Consecutive Observational Study to Validate EuroSCORE II Performances on a Single-Center, Contemporary Cardiac Surgical Cohort.

OBJECTIVE: To compare and validate the original EuroSCORE risk stratification models with the renewed EuroSCORE II model in a contemporary cardiac surgical practice. DESIGN: A consecutive observational study to validate EuroSCORE II performances, conducted as retrospective analysis of prospectively collected data. SETTING: A tertiary university institute for cardiovascular diseases. PARTICIPANTS: Adult patients undergoing cardiac surgery between January and December 2012. METHODS: One thousand eight hundred sixty-four consecutive patients were scored preoperatively using additive and logistic EuroSCORE as well as EuroSCORE II. The discriminative power of the EuroSCORE models was tested by calculating the area under the receiver operating characteristic curve (AUC). The calibration of the models was assessed by Hosmer-Lemeshow statistics and with observed-to-expected mortality ratio. MEASUREMENTS AND MAIN RESULTS: The in-hospital overall mortality was 3.65%, with predicted mortalities according to additive EuroSCORE, logistic EuroSCORE, and EuroSCORE II of 5.14%, 6.60%, and 3.51%, respectively. The observed-to-expected (O/E) mortality ratio confirmed good calibration for the entire cohort only for EuroSCORE II (1.05, 95% confidence interval 0.81 - 1.29). Hosmer-Lemeshow test confirmed overall good calibration only for additive EuroSCORE (p = 0.129). The EuroSCORE II confirmed very good discriminatory power for a prolonged intensive care unit (ICU) stay of>2 days and>5 days (AUCs>0.75). Acceptable discriminatory power was confirmed for a prolonged postoperative stay of>7 days and>12 days (AUCs>0.70). CONCLUSION: EuroSCORE II confirmed very good discriminatory capacity, good calibration ability (O/E mortality ratio), and good capability to predict prolonged ICU and postoperative stays in a contemporary patient cohort undergoing cardiac surgery.