Medicare Advantage Enrollment and Disenrollment Among Persons With Alzheimer Disease and Related Dementias

This cross-sectional study assesses enrollment and disenrollment patterns for Medicare Advantage plans from 2013 to 2018 among beneficiaries with complex health conditions, including Alzheimer disease and related dementias.


Introduction
More than 6 million individuals 65 years or older in the US are living with Alzheimer disease and related dementias (ADRD), 1 for whom total payments for health care in 2021 were estimated at $355 billion. 2,35][6][7] Annual mean per capita spending for beneficiaries with ADRD is estimated to be more than 3 times the mean costs for beneficiaries without ADRD, with out-of-pocket costs that are at least twice as high. 6,8,9ditionally, persons with ADRD experience more potentially avoidable hospitalizations and are likely to benefit from care management programs that are designed to prevent avoidable admissions. 10e Medicare Advantage program may be one avenue to improve the efficiency of health spending for persons with ADRD.Medicare Advantage plans receive capitated payments to manage care for their enrollees and therefore have financial incentives to improve the value of care delivery.
Unlike traditional Medicare, Medicare Advantage plans must provide an annual patient out-of-pocket cost limit.They also frequently offer additional plan benefits (eg, nonemergency transportation) for services that are not typically covered by Medicare but may be valued by persons with ADRD and their caregivers. 11,12Moreover, enrollment in Medicare Advantage plans may be incentivized to encourage participation from healthier beneficiaries, on average, and restrict care that will not diminish overall quality. 13Prior research has identified that people with complex health care needs are disproportionately more likely to disenroll from Medicare Advantage in favor of traditional Medicare. 14,15Higher rates of voluntary disenrollment can be indicative of poor beneficiary experience and are important to evaluate as a signal of program success in meeting the needs of beneficiaries enrolled in the Medicare Advantage program.
Enrollment in Medicare Advantage is projected to surpass traditional Medicare enrollment nationally in 2023. 16Despite this tremendous growth, little is known about enrollment and disenrollment patterns over time among beneficiaries with complex health conditions, particularly ADRD.Prior studies of disenrollment evaluated data from a limited number of Medicare Advantage payers 17 or a single year. 18National estimates characterizing Medicare Advantage enrollment among beneficiaries with ADRD are unavailable.
Differences in disenrollment by health status can indicate that a plan is not equipped to meet the needs of patients living with a particular condition.To the extent that differences in disenrollment are observed by health status, there may be cause for concern that Medicare Advantage plans are not serving the needs of the entire Medicare population.In this study, we aimed to evaluate patterns in Medicare Advantage enrollment and disenrollment among beneficiaries with or without ADRD between 2013 and 2018.

Methods
The Brown University Institutional Review Board approved the cross-sectional study and waived the informed consent requirement owing to the study's inability to contact enrollees with deidentified

Measures
The 2 primary study outcomes were Medicare Advantage enrollment and disenrollment.We defined enrollment in Medicare Advantage as enrollment by December in each study year.We also characterized enrollment by other definitions, such as at least 1 month of enrollment, enrollment for most months of the year, and full enrollment, and the findings remained consistent.We defined disenrollment from Medicare Advantage as enrollment in December of the baseline year and enrollment in traditional Medicare in December of the following year.
For disenrollment analyses, the eligible population for study included enrollees in a Medicare Advantage contract in December of a baseline year who remained alive in December of the subsequent year, who did not move residence (as reflected by a change in the county indicator from the MBSF across years; <4% per year), and whose baseline year contract was not terminated administratively (<0.05% per year).For each study year, we assessed whether enrollees in Medicare Advantage in December of a baseline year subsequently (1)   3.9%-3.9%) of beneficiaries without ADRD in unadjusted analyses.After adjusting for demographic characteristics, health status, and utilization, the difference in disenrollment rates remained significantly higher among beneficiaries with ADRD.In adjusted analyses, 5.0% (95% CI, 4.9%-5.0%) of beneficiaries with ADRD disenrolled to traditional Medicare compared with 4.0% (95% CI, 4.0%-4.0%) of beneficiaries without ADRD.Between 2013 and 2018, disenrollment rates ranged from 4.4% (95% CI, 4.4%-4.5%)to 6.1% (95% CI, 6.1%-6.2%)for beneficiaries with ADRD and from 3.2% (95% CI, 3.2%-3.2%)to 4.8% (95% CI, 4.8%-4.8%)for beneficiaries without ADRD.From 2017 to 2018, beneficiaries with ADRD who were enrolled in Medicare Advantage were 1.4 times as likely to disenroll to traditional Medicare compared with beneficiaries without ADRD (4.4% vs 3.2%; P < .001)(Figure 2).
Beneficiaries with ADRD had higher rates of disenrolling to traditional across all racial and ethnic categories and dual-eligibility enrollment status.Disenrollment rates among beneficiaries with ADRD ranged from 4.9% (95% CI, 4.8%-4.9%)among non-Hispanic White beneficiaries to

Discussion
This study presented 3 key findings.First, from 2013 to 2018, there was substantial growth in enrollment in Medicare Advantage among beneficiaries with ADRD, which was similar in magnitude to the increase in enrollment among beneficiaries without ADRD.Second, beneficiaries with ADRD were more likely to leave their Medicare Advantage contract compared with beneficiaries without ADRD.Beneficiaries with and without ADRD had similar rates of switching contracts within the Medicare Advantage program, but during the study period, beneficiaries with ADRD were more likely to disenroll to traditional Medicare.Third, there were differences in contract exit rates associated with a beneficiary's race and ethnicity and dual-eligibility enrollment status for Medicaid.
Hispanic beneficiaries with ADRD had the highest rate of exiting their Medicare Advantage contract compared with their counterparts in other racial and ethnic groups.This finding could be associated with the greater likelihood of lower-income populations to enroll in lower-performing Medicare Advantage plans [21][22][23] or plans with more restrictive benefits or networks, which are often less expensive for beneficiaries to purchase.The findings for individuals with dual-eligibility enrollment in Medicaid are aligned with results of prior work showing higher rates of disenrollment for this population, particularly among beneficiaries with acute or postacute care utilization. 14is study builds on prior work by evaluating a longer timespan to better understand the changes over time in Medicare Advantage enrollment and disenrollment for beneficiaries with ADRD and potential differences based on dual-eligibility enrollment status and race and ethnicity.A study 17 using data from 3 Medicare Advantage payers found higher all-cause disenrollment (switching plans, disenrolling to traditional Medicare, and death) among beneficiaries with ADRD but could not disaggregate the reasons for disenrollment due to data limitations.Another study 18 found higher disenrollment rates among beneficiaries with ADRD and higher rates of disenrollment among beneficiaries with poorer cognitive functioning but was only able to evaluate 1 year of data.
Despite similar patterns in increasing Medicare Advantage enrollment, beneficiaries with ADRD enrolled in a Medicare Advantage contract had consistently higher rates of contract exit and switch to traditional Medicare compared with beneficiaries without ADRD.This finding suggests that for a proportion of beneficiaries with ADRD, their Medicare Advantage contract may not meet their needs.
This finding was contrary to that of a prior study that used a smaller sample of Medicare beneficiaries and reported that beneficiaries with ADRD who were enrolled in Medicare Advantage had lower rates of health care utilization that did not adversely affect care satisfaction and health status, but the study did not evaluate disenrollment. 24Beneficiaries with ADRD and their caregivers may face challenges in navigating the narrow networks common in the Medicare Advantage program as well as more complicated plan designs that may require prior authorization and more administrative barriers to access care. 25Given the heterogeneity of plan design and the volume of plans available in some US counties, it is possible that beneficiaries with ADRD and their caregivers may face substantial challenges in identifying a suitable Medicare Advantage plan; beneficiaries with ADRD who lack sufficient social support may be further disadvantaged.The findings of this study are consistent with those of prior work demonstrating that persons with complex health care needs have higher rates of disenrolling from Medicare Advantage in favor of traditional Medicare. 14,15

Limitations
This study has 3 main limitations.First, our ability to identify beneficiaries with ADRD in the Medicare program relied on beneficiaries' use of hospital, nursing home, home health, or inpatient rehabilitation care.This means that the study population was likely to be sicker on average (regardless of ADRD status), and therefore the results may not be generalizable to all Medicare beneficiaries.Second, we lacked information on beneficiaries' specific reasons for Medicare Advantage enrollment to confirm the potential mechanisms behind these enrollment patterns over time.Third, we lacked information on beneficiaries' cohabitation or marital status, which may be an important factor in both enrollment and disenrollment decisions.

Conclusions
In this repeated cross-sectional study, we identified a pattern of growth in Medicare Advantage enrollment among Medicare beneficiaries with ADRD that mirrors the increasing enrollment pattern among beneficiaries without ADRD.Across all study years, beneficiaries with ADRD were consistently more likely to disenroll from Medicare Advantage or leave their existing Medicare Advantage contract compared with persons without ADRD.These findings highlight the urgency to better understand the factors associated with higher disenrollment rates for beneficiaries with ADRD and to determine whether higher disenrollment rates reflect challenges with access to care or quality of care for this high-cost, high-need population.Policy responses that heighten the oversight of plans with higher rates of disenrollment among enrollees with chronic illness (in addition to overall measures of disenrollment) and increased oversight and enforcement of existing network adequacy rules are essential to ensure that the Medicare Advantage program is meeting the needs of all enrollees.Further research is needed to identify Medicare Advantage plans that can best support the dynamic needs of persons with ADRD and their caregivers.
19dicare Advantage Enrollment and Disenrollment in Persons With Alzheimer Disease and Related DementiasWe used 6 sources of individual-level national data for Medicare Advantage and traditional Medicare beneficiaries from January 1, 2011, to December 31, 2018.The Medicare Provider Analysis and Review file provided information on inpatient stays in acute care hospitals.The Minimum Data Set provided information on nursing home care for all patients who were admitted to nursing homes.The These data sets are collected for both Medicare Advantage and traditional Medicare enrollees and enable the study of a representative population of beneficiaries across both programs.19Each of these data sources includes diagnosis codes that reflect beneficiary health status and can be used to identify groups of beneficiaries based on the presence of a health condition.In this study, we focused on persons with an ADRD diagnosis present in any of the utilization data sets.

Table 1 .
disenrolled from Medicare Advantage to enroll in traditional Medicare, (2) stayed in Medicare Advantage but enrolled in a different Medicare Advantage contract, or (3) remained in the same Medicare Advantage contract in January of the following year.A secondary outcome was contract exit, defined as either exiting an existing Medicare Advantage contract for a different Medicare Advantage contract or for traditional Medicare in the following year.Covariates derived from the MBSF included age, sex, race and ethnicity (self-reported as Asian or Pacific Islander, Black, Hispanic, non-Hispanic White, and other [including American Indian or Alaska Native, other, or unknown]), dual-eligibility enrollment status, and reason for entitlement.Race and ethnicity data were collected and included in analysis because enrollment in Medicare Advantage has varied according to racial and ethnic identity.Other covariates were derived from the utilization data sources to reflect additional indicators of health status (indicators for diabetes, heart failure, chronic obstructive pulmonary disease, acute myocardial infarction, and schizophrenia and other serious mental illness) and binary indicators of inpatient, nursing home, and home health utilization in each study year.The study included 32 796 872 unique Medicare beneficiaries with acute or postacute care utilization between 2013 and 2018; of these beneficiaries, 18 228 513 were females (55.6%) and 14 568 359 were males (44.4%) with a mean (SD) age of 74.0 (12.5) years.For each study year, there were approximately 16 529 641 beneficiaries meeting the overall inclusion criteria, and a mean (SD) 14.5% (0.1%) of the eligible study population in each year had evidence of ADRD.Beneficiaries with ADRD Sample Characteristics of Medicare Enrollees by ADRD Status for 2013 and 2018 beneficiaries with ADRD, unadjusted enrollment in Medicare Advantage increased by 9.5 percentage points from 22.6% in 2013 to 32.1% in 2018.Unadjusted enrollment in Medicare Advantage among those without ADRD increased by 8.0 percentage points from 28.0% to 36.0% over the study period.Among those with ADRD, adjusted enrollment in Medicare Advantage increased from 24.7% (95% CI, 24.7%-24.8%) in 2013 to 33.0% (95% CI, 32.9%-33.1%) in 2018, an absolute increase of 8.3 percentage points and a 33.4% relative increase over the study period.
After defining the cohort, we descriptively compared the characteristics of beneficiaries enrolled in Medicare Advantage vs traditional Medicare by ADRD status.We then evaluated enrollment in Medicare Advantage as a proportion of the total eligible population of beneficiaries with or without ADRD who were enrolled in a Medicare Advantage contract in December of each study year (2013 to 2018).We used Medicare Advantage enrollment indicators from the MBSF, adjusting for potential differences in age, sex, race and ethnicity, dual-eligibility enrollment status, additional health conditions, and acute and postacute care utilization in a multivariable linear regression model with county fixed effects to account for the differential opportunity to enroll in Medicare Advantage by county based on plan offerings.Beneficiary-level demographic characteristics, health status indicators, and utilization indicators were included to adjust for differences based on ADRD status.We then plotted patterns in adjusted enrollment in Medicare Advantage over time for beneficiaries with or without ADRD.We also estimated overall growth in enrollment using a pooled multivariable linear regression model with year fixed effects.We estimated adjusted disenrollment rates using a multinomial logistic regression model, including covariates for age, sex, race and ethnicity, dual-eligibility enrollment status, health status, and utilization indicators for each study period(2013-2014, 2014-2015, 2015-2016, 2016-2017, and   2017-2018), among beneficiaries enrolled in Medicare Advantage in the baseline year who remained alive in the following enrollment year.We calculated the adjusted overall disenrollment rate for the entire study period separately in a pooled multinomial logistic regression model that also included indicators for each study period.In sensitivity analyses, we evaluated a binary outcome for contract exit (stayed vs exited the Medicare Advantage contract) by dual-eligibility enrollment status, sex, and race and ethnicity while controlling for all other demographic variables, such as age and other health status and utilization indicators.In further sensitivity analyses, we assessed the data sources of ADRD diagnoses (eFigure 2 in Supplement 1) and tested alternative look-back periods for an ADRD diagnosis (eFigure 3 in Supplement 1).benefits;ESKD, end-stage kidney disease; OASI, Old Age and Survivors Insurance; SMI, schizophrenia and other serious mental illness.aRaceandethnicity were self-reported by beneficiaries and obtained from the Medicare Beneficiary Summary File.bOther included American Indian or Alaska Native, other, or unknown.cUtilization reflects beneficiaries with any utilization during the study year and the prior 2 years (eg, 2013 study population included those with utilization between 2011 and 2013) from the Medicare Provider Analysis and Review, Outcome and Assessment Information Set, Inpatient Rehabilitation Facility Patient Assessment Instrument, and Minimum Data Set.Beneficiaries commonly have multiple utilization events during a period across different data sources.Among

Table 2 .
Estimated Medicare Advantage Enrollment Stratified by Race and Ethnicity and Dual-Eligibility Enrollment Status by Alzheimer Disease and Related Dementias (ADRD) Status for 2013 and 2018 a Estimates were calculated using margins following estimation of a multivariable linear regression model, including covariates for age; sex; indicators for diabetes, heart failure, chronic obstructive pulmonary disease, acute myocardial infarction, and schizophrenia and other serious mental illness; and yearly indicators for inpatient, nursing home, and home health utilization.bRaceand ethnicity data relied on the Research Triangle Institute race variable in the Medicare Beneficiary Summary File, which enhances the beneficiary race code variable to more accurately identify beneficiaries as Hispanic or Asian based on surname and other geographic indicators.cOther included American Indian or Alaska Native, other, or unknown.
Figure 2. Estimates of Adjusted Percentage of Medicare Advantage Enrollees Disenrolling to Traditional Medicare by Alzheimer Disease and Related Dementias (ADRD) Status From 2013 to 2018 Beneficiaries had similar rates of changing their contract but staying in Medicare Advantage (11.1%) regardless of ADRD status.Given the similarities in the proportion of beneficiaries switching to a new contract within the Medicare Advantage program, the difference in the contract exit outcome by ADRD status was primarily associated with differences in disenrollment to traditional Medicare.Over the study period, 16.3% (95% CI, 16.2%-16.3%)of beneficiaries with ADRD exited their Medicare Advantage contract compared with 15.1% (95% CI, 15.1%-15.1%)among beneficiaries without ADRD (eTable 3 in Supplement 1).We observed variation in the contract exit rate by beneficiary race and ethnicity.Hispanic beneficiaries with ADRD had the highest contract exit rate at 18.2% (95% CI, 18.1%-18.2%),which was 1.2 times greater than the rate of contract exit among beneficiaries with ADRD with the lowest contract exit rate (eTable 4 in Supplement 1).For Figure 3. Estimates of Adjusted Percentage of Medicare Advantage Enrollees Exiting Medicare Advantage Contracts by Alzheimer Disease and Related Dementias (ADRD) Status From 2013 to 2018