Disenrollment from Medicare HMOs

Background: Since the program's inception, there has been great interest in determining whether beneficiaries who enter and subsequently leave Medicar e health maintenance organizations (HMOs) are more or less costly than thos e remaining in fee-for-service (FFS) Medicare. Objectives: To examine whether relatively high-cost beneficiaries disenroll from Medicare HMOs (disenrollment bias) and whether disenrollment bias var ies by Medicare HMO market characteristics, In addition, we compare rates o f surgical procedures and hospitalizations for ambulatory care-sensitive co nditions for disenrollees and continuing FFS beneficiaries. Design: Cross-sectional analysis of 1994 Medicare data. Participants and Methods: Medicare beneficiaries were first sampled from th e 124 counties with at least 1000 Medicare HMO enrollees. From this pool, H MO disenrollees and a sample of continuing FFS beneficiaries were drawn. Th e FFS beneficiaries were assigned dates of "pseudodisenrollment." Expenditu res and inpatient service use were compared for 6 months after disenrollmen t or pseudodisenrollment. Results: The HMO disenrollees were no more likely than the continuing FFS b eneficiaries to have positive total expenditures (Part A plus Part B) or Pa rt B expenditures in the first 6 months after disenrollment. However, disen rollees were more likely to have Part A expenditures. Among beneficiaries w ith spending, disenrollees had higher total and Part B expenditures than co ntinuing FFS beneficiaries. Moreover, the disparity in total and Part B spe nding between disenrollees and continuing FFS beneficiaries increased with HMO market penetration. Although Part A spending was higher for disenrollee s with spending, it was not sensitive to changes in market share. The HMO d isenrollees received more surgical procedures and were hospitalized for mor e of the ambulatory care-sensitive conditions than the FFS beneficiaries. Conclusions: On several measures, Medicare HMOs experienced favorable disen rollment relative to continuing FFS beneficiaries as recently as 1994, whic h increased as HMO market share increased.