The Case for Restructuring the Medicare Prescription Drug Benefit

Implemented a dozen years ago, the Medicare's prescription drug benefit (Part D) has outgrown its initial design and is in need of restructuring. Created as a public–private partnership, clear signs have emerged that changes in the market and private firms’ responses to incentives in the Part D reinsurance program act to inefficiently increase government expenditures. However, a few reforms could shrink this source of waste while protecting and supporting aspects of the program that benefit enrollees.

Part D provides Medicare's prescription drug benefit through private insurers with government subsidization and risk sharing. Private insurers administer the benefit either through stand‐alone prescription drug plans (PDPs) or bundled with other health care benefits as part of Medicare Advantage (MA‐PD plans). Reflecting concerns about whether private firms would be willing to offer drug coverage, Part D's authorizing legislation (the Medicare Modernization Act of 2003) provides considerable support for the market that mitigates insurer risk. From original implementation in 2006 through 2013, most of the government's support came in the form of a “direct subsidy” for coverage below a catastrophic cap, which was set at $5,000 of out‐of‐pocket costs in 2018 and is indexed to the growth of per capita Part D spending.

However, another insurer risk‐mitigation program—reinsurance for enrollees that reach this catastrophic level of drug costs—now accounts for twice as much government spending on Part D plans as the direct subsidy. At nearly 20 percent per year between 2007 and 2015 (MedPAC 2016), this catastrophic range of Part D is the fastest growing component of the program's costs. There is substantial evidence that the rapidly rising taxpayer cost of Part D reinsurance is a consequence of a program design that has not kept up with the changes in the prescription drug market, as well as with insurers’ efforts to exploit the program's structure in ways that benefit them but are inefficient from a broader, social perspective. Unfortunately, these troubling facts have been largely overshadowed by others that make Part D appears to be functioning more efficiently than it actually is.

Indeed, much commentary about the Part D program praises it for the establishment of a popular benefit (about 70 percent of Medicare beneficiaries take advantage of it) within a competitive market. Plan choices are numerous, and competition has kept premiums stable at about $30 per month between 2010 and 2017 (Schmidt and Suzuki 2018). However, there is a connection between Part D's low premium growth and high reinsurance cost growth. Competitively bid premiums do not accurately reflect the 80 percent of reinsurance cost that is borne by taxpayers (15 percent is borne by plans and 5 percent by enrollees). Therefore, insurers have a relatively weak incentive to minimize enrollee entry into and costs within the catastrophic range.

In fact, analysis by the Medicare Payment Advisory Commission (MedPAC) shows that under the current program structure, insurers have an incentive to cover more expensive medications (MedPAC 2017). This incentive arises due to the way that rebates from manufacturers interact with Part D reinsurance. When high‐cost medications are largely paid for by taxpayers in the catastrophic range, but enough of the rebate flows to insurers (through a complex formula, some is shared with Medicare), it is to insurers’ financial advantage to cover those medications more generously (or pay higher sticker prices for them) even if lower‐cost ones would suffice.

In part, this explains why the Part D reinsurance program has grown well beyond what is customary for other health insurance products. For example, in commercial insurance markets, about 1 percent of policyholders reach a catastrophic level of expenditures at which reinsurance kicks in, and reinsurance costs account for 10 percent of spending (Milliman 2015). But in Part D, about 8 percent of enrollees (and growing) trigger reinsurance payments, which accounted for about one‐third of the nearly $100B combined enrollee and government spending through Part D in 2016 (MedPAC 2016; Boards of Trustees 2017).

These findings are consistent with the claim that the Part D reinsurance program is not functioning as insurance at all, which should mitigate against unpredictable risk. Instead, high drug spending exhibits considerable persistence (MedPAC 2016). Part D insurers can predict it to some degree and craft coverage and utilization management policies that shift liability from them to the reinsurance program.

The study in this issue by Jung and Feldman (2018) adds to evidence that the Part D reinsurance program is due for restructuring. One key finding is that, after adjusting for enrollee risk and plan fixed effects, excess reinsurance payments exhibit wide variation for a specific premium. That is, relatively lower premium plans can also account for relatively higher reinsurance payments. Furthermore, this disconnection between premiums and reinsurance payments has increased over time (from 2010 to 2015). The consequence is that premiums are an increasingly weaker signal of plans’ costs. Instead of taking comfort that relatively flat premiums suggest the program is performing efficiently, we should be concerned that they are concealing exploding program growth. The study also found that plans that used fewer utilization management tools for high‐cost drugs had higher reinsurance payments. This is consistent with incentives to increase the chances enrollees reach the catastrophic range of spending, for which plans are liable for relatively little.

There are other contributors to growing reinsurance costs (Schmidt and Suzuki 2018). First, enrollment in Part D is growing, fueled by the baby boom generation reaching retirement age. For this reason alone, beneficiaries are reaching the Part D catastrophic range in higher numbers. For example, between 2010 and 2015, the number of Part D enrollees reaching the catastrophic range of drug spending grew 50 percent, from 2.4 million to 3.6 million.

Second, drug prices are growing as fast or faster than any other component of health care (MedPAC 2017; Health Care Cost Institute 2018). This stems from increasing numbers and use of high‐priced specialty drugs, not greater per person drug volume (Cubanski, Neuman, and Orgera 2017). Although Part D's structure—in particular the level of the catastrophic cap—is indexed to per capita Part D spending, that index has not offset the growth in the number of beneficiaries using high‐cost drugs that reach the catastrophic range.

Third, and contributing to the inadequacy of the catastrophic cap's growth index, by virtue of provisions of the Affordable Care Act (which have been further enhanced by the Bipartisan Budget Act of 2018), Part D's standard benefit gradually relieves enrollees of most of the cost sharing they had been subject to prior to the catastrophic cap, in the so‐called “donut hole.” This is largely accomplished through manufacturer discounts that count as enrollee out‐of‐pocket spending (even though it is not), which hastens enrollees toward the catastrophic cap. This is of great benefit to enrollees with medium‐to‐high drug needs, but it also has the effect of pushing more people onto reinsurance.

Naturally, there are ways to address these issues and their effects on Part D spending. For example, Jung and Feldman (2018) suggest that the catastrophic cap be reset to the 99th percentile of Part D spending, bringing the reinsurance program more in line with that of commercial market plans. Another option they suggest is changing the government's reinsurance liability from the current 80 percent to 20 percent. This echoes a similar recommendation from MedPAC, which pairs it with relieving enrollees’ from any catastrophic liability, currently set at 5 percent (Schmidt and Suzuki 2018). This MedPAC recommendation is part of a suite of ideas that also includes not treating manufacturer discounts in the donut hole as enrollee out‐of‐pocket costs, which would push fewer people into the catastrophic range. Several of these ideas were included in President Trump's recent budget, suggesting that they are becoming more accepted as viable policy proposals (Sachs 2018).

Adding a drug benefit to Medicare in 2006 reflected a crucial modernization of the program. But it was done so in a climate in which we had no experience with stand‐alone drug plans. In that context, the risk‐sharing provisions of the original law made eminent sense, and they helped support the development of what has become a vibrant market for drug coverage for Medicare beneficiaries. However, now, with nearly a dozen years of experience, the market has matured and the original risk‐sharing provisions are no longer needed in their current form. Worse, as documented by Jung and Feldman, as well as MedPAC, that form is being exploited by insurers to boost program costs in inefficient ways—a form of moral hazard.

There is always a trade‐off between risk mitigation and moral hazard. However, it is abundantly clear that the current Part D structure no longer strikes a sensible balance between the two.

Supporting information

Appendix SA1: Author Matrix.