Volume 39 | Number 5 | October 2004

Abstract List

Eric A. Coleman M.D., M.P.H., Sung‐joon Min, Alyssa Chomiak, Andrew M. Kramer


To (1) describe patterns of posthospital care transitions; (2) characterize these patterns as uncomplicated or complicated; (3) identify those at greatest risk for complicated transitions.

Data Sources/Study Setting

The Medicare Current Beneficiary Survey was used to identify beneficiaries aged 65 and older who were discharged from an acute care hospital in 1997–1998.

Study Design

Patterns of posthospital transfers were described over a 30‐day time period following initial hospital discharge. Uncomplicated posthospital care patterns were defined as a sequence of transfers from higher‐to lower‐intensity care environments without recidivism, while complicated posthospital care patterns were defined as the opposite sequence of events. Indices were developed to identify patients at risk for complicated transitions.

Principal Findings

Forty‐six distinct types of care patterns were observed during the 30 days following hospital discharge. Among these patterns, 444 episodes (61.2 percent) were limited to a single transfer, 130 episodes (17.9 percent) included two transfers, 62 episodes (8.5 percent) involved three transfers, and 31 episodes (4.3 percent) involved four or more transfers. Fifty‐nine episodes (8.1 percent) resulted in death. Between 13.4 percent and 25.0 percent of posthospital care patterns in the 1998 sample were classified as complicated. The area under the receiver operating curve was 0.771 for a predictive index that utilized administrative data and 0.833 for an index that used a combination of administrative and self‐reported data.


Posthospital care transitions are common among Medicare beneficiaries and patterns of care vary greatly. A significant number of beneficiaries experienced complicated care transitions—a finding that has important implications for both patient safety and cost‐containment efforts. Patients at risk for complicated care patterns can be identified using data available at the time of hospital discharge.