Weird Science: Projecting the Effects of Medicare's Odd Drug Benefit Design

Report Health Care Reform

Weird Science: Projecting the Effects of Medicare's Odd Drug Benefit Design

March 3, 2005 7 min read
Edmund Haislmaier
Senior Research Fellow
Ed is an expert in health care policy and frequently is asked to help lawmakers design and draft reforms to the health systems.

Millions of seniors are in for a surprise.

On January 1, 2006, the new Medicare Part D prescription drug benefit is scheduled to go into effect. At its core is one of the oddest coverage designs ever created for a health insurance benefit. The coverage consists of a low initial deductible, followed by generous reimbursement for remaining front-end expenses, then followed by no coverage at all for mid-range expenses, and ending with almost full reimbursement of major expenses (see Table 1).

Table 1

Apparently, Congress conceived this odd design as a way to reconcile the differences between its political objectives and fiscal reality. Congress's evident political intention was to win support from seniors by expanding Medicare to cover one of the largest heath expenses not yet covered by the program - outpatient prescription drugs. But the fiscal reality is that the federal government can't possibly afford to cover all such expenses. The Congressional Budget Office (CBO) projects that, even without the new program, spending on prescription drugs by or on behalf of Medicare enrollees in 2006 will total $134 billion, with annual spending rising to $284 billion by 2013.[1]

Faced with the reality that providing "everything to everybody" was fiscally impossible, Congress' political imperative became to provide at least something to most Medicare beneficiaries. Thus, the resulting coverage design included a low deductible of $250 and 75 percent reimbursement for the next $2,000 of each beneficiary's drug spending . Political and policy considerations also meant that Congress had to provide additional relief for the small number of beneficiaries with catastrophic drug expenses, as well as further assistance to low-income beneficiaries. So Congress set a per beneficiary limit of $3,600 in out-of-pocket expenses, beyond which Medicare will pay at least 95 percent of all additional drug costs, and also established a set of subsidies for low-income beneficiaries to further reduce their out-of-pocket costs and coverage premiums.

But the unavoidable result of this design was that it includes a large "coverage gap" or "doughnut hole" in the mid-range of drug spending. Once a beneficiary has spent $2,250 on prescription drugs, with Medicare reimbursing $1,500 of those costs, he or she will receive no further Medicare reimbursement until reaching a total of $3,600 in out-of-pocket drug costs. That means the beneficiary must pay entirely out-of-pocket for the next $2,850 of drugs, after the first $2,250. Of course, this "coverage gap" or "doughnut hole" feature of the Part D benefit design is likely to prove highly unpopular with Medicare enrollees

Volatile Variables

The interesting question then, is how might the conflicting political dynamics - particularly, beneficiary support for generous front-end drug coverage and beneficiary dissatisfaction with the coverage gap - play out next year when the Part D benefit is scheduled to take effect? Given that 2006 will not only be the first year of the new program but also an election year, it will be a test of whether Congress got its political calculations right.

The key variables are:

  • The number of enrollees likely to perceive that they have benefited from the generous front-end coverage of the benefit design and/or the program's low-income subsidies;
  • The number of enrollees likely to not benefit at all from the new program;
  • The number of enrollees likely to experience the "coverage gap;" and
  • The timing at which some enrollees will experience the coverage gap and reach the out-of-pocket spending limit.

CBO projects that in 2006, out of a total of 42.6 million eligible Medicare beneficiaries, 29 million will participate in the new Part D program.[2] A study conducted last fall by the Actuarial Research Corporation and the Kaiser Family Foundation, and using CBO's assumptions, further estimated that of those 29 million enrollees, 6.9 million would reach the coverage gap, another 15.1 million would benefit from the program's front-end coverage and low-income subsidy features but not reach the coverage gap, and the remaining 6.0 million would not have enough drug spending during the year to benefit from the program at all (see Table 2).

Table 2

While the ARC/Kaiser study estimates the distributional effects on the enrollee population of the Part D coverage design and subsidies, it doesn't offer an analysis of the timing of those effects. Of particular interest is how the widely discussed drug "coverage gap" can be expected to effect the enrollee population over the course of 2006, the first year of the entitlement's implementation.

To assess the timing effect, one must start with data on the distribution of enrollees by total drug spending from all sources, prior to the inception of the new Part D program. By applying the benefit design to each spending level, one can calculate if and when beneficiaries within each spending level will reach the coverage gap and out-of-pocket limit. The distribution of enrollees by spending level must then be adjusted to account for projected Part D enrollment and for the effects of Part D subsidies, in order to generate revised estimates for the number of enrollees in each spending level. Table 3 shows the results of such an analysis.

Table 3

The estimates derived from this analysis (Table 3) are close to those reported by the ARC/Kaiser study (Table 2) in three instances, including two at the opposite ends of the distribution. First, both analyses conclude that about 3 million enrollees will have no drug spending at all in 2006. Second, the ARC/Kaiser study estimates that 3.1 million beneficiaries will have catastrophic drug expenses sufficient to trigger the out-of-pocket spending limit in 2006, while the analysis presented in Table 3 estimates a slightly higher figure of 3.3 million enrollees. Finally, the ARC/Kaiser study estimates that 16.1 million beneficiaries will have some of their drug costs reimbursed by Medicare, but will not experience the coverage gap, while the analysis presented in Table 3 estimates a somewhat higher figure of 16.9 million enrollees.

However, the two analyses differ substantially as to the number of beneficiaries that will experience the coverage gap but will not trigger the out-of-pocket spending limit in 2006. The ARC/Kaiser study projects the number of such beneficiaries to be 3.8 million while the analysis presented in Table 3 puts the number at 5.8 million. The other significant difference is that the ARC/Kaiser study is able to estimate that 3.0 million unsubsidized beneficiaries will have drug expenses that do not exceed the $250 deductible and thus will also not have any expenses reimbursed by Medicare.

These differences in estimates can be attributed to the greater precision that can be achieved by a more thorough econometric analysis of the kind conducted by ARC/Kaiser. However, the analysis presented in Table 3 is still useful for projecting the rough timing effect of the coverage gap and out-of pocket spending limits in the new Medicare Part D program. Chart 1 graphs the cumulative effects of those provisions using the data presented in Table 3.

Chart 1

The Data Show
With the caveats that the underlying data both likely overestimates the number of enrollees who will experience the coverage gap in 2006 and lacks sufficient precision to plot smoother curves, Chart 3 does show that some reasonable conclusions can be made about the dynamics of these provisions.

First, some enrollees will enter the coverage gap during the first quarter, and their numbers will grow steadily until they peak near the end of the year. Second, those enrollees with the highest costs will begin triggering the out-of-pocket limit at mid-year and their numbers will climb gradually throughout the second half of the year.

Intuitively, these conclusions make sense. Drug costs for Medicare beneficiaries with mid-range to high annual drug spending are in most instances attributable to maintenance therapies - in other words medicines taken regularly over long periods of time to treat chronic conditions. Any changes in their monthly drug expenses will most likely come from adding or switching maintenance therapies. Therefore, their actual monthly drug expenditures are likely to closely track any estimate made by dividing their annual expenditures by twelve. This means that most will enter and exit the coverage gap on a steady, predictable path over time. The higher their annual drug expenses, the quicker they will enter and the quicker they will exit the coverage gap. Conversely, the lower their annual drug expenses, the longer it will take to reach the coverage gap and the longer it will be before they exit the coverage gap by reaching the out-of-pocket limit.

Third, there will be a year-end "pile up" in the Medicare drug "dough-nut" hole. As the experiences of the different drug spending cohorts overlap, the number of enrollees in the coverage gap and the number of enrollees whose drug spending exceeds the out-of-pocket limit will both "pile-up" towards the end of the year.

Furthermore, this "piling up" effect will be exacerbated by the fact that there are more enrollees in the lower-spending cohorts reaching the coverage gap and out-of-pocket limit later in the year than there are in the higher-spending cohorts who reach those thresholds earlier in the year, as can be seen in Table 3.

During the second half of the year, the effect will be somewhat dampened as those with the highest drug spending exit the coverage gap by reaching the catastrophic out-of-pocket limit. However, those in the lower part of the mid-range - with annual drug expense between $2,250 and $5,100 - will enter the coverage gap beginning in mid-year but not exit it before the end of the year.

Of course, at the beginning of each year the benefit calculation is reset to zero and the pattern starts over again.

The data presented in Tables 2 and 3 clearly show that the odd coverage design of the new Part D benefit is likely to achieve Congress's intended effect of creating a large number of "winners." However, the data also indicate that a significant share of enrollees are likely to be less than thrilled with the effects of that coverage design.

It is axiomatic among health care policy analysts that those who benefit from health care changes rarely voice their appreciation while those who feel disadvantaged by the changes always complain loudly. For Congress, that will be a problem.

Edmund F. Haislmaieris a Visiting Research Fellow in the Center for Health Policy Studies at The Heritage Foundation.

[1] "Projected spending for prescription drugs by and on behalf of Medicare enrollees," CBO Memo to interested parties, February 3, 2003, at

[2] A Detailed Description of CBO's Cost Estimate for the Medicare Prescription Drug Benefit," CBO, July, 2004, at


Edmund Haislmaier
Edmund Haislmaier

Senior Research Fellow