Predicting hospital readmissions in the Medicare population.

Using a nationally random sample of Medicare beneficiaries, we developed a multivariate logistic model that identified a series of factors that predict readmission to an acute care hospital within 60 days of discharge. Our regression results show that 10 variables are statistically significant predictors of readmissions, with the patient's disease history and diagnosis among the best predictors. Our results can be used by peer review organizations to evaluate the quality of care provided by different hospitals as well as by physicians and social workers to improve discharge planning.