Voluntary Disenrollment From Medicare Advantage Plans: Valuable Signals of Market Performance

Published on: 
, , , , , ,
The American Journal of Managed Care, December 2007, Volume 13, Issue 12

Objective: To examine 2000-2005 trends in the reasons Medicare beneficiaries gave for disenrolling from their Medicare Advantage (MA) plans.

Study Design: We used data from 6 consecutive years of Consumer Assessment of Health Plans surveys, which asked about 33 possible reasons for disenrollment, including problems with plan information, out-of-pocket costs, plan benefits, and coverage. Respondents numbered more than 50 000 beneficiaries each year from a variety of MA plan types providing full Medicare benefits in place of traditional fee-for-service Medicare. The survey also collected demographic and health status information.

Methods: We classified reasons for disenrollment into 2 key groups: (1) reasons related to plan information and (2) reasons related to cost/benefits problems. We examined whether disparities existed between vulnerable and less vulnerable populations that might reflect different experiences by these groups over time.

Results: Disparities between vulnerable and less vulnerable groups were present but generally diminished over time as competition intensified, with noticeable differences between African American and Hispanic subpopulations regarding problems with plan information.

Conclusions: The premise of the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 was that more plans would increase competition, resulting in higher-quality healthcare services. However, an increased number of plan choices complicates the health plan decision-making process for beneficiaries. With further expansion of plans and choices following implementation of Part D, efforts must continue to direct informational materials to all beneficiaries, particularly those in vulnerable subgroups. More help in interpreting the information may be required to maximize consumer benefits.

(Am J Manag Care. 2007;13:677-684)

Following the Medicare Modernization Act of 2003, the elderly had many new choices and opportunities to find better coverage. As a result, they disenrolled from current plans to join other Medicare Advantage plans at unprecedented rates.

Plan-related information problems increased, highlighting the importance of ongoing efforts to develop and disseminate consumer-directed information products for the elderly.

As plans used enhanced benefits and lower cost-sharing to compete for enrollees, disparities narrowed among vulnerable and less vulnerable populations in their propensity to cite problems related to costs and benefits.

The Centers for Medicare & Medicaid Services (CMS) sponsored the annual Medicare Consumer Assessment of Health Plans (CAHPS) disenrollment surveys of voluntarily disenrolled beneficiaries beginning in 2000 and ending in 2005, and used the survey results to provide information to consumers and health plans about why beneficiaries chose to leave their Medicare Advantage (MA) plans. (For example, see the Medicare Personal Plan Finder–Why People Leave Plans at This paper analyzes data from 6 annual implementations of the survey in terms of how respondents' reasons for leaving their MA plans reflected changes occurring in the Medicare managed care market. In particular, we focus on reasons related to information about plan benefits and prescription drug coverage issues.

The Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003 resulted in sweeping changes to the Medicare program, including a new Medicare prescription drug benefit. After the MMA was passed, payment rates to plans and physicians increased and there was unprecedented growth in the number of plan options, especially for Medicare beneficiaries in urban areas.1—5 The first wave of expansion occurred in 2003 with the launch of the Medicare preferred provider organization (PPO) demonstration project, with 35 new Medicare PPO plans. Another wave occurred in the latter months of 2005, as the managed care industry rushed to launch new local plans before the moratorium on these began in 2006.3 By January 2007, 589 MA plans were offered, more than double the number (n = 222) available in 2002.6,7

The premise of the MMA was that more plans would increase competition, resulting in higher-quality healthcare services. However, with increases in the number of plan choices, the health plan decision-making process for beneficiaries became more complex. When faced with too many choices or too much uncertainty, people may choose not to make any changes to their health plan or may have difficulty making informed choices.8-11 Vulnerable populations are at greater risk of making less-thanoptimal decisions, particularly given their lower levels of education and health literacy.12-14 Thus, there is an economic trade-off–more plans increase competition, but in the absence of adequate information, vulnerable populations may fare worse as the markets change.12,15-19

CMS engages in a wide range of activities to improve the health insurance information available to beneficiaries to promote informed decision making about health plan choice. The goal is to raise awareness and encourage people to make the most of their available coverage to maintain and manage their health.20 Educational media include (but are not limited to) the Medicare & You handbook, a CMS-sponsored Web site ( providing basic and comparative information on health insurance options and quality-of-care measures, and the Medicare beneficiary hotline.

Extensive and continued evaluation and testing of beneficiary educational and informational interventions are part of CMS's agenda. CMS's partners are routinely asked for the type and formats of information they want, as well as their assessment of the available materials and resources. CMS also is focusing on developing information and educational opportunities targeted at specific topics and at beneficiaries with specific interests.

When analyzing trends in the survey data from 2000 to 2005, we expected to find 3 things consistent with market changes:- Vulnerable subgroups would have a greater propensity to cite plan-related information, cost, or benefit problems throughout the period.

- The number of beneficiaries citing problems with information as a reason for disenrolling would increase after implementation of the MMA.


The surveys covered a variety of MA plans that provide full Medicare benefits in place of traditional fee-for-service (FFS) Medicare. The number of managed care plans included in the survey each year reflects the availability of Medicare plans (Figure 1). Survey participants were enrolled in a variety of MA plan types, including health maintenance organizations, PPOs, private FFS plans, provider-sponsored organizations, and demonstration plans.21 Figure 1 plots the number of MA plan contracts in January of each year, from 2000 through 2007.


The sample size in each year was large–more than 50 000 beneficiaries. (The sample sizes, by year, were 87 465 in 2000; 64 430 in 2001; 53 241 in 2002; 59 072 in 2003; 67 146 in 2004; and 55 730 in 2005.) Except for a slight increase in the proportion of minority populations over time, the characteristics of the survey respondents and the survey response rates (averaging 64% over time) were quite comparable during the 6 years of survey implementation.13

The surveys included a series of screening questions to verify that respondents were voluntary disenrollees, questions about reasons for leaving their former health plan, questions asking respondents to rate their former health plan on the care they received from the plan and their overall experience with the plan, and questions about health status and demographic characteristics. The surveys provided information to 3 major constituents:- CMS, to aid in fulfillment of its legislative mandate to present disenrollment rates to Medicare beneficiaries and to help monitor the quality of the services for which CMS contracts.

- Medicare beneficiaries, to help them make more informed health plan choices.

The sampling frame for each annual survey consisted of all Medicare beneficiaries who had voluntarily disenrolled from a managed care organization under contract with CMS during the calendar year. To be included in the survey, health plans were required to have been in operation for at least 1 full year prior to the beginning of the survey year.We did not oversample members of any particular subgroup, such as dual eligibles or the disabled. There was some overlap in these 2 groups. About 12% of the sample were dual eligibles, 3.5% of the sample were dual eligibles and disabled/under age 65 years, and another 6.5% were not dual eligibles but were disabled/ under age 65 years (so about 10% were disabled/under age 65 years). These proportions were quite stable over time. However, there was an increase over time in the proportion of the sample who were Hispanic (from 8% in 2000 to 21% in 2005) and African American (from 11% in 2000 to 16% in 2005), which reflects changes in the enrollment populations across the plans sampled.

We used descriptive statistics from these 6 annual surveys to assess whether the survey responses were consistent with our hypotheses. We first calculated the weighted proportion of respondents who cited 1 of 4 information-related reasons for disenrollment (Figure 2) or 1 of 6 reasons related to drug coverage, premiums, or costs (Figure 3). (Sample design weights in the CAHPS survey account for differences in the proportion of plan members surveyed across plans of different sizes.) Given the large sample size, these sample proportions were viewed as robust point estimates of these prevalences within the disenrollee population each year, for each of the reasons cited.

We recognized that vulnerable groups include several subpopulations.15,22,23 For our analysis, we defined this vulnerable population as consisting of the under-age-65 and disabled group, the oldest-elderly subgroup (age =80 years), beneficiaries dually eligible for Medicare and Medicaid (ie, dual eligibles), racial or ethnic minority groups (eg, blacks, Hispanics), and persons with worse self-reported health (fair/poor rather than good). These vulnerable subgroups together comprised about 40% of the disenrollee survey respondent sample in 2000, rising to about 60% of the sample in 2005.13

Table 1 and Table 2 show the propensities of each of the above-mentioned vulnerable subgroups to cite each of the information- or cost-related reasons. The full-sample propensities (those shown in Figures 2 and 3) are highlighted in the tables as benchmarks. These tables allowed us to assess observed disparities between vulnerable subgroups and others over time.


Figure 2 shows the trend pattern for information problems for the sample as a whole (ie, the benchmark). In Figure 2 we see that problems with information about the plan first exhibited a declining pattern (up to about 2003) and then an increasing pattern. This suggests that for the sample population as a whole, plan-related information problems worsened after the passage of the MMA, which was followed by a proliferation of both the number and types of plans. This is consistent with findings from the literature that, during this period, beneficiaries did not have good knowledge regarding Medicare plan choices and options.24-30

Although Figure 2 shows trends in information problems for the sample as a whole, it masks important disparities across the sample population subgroups. Table 1 shows the propensities by vulnerable subgroup to cite each of the information reasons. The data show that, with the exception of the oldestelderly cohort, the propensity to cite every plan-related information reason was generally higher among the vulnerable subpopulations than among the sample as a whole. For example, the reason “thought you were given incorrect/incomplete information,†had a benchmark in 2000 of about 13%, whereas 19.5% of dual eligibles cited this reason (about a 6.5% disparity). In 2005, the benchmark for that reason was 16.6%, whereas 24.4% of dual eligibles cited this reason (an almost 8% disparity). This suggests that these vulnerable subgroups had more problems with information than the average sample member in all time periods, particularly after 2003. There were some notable differences in the proportions between the African American and Hispanic subpopulations. African Americans had higher propensities to cite problems with incorrect/incomplete information than Hispanics in every year, and also were generally more likely to cite “plan wasn't as expected.†The reason “plan information was not helpful†was cited at about the same rate by Hispanics and African Americans. Hispanics were more likely to cite “plan's customer service staff were not helpful†than African Americans during 2000-2003, but African Americans appeared more likely than Hispanics to cite this problem at the end of the period.

Next, we look at experiences with cost or benefits as reasons for disenrollment. We see in Figure 3 that the propensities to cite reasons related to costs, drug coverage, or other benefits generally declined after 2002 for the sample as a whole (ie, the benchmark). However, the tendency to leave a plan because the plan charged a monthly premium declined throughout the period, from about 40% in 2001 to 16% in 2005, suggesting that “zero premium†MA plans were becoming increasingly more common. After about 2002, there was a decline in the propensity to cite problems with increased copayments for office visits and problems with not being able to pay the monthly premium. These trends are consistent with the increased competition among plans, which were receiving higher payment rates from Medicare and using these to enhance drug coverage benefits and reduce beneficiary cost-sharing.2

Although Figure 3 shows trends for the sample as a whole (the benchmark) in the propensities to cite problems related to cost or coverage, it also masks important disparities between more and less vulnerable groups. Table 2 shows the propensities to cite each of the cost or benefits reasons by vulnerable subgroup, with the full-sample propensities given as a benchmark. The propensity to cite problems with paying the monthly premium fell over time among all vulnerable subgroups, but remained higher than that among the overall sample. However, the disparity diminished over time. For example, in 2001, for the reason “could not pay monthly premium,†the benchmark was about 29%, whereas 48% of dual eligibles cited this reason (about a 19% disparity). In 2005, the benchmark for that reason had fallen to 14%, whereas 20% of dual eligibles cited this reason (about a 6% disparity). Thus, disparities between dual eligibles and others in the ability to pay the monthly premium diminished as plan choices expanded.

The only subgroups with a persistently higher propensity to cite increased copayments for office visits as a reason for disenrolling (relative to the benchmark) were persons under age 65 years and disabled and persons in worse health. However, this disparity declined toward the end of the period, suggesting that some benefits from competition were reaped by these groups as well as by the other vulnerable beneficiaries. By contrast, the under-age-65 and disabled group showed the greatest disparity in the propensity to cite problems related to plans not covering medications (relative to the benchmark), and these disparities did not diminish after 2003. Together, these findings suggest that the vulnerable subgroups may have seen less benefit than the sample as a whole from changes occurring as a result of the increased competition.


This work is timely because the annual MA plan voluntary disenrollee surveys ended in 2005, 2 years after the first expansion and immediately prior to the largest expansion in plan choices and options. Given the changes in MA markets following implementation of the MMA, we had several expectations regarding how the pattern of survey responses would reflect market activity. All 3 hypotheses were confirmed.

Our first expectation was that more vulnerable subgroups (eg, the disabled, dual eligibles, the elderly, those in worse health, minority populations) would have a greater number of problems overall compared with less vulnerable groups throughout the period. This expectation was verified as noted below for both the information- and cost-related reasons for disenrollment, with one exception. The oldest-elderly (age =80 years) group was generally less likely than the sample as a whole to cite any of these problems.

Our second expectation was that the number of beneficiaries citing cost or drug coverage/ benefits as reasons for disenrolling from their MA plan would decrease after implementation of the MMA. This expectation was consistent with the fact that more generous payments to plans and physicians following implementation of the MMA allowed plans to compete more vigorously on benefits and coverage. This hypothesis was confirmed by the survey data, as the number of beneficiaries citing problems related to out-of-pocket costs or drug coverage diminished after 2003. We also found diminishing disparities over time between more and less vulnerable groups' ability to pay the monthly premium.

Our findings suggest that the increased competition in the MA market helped narrow the disparity gap. The 2 subgroups with the highest utilization of doctor visits and healthcare resources–those under age 65 years and disabled, and beneficiaries in worse health–showed a persistently higher propensity to cite increased copayments for office visits as a reason for disenrolling, but this disparity diminished toward the end of the period. However, the under-age-65 and disabled group also showed the greatest disparity in the propensity to cite problems related to plans not covering prescription drugs (relative to the benchmark), and this disparity did not diminish over the period. Taken together, these findings suggest that cost issues for vulnerable subgroups, while improving, are somewhat more constraining than cost issues for less vulnerable groups.

Our third expectation was that the number of beneficiaries citing problems with information as reasons for disenrollment would increase after implementation of the MMA. In 2003, the number of plan choices increased and a new type of plan–the PPO–was introduced as an option. Also, the Part D drug plan was announced as a forthcoming event, and there was considerable media coverage and beneficiary interest in what the new coverage would entail. In this rapidly changing environment, beneficiaries needed good, plan-specific information to make informed choices. We found that plan-related information issues seemed to be cited more frequently after passage of the MMA, especially among the more vulnerable subgroups. The results suggest that information campaigns aimed at Hispanic populations may have been relatively successful, in that the Hispanic group grew to compose a larger proportion of the sample over time, yet this group cited fewer information problems on average than other vulnerable groups. However, other vulnerable groups (African Americans, those dually eligible for Medicare and Medicaid, persons with disabilities) had generally higher propensities to cite information problems than all sample members (ie, the benchmark) in all time periods. These findings suggest that there were still opportunities for improvement in the information comprehended by many in vulnerable subgroups at the eve of Part D implementation.

Although CMS has historically provided beneficiaries with information about coverage options in several forms, further efforts are still needed, particularly with regard to improving the flow of plan-specific comparative information to these most vulnerable subgroups. At present, CMS is planning on making available comparative plan-level summary data for the MA and Freestanding Prescription Drug Plans in a consumerfriendly format. These comparison data will include several new composite measures displaying overall plan performance as well as cost and benefit information on the same page, which will allow for 1-page side-by-side plan comparisons. This comparative data also will be easy to locate on the Web site, which also will provide the capacity to drill down on these new measures for more detailed information (A. Heller, PhD, oral communication with CMS staff March 2007). In addition to CMS's efforts, support from community groups that provide more personal, one-on-one counseling is perhaps needed, to help vulnerable beneficiaries make use of all the comparative information provided.

In conclusion, to ensure continued access to and utilization of the many plans that exist today in the very competitive MA market, efforts must continue to direct informational materials to all beneficiaries, particularly those in vulnerable subgroups. If that happens, the elderly will have increased opportunities to receive better benefits at lower cost, and the premise of the MMA–to increase options and competition, thus improving quality–will be fully realized. These efforts are especially important with the new benefits offered under Part D plans–which are quite different from anything previously offered by Medicare, with choices more numerous than ever before.

2. Achman L, Harris L. Early effects of the Medicare Modernization Act: benefits, cost sharing, and premiums of Medicare Advantage Plans, 2005. Research report. April 2005. AARP Public Policy Institute publication 2005-02. Available at: Accessed November 28, 2007.

4. Lambrew J, Davenport K. Has Medicare been privatized? Implications of the Medicare Modernization Act, beyond the drug benefit. Center for American Progress. February 8, 2006. Available at: Accessed April 1, 2007.

6. Peterson S, Gold M. Tracking Medicare Health and Prescription Drug Plans: Monthly Report for January 2006. Kaiser Family Foundation. February 1, 2006. Available at: Accessed November 28, 2007.

8. Iyengar S, Lepper M. When choice is demotivating: can one desire too much of a good thing? J Pers Soc Psychol. 2000;79:995-1006.

10. Hibbard JH, Slovic P, Peters E, Finucane ML, Tusler M. Is the informed-choice policy approach appropriate for Medicare beneficiaries? Health Aff. May/June 2001. Available at: Accessed April 1, 2007.

12. Hall J, Kurth N, Moore J. Transition to Medicare Part D: an early snapshot of barriers experienced by younger dual eligibles with disabilities. Am J Manag Care. 2007;13:14-17.

14. Mobley L, McCormack L, Booske B, et al. Voluntary disenrollment from Medicare managed care: market factors and disabled beneficiaries. Health Care Financ Rev. 2005;26:45-62.

16. Biles B, Nicholas L, Guterman S. Medicare Beneficiary Out-of-Pocket Costs: Are Medicare Advantage Plans a Better Deal? New York: Commonwealth Fund; May 2006. Publication 927. Available at: Accessed October 30, 2007.

18. Pollack R. Low-income enrollment deeply disappointing. Testimony before the House Committee on Ways and Means. June 14, 2006. Available at: Accessed November 28, 2007.

20. Goldstein E, Teichman L, Crawley B, Gaumer G, Joseph C, Reardon L. Lessons learned from the National Medicare & You education program. Health Care Financ Rev. 2001:23:5-20.

22. Vulnerable populations: who are they? Am J Manag Care. 2006; 12(13 suppl):S348-352.

24. Cihak J. When You’re 64: What Consumers Don’t Know About Medicare. California Health Care Foundation Issue Brief. February 2006. Available at: Accessed October 30, 2007.

26. Government Accountability Office. Medicare: Communications to Beneficiaries on the Prescription Drug Benefit Could Be Improved. Report to Congressional Requesters. Washington, DC: Government Accountability Office; May 2006. Report GAO-06-654.

28. Medicare Payment Advisory Commission. Report to Congress: Increasing the Value of Medicare. June 2006. Available at: Accessed November 28, 2007.

30. Uhrig JD, Bann CM, McCormack LA, Rudolph N. Beneficiary knowledge of original Medicare and Medicare managed care. Med Care. 2006;44:1020-1029.