THE RELATIVE PERFORMANCE CHARACTERISTICS AND CORRELATION OF THE LOGISTIC EUROSCORE AND THE SOCIETY OF THORACIC SURGERY SCORE IN THE PARTNER TRIAL

Abstract

Background: The Logistic EuroSCORE (LES) and the Society of Thoracic Surgery score (STS) are validated risk scores predicting 30-day outcomes in patients undergoing surgical aortic valve replacement with or without coronary artery bypass grafting. The performance characteristics of these scores applied to transcatheter aortic valve replacement (TAVR) are controversial. Methods: We compared predicted and observed 30 day/in-hospital mortality of patients undergoing TAVR in the PARTNER inoperable and high risk randomized trials and continued access registries (n=2552). The performance of the LES and STS scores was evaluated using standard assessments of discrimination and calibration. Results: Among patients undergoing TAVR, the observed mortality was 6.5% while the predicted mortality was significantly higher by both STS (11.4%±3.9%) and LES (26.6±16.2%). STS and LES were weakly correlated (r2=0.12). The STS demonstrated greater discriminatory power than the LES score (AUC 0.60 vs 0.53), but the overall discriminatory power for both scores was not robust (Figure). Calibration of the scores to clinical outcomes (Hosmer-Lemeshow test), was good for STS (p=0.3) but poor for LES (p<0.001). Conclusions: Existing surgical risk scores are poor discriminators of 30 day/in-hospital outcomes of TAVR, but exhibit some degree of calibration to these outcomes (specifically STS). TAVR-specific risk models need to be developed to optimize patient selection. (Figure Presented).