Abstract
This study examines the association of partner Medicare Advantage plan status over 1 year with beneficiary and plan characteristics.
In 2023, 50% of Medicare beneficiaries were enrolled in Medicare Advantage (MA).1 While enrollment is common, so is disenrollment: between 2011 and 2018, 12% to 20% of enrollees left their MA contract within 1 year.2 Beneficiary- and plan-level characteristics, including Medicare-Medicaid dual eligibility, functional limitations, health care needs, and plan star rating, are associated with disenrollment.3,4,5,6 For working-age adults, spouses or partners (hereafter, partners) often share insurance coverage under the same plan; whether this pattern occurs in MA is unknown. This study considered dyadic MA enrollment decisions, examining the association of respondent plan changes with partner plan changes and other beneficiary and plan characteristics.
Methods
We used the 2010 to 2018 waves of the Health and Retirement Study (HRS) with linked annual Medicare Beneficiary Summary File (MBSF) data and publicly available plan characteristics. The HRS is a longitudinal, nationally representative study fielded biennially via face-to-face or telephone interviews of adults aged 51 years or older and household partners in all US states and the District of Columbia; response rates ranged from 74% to 89%. The Michigan Medicine institutional review board approved this study and provided a waiver of informed consent.
We used the HRS to identify respondent dyads. We identified dyad waves in which both partners were aged 65 years or older, enrolled in MA in January of the survey year, and did not change county of residence during the observation year. For a dyad in a given wave, we randomly selected 1 member of the dyad for analysis; if a given dyad member was selected in multiple waves, we randomly selected 1 of those waves. Therefore, a dyad contributed to analysis in no more than 2 waves, with each partner contributing from only 1. Data were weighted using the HRS analytic weights; Stata svy commands were used to incorporate strata of the sample design and clustering around dyads to account for correlation within dyads that were observed twice when calculating SEs of the estimates.
For MA-enrolled dyads at the start of the year, we used the MBSF to determine changes in MA enrollment during the following 12 months (ie, through January of the following calendar year), which we modeled using multinomial logistic regression for the following plan-level outcomes: remaining in the same plan, switching to another MA plan, and disenrolling to fee-for-service (FFS) Medicare. Covariates included partner’s MA enrollment change during the observation year; respondent demographic, socioeconomic, and health characteristics from the selected HRS wave; and MA plan characteristics at the beginning of the observation year (Table 1). To facilitate interpretation, we calculated the predicted probability of MA enrollment change and marginal effects for each characteristic. Statistical significance was set at 2-tailed P < .05; analyses were performed using Stata, version 17.0 (StataCorp LLC).
Table 1. Characteristics of Partnered Respondents and Their Medicare Advantage (MA) Plans by Change in MA Enrollment Statusa.
| Characteristic | Respondents, No. (%)b | P valuec | ||
|---|---|---|---|---|
| Remained in same MA plan | Switched to another MA plan | Disenrolled from MA to FFS | ||
| Sample population | 1410 (79.7) | 352 (17.9) | 50 (2.4) | |
| Partner MA enrollment status change | ||||
| Remained in same MA plan | 1341 (95.5) | d | d | <.001 |
| Switched to another MA plan | 57 (3.9) | 293 (85.6) | d | |
| Disenrolled from MA to FFS | 12 (0.7) | d | 32 (69.0) | |
| Sex | ||||
| Female | 706 (48.9) | 166 (46.7) | 24 (48.6) | .83 |
| Male | 704 (51.1) | 186 (53.3) | 26 (51.4) | |
| Age, y | ||||
| 65-74 | 627 (53.1) | 159 (51.8) | 18 (48.7) | .49 |
| 75-84 | 633 (37.4) | 163 (40.3) | 21 (35.0) | |
| ≥85 | 150 (9.5) | 30 (7.9) | 11 (16.3) | |
| Race and ethnicitye | ||||
| Hispanic | 199 (10.1) | 43 (9.5) | d | .23 |
| Non-Hispanic Black | 130 (5.2) | d | d | |
| Non-Hispanic White | 1048 (81.8) | 270 (80.9) | 34 (71.4) | |
| Otherf | 33 (2.9) | d | d | |
| Nursing home resident | 27 (1.4) | d | d | .002 |
| Medicaid dual eligibility | 112 (6.1) | 34 (6.8) | 13 (28.1) | <.001 |
| Health characteristics | ||||
| Limitations in activities of daily living, No.g | ||||
| 0 | 1123 (82.2) | 284 (82.9) | 32 (71.5) | .11 |
| 1-2 | 187 (12.3) | 50 (12.8) | d | |
| 3-6 | 100 (5.4) | 18 (4.3) | d | |
| Limitations in instrumental activities of daily living, No.h | ||||
| 0 | 1161 (85.1) | 294 (86.8) | 27 (60.4) | <.001 |
| 1-2 | 168 (10.8) | 41 (9.0) | 12 (21.5) | |
| 3-5 | 81 (4.1) | 17 (4.2) | 11 (18.0) | |
| Hospitalization in past 2 y | 436 (28.6) | 93 (25.1) | 15 (30.2) | .51 |
| MA plan star ratingi | ||||
| Unrated | 49 (3.7) | 47 (11.1) | 10 (15.5) | <.001 |
| 2-2.5 | 55 (2.9) | 25 (5.9) | d | |
| 3-3.5 | 453 (28.1) | 178 (47.5) | 24 (57.4) | |
| 4-5 | 853 (65.3) | 102 (35.5) | d | |
| Type of MA plan | ||||
| HMO | 923 (62.8) | 143 (42.7) | 22 (48.9) | <.001 |
| PPO | 392 (30.5) | 140 (40.8) | 14 (26.9) | |
| Otherj | 95 (6.7) | 69 (16.4) | 14 (24.2) | |
| Special needs plan | 66 (4.3) | 26 (5.3) | d | .26 |
| MA plan market share in the countyk | ||||
| Tertile 1: 0%-4.7% | 416 (31.1) | 152 (43.3) | 22 (35.8) | <.001 |
| Tertile 2: 4.8%-15.1% | 473 (33.3) | 116 (32.1) | 16 (43.1) | |
| Tertile 3: 15.2%-100% | 521 (35.6) | 84 (24.5) | 12 (21.0) | |
Data were weighted using Health and Retirement Study (HRS) analytic weights. We also examined respondent education years, total family income, number of family and unpaid caregivers, metropolitan statistical area, comorbidities, and survey years (not reported for brevity).
Of 1812 partnered respondents, 1499 (83.8%) were enrolled in the same MA plan with their partner at the beginning of the survey year.
Rao-Scott χ2 tests for categorical characteristics compared partnered respondents and their MA plans by change in MA enrollment status.
Estimates of <10 cases suppressed per HRS reporting instructions. For variables with ≥3 groups, if the estimate for 1 group was suppressed, the estimate for the group with the second-lowest number of cases was also suppressed to avoid derivation of suppressed estimates.
Race and ethnicity are collected from all HRS respondents and were included because of earlier analyses showing variation in MA disenrollment.2
Alaska Native, American Indian, Asian, Native Hawaiian, Pacific Islander, and other (specific subcategory masked in the publicly available HRS data).
Respondent performance of 0-6: getting in or out of bed, walking across a room, bathing, dressing, eating, and toileting.
Respondent performance of 0-5: shopping, preparing meals, using a telephone, handling banking, and managing medications.
Calculated at the contract level. Unrated plans were too new or too small.
Includes private fee-for-service (FFS), 1876 cost plans, Medicare-Medicaid plans, and provider-sponsored organizations operated by a group of physicians and hospitals that form a network.
The percentage of all MA beneficiaries in the county enrolled in that plan.
Results
Of 1812 partnered respondents, 80% stayed in the same MA plan, 18% switched MA plans, and 2% disenrolled to FFS during the observation year (Table 1). Respondents remaining in the same MA plan were less likely to be dual eligible and had fewer functional limitations.
Respondents’ MA enrollment status mirrored their partners’. If a partner remained in the same MA plan, 95.7% (95% CI, 94.6%-96.9%) of respondents also stayed, 3.5% (95% CI, 2.4%-4.6%) switched to another MA plan, and 0.8% (95% CI, 0.3%-1.2%) disenrolled (Table 2). If a partner switched to another MA plan compared with remaining, respondents were 67.3 (95% CI, 59.6-75.1) percentage points more likely to also switch MA plans. When a partner disenrolled from MA to FFS compared with remaining, respondents were 57.3 (95% CI, 42.1-72.5) percentage points more likely to also disenroll to FFS. The associations of other respondent and plan characteristics with plan switch or disenrollment were either not significant or markedly smaller.
Table 2. Association of Partner Medicare Advantage (MA) Enrollment Status Change and Other Respondent and Plan Characteristics With Respondent Enrollment Status Changea.
| Characteristic | Respondent MA enrollment status change | |||||
|---|---|---|---|---|---|---|
| Predicted probability, % (95% CI) | Marginal effects, percentage points (95% CI) | |||||
| Remained in same MA plan | Switched to another MA plan | Disenrolled from MA to FFS | Remained in same MA plan | Switched to another MA plan | Disenrolled from MA to FFS | |
| Partner MA enrollment status change | ||||||
| Remained in same MA plan | 95.7 (94.6-96.9) | 3.5 (2.4-4.6) | 0.8 (0.3-1.2) | [Reference] | [Reference] | [Reference] |
| Switched to another MA plan | 28.2 (20.7-35.7) | 70.8 (63.3-78.4) | 0.9 (0.1-1.8) | −67.5 (−75.3 to −59.8)b | 67.3 (59.6 to 75.1)b | 0.2 (−0.8 to 1.1) |
| Disenrolled from MA to FFS | 33.7 (18.9-48.5) | 8.3 (2.0-14.5) | 58.0 (43.0-73.1) | −62.0 (−77.0 to −47.1)b | 4.8 (−1.6 to 11.1) | 57.3 (42.1 to 72.5)b |
| Sex | ||||||
| Female | 79.5 (77.2-81.8) | 17.9 (15.7-20.1) | 2.6 (1.6-3.5) | −0.4 (−2.5 to 1.8) | 0.0 (−2.1 to 2.1) | 0.3 (−0.5 to 1.2) |
| Male | 79.9 (77.5-82.3) | 17.9 (15.6-20.2) | 2.2 (1.3-3.1) | [Reference] | [Reference] | [Reference] |
| Age, y | ||||||
| 65-74 | 80.5 (78.0-82.9) | 17.4 (15.0-19.8) | 2.1 (1.1-3.1) | [Reference] | [Reference] | [Reference] |
| 75-84 | 78.5 (76.1-81.0) | 19.0 (16.6-21.3) | 2.5 (1.6-3.4) | −1.9 (−4.4 to 0.5) | 1.6 (−0.8 to 4.0) | 0.4 (−0.5 to 1.3) |
| ≥85 | 80.8 (76.5-85.2) | 16.4 (12.2-20.5) | 2.8 (1.4-4.2) | 0.4 (−4.1 to 4.9) | −1.0 (−5.4 to 3.3) | 0.7 (−0.8 to 2.1) |
| Nursing home resident | ||||||
| No | 79.7 (77.6-81.8) | 17.9 (15.9-19.9) | 2.4 (1.5-3.2) | [Reference] | [Reference] | [Reference] |
| Yes | 79.8 (71.8-87.8) | 17.3 (9.7-24.8) | 2.9 (−0.2 to 6.0) | 0.1 (−7.8 to 8.1) | −0.6 (−8.1 to 6.9) | 0.5 (−2.7 to 3.7) |
| Medicare-Medicaid dual eligibility | ||||||
| No | 80.3 (78.2-82.4) | 17.6 (15.6-19.7) | 2.1 (1.3-2.9) | [Reference] | [Reference] | [Reference] |
| Yes | 73.0 (66.3-79.7) | 21.9 (15.6-28.1) | 5.1 (2.5-7.8) | −7.3 (−14.0 to −0.6)c | 4.3 (−2.0 to 10.5) | 3.0 (0.5 to 5.6)c |
| Limitations in activities of daily living | ||||||
| 0 | 79.2 (77.0-81.4) | 17.8 (15.8-19.9) | 3.0 (1.8-4.1) | [Reference] | [Reference] | [Reference] |
| 1-2 | 79.8 (76.5-83.1) | 18.9 (15.7-22.1) | 1.3 (0.3-2.4) | 0.6 (−2.6 to 3.8) | 1.1 (−2.0 to 4.1) | −1.6 (−3.1 to −0.2)c |
| 3-6 | 83.0 (77.6-88.5) | 15.0 (9.6-20.3) | 2.0 (0.8-3.2) | 3.8 (−1.7 to 9.3) | −2.9 (−8.2 to 2.4) | −1.0 (−2.5 to 0.5) |
| Limitations in instrumental activities of daily living | ||||||
| 0 | 79.1 (76.7-81.6) | 19.2 (16.8-21.6) | 1.7 (0.9-2.4) | [Reference] | [Reference] | [Reference] |
| 1-2 | 81.0 (76.0-86.1) | 14.0 (9.6-18.5) | 4.9 (1.2-8.7) | 1.9 (−3.7 to 7.5) | −5.2 (−10.4 to 0.0) | 3.3 (−0.3 to 6.9) |
| 3-5 | 77.3 (64.6-90.0) | 8.9 (3.0-14.8) | 13.8 (−0.2 to 27.8) | −1.8 (−14.8 to 11.2) | −10.3 (−16.9 to −3.6)d | 12.1 (−1.9 to 26.1) |
| Hospitalization in the past 2 years | ||||||
| No | 79.6 (77.4-81.8) | 17.9 (15.7-20.0) | 2.5 (1.6-3.4) | [Reference] | [Reference] | [Reference] |
| Yes | 79.9 (77.2-82.7) | 18.0 (15.3-20.6) | 2.1 (1.1-3.1) | 0.3 (−2.3 to 2.9) | 0.1 (−2.4 to 2.6) | −0.4 (−1.4 to 0.6) |
| MA plan star rating | ||||||
| Unrated | 81.3 (75.3-87.4) | 16.7 (10.9-22.5) | 2.0 (0.8-3.1) | 3.3 (−4.7 to 11.2) | −1.3 (−8.7 to 6.2) | −2.0 (−4.4 to 0.5) |
| 2-2.5 | 78.1 (72.5-83.7) | 18.0 (13.0-23.0) | 3.9 (1.4-6.5) | [Reference] | [Reference] | [Reference] |
| 3-3.5 | 74.0 (70.6-77.5) | 21.3 (18.4-24.1) | 4.7 (2.4-7.1) | −4.1 (−9.7 to 1.6) | 3.3 (−1.7 to 8.3) | 0.8 (−2.1 to 3.7) |
| 4-5 | 83.4 (81.0-85.9) | 15.6 (13.1-18.0) | 1.0 (0.2-1.7) | 5.4 (−0.6 to 11.3) | −2.4 (−7.7 to 2.9) | −3.0 (−5.8 to −0.1)c |
| MA plan market share in the county, by tertiles | ||||||
| 0%-4.7% | 77.4 (74.6-80.1) | 20.5 (17.8-23.2) | 2.2 (1.2-3.1) | [Reference] | [Reference] | [Reference] |
| 4.8%-15.1% | 80.3 (77.7-82.9) | 16.9 (14.5-19.3) | 2.8 (1.5-4.1) | 2.9 (0.0 to 5.9)c | −3.6 (−6.4 to −0.8)c | 0.6 (−0.6 to 1.9) |
| 15.2%-100% | 81.9 (79.2-84.5) | 15.8 (13.3-18.3) | 2.3 (1.3-3.4) | 4.5 (1.5 to 7.5)d | −4.7 (−7.6 to −1.7)d | 0.2 (−0.8 to 1.2) |
We used multinomial logistic regression to model respondent changes in MA enrollment during the observation year: remaining in the same plan, switching to another MA plan, and disenrolling to fee-for-service (FFS) Medicare. Covariates included partner’s MA enrollment change; respondent demographic, socioeconomic, and health characteristics; and MA plan characteristics presented in Table 1, with respondent education, total family income, number of family and unpaid caregivers, metropolitan statistical area, comorbidities, and year fixed effects. To facilitate interpretation, we calculated the mean predicted probability of MA plan enrollment change associated with each characteristic using estimates from the multinomial logistic regression and calculated the average marginal effects for each characteristic to compare whether the predicted probabilities of MA plan enrollment change differed by each characteristic. Data were weighted using the Health and Retirement Study analytic weights; Stata svy commands were used to incorporate strata of the sample design and clustering around dyads to account for correlation within dyads that were observed twice when calculating SEs of the estimates. For brevity, we report a subset of factors; other factors were either not associated with MA plan enrollment change or had very small marginal effects for the outcome.
P < .001. cP < .05. dP < .01.
Discussion
Among adults aged 65 years or older, individuals’ MA enrollment decisions were associated with those of their partner and the magnitude of this association was greater than for other respondent and plan characteristics examined. Analyses of insurance enrollment of partnered Medicare beneficiaries that do not consider partners are missing an important aspect of plan selection. Policy initiatives to help beneficiaries optimize plan choice should consider that dyads frequently make enrollment choices together. Limitations include that the most recent data were from 2018, both partners were initially MA enrolled, the partner responsible for enrollment decisions could not be identified, only 1 year of coverage was included, and sample sizes in some cells were small.
Section Editors: Kristin Walter, MD, and Jody W. Zylke, MD, Deputy Editors; Karen Lasser, MD, Senior Editor.
Data Sharing Statement
References
- 1.Biniek JF, Freed M, Damico A, Neuman T. Half of all eligible Medicare beneficiaries are now enrolled in private Medicare Advantage plans. KFF. Accessed July 10, 2023. www.kff.org/policy-watch/half-of-all-eligible-medicare-beneficiaries-are-now-enrolled-in-private-medicare-advantage-plans/
- 2.Meyers DJ, Ryan AM, Trivedi AN. Trends in cumulative disenrollment in the Medicare Advantage program, 2011-2020. JAMA Health Forum. 2023;4(8):e232717-e232717. doi: 10.1001/jamahealthforum.2023.2717 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Meyers DJ, Belanger E, Joyce N, et al. Analysis of drivers of disenrollment and plan switching among Medicare Advantage beneficiaries. JAMA Intern Med. 2019;179(4):524-532. doi: 10.1001/jamainternmed.2018.7639 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ankuda CK, Ornstein KA, Covinsky KE, et al. Switching between Medicare Advantage and Traditional Medicare before and after the onset of functional disability. Health Aff (Millwood). 2020;39(5):809-818. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Rahman M, Keohane L, Trivedi AN, Mor V. High-cost patients had substantial rates of leaving Medicare Advantage and joining traditional Medicare. Health Aff (Millwood). 2015;34(10):1675-1681. doi: 10.1377/hlthaff.2015.0272 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Raver E, Jung J, Xu WY. Medicare Advantage disenrollment patterns among beneficiaries with multiple chronic conditions. JAMA. 2023;330(2):185-187. doi: 10.1001/jama.2023.10369 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Sharing Statement