The role of networks and serious illness in Medicare Advantage disenrollment

Abstract

Medicare Advantage (MA) plans provide the Medicare benefit to beneficiaries through private plans. These plans are paid capitated payments to deliver care to their enrollees and thus use various strategies to control the cost of providing care. Understanding whether the MA program meets the needs of enrollees is important for beneficiaries with serious illnesses. We used novel data on MA networks from Ideon and data from Surveillance, Epidemiology, and End Results (SEER)-Medicare to answer various questions about the MA program for enrollees needing specialty care. We examined how changes in special enrollment rules that gave beneficiaries more flexibility in changing their MA coverage effected MA disenrollment for beneficiaries with new cancer diagnoses. Using 2016-2019 SEER-Medicare data, we compared outcomes among beneficiaries diagnosed with cancer in January and March (treatment groups) to those diagnosed in April (control group). We found that more flexible enrollment rules had no effect on beneficiaries diagnosed with cancer in January but led to small increases in switching coverage for beneficiaries diagnosed in March. We also wanted to understand how network breadth for specialized cancer care was associated with changes in MA coverage after a new cancer diagnosis. The analysis stratified by three plan types: non-employer coverage; zero premium, non-employer coverage; premium charged, and employer coverage. Using data from SEER-Medicare (2019-2020) and Ideon (2019), we find that increases in network breadth are associated with decreases in the probability of switching coverage for enrollees in non-employer plans that charged a premium. There was no association between network breadth and changes in enrollment for those enrolled in zero premium non-employer plans or employer coverage. Lastly, we explored how MA plans construct networks for skilled nursing facility (SNF) care. We used 2022 data from Ideon in this analysis and various publicly available data files to contruct measures for network breadth. Our analysis found that plans included less than half of SNFs in the service area in network.