Can EuroSCORE predict direct costs of cardiac surgery?

Abstract

The aim of this study is to determine if a preoperative risk stratification model can identify different surgical costs. Four hundred and eighty-eight patients undergoing open heart surgery between March 2000 and March 2001 were classified with the EuroSCORE model. Direct variable costs were prospectively collected, surgical team costs excluded. The multivariate analysis was used to find variables independently associated with costs. Of the 488 patients enrolled 342 (70%) were males, mean age 65+/-10 years, 57 (12%) had myocardial infarction, 20 (4%) had ejection fraction <30%, 56 (11%) were operated in emergency, 26 (5%) had a re-operation. 113 (23.2%) were operated for valvular disease, 30 (6.1%) were operated for thoracic aortic surgery, one (0.2%) was operated for interatrial septal defect, 79 (16.2%) were operated for other intervention in addition to coronary bypass and 265 (54.3%) for isolated coronary bypass. The mean intensive care unit length of stay (ICU-LOS) was 2.3+/-4.1 days and the postoperative LOS was 8.2+/-5.3 days. According to EuroSCORE, 117 patients (24%) were at low, 187 (38%) at medium, and 184 (38%) at high risk. Costs were significantly and directly correlated with preoperative risk model with a correlation coefficient of 0.47 and an increase of costs of 3.5% (95% CI 2.3-4.7, P<0.0001) for each single rise of risk score. The relationship EuroSCORE vs. direct costs is, respectively: EuroSCORE 0-2 ==> 6863+/-861 Euro; 3-4 ==> 8292+/-3714 Euro; 5-6 ==> 8908+/-3480 Euro; 7-8 ==> 10,462+/-6123 Euro; 9-10 ==> 13,711+/-12,634 Euro; >10 ==> 21,353+/-18,507 Euro. Excluding EuroSCORE from the preoperative logistic model, age, preoperative creatinine, critical condition, ejection fraction, re-operation and sex were independently correlated with costs. From our data the EuroSCORE model developed to predict (30-day postoperative) hospital mortality could be used to predict direct operative costs and identify patients with different levels of resource consumption.