Performance of EuroSCORE II and SinoSCORE in Chinese patients undergoing coronary artery bypass grafting

Abstract

OBJECTIVES The European System for Cardiac Operative Risk Evaluation II (EuroSCORE II) was developed to update EuroSCORE and incorporated refinement and modification of several risk factors. SinoSCORE was designed by Chinese scholars based on 9839 patients who underwent coronary artery bypass grafting (CABG) at 43 participating institutions. This study was designed to validate the EuroSCORE II and SinoSCORE in Chinese patients undergoing CABG and to compare their performance overall and per subgroup. METHODS A total of 4507 adult receiving CABG at our institution between January 2010 and April 2014 were included in this retrospective study. Patients were stratified for cardiovascular risk using EuroSCORE II and SinoSCORE. The performance of EuroSCORE II and SinoSCORE was analysed with a focus on discrimination power and calibration. RESULTS The in-hospital mortality rate for the entire cohort was 1.4%, while the mortality rate predicted by EuroSCORE II was 1.47 ± 1.2% (95% CI 1.43-1.50) and by SinoSCORE was 2.86 ± 3.5% (95% CI 2.76-2.96). The C-statistics of EuroSCORE II and SinoSCORE were 0.728 and 0.716, respectively. The Hosmer-Lemeshow test indicated that EuroSCORE II had poor goodness of fit while SinoSCORE performed slightly better. When patients were divided into quartiles based on predicted risk, respectively defined as group I, II, III and IV, EuroSCORE II underestimated mortality rates of patients scored IV, but overestimated mortality rates in all other groups; SinoSCORE underestimated mortality rates of patients scored I and overestimated mortality rates in all other groups. EuroSCORE II only achieved good discrimination for patients scored I (area under the receiver operating characteristic curve, AUC = 0.707 > 0.70), and SinoSCORE achieved poor discrimination for all subgroups except group II (AUC = 0.754 > 0.70). EuroSCORE II overestimated the mortality rate in the isolated CABG group and underestimated mortality rates in patients with other cardiac surgeries. SinoSCORE overestimated mortality rates in all pathology subgroups. The AUC values of EuroSCORE II and SinoSCORE were 0.694 and 0.687, respectively, for isolated CABG. The AUC values of EuroSCORE II and SinoSCORE were 0.772 and 0.669 for combined cardiac surgery CABG. CONCLUSIONS EuroSCORE II could predict mortality in the entire group and in the low-middle risk group, but not in the high-risk group, in which it underestimated mortality. SinoSCORE overestimated mortality rates in the entire group and in all subgroups. Risk models should be targeted to different heart diseases, and the statistical methods of established risk systems should be improved.