Predictors of hospital readmissions in internal medicine patients: Application of Andersen's Model

This study aimed to identify predictors of internal medicine patients' readmission to hospital, using Andersen's behavioral model.

This prospective cohort study included 2622 patients aged ≥18 years, who were admitted to internal medicine wards at a university hospital between 1 February 2015 and 31 January 2016. Independent variables were divided into four groups (predisposing, enabling, need, and utilization), based on Andersen's model, and included in stepwise logistic regression analysis.

Younger age, male sex, a main diagnosis of neoplasm, longer length of stay, higher comorbidity scores, and weaker coping ability predicted all readmission. Predictors of unplanned readmission included having someone to help at home following discharge, comorbidity scores, and length of stay. Predictors of unplanned, related, and preventable readmissions included having someone to help at home following discharge, having a regular physician, and the main diagnosis at discharge. The most powerful predictors influencing readmission were need-related variables.

Although some predictors of readmission were unalterable, they could be used to identify high-risk patients. Innovative approaches targeting discharge planning and postdischarge care for patients with high comorbidity scores and long length of stay could reduce internal medicine patients' unplanned readmission.