Abstract
This study tests whether the likelihood of hospital readmission within 30 days of discharge is different for enrollees in Medicare Advantage plans versus the standard fee-for-service program. A key requirement is to control for self-selection into Advantage plans. The study uses statewide inpatient databases maintained by the Agency for Healthcare Research and Quality for five states in 2006. The type of Medicare coverage is known, along with an encrypted patient identifier. We identify eligible first discharges and the first readmission within 30 days. We use selected area characteristics as instrumental variables for enrollment in Advantage plans and apply a bivariate probit analysis. Descriptively, there is a slightly lower likelihood of readmission for Advantage plan enrollees. However, the Advantage plan patients are younger and less severely ill. After risk adjustment and control for self-selection, the enrollees in Advantage plans have a substantially higher likelihood of readmission. Recognizing caveats and limitations, the study supports informing Medicare beneficiaries about the rates of readmission for Advantage plans in their area. Analytical methods to adjust for self-selection into particular plans or plan types should be considered when possible.