High-Risk Cardiac Surgery in an Australian Population is Better Defined by the STS Score and EuroSCORE II Than the AusSCORE

Introduction: Risk scoring models (RSM) are commonly implemented to estimate the postoperative mortality risk of cardiac surgery patients. Studies have reported inaccurate mortality prediction by the RSM. Furthermore, the discrimination and calibration of the RSM to the Australian population remains unknown. Methods: The study population included cardiac surgery patients operated at Liverpool Hospital between January 20 13 and December 2014. Patient's demographics were routinely collected and entered into a prospective database. Mortality risks were estimated using the Additive EuroSCORE (AS), Logistic EuroSCORE (LES), EuroSCORE II (ES2), Society of Thoracic Surgeon Score (STS), AusSCORE total score (AS) and AusSCORE Multirisk (ASMR). Discrimination and calibration of the RSM was evaluated to determine the prediction accuracy to the population. High-risk cut-off points for ES2 and STS were derived using Youden's J statistics. Results: Overall, 738 surgical patients were used for the RSM analysis. Observed mortality of the population was 3.78%. ES2 has the best discrimination and acceptable calibration (AUC = 0.850; p = 0.053) while STS has good discrimination and calibration (AUC = 0.806; p = 0.652). AES, LES, ES2 and AS over-predicted while the ASMR and STS under-predicted mortality. Estimated mortality of ES2 was closest to observed mortality (observed to expected ratio = 0.909). High-risk cut-off points for ES2 and STS were 3.00% and 1.705% respectively. Conclusion: The final decision regarding the benefits and risks of surgery should be based on clinical judgment and the Heart Team recommendation. RSM should be used as a guide for identification of high-risk patients. Calibration of RSM to the population is essential prior to use.