Does the Hospital Predict Readmission? A Multi-level Survival Analysis Approach

Time to psychiatric rehospitalization was predicted for a sample of 1473 Medicaid-insured youth in Illinois in 2005 and 2006. A multi-level model statistical strategy was employed to account for the fact that youth days to rehospitalization were nested within hospital and to test the hypothesis that hospitals would vary significantly in return rates, controlling for individual-level (e.g., symptom, demographic) variables. Hospitals did not vary significantly in days to rehospitalization. At the individual-level, level of externalizing behavior and residential treatment placement predicted a faster return to the hospital. These results support the perspective that hospital outcomes are best operationalized using variables tied more directly to the inpatient episode (e.g., LOS, reductions in acuity).