Background: The European System for Cardiac Operative Risk Evaluation (EuroSCORE) is one of the most widely used risk models for the predicting mortality after cardiac surgery. The aim of this study was to validate the EuroSCORE model for predicting operative mortality in heart valve surgery on a Chinese multicenter database and comparing the performance of EuroSCORE with our new risk stratification system, the Sino System for Coronary Operative Risk Evaluation (SinoSCORE).
Methods: Data from patients undergoing heart valve surgery between January 2007 and December 2008 were retrospectively collected, from 43 hospitals in China. The EuroSCORE and the SinoSCORE were calculated for each patient. Mortality was defined as any in-hospital death. Area under the receiver operating characteristics curve (AUC) was used to study the discriminatory abilities of the models. The Hosmer-Lemeshow (H-L) goodness-of-fit test was used to study the calibration of the predictive models.
Results: A total of 15 367 patients were analyzed. For the entire cohort, the observed mortality was 2.34%, the predicted mortality was 3.71% (additive), 3.19% (logistic) and 3.66% (SinoSCORE). AUC was 0.747 for SinoSCORE, 0.699 additive and 0.696 for logistic EuroSCORE. Calibration of SinoSCORE and additive EuroSCORE was good (H-L: P = 0.250 and P = 0.051, respectively), but the logistic EuroSCORE model had a poor calibration (H-L: P < 0.05). The discriminatory ability and calibration of the SinoSCORE were good in low- and high-risk patients. However, the discriminatory ability of the EuroSCORE model was poor in all risk deciles.
Conclusions: The EuroSCORE does not accurately predict mortality in Chinese patients with heart valve surgery, and the SinoSCORE is superior to the EuroSCORE at predicting in-hospital mortality in Chinese heart valve surgery patients.