Prediction of Costs and Length of Stay in Coronary Artery Bypass Grafting

Abstract

Listen

Translate article

Although more than 200,000 bypass operations are performed in the United States annually, few data exist on the predictors of costs and resource use for this procedure. Questions related to clinical outcomes, costs, and resource use in coronary artery bypass grafting (CABG) were addressed. In a multiinstitutional statewide database, patient level data from 42,839 patients undergoing isolated CABG were combined with cost data. After adjustment for cost-to-charge ratios and inflation, the association of length of stay and costs with the Society ofThoracic Surgeons-Predicted Risk of Mortality (STS-PROM) was analyzed. Patients were randomly divided into development (60%) and validation (40%) cohorts. Regression models were developed to analyze the impact of patient characteristics, comorbidities, and adverse events on postoperative length of stay and total costs. Postoperative length of stay and total direct costs for CABG averaged 6.9 days and $38,847. Length of stay and costs increased from 5.4 days and $33,275 in the lowest-risk decile (mean STS-PROM of 0.6%) to 13.8 days and $69,122 in the highest-risk decile (mean STS-PROM 19%). Compared with adverse events, patient characteristics had little impact on length of stay and costs. onvalidation, the models that combined preoperative and postoperative variables explained variance better (R(2)=0.51 for length of stay; R(2)= 0.47 for costs) and were better calibrated than the preoperative models (R(2)= 0.10 for length of stay; R(2)= 0.14 for costs). The STS-PROM and preoperative regression models are useful for preoperative prediction of costs and length of stay for groups of patients, case-mix adjustment in hospital benchmarking, and pay for performance measures. The combined preoperative and postoperative models identify incremental costs and length of stay associated with adverse events and are more suitable for prioritizing quality improvement efforts.