Out-Of-Network Utilization and Plan Selection Among Medicare Advantage Cost Plan Enrollees.

objectiveTo understand how Medicare Advantage (MA) networks impact utilization patterns and plan choices, using the 2019 discontinuation of MA 1876 Cost plans as a natural experiment. STUDY SETTING AND

designWe study 1876 Cost plans, MA plans for which out-of-network care is covered through traditional Medicare (TM) and many of which CMS discontinued in 2019. We characterize the proportion of Cost plan enrollees who utilized out-of-network care in 2018 from different types of medical specialties. We then study how enrollees in discontinued plans selected into new plans in 2019. We use regression analysis to characterize whether higher risk enrollees selected into TM at higher rates. DATA SOURCES AND ANALYTIC SAMPLE: We identify discontinued plans using public MA plan data. We employ administrative Medicare enrollment and TM claims data to identify 2018 enrollees of discontinued plans, their 2018 out-of-network utilization, and their subsequent 2019 enrollment decisions. PRINCIPAL

findingsAmong Cost plan enrollees, 69% utilized non-emergency room related care out of network in 2018. Out-of-network utilization was distributed across several types of specialties: 43% of Cost plan enrollees had at least one out-of-network claim with a primary care physician and over 20% had a claim with a medical specialist, surgical specialist, or nurse practitioner. We find evidence of adverse selection among enrollees of discontinued Cost plans in 2019. Conditional on one's 2018 Cost plan and county of residence, a standard deviation increase in risk score was on average associated with a 26.35% (95% CI, 25.57%-27.12%) increased likelihood of enrolling in TM.

conclusionThe high rate of out-of-network utilization suggests that MA enrollees value access to care outside of standard MA networks. Subsequent selection patterns indicate that preferences for broader networks and subsequent enrollment in TM is highest among higher risk enrollees, suggesting limited networks may induce extensive margin selection.