Risk Calculator to Predict 30-Day Readmission After Coronary Artery Bypass: A Strategic Decision Support Tool

Abstract

Re-admission is an important source of patient dissatisfaction and increased hospital costs. A simple calculator to determine the probability of re-admission may help guide patient dismissal planning. Using the national readmissions database (NRD), we identified admissions for isolated primary coronary artery bypass (CABG) and stratified them according to 30-day readmission. Including pre, intra and postoperative variables, we prepared a logistic regression model to determine the probability for re-admission. The model was tested for reliability with boot-strapping and 10-fold cross-validation. From 135,699 procedures, 19,355 were readmitted at least once within 30days of dismissal. Patients who were readmitted were older (67±10 vs 65 ± 10 years, p<0.01), females (32% vs 24%; p<0.01) and had a higher Elixhauser comorbidity score (1.5±1.4 vs 1.1±1.2; p<0.01). Our final model (c- statistic=0.65) consisted of 16 pre and three postoperative factors. End-stage renal disease (OR 1.79 [1.57-2.04]) and length of stay>9days (OR 1.60 [1.52-1.68]) were most prominent indicators for readmission. Compared to Medicaid beneficiaries, those with private insurance (OR 0.62 [0.57-0.68]) and Medicare (OR 0.85 [0.79-0.92]) coverage were less likely to be readmitted. Our simple 30-days CABG readmission calculator can be used as a strategic tool to help reduce readmissions after coronary artery bypass surgery.