Mortality prediction after transcatheter treatment of failed bioprosthetic aortic valves utilizing various international scoring systems: Insights from the Valve-in-Valve International Data (VIVID).

Background: Transcatheter Aortic Valve Implantation (TAVI) is commonly used to deploy new bioprosthetic valves inside degenerated surgically implanted aortic valves in high risk patients. The three scoring systems used to assess risk of postprocedural mortality are: Logistic EuroSCORE (LES), EuroSCORE II (ES II), and Society of Thoracic Surgeons (STS).
Objective: The purpose of this study is to analyze the accuracy of LES, ES II, and STS in estimating all-cause mortality after transcatheter aortic valve-in-valve (ViV) implantations, which was not assessed before.
Methods: Using the Valve-in-Valve International Data (VIVID) registry, a total of 1,550 patients from 110 centers were included. The study compared the observed 30-day overall mortality vs. the respective predicted mortalities calculated by risk scores. The accuracy of prediction models was assessed based on calibration and discrimination.
Results: Observed mortality at 30 days was 5.3%, while average expected mortalities by LES, ES II and STS were 29.49 (± 17.2), 14.59 (± 8.6), and 9.61 (± 8.51), respectively. All three risk scores overestimated 30-day mortality with ratios of 0.176 (95% CI 0.138-0.214), 0.342 (95% CI 0.264-0.419), and 0.536 (95% CI 0.421-0.651), respectively. 30-day mortality ROC curves demonstrated that ES II had the largest AUC at 0.722, followed by STS at 0.704, and LES at 0.698.
Conclusions: All three scores overestimated mortality at 30 days with ES II showing the highest predictability compared to LES and STS; and therefore, should be recommended for ViV procedures. There is a need for a dedicated scoring system for patients undergoing ViV interventions.
(© 2018 Wiley Periodicals, Inc.)