Development and Validation of a Score to Predict the Risk of Readmission After Adult Cardiac Operations

Background The purpose of this study was to develop and validate a risk score for readmissions after cardiac operations. Methods Adults surviving to discharge after cardiac operations at a single institution from 2008 to 2013 were randomly divided 3:1 into training and validation cohorts. The primary outcome was readmission within 30 days of discharge. A multivariable model was constructed in the training cohort incorporating variables associated with 30-day readmission in univariate logistic regression. Points were assigned to predictors in the multivariable model proportional to their odds ratios. Results Among 5,193 patients undergoing cardiac operations and surviving to discharge, the 30-day readmission rate was 10.3% (n = 537). The most common reasons for readmission were volume overload (24%; n = 131) and infection (21%; n = 113). The risk score incorporated 5 multivariable predictors and was out of 20 possible points. The predicted rate of 30-day readmission based on the training cohort ranged from 5.9% (score = 0) to 54.7% (score = 20). Patients were categorized as low (score = 0; readmission 5.7%), moderate (score 1-7; readmission 11.0%), and high risk (score >7; readmission 24.2%) (p < 0.001). Thirty-day readmission rates based on these score categories were similar in the validation cohort (low 6.4%, moderate 11.0%, high 17.4%; p < 0.001). There was a robust correlation between predicted rates of readmission in the training cohort based on the composite risk score and actual rates of readmission in the validation cohort (r = 0.95; p < 0.001). Conclusions We developed and validated a risk score for readmission after cardiac operations that may have utility in targeting interventions and modifying risk factors in high-risk populations.