The relative performance characteristics of the logistic European System for Cardiac Operative Risk Evaluation score and the Society of Thoracic Surgeons score in the Placement of Aortic Transcatheter Valves trial

Objectives: The logistic European System for Cardiac Operative Risk Evaluation (LES) score and the Society of Thoracic Surgeons (STS) score are validated to predict 30-day outcomes following surgical aortic valve replacement (SAVR) with or without coronary artery bypass grafting. Their performance when applied to patients undergoing transcatheter aortic valve replacement (TAVR) is controversial. Methods: We compared predicted and observed 30-day/in-hospital and 1-year mortality of patients undergoing TAVR in the first Placement of Aortic Transcatheter Valves trial and continued access registry (N = 2466). The performance of the LES and STS scores (prospectively calculated) was evaluated using standard assessments of discrimination and calibration. Performance of STS and LES scores among 307 patients undergoing SAVR from the high-risk cohort of the randomized trial were also examined. Results: In patients undergoing TAVR, the observed 30-day/in-hospital mortality was 6.5%, whereas the predicted 30-day mortality was higher by both STS score (11.4% +/- 3.9%) and LES score (26.6% +/- 16.2%). The discrimination for both scores was poor for 30-day/in-hospital and 1-year mortality. Calibration was better for STS score than for LES at 1 year but poor for both at 30 days among TAVR cohort. These results were consistent among the subgroups of patients undergoing transfemoral and transapical access; however, the STS score had better performance among the high-risk patients who underwent SAVR at 30 days but not 1 year. Conclusions: The STS and LES surgical risk scores overestimated 30-day/in-hospital mortality and were poor discriminators of post-TAVR mortality, but the calibration of the STS score was better in these high-risk patients. These data highlight the need for TAVR-specific risk models to optimize patient selection.