Estimating the effects of prescription drug coverage for Medicare beneficiaries

Health Services Research, June, 2007 by Dennis G. Shea, Joseph V. Terza, Bruce C. Stuart, Becky Briesacher

The recent experience enrolling Medicare beneficiaries in Part D prescription drug plans raises important questions about the stability and affordability of the new coverage. The Part D benefit relies on Medicare beneficiaries voluntarily choosing from a large selection of private stand-alone prescription drug plans (PDPs) and Medicare Advantage (MA) plans. Previous work (Smart, Shea, and Briesacher 2001) showed considerable churning in prescription drug coverage among Medicare beneficiaries before Part D. If that experience continues under the new benefit, it will raise potentially serious questions regarding adverse selection and spiraling premiums. A related question is how much demand will be stimulated by new Part D enrollments. The price sensitivity of beneficiaries' demand for drugs--referred to as price elasticity or moral hazard--is central to the debate over Medicare Part D costs. Work by Shea, Smart, and Briesacher (2003/2004) demonstrated the effect that price elasticity assumptions have on projected prescription drug benefit costs. This paper estimates the interrelated impacts of selection and price effects on prescription drugs utilization using recent data from the Medicare Current Beneficiary Survey (MCBS) and econometric methodology.

FRAMEWORK AND BACKGROUND

Recent discussions of the Medicare drug benefit have not often been clear about how theoretical and empirical research relates to benefit design. In this section, we review how empirical estimates can lead to differences in preferred benefit design. In short, we outline the implications of empirical results for prescription drug policy.

According to economic theory, beneficiaries will select prescription drug coverage based on estimated costs and benefits of insuring against risk. Individuals who judge risk to be sufficiently high will choose to purchase insurance, all other things equal. If insurers have similar information, they price insurance based on expected risk. Empirically, if prescription drug use is related to observable characteristics, we can expect that private insurers will price coverage based on risk. As risk is often highly correlated to poor health and low income, pricing based on observable risks can lead to an affordability problem. Public policy addresses this by subsidizing coverage and either requiring community rating, risk adjusting plan premiums or reinsuring high cost cases. Community rating can lead insurers to avoid risky individuals or skimp on services provided. Reinsurance or risk-adjusted payments are intended to limit this behavior. However, adverse selection can occur in the presence of these mechanisms if individuals have information about their true risk not available to insurers. Adverse selection can make it impossible for a private insurance market to operate or require administrative mechanisms that limit switching in and out of coverage (Feldman and Dowd 1991).

Price sensitivity also raises questions for benefit design. The greater is the impact of insurance on drug use, the greater is the need for insurers to control prices or quantities. When demand elasticity is high, insurers use deductibles, coinsurance, copayments, coverage caps, formularies, and other utilization control features to manage costs. While each seeks the same end, means to that end differ. Deductibles create an "all or nothing" effect focused on whether an individual decides to use any care. Coinsurance and copayments may impact the amount of services by directly altering the covered price. Formularies shift use of drug products to those on the plan's list. Coverage caps force individuals to face the full cost of their drug spending after reaching the cap threshold.

The design of the standard Medicare benefit suggests some beliefs of policy makers about the nature of selection and drug price sensitivity among the elderly and disabled. The standard benefit effective in 2006 includes a $250 deductible with 25 percent coinsurance for spending between $250 and $2,250. Between $2,250 and $5,100 in total drug costs, coinsurance is 100 percent, creating the so-called "doughnut hole." Above $5,100 in total drug costs, the beneficiary pays just 5 percent of drug costs. The Part D premiums for the standard benefit are subsidized at approximately 75 percent for all individuals. Individuals who are dual eligible for Medicare and Medicaid and those with low incomes and assets are eligible for additional subsidies that reduce the deductible and coinsurance, remove the doughnut hole, and reduce or eliminate the premium.

The low deductible in the standard benefit suggests policy makers think nearly all beneficiaries will use drugs; i.e., that discouraging use in total is ineffective. Low coinsurance compared with previously marketed private prescription drug coverage (Medigap plans H, I, and J) suggests a belief that price elasticity is low and substantial cost-sharing will not result in major savings. The "doughnut" hole and catastrophic coverage suggest an emphasis on controlling midrange costs through drug choice and increased cost-sharing, with protection for those with extremely high drug costs and low incomes. Concerns about affordability are handled with substantial individual subsides for the lowest income groups, a significant system of reinsurance, and some risk adjustment in payments. Taken together these policy decisions suggest policy makers tend to think that drug insurance choice is driven by observable risks. The policy's reliance on private plans also suggests a belief that adverse selection will not be a significant problem, although the policy does have substantial late enrollment penalties.