principal ICD-9 diagnosis of 441.4 and an ICD-9 procedure code of 38.44 or 39.71. Elective surgery was determined by a physician referred admission code on the inpatient surgery claim. Beneficiaries were at least 65 years old and had continuous Part A coverage for at least four quarters prior to surgery. Co-morbidity was determined using the Elixhauser method. High volume was defined as a hospital that performed 10 or more AAA repairs on Medicare beneficiaries per year. Principal Findings: Beneficiaries who received either open (n=1,466) or endovascular (n=1,392) repairs were similar in terms of demographics. The endovascular repair group had a significantly higher proportion of the following comorbidities: congestive heart failure (3% open; 4 % endo), hypertension (19%; 23%), neurological disorders (0.6%; 1%), diabetes (3%; 6%), and fluid and electrolyte disorders (4%; 6%). A larger proportion of endovascular repair was performed in high volume hospitals (71% open; 88% endo). Significantly more beneficiaries who received open repair were discharged to a SNF compared to those who received endovascular repair (open: 10%, n=143; endo: 3%, n=47; p<0.001). When controlling for other factors, a beneficiary who underwent open repair was 3.52 times more likely to be discharged to a SNF than a beneficiary who underwent endovascular repair (OR=3.52.; 95% CI 2.413-5.135; p<0.0001). Other factors that predicted discharge to a SNF include: prior SNF admission (OR=20.20; p<0.0001), age (OR=1.032; p=0.03), history of solid tumor (OR=6.12; p<0.01), obesity (OR=4.57; p=0.01), and depression (OR=3.31; p=0.02). Conclusions: An open repair of AAA significantly increases the likelihood of a discharge to a SNF compared to an endovascular repair. This study used administrative data and was unable to control for the size of the AAA. Implications for Policy, Delivery, or Practice: Repair of AAA is important to Medicare because of the prevalence, mortality, and cost associated with the condition. There is some debate over whether the advantages of endovascular over open repair of AAA justify the difference in hospital related costs for the procedure. The cost and decreased quality of life that may be associated with a SNF stay should be considered in evaluating the benefits of endovascular repair of AAA and ensuring access to this treatment for Medicare beneficiaries. Primary Funding Source: WL Gore & Associates ●Evaluating the Generosity of Medicare Prescription Drug Plans among Beneficiaries with Hypercholesterolemia Janice Tzeng, BSPH, Shelby Reed, Ph.D, Lesley Curtis, Ph.D, Joelle Friedman, M.P.A. Presented By: Janice Tzeng, BSPH, Data Technician, Center for Clinical and Genetic Economics, Duke Clinical Research Institute, P.O. Box 17969, Durham, NC 27715; Tel: (919) 6685957; Fax: (919) 668-7124; Email: janice.tzeng@duke.edu Research Objective: This study explores the generosity of the Medicare prescription drug plans offered in North Carolina, and examines their impact on the estimated out-of-pocket (OOP) spending among enrollees who require the use of a statin, a widely-recommended class of cholesterol-lowering drugs. Study Design: We evaluated ‘generosity’ by examining the formulary inclusion of statins for each fee-for-service prescription drug plan, measuring the mean monthly OOP costs across all nine statins for each plan, and measuring the number of individual statins available within a plan at various levels of monthly OOP costs. The final measure allowed us to examine the extent to which enrollees within a plan could switch from one statin to another while maintaining a given level of monthly OOP costs. Population Studied: We limited the hypothetical population to Medicare fee-for-service enrollees in the state of North Carolina, who require the chronic use of a statin and no other prescription drug. Principal Findings: All 38 prescription drug plans in North Carolina, representing 13 organizations, offered at least one of the following statins: lovastatin, rosuvastatin, fluvastatin, atorvastatin, pravastatin, and simvastatin. The three most covered drugs were brand atorvastatin (94.7% of plans), generic lovastatin (92.1%), and brand simvastatin (86.8%). Every plan included at least two of the nine statins on their formularies. Ten plans (26.3%) included all nine statins on their formularies, seven plans (18.4%) included five to eight statins, and 16 plans (42.1%) included four or fewer statins. The mean monthly OOP cost per plan (across all nine statins) ranged from $51.46 to $120.35, and correlated negatively (0.53) with the number of statins included on each plan’s formulary. In addition, the five plans with the lowest monthly OOP costs each included at least seven statins on their formularies. The five plans with the highest monthly OOP costs included six or fewer statins on their formularies. Seven plans offered only one statin for less than $40 in monthly OOP costs; none of the plans offered more than one statin at this cost level. Twenty-nine (76.3%) plans offered one to three statins for less than $80 in monthly OOP costs, six (15.8%) plans offered four to six statins, and three (7.9%) offered seven to nine statins. Conclusions: Although hypercholesterolemia is a highly prevalent condition among seniors and aging baby boomers, the average OOP costs for individual statins can vary considerably across plans and within plans. Many patients who select plans on the basis of their OOP costs for an individual statin may experience a significant increase in OOP expenditures if they decide to switch to an alternative statin. Implications for Policy, Delivery, or Practice: This study demonstrates a significant need for improvement in policies affecting the design and generosity of Medicare prescription drug plans. It presents policy-makers with the likely impact of plan generosity on the OOP expenditures for Medicare beneficiaries, particularly for those who are locked in to their plan after enrollment and require midyear changes in their prescription drug regimens. Primary Funding Source: No Funding ●Medicare Advantage: Costs, Coverage and Economic Theory Greg Watson, M.S. Presented By: Greg Watson, M.S., Principle, The Moran Company, 1655 N Fort Myer Dr, #1250, Arlington, VA 22209; Tel: (703)841-8404; Email: gjwatson@themorancompany.com Research Objective: This analysis first projects how MMA payment polices for MA plans may effect private plan decisions regarding plan participation and marketing strategies. The analysis then projects how decisions by MA plans may effect: (1) Medicare costs; and (2) decisions by Medicare beneficiaries to enroll in MA plans. Study Design: MMA policies provide that MA plans will be paid different amounts: (1) in geographic areas across the nation at the county level; and (2) in each of the next five years. MA payment rules give plans the potential to maximize revenues based on the differential between local plan and regional PPO payment rates. This study is based on a county level MA plan enrollment model incorporating CMS data on MA payment rates, risk characteristics, and fee-for-service (FFS) costs. The study includes simulations that assume plans can offer both local and regional plans with generally only modest differences in health care costs (consistent with “CBO’s Analysis of Regional Preferred Provider Organizations under the Medicare Modernization Act”, October, 2004). The analysis assumes that enrollment is disproportionately more likely in those counties where the model projects that MA plans payment rates are more than risk adjusted FFS costs. Applying rational decision rules, the study analyzes the impact including costs to Medicare of alternative budget neutrality assumptions, enrollment scenarios, and selection incentives into local versus regional PPO plans. The study also explores the Medicare costs of alternative enrollment paths, with the projections of the Medicare Office of the Actuary providing the upper bound and of CBO the lower bound. Population Studied: The study is based on a MA plan enrollment model that incorporates CMS data on MA payment rates, risk characteristics, and fee-for-service (FFS) costs. Principal Findings: Study findings are consistent with economic theory that suggests that MA private plans will (a) not enter markets where costs exceed revenues, and (b) maximize net revenues when confronted with a choice between local MA and regional rates by steering enrollment to the most profitable area and plan type which will generally be the one with the higher reimbursement. Conclusions: The study projects that Medicare costs will increase significantly if MA plans opt to maximize revenue by steering enrollment into geographic areas and plans with higher payments. MA plan coverage may also be less available, or less actively marketed, to beneficaries in some areas of the nation than has been projected. Implications for Policy, Delivery, or Practice: The study suggests that the great variation in Medicare payments to MA plans across the nation may lead private plans to target geographic areas where they participate and seek to enroll beneficiaries. This behavior, consistent with economic theory, may result in a MA program with higher overall costs to Medicare and with enrollment focused in specific geographic areas. Primary Funding Source: Private funding ●Assessing the Feasibility HEDIS® quality of care reporting by Medicare PPOs Amanda Zides, M.H.S., Lok Wong, M.H.S., Joachim Roski, Ph.D., Karen Onstad, M.P.H./M.P.P., Russell Mardon, Ph.D. Presented By: Amanda Zides, M.H.S., Senior Health Care Analyst, Quality Measurement, NCQA, 2000 L Street, NW, Suite 500, Washington, DC 20036; Tel: 2029551716; Fax: 2029553599; Email: zides@ncqa.org Research Objective: This study assesses the feasibility of HEDIS® quality of care reporting by Preferred Provider Organizations (PPO). Results of this study will assist the Centers for Medicaid and Medicare Services (CMS) in determining new quality improvement program requirements for PPO health plans in the Medicare Advantage program starting in 2006. Study Design: A comprehensive assessment of PPOs’ ability to report HEDIS quality of care measures was conducted. We identified a cross-section of commercial and Medicare PPO health plans to provide feedback in semi-structured interviews and submit health plan data for analysis on select HEDIS measures using administrative claims data only and two HEDIS patient experience surveys. Population Studied: A total of 30 plans - 19 Medicare and 11 commercial submitted HEDIS data in 2005; 13 participated in qualitative interviews. PPO plans’ enrollment varied from 30,000 to 4 million in commercial, and from 1000 to 400,000 in Medicare. Principal Findings: Most PPOs interviewed indicated no technical barriers to reporting HEDIS measures using existing administrative data. Some PPOs indicated issues with HEDIS reporting due to small eligible population, coding, and different database systems for PPOs. The number of measures reported varied by plan; the highest number reported was 18 out of 22 Medicare-relevant measures and 17 out of 20 commercial-relevant measures. A third of all plans submitted at least one patient experience survey. Most Medicare and commercial plans were able to report a majority of selected measures with complete data, especially the 7 clinical care measures addressing diabetic care, mental health and breast cancer screening). Less plans reported cardiovascular care measures. Plans indicated less interested in reporting utilization or access measures. The range of performance among PPOs in the study was wide and appeared generally to be lower in commercial PPOs compared to average national HMO performance, but was more similar for Medicare PPOs and HMOs. However rates reported are not reliable as PPO data was not audited. Statistical differences between HMOs and PPOs, and among PPOs could not be determined due to the small number of plans. Therefore limited conclusions can be drawn on the relative performance of PPOs in the study compared to HMOs. Conclusions: Quantitative and qualitative findings demonstrated the feasibility of PPOs’ to report multiple HEDIS measures using a variety of administrative data, therefore we conclude PPOs can feasibly report all 41 administrative HEDIS measures. Although PPOs did not anticipate technical barriers to reporting quality measures using clinical data from medical records, further investigation is needed to assess feasibility of reporting measures not included in this study that rely on abstracting enrollees’ medical records from their primary care provider. Implications for Policy, Delivery, or Practice: PPOs' participation in the Medicare Advantage program has increased interest for previously unavailable information on quality of care in PPOs to inform Medicare beneficiaries’ choice of HMO and PPO plans. Study results will have a significant impact on federal policy in defining initial quality reporting requirements for Medicare Advantage PPO plans. Primary Funding Source: CMS