Medicare Advantage plan characteristics associated with sorting their beneficiaries to providers that generate fewer avoidable hospital stays

Objective: To examine whether certain Medicare Advantage (MA) plan characteristics are associated with driving beneficiaries to providers that generate fewer avoidable hospital stays. Data Sources: This paper primarily used 2018–2019 MA encounter data and traditional Medicare (TM) claims data for a nationally representative 20% sample of Medicare beneficiaries. Study Design: For each plan design aspect—plan type, carrier, star rating, and network breadth—we estimated two adjusted Poisson regressions of avoidable hospital stays: one without clinician fixed effects and the other with. We calculated the difference between the coefficients to evaluate the extent to which patient sorting affected avoidable hospital stays relative to TM. Data Extraction Methods: Our sample included Medicare beneficiaries 65 years and older who were continuously enrolled in either MA or TM during 2018–2019. Beneficiaries in our sample had one or more chronic, ambulatory care-sensitive conditions. Principal Findings: Patient sorting can be attributed to certain characteristics of plan design aspects. For plan type, HMOs account for 86%, with PPOs accounting for only 14%. For carriers, Humana and smaller carriers account for 89%. For star ratings, high-star contracts account for 94%, with other stars only accounting for 6%. By network design, narrow network plan-counties explained 20% of the patient sorting effect. Conclusions: While MA plans were found to be associated with driving beneficiaries to providers that generate fewer avoidable hospital stays, the effect is not homogeneous across the characteristics of MA plans. HMOs and high-star contracts are drivers of this MA phenomenon.