A nomogram for estimating the risk of unplanned readmission after major surgery

Abstract

Unplanned hospital readmissions among surgical patients are the target of multiple efforts to improve patient outcomes and to decrease avoidable costs. This study presents an analysis of unplanned readmissions for adult patients who undergo major surgery and the associated risk presented by clinical characteristics of the individual patients known before discharge. Multivariable logistic regression analysis was used to develop and validate a model for estimating risk of readmission using data from the participant use data file of the American College of Surgeons National Surgical Quality Improvement Program. Unplanned readmission occurred in 5.3% of major surgery cases for patients who were discharged alive. A total of 48 candidate predictors of unplanned readmission were evaluated. A reduced model was developed that included the 10 covariates that provide the greatest contributions to the full model. The reduced model demonstrated good statistical performance (validated C statistic=0.70) and demonstrated excellent calibration in an independent dataset of patients undergoing major surgery in 2012. The predictive equation from the reduced model is presented as a nomogram and formula for calculating individual patient risk of unplanned readmission. Accurate identification of patients at high risk for unplanned readmission can be conducted using selected patient characteristics known before discharge. A nomogram reflecting the effects of these key patient characteristics can be used to calculate accurately a patient's individual risk of readmission. The availability of these estimates before discharge could improve the efficacy of discharge planning efforts and related programs coordinating care seeking to prevent avoidable readmission.