The PARTNER group recently published a comparison between the latest generation SAPIEN 3 transcatheter aortic valve implantation (TAVI) system (Edwards Lifesciences, Irvine, CA, USA) and surgical aortic valve replacement (SAVR) in intermediate-risk patients, apparently demonstrating superiority of the TAVI and suggesting that TAVI might be the preferred treatment method in this risk class of patients. Nonetheless, assessment of the non-randomized methodology used in this comparison reveals challenges that should be addressed in order to elucidate the validity of the results. The study by Thourani and colleagues showed several major methodological concerns: suboptimal methods in propensity score analysis with evident misspecification of the propensity scores (PS; no adjustment for the most significantly different covariates: left ventricular ejection fraction, moderate-severe mitral regurgitation and associated procedures); use of PS quintiles rather than matching; inference on not-adjusted Kaplan-Meier curves, although the authors correctly claimed for the need of balancing score adjusting for confounding factors in order to have unbiased estimates of the treatment effect; evidence of poor fit; lack of data on valve-related death.These methodological flaws invalidate direct comparison between treatments and cannot support authors' conclusions that TAVI with SAPIEN 3 in intermediate-risk patients is superior to surgery and might be the preferred treatment alternative to surgery.
Keywords: Aortic valve replacement; Propensity score analysis; Statistical analysis; Transcatheter valve therapy; Valve disease.
© The Author 2017. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.