Factors and Prediction Models for Unplanned Hospital Readmissions at a Pediatric Tertiary Centre

Abstract

Unplanned Hospital Readmissions (UHRs) are associated with increased morbidity and mortality, and may be preventable. This study identified factors associated with pediatric UHRs and developed prediction models. UHRs for pediatric patients from 2007-2009 and 2017-2019 at British Columbia Children’s Hospital were retrospectively reviewed. Factors for UHRs were analyzed, and prediction models were derived and tested. 5.26% (411/8387) of patients from 2007-2009 and 3.95% (329/8316) from 2017-2019 experienced at least one UHR. Varying by time period, factors for UHRs included: home health authority, age, previous ER visits, preadmission comorbidities, admission type, in-hospital interventions, and intensive care unit stay. Prediction models had areas under the receiver operating characteristic curve of .61 (2007-2009) and .67 (2017-2019). This study identified variables associated with UHRs. Differences in predictor variables between two time periods suggest that UHRs may not reflect quality of care, and future prediction models need to be iteratively refined.