Should I stay or should I go again: Multiple switching between fee-for-service Medicare and Medicare advantage among older beneficiaries

Objective: To evaluate whether having previously disenrolled from Medicare Advantage (MA) is associated with lower hazards of future MA enrollment. Data Sources and Study Setting: Secondary data from Medicare. Study Design: We examined beneficiaries with baseline FFS enrollment from 2017-2019 using a 20% sample of Medicare claims. Cox proportional hazard models were used to examine the association of prior MA enrollment (in the three years prior to baseline FFS enrollment) with MA re-enrollment, and whether this association is modified by Alzheimer's Disease and Related Dementias (ADRD), prior nursing home use, chronic illness, dual eligible status, and availability of MA plans and quality. Data Collection: Not applicable. Principal Findings: Overall, 3.3% of beneficiaries switched to MA annually. Of those with prior MA enrollment, MA switching percentages were 9.0%, 4.6%, and 6.8% for those whose most recent MA enrollments were 1, 2, and 3 years prior to their baseline FFS year. Comparatively, the switching percentages was 3.2% for those with no prior MA enrollment. The hazards of switching to MA were 2.73 (p < 0.001), 1.29 (p < 0.001), and 1.97 (p < 0.001) times greater than remaining in FFS for beneficiaries whose most recent MA enrollments were one, two, and three years prior to their baseline FFS year. Hazards of switching were generally similar between those with and without ADRD, stratified by recency in prior MA experience, except those with dual eligibility. Among those with ADRD, switching hazards were greatest for 3 years prior MA enrollees in counties with the fewest available (HR: 3.84, p < 0.001) and lowest-rated plans (HR: 4.02, p < 0.001). Conclusions: Recency of switching from MA to FFS was the strongest predictor of a FFS-to-MA switch, identifying a population of beneficiaries who multiply switch regardless of health status or MA access. Future health policy considerations should more closely examine the vulnerabilities and long-term outcomes of this population.