|
D A T A W A T C H :
M E D I C A R E P A R T D
1 August 2006
Who Failed To Enroll In
Medicare Part D, And Why?
Early Results
Survey results find gains in coverage among seniors
but indicate some dissatisfaction and doubts about
its adequacy in the long run.
by Florian Heiss, Daniel McFadden, and Joachim Winter
ABSTRACT:
Early results on the Medicare Part D prescription drug program, from a survey of people age sixty-five and older who were interviewed just before enrollment started and just after it ended, indicate that Medicare has met its target of 90 percent coverage. Enrollment rates in vulnerable subpopulations—poor health, low income, or cognitive impairment—are almost high enough to offset lower rates of other coverage. However, sizable numbers of elderly people remain uncovered, contrary to their self-interest. Seniors give Part D mixed reviews, and majorities are less satisfied with Medicare and with the government as a result of their experience with this program. [Health Affairs 25 (2006): w344–w354; 10.1377/hlthaff.25.w344]
Enrollment in the first outpatient drug coverage benefit in Medicare’s history began 1 January 2006, following the passage of the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003. A much-publicized enrollment deadline passed 15 May 2006, after which most seniors face penalties for late enrollment. This paper presents early results on the demographics of enrollment in the Medicare Part D prescription drug program, based on the Retirement Perspectives Survey (RPS) conducted by our research group before and after the enrollment deadline.1
Study Data And Methods
The RPS respondents were 2,137 seniors who were interviewed 7–15 November 2005, immediately before the inaugural open enrollment period for Part D, and reinterviewed between 16 May and 2 June 2006, immediately after open enrollment closed.2 Respondents were asked about their Part D knowledge and intentions in the first interview and about their enrollment-process choices and opinions in the second interview. We present results primarily from a core of 1,571 respondents who were age sixty-five and older in May 2006, were eligible for Part D, were interviewed in both waves, and had no item nonresponse on key variables.
The RPS was administered to a panel of subjects enrolled by Knowledge Networks, a commercial survey firm. Panel members were representative of the U.S. resident population in terms of demographics and socioeconomic status, and sample weighting was used to adjust for subsequent nonresponse.3 They were interviewed using supplied Web TV hardware. Respondents are somewhat more computer-literate than the underlying population: About half of the panel members use the Internet, compared with about a third in the corresponding population.4 Panel members were compensated for participation.
In addition to questions about Medicare Part D, each RPS interview contained questions and embedded experiments regarding health status and conditions, long-term care choices, prescription drug use and cost, and attitudes toward risk.5 Additional socioeconomic and demographic variables were provided by Knowledge Networks as background on panel members. We also constructed a simple index of cognitive impairment.6
Study Results
Of the 35.84 million people age sixty-five and older who reside in the United States and are eligible for Medicare Part D, data from the Centers for Medicare and Medicaid Services (CMS) suggest that as of 14 June 2006, 24.67 million (68.9 percent) were enrolled in Part D, 8.50 million (23.7 percent) had other creditable prescription drug coverage, and 2.66 million (7.4 percent) had no creditable coverage.7 The CMS classifies prescription drug coverage as “creditable” if it is at least as good as the Part D Standard Plan.8 RPS tabulations are consistent with these CMS estimates: Among the 1,571 people in the sample who were eligible for Part D, 68.1 percent were enrolled in Part D, 24.5 percent had other creditable coverage, and 7.4 percent had no creditable coverage as of 1 June 2006.9 Thus, subject to the caution that matching of RPS responses to categories covered in CMS tabulations is imperfect, our estimates based on RPS survey data and CMS population estimates both confirm that the actuarial target of a 90 percent prescription drug coverage rate in the population age sixty-five and older has been achieved.
A sizable fraction of the elderly are automatically enrolled in Part D or other creditable drug coverage, related to their employer or union plans coordinated with Part D; Medicaid; or programs such as federal employees, military, and veterans coverage. In the RPS sample, 48.1 percent of the 1,484 respondents who were sure of their status indicated that they had a choice.10 Of these, 23.8 percent did not enroll in Part D or obtain other creditable coverage. This number suggests that the CMS was reasonably successful in enrolling the elderly who had a choice, but some are still left without coverage.
Demographics. The healthy have lower Part D enrollment rates than the unhealthy, and sharply lower rates for other creditable coverage (Exhibit 1). Joachim Winter and colleagues estimate that about 24.4 percent of the population age sixty-five and older are healthy enough that it might have been rational for them to delay enrollment at the monthly premium of $37 originally projected by the CMS.11 It is a major advantage for the CMS and its insurance providers in Part D that 75–80 percent of the very healthy now have prescription drug coverage. This greatly reduces the risk of unraveling of the market as a result of adverse selection.
A concern prior to opening Part D enrollment was that vulnerable segments of the population age sixty-five and older would not have the awareness and acuity needed to make rational enrollment and plan decisions. Making contact with vulnerable populations and counseling them was a major element of the CMS’s strategy for launching Part D. This strategy was largely successful; it resulted in coverage rates among vulnerable populations that are comparable to coverage rates in the overall population for low-income and cognitively impaired groups (Exhibit 2). The most problematic groups are widows and unmarried women and the less educated, among whom increased Part D enrollment is insufficient to greatly offset their lower rates of having other creditable coverage.
Exhibit 3 provides a breakdown of the Medicare population by socioeconomic status and number of prescription drugs used in November 2005.12 Two groups that have failed to enroll in Part D are of particular concern to health policymakers: people who are healthy now and using no prescription drugs, who could reasonably delay enrollment but might be unaware of the potential future benefits of coverage; and the less healthy who are currently using prescription drugs and are uncovered by Part D (or other insurance), against their own immediate interest. At a monthly pharmacy cost of about $52 per prescription and a Part D Standard plan monthly premium of $24 or less available to most of the elderly, the immediate Standard Plan annual benefit net of premium is about break-even for a person with one prescription, and most likely immediately beneficial for people with two or more prescriptions. Since the Standard Plan also has value as insurance against catastrophic drug costs, most people with one or more prescriptions are better off enrolling.
About two million nonenrollees use one or more prescriptions (8.5 percent of 9.1 million plus 5.2 percent of 23.0 million), which means that their choice is in most cases contrary to their immediate self-interest. In this group, there is not much variation by income or education in the percentage uncovered. There is a fairly small group of about 0.5 million uncovered people with low incomes and one or more prescriptions who can qualify for subsidies and avoid late enrollment penalties, but about 1.5 million uncovered people with one or more prescriptions have incomes high enough to be subject to these penalties. Many of the nonenrollees who use one or more prescriptions have no more than a high school education and might have failed to enroll as a result of procrastination or confusion.
The second vulnerable group in Exhibit 3 are people with zero prescriptions for whom Part D is not an immediate benefit but might be beneficial as insurance against future drug costs if their health deteriorates. Coverage rates in this group are particularly low in the low-income segment, with 38.2 percent uncovered, and the low-education segment, with 22.8 percent uncovered. We speculate that this arises because of constraints or perceptions that make it difficult for people in these groups to account for future benefits.
Many seniors had some form of prescription drug insurance before Part D was introduced and consequently enrolled in Part D through coordination of their existing insurance with Part D or through Part D plans offered by the same provider. Enrollment rates are much lower for those without prior coverage (Exhibit 4). A major concern prior to the inaugural Part D enrollment period was that consumers lacked information on what promised to be a complex decision, and many had formed intentions to not enroll without carefully assessing their interests. Exhibit 5 shows that coverage is lower among those who had little knowledge, or expressed intentions to not enroll, in November 2005. However, majorities did enroll in Part D even among those with little knowledge or contrary intentions.
Enrollment process. We asked 776 respondents in the sample who were not automatically enrolled in Part D or other prescription drug coverage, and thus had to decide whether to enroll, about their experience with the enrollment process (Exhibit 6). Substantial majorities were able to get their questions answered and determine whether or not they would benefit from enrollment. Those who did not enroll had difficulty understanding how Part D works and found the enrollment process complicated. Majorities liked having alternative insurers from which to choose, even though they felt that they faced too many alternatives. Many had doubts about plans’ formularies (whether or not their needed drugs would be covered), although most stated that they were able to get coverage for the prescription drugs they needed. The concerns of the minority about getting specific needed drugs from plan formularies might be amplified by copayment disincentives for brand-name drugs and by classification of some treatments as medically unnecessary. Since nonenrollees in particular have had no opportunity to gain firsthand experience with Part D, their responses provide some indication of the extent to which opinions are colored by general attitudes.
Respondents’ assessments of Medicare Part D. We asked all 2,137 respondents to the May 2006 RPS, including those ages 63–65 who are not yet eligible for Part D as well as many age sixty-five and older whose prescription drug coverage was automatic, about their overall evaluation of the Medicare Part D program. (Numbers are qualitatively similar if the analysis is restricted to those 1,571 respondents whose drug coverage was analyzed in earlier exhibits.) Exhibit 7 presents satisfaction ratings on various dimensions. Majorities agree that Medicare Part D is a major benefit to seniors and works well once enrollment is completed; these findings are consistent with those of earlier surveys.13 However, only about a third feel that the program is well designed overall. People are conflicted and somewhat inconsistent regarding the organization of Part D as a market with competing plan providers. Majorities like the competition among plan providers and choose widely among the options available, but they also say that they would have preferred a program more like traditional Part A/B Medicare coverage. Respondents dislike the deductible and gap features of Part D plans and the unpredictability of formularies. Since seniors can, and in many cases do, avoid some of these features by choosing plans with deductible and gap coverage, these dissatisfactions apparently arise because of the remaining uncertainties in prescription drug coverage or because optional payments require extra attention or offend consumers’ sense of entitlement.
Discussion And Conclusions
We conclude from our early results from the RPS survey, and from CMS enrollment figures, that the primary objective of the new Part D program—to provide most seniors with affordable coverage for the medications they need—has been largely achieved. An unanswered question is whether provision of drug coverage by Medicare without private insurance intermediaries would have been equally or more effective in controlling drug costs. The information and administrative problems that arose early in the inaugural enrollment period appear to have been overcome, although they might have colored attitudes toward the program. Concerns that large numbers of seniors, particularly among vulnerable groups, would fail to enroll in the program are largely allayed. However, there remains a substantial core of seniors who need to be educated that Part D is in their own interest, and reaching them should continue to be a health policy priority.
The CMS has an extensive outreach program (the Low-Income Subsidy, or LIS) and a Special Enrollment Period without a late enrollment penalty for low-income nonenrollees. These mechanisms need to be effective in bringing the 0.8 million remaining nonenrolled low-income people into the program. More effort is needed to reach the 1.5 million nonenrolled unhealthy people who should be enrolled in their own interest and whose incomes are too high to qualify for the LIS. Healthy people with low education should also be targeted with counseling on the future benefits of coverage. A well-publicized penalty forgiveness window could help bring some of the nonenrolled into the program.
Consumers’ opinions about Part D are mixed. Majorities are troubled by the deductible and gap provisions of Standard Part D coverage and find it difficult to determine the current and future formularies of the plans they evaluate. There are likely to be complaints in the future regarding changes in available providers, jumps in premiums, and difficulties in plan switching. The CMS could address many of these problems by encouraging people to choose enhanced plans with gap coverage and by requiring “truth in insuring” so that information on plan features and costs is available to decisionmakers in a common, easily understood format. At this point, the Part D market does not have a major problem of adverse selection, in which large numbers of the most healthy fail to enroll. However, the market should be monitored in the future, particularly if premiums rise, to ensure that it does not begin to unravel. Monitoring is also needed to determine the extent of moral hazard, in which relatively low copay rates encourage heavier use of optional medications, and if necessary to provide incentives to physicians and patients to be conservative in the use of medications.
Consumers are often skeptical about markets and suspicious of their organizers.14 This seems to be the case for Part D. Asked the question, “Does your experience with Medicare Part D leave you more satisfied or less satisfied with the Medicare program?,” 58.1 percent of the 2,027 people responding said that they were less satisfied. Asked the question, “Does your experience with Medicare Part D leave you more satisfied or less satisfied with the political process in Washington that produced this program?,” 74.7 percent of the 2,023 responding said that they were less satisfied. The responses indicate substantial dissatisfaction with the design and administration of the program. An absence of negative experiences over an extended period might be needed to persuade a majority of the elderly that the Part D market for prescription drug coverage operates to their advantage.
Careful attention to explaining the Part D program to consumers—particularly flashpoints such as provider and premium changes, gap coverage, and the stability and predictability of formularies—and streamlining and simplifying the enrollment process are important to ensure the program’s future success.
The authors thank Carol Kelly and Audrey McDowell of the Centers for Medicare and Medicaid Services (CMS) and Richard Suzman of the National Institute on Aging (NIA) for information and comments; the authors are solely responsible for the results and conclusions offered in this paper. This research was supported by the Behavioral and Social Research program of the NIA, with additional support from the E. Morris Cox Fund at the University of California, Berkeley.
NOTES
1. In addition to the authors, research team members participating in the RPS study were Francis Caro, Gerontology Institute, University of Massachusetts, Boston; Byung-hill Jun, Department of Economics, University of California, Berkeley; Rosa Matzkin, Department of Economics, Northwestern University; and Rowilma Balza, Department of Economics, University of California, Berkeley.
2. The November 2005 survey contacted 5,879 subjects age fifty and older, of whom 4,738 (80.6 percent) responded. Of these, 1,884 were age sixty-five and older. The May 2006 resurvey contacted 2,598 November respondents age sixty-three and older, of whom 2,137 (82.3 percent) responded. Of these, 1,706 were age sixty-five and older.
3. The panel was recruited from the population by random-digit dialing. The sample is selective in that it excludes people who do not have TV sets or are unable or unwilling to participate in the panel. Details can be found in J.A. Krosnick and L. Chang, “A Comparison of the Random Digit Dialing Telephone Survey Methodology with Internet Survey Methodology as Implemented by Knowledge Networks and Harris Interactive,” April 2001, http://www.knowledgenetworks.com/ganp/docs/OSUpaper.pdf (accessed 7 July 2006).
4. Because of its random recruitment, the Knowledge Networks panel is closely comparable on key demographic and health status variables to the Health and Retirement Study (HRS) panel, which uses best practices for sample design and recruitment. This is in sharp contrast to opportunistically recruited Internet samples, where participation is biased to younger, higher-income, and better-educated respondents; see A. Börsch-Supan et al., “Reducing the Participation Bias of an Internet Survey” (Working paper, Mannheim Research Institute for the Economics of Aging, July 2006). Evidence from our studies and from the RAND HRS Internet surveying project indicates that interview mode effects between Internet, Web TV, computer-assisted telephone interviewing (CATI), and mailback are small, and Internet modes have substantial advantages in controlled presentation of visual information.
5. One objective of the RPS is to demonstrate the effectiveness of “fast and cheap” Internet survey methods for social science experimentation, response error control, and policy analysis. The RPS data were collected at a per case cost well below the typical cost of traditional surveys, and the complete data set was available for analysis four days after the survey closed.
6. A cognitive-impairment scale is constructed by summing instrumental activities of daily living (IADL) indices for difficulty with money management and taking medications and the number of errors made on five items that test numerical and logical skills; see J. Winter et al., “Medicare Prescription Drug Coverage: Consumer Information and Preferences,” Proceedings of the National Academy of Sciences (U.S.) 103, no. 20 (2006): 7929–7934.
7. The U.S. resident population age sixty-five and older on 1 January 2006, by census projections, was 37.02 million. Of these, 1.18 million (3.2 percent) are not enrolled in either Medicare Part A or B and are hence ineligible for Part D. We estimated enrollment status in the population age sixty-five and older by starting from CMS data released 14 June 2006 on all eligible beneficiaries and backing out enrollment estimates for the 6.84 million people who qualify for Medicare as a result of disability. We treated Retiree Drug Subsidy (RDS) participants as enrollees in Part D and treated federal retiree coverage (Federal Employees Health Benefit, or FEHB, and TRICARE) as being in the “other creditable coverage” category. We adopted CMS estimates that among the disabled under age sixty-five, 68.8 percent are enrolled in Part D, and 4.9 percent have other creditable coverage. To estimate the number of disability-eligible without creditable coverage in the subpopulations reported by the CMS, we adopted CMS assumptions that 85 percent of the 2.01 million eligible veterans, 95 percent of the 0.1 million eligible people covered by the Indian Health Service and the 0.59 million people covered by State Pharmaceutical Assistance Programs, and all of the 2.57 million active workers with Medicare as a secondary payer and 0.1 million other non-RDS retirees are age sixty-five and older.
8. The Standard Plan covers 75 percent of pharmacy costs above $250, up to $2,250, and 95 percent of pharmacy costs above $5,100 but provides no coverage in the gap between $2,250 and $5,100 (the so-called doughnut hole). At 2005 estimates of the distribution of pharmacy costs in the population age sixty-five and older, the actuarial value of benefits under the Standard Plan was $1,542, less an annual premium of $444. A creditable plan is described as having formulary rules and expected benefits at least as good as those of the Standard Plan. Actual experience with plans on the market shows that both premiums and the prices of medications are lower than the CMS anticipated in 2005.
9. The 1,571 RPS respondents exclude 136 people age and older who were either not asked (due to Medicare ineligibility) or refused to identify a source of prescription drug coverage. There were 205 respondents who were unsure if their coverage is coordinated with Part D, and we treated them as enrolled in Part D. There were 101 respondents who were unsure if their medical insurance includes drug coverage. For this calculation only, we allocated these respondents between “other coverage” and ”no coverage” in the same proportion as these categories appear among respondents who are sure of their status; in all other tabulations, we treated them as uncovered. If we had instead omitted the cases with uncertain status, then among the 1,265 certain respondents, 67.5 percent are in Part D, 25.1 percent have other creditable coverage, and 7.4 percent have no coverage. Sample counts are from the weighted sample; use of unweighted sample counts gives nearly the same results.
10. Among the automatically enrolled, 10.7 percent use no prescriptions and 67.6 percent use three or more, while among those with choice, 8.9 percent use no prescriptions and 61.9 percent use three or more.
11. Winter et al., “Medicare Prescription Drug Coverage,” uses a computational approximation to the dynamic stochastic program of a risk-neutral healthy person facing the probability of future medication needs and premium penalties, and it calculates the proportion of the population age sixty-five and older for whom expected present value of net benefits is higher with an enrollment delay until health conditions warrant. At sufficiently low monthly premiums, below about $10 per month, delay is never optimal.
12. Exhibit 3 was obtained by scaling weighted RPS sample observations to the eligible age population age sixty-five and older, with weights adjusted by iterative proportional fitting to match our estimates of Part D enrollment derived from CMS data.
13. The U.S. Chamber of Commerce released survey results on 25 April 2006 indicating that 84 percent of seniors enrolled in the Medicare prescription drug program are satisfied with their drug coverage and that 52 percent say that they are enjoying a significant cost saving. U.S. Chamber of Commerce, “Nationwide Poll of Seniors Shows High Level of Satisfaction with Medicare Prescription Drug Plan,” Press Release, 25 April 2006, http://www.uschamber.com/press/releases/2006/April/06-71.htm (accessed 10 July 2006). In March 2006, America’s Health Insurance Plans (AHIP) surveyed equal numbers of Medicare Part D beneficiaries who self-enrolled and those who were autoenrolled. The survey found that the vast majority of seniors enrolled in Part D have had no problems with their plan and say that the plan covers the drugs they need, and among those who signed up for the plan themselves, two-thirds recommend that other seniors sign up as well. AHIP, “Ninety Percent of Low-Income Seniors Surveyed Say No Problems Using Medicare Drug Benefit,” Press Release, 13 March 2006, http://www.ahip.org/content/pressrelease.aspx?docid=15333 (accessed 10 July 2006).
14. D. McFadden, “Free Markets and Fettered Consumers,” American Economic Review 96, no. 1 (2006): 5–29.
Florian Heiss is a postdoctoral fellow in the Department of Economics, University of Munich, Germany. Daniel McFadden (mcfadden{at}econ.berkeley.edu) is the E. Morris Cox Professor, Department of Economics, University of California, Berkeley. Joachim Winter is a professor of economics in the Department of Economics, University of Munich. All three authors are affiliates of the RAND HRS program project on Internet surveying. McFadden is also a faculty affiliate of the University of Michigan Survey Research Center and a principal investigator in the National Bureau of Economic Research (NBER) project on the economics of aging. He is past president of the American Economic Association and a recipient of the Clark, Frisch, Laffont, Nemmers, Nobel, and Stone prizes.
DOI: 10.1377/hlthaff.25.w344
©2006 Project HOPE–The People-to-People Health
Foundation, Inc.
|
