Medicare Part D Formulary Coverage Since Program Inception: Are Beneficiaries Choosing Wisely?

Published on: 
The American Journal of Managed Care, November 2008 - Special Issue, Volume 14, Issue 11 SP

Analysis of Medicare Part D formulary composition since program inception suggests beneficiaries may not be using their open-enrollment periods to reevaluate available plan offerings.

Objective: To evaluate how Medicare Part D formulary composition has changed since program inception, including comparison of plans eligible for full premium subsidy (ie, benchmark plans) with their counterparts.

Study Design and Methods: The study used publicly available data released by the Centers for Medicare & Medicaid Services to generate snapshots of formulary coverage and enrollment levels in each plan year. The analysis included all Part D plans and tracked formulary coverage of 152 of the most common brand name and generic drugs prescribed to seniors.

Results: Since 2006, the number of products available without restriction has increased and the number of drugs not on formulary has decreased. However, it appears that beneficiaries (subsidized beneficiaries in particular) may not be using their open-enrollment periods to reevaluate the available plan offerings.

Conclusions: Beneficiaries need to reevaluate the Part D options available on an annual basis to maintain enrollment with the most appropriate plan available. Although all plans meet the proscribed formulary requirements, some plans offer richer drug coverage with more drugs available on an unrestricted basis. Benchmark plan status allows Part D plans to maintain or gain significant Medicare enrollment from year to year. Careful oversight should be provided to ensure that the level of formulary coverage offered at benchmark and other plans remains consistent.

(Am J Manag Care. 2008;14(11 Spec No.):SP29-SP35)

A review of coverage for 152 commonly prescribed drugs suggests beneficiaries may not be using their open-enrollment periods to reevaluate the available plan offerings.

  • Other beneficiaries may be renewing their original plan without considering alternatives.

Effective January 2006, Medicare Part D expanded the Medicare program to include coverage for prescription drugs. Prior to 2006, low-income Medicare beneficiaries who qualified for their state Medicaid program received drug coverage through Medicaid, while Medicare remained the primary payer for most other medical services. Individuals who are eligible for both Medicare and Medicaid are called dual-eligible beneficiaries. Under the Medicare drug benefit, dual-eligible beneficiaries qualify for reduced cost-sharing requirements and are exempt from premium payments if they are enrolled in basic plans with a premium below the minimum premium benchmark, designated as benchmark plans. Other Medicare beneficiaries with low income and minimal assets also may qualify for reduced cost-sharing requirements and are referred to as the other low-income subsidy (LIS) population in this report. If they do not choose a Medicare Part D plan, full-benefit LIS and dual-eligible beneficiaries are automatically enrolled in a benchmark plan based on a random assignment.

A May 2007 report by the Government Accountability Office acknowledged the impact beneficiary-specific assignments had in a pilot program launched in Maine, but concluded that the existing random assignment process should be maintained.1 The report found that the dual-eligible beneficiaries’ right to change plans at any time and the minimum proscribed formulary requirements provide beneficiaries with the flexibility to meet their coverage needs.

This study was conducted to assess (1) whether there are any differences in the formulary composition of the benchmark plans versus other plans and (2) whether it appears that dual-eligible and other LIS beneficiaries are actively making enrollment decisions based on formulary composition.


The analysis included stand-alone prescription drug plans (PDPs) and Medicare Advantage plans that provide a Part D benefit (MAPDs). The analysis was limited to non–employer-sponsored plans to be consistent with the formulary information available in the PUF.

For the purposes of counting plans, we defined a plan as the unique combination of contract ID and plan ID. The contract ID and plan ID are referenced in plan documents published by the Part D plan, and they are used in the PUF files and enrollment reports to track carrier or plan information. The contract ID corresponds to the carrier providing the coverage (eg, Humana Insurance Company of New York). The plan ID corresponds to the benefit offered (eg, Silver, Gold, or Bronze). PDP carriers may offer no more than 3 plans in any region (with some exceptions due to mergers or acquisitions). As such, a PDP offered in all regions would have no more than 114 unique combinations of contract ID and plan ID filed with CMS (ie, maximum of 3 plans filed in 38 regions). There may be many more combinations of contract ID and plan ID associated with the Part D component of MA-PD plans, however. Most MA-PD plans are filed at the county level (rather than the multistate region level, like PDPs). Approximately 1% of Part D enrollment is in regional MA-PD plans that are filed for a multistate area.

We reviewed the top 168 drugs dispensed to persons age >65 years as of October 2007, as identified by Verispan based on script volume.6 The drugs were stratified into the following categories:

• Drugs in excluded therapeutic classes were not included in the study. Eight of the most common 168 drugs were in this category: alprazolam, lorazepam, folic acid, clonazepam, temazepam, diazepam, Viagra, and enteric-coated aspirin.

• Four of the most common drugs were the generic form of a brand drug also listed among the most common drugs. These products had recent patent expirations. Coverage for the drugs in each year was included at the best coverage available for either the brand or generic form. The 4 pairs of drugs were Norvasc/amlodipine besylate, Toprol XL/metoprolol succinate, Ambien/zolpidem tartrate, and Lotrel/amlodipine besylate and benazepril HCl.


Table 1 summarizes the total enrollment and count of plans included in the analysis. The enrollment associated with benchmark plans includes all enrollees, not limited to those who qualify for the subsidy. As seen in Table 1, the enrollment in the benchmark plans decreased significantly between 2007 and 2008. The number of individuals eligible for the subsidy did not decline over this time. There were 9.2 million Part D enrollees eligible for the subsidy in 2007 and 9.4 million Part D enrollees eligible for the subsidy in 2008.

The decline in enrollment in the benchmark plans shown in Table 1 is the result of increased disruption in the plans with benchmark status in 2008. Between plan years 2006 and 2007, few plans lost benchmark status (notifications of premium increase were mailed to 0.6 million beneficiaries7). A greater number of plans lost benchmark status between plan years 2007 and 2008 (notifications of premium increase were mailed to 2.6 million beneficiaries8). From year to year, the number of beneficiaries who change plans has remained consistent. The CMS indicated that 10% of Part D enrollees changed plans between 2006 and 20079 and 12% of Part D enrollees changed plans between 2007 and 2008.10 Between 2007 and 2008, it is likely that many of the nonsubsidized enrollees renewed with their plans, even when the plan lost benchmark status. As a result, it appears the nonbenchmark plans gained enrollment, although most of the movement in enrollment would have been between plans with benchmark status.

Figure 1

As illustrated in , drugs were categorized into 3 levels of coverage: (1) not listed, (2) restricted, and (3) unrestricted. Drugs that are not listed in the PUF for a plan are generally not covered, although beneficiaries may appeal to the plan for a coverage exception. A drug was classified as restricted if a prior authorization or step therapy requirement was reported in the PUF for a plan. A drug was classified as unrestricted regardless of its tier status if there were no reported prior authorization or step therapy requirements. There may be quantity limits associated with these drugs. As shown in Figure 1, across all plans, the number of study drugs available without restriction peaked in 2007. The peak in unrestricted study drugs in 2007 appears to be related to expanded formulary coverage between 2006 and 2007. Plans may have been conservative when they filed their 2006 benefits. On average, the plans that renewed between 2006 and 2007 increased the number of drugs available on an unrestricted basis by 4.5 (MA-PDs) and 5.5 (PDPs). From 2006 to 2008, the number of study drugs available on an unrestricted basis increased by 2 or more and the number of drugs not listed decreased by more than 2 drugs. The number of drugs not listed on the formulary decreased most significantly at MA-PD plans (from 12 to 6) and nonbenchmark PDPs (from 10 to 6).

The values in Figure 1 were calculated on a straight-average basis. That is, all plans were equally weighted. This calculation is analogous to a beneficiary reviewing the PDP offerings during the open-enrollment period. An enrollment-weighted approach counts each plan offering in proportion to its enrollment levels. This reflects actual beneficiary selection from the available plan offerings.

Figure 2 illustrates the number of study drugs on a straight-average and enrollment-weighted basis at 2006 and 2008 by plan type. Enrollees in the nonbenchmark plans exhibited the strongest preference for less restrictive formularies, as measured by the difference in the count of unrestricted study drugs on a straight-average basis and an enrollment-weighted basis. Enrollees in MA-PD plans showed almost no preference for less restrictive formularies using the same metric. Figure 2 also suggests the preference for PDPs with unrestricted coverage was stronger at program inception (as observed by larger differentials in 2006 than in 2008). These results hold true at the region level as well.

Enrollees in the nonbenchmark PDPs were specifically choosing prescription drug coverage (compared with MAPD enrollees who had to consider medical providers as well as benefit designs). All the individuals in the nonbenchmark plans in 2006 elected that coverage; none were auto-assigned. The higher count of unrestricted drugs on an enrollment-weighted basis reflects their selection process. The enrollment in the benchmark plans, however, reflects a mix of beneficiaries who elected the plan as well as those who were auto-assigned to the plan. The auto-assigned individuals would be expected to have coverage that is most similar to the straight-average basis (as this is reflective of the random assignment process).

Figure 1 indicates that the average number of unrestricted study drugs remained fairly stable from year to year. However, the averages shown in Figure 1 mask some significant changes in formulary composition at the plan level. The extent of plan-level changes was quantified by isolating the PDPs that maintained the same plan number in 2006, 2007, and 2008. Plans that were available for all 3 years account for approximately 96% of the 2006 and 2007 PDP enrollment and approximately 88% of the 2008 PDP enrollment.

Table 2 stratifies the renewing plans by the change in the number of unrestricted drugs between 2006 and 2008. Table 2 indicates that 39.3% of the PDPs that were available in all 3 plan years added or removed 7 or more drugs to or from unrestricted access. Of the renewing PDPs that removed 7 or more products from unrestricted status, the average number of drugs removed was 18.4 (from 145.2 to 126.8). Of the renewing PDPs that added 7 or more drugs to unrestricted status, the average number of drugs added was 16.1 (from 123.4 to 139.5). This high volume of renewing plans with significant formulary changes suggests that at open enrollment beneficiaries would need to review their current plan to determine whether it or a competitor plan would better meet their needs.

Beneficiaries associated with the plans that removed 3 or more products from unrestricted status constituted 27% of PDP enrollment in 2006 and 21% of PDP enrollment in 2008. Beneficiaries associated with the plans that added 7 or more products to unrestricted status constituted 22% of PDP enrollment in 2006 and 23% of PDP enrollment in 2008.


The changes in formulary composition from year to year, whether positive or negative, suggest that beneficiaries should review their plan options carefully each year. Beneficiaries cannot assume their formulary coverage will be maintained from one year to another. Their current plan may become more restrictive, or a competitor plan may become more attractive. Beneficiaries must balance formulary access, premium costs, and cost-sharing levels to make the most appropriate selection.

When plans are filed with the CMS, it is unknown whether they will fall below the benchmark premium; that status is determined after a review of all approved plans. The early experience indicates that the formulary composition of benchmark and other plans has been consistent. However, as the number of plans with auto-enrollment status declines (due to changes in the methodology for identifying the minimum premium benchmark) and the potential gain or loss of a significant block of enrollment increases, the CMS and other interested parties should consider whether the motivation to be a benchmark plan has the potential to introduce increasingly more restrictions on formulary access.

When Medicare Part D experience becomes available to researchers, it will be important to evaluate whether differences in coverage levels (including the presence of restrictions) have an impact on the utilization levels of the covered enrollees. Moreover, it will be important to evaluate whether the impact disproportionately affects the dual-eligible and other LIS populations.

Author Affiliations: From Milliman, Inc (EAJ), Indianapolis, IN; and Pfizer, Inc (KJA), New York, NY.

Funding Source: This analysis was funded by Pfizer, Inc.

Author Disclosure: Ms Jackson is an employee of Milliman, Inc, a healthcare consultant firm, and was retained by Pfizer, Inc to perform this analysis. Ms. Axelsen is an employee of Pfizer, Inc, the sponsor of this study.

Authorship Information: Concept and design (EAJ, KJA); acquisition of data (KJA); analysis and interpretation of data (EAJ); drafting of the manuscript (EAJ); critical revision of the manuscript for important intellectual content (EAJ, KJA); statistical analysis (EAJ); obtaining funding (KJA); and

2. Centers for Medicare & Medicaid Services. Prescription drug plan formulary and pharmacy network files. Accessed October 6, 2008.

4. Centers for Medicare & Medicaid Services. Medicare Advantage/Part D contract and enrollment data. Monthly enrollment by plan. Accessed October 6, 2008.

6. Verispan Vector One® datasource. Custom analysis prepared for Pfizer, listing of the top 200 drugs dispensed in the retail setting to people age 65 or older, moving annual total October 2007. Accessed October 21 2008.

8. Centers for Medicare & Medicaid Services. Limited income & resources. Overview. Accessed October 6, 2008.

10. Centers for Medicare & Medicaid Services. Medicare Prescription Drug Benefit’s Projected Costs Continue To Drop: Part D Attracts New Beneficiaries and Achieves High Rates of Satisfaction. CMS press release. January 31, 2008. Accessed October 21, 2008.