A number of research studies have examined the potential use of early readmission rate as an indicator for hospital quality of care. In spite of the volume of research focusing on this issue, there is as yet no evidence that differences in early readmission rates do in fact relate to quality differences among hospitals. The overall objective of the study proposed here is to determine whether early readmission rates,risk- adjusted to account for differences in the clinical characteristics (mix) of hospitals' patients, provide valid information on hospital quality of care. A related objective is to assess alternative definitions for """"""""early;"""""""" we propose to determine whether the validity of relationships between risk-adjusted readmission rates and quality depend on readmission time windows (e.g., 15 days, 30 days), and whether different time windows are appropriate for different clinical conditions. We will also examine risk models that do not depend on specified time windows, but instead estimate readmission risk as a cumulative function of days since discharge. Two data sets covering Medicare hospitalizations in Michigan during 1989- 1991 will be used in the study. For each of 12 clinical conditions, inpatient claims records will serve as the basis for modeling readmission risk. After developing risk models, we will examine relationships between risk-adjusted readmission rates and quality of care using peer review- result records provided by Michigan Peer Review Organization, Inc.
