Abstract
Background. Before transcatheter aortic valve implantation (TAVI), predicted prosthesis–patient mismatch (PPM) is graded from a reference effective orifice area (EOA) taken from interchangeably-used sources that were never compared.Methods. In 1,419 consecutive transfemoral native-valve TAVI patients, predicted EOA was derived four ways: Transcatheter Valve Therapy (TVT) registry values; manufacturer/bench values (IFU/Bench); a CT-derived EOA from the CT-indicated size; and the CT annular area directly (PPMp2), graded by VARC-3. With no method as reference, the four were compared against each other, the residual gradient and mortality.Results. No patient had severe PPM; moderate PPM ranged 0.6–11.7%. The methods were not equivalent (Cochran Q = 78.9, p < 0.001), five of six pairs differing. For a gradient ≥20 mmHg, TVT registry and CT-derived-EOA were indistinguishable (ΔAUC 0.019, p = 0.103), whereas TVT registry outperformed IFU/Bench (0.057, p = 0.010) and PPMp2 (0.106, p = 0.005). Over a median 36.7 months, 363/1,177 died; no method predicted mortality (hazard ratios 0.99–1.13, all p > 0.45; C-indices 0.51–0.53).Conclusions. The sources differ and rank against the residual gradient, yet none is prognostic. Predicted PPM is a measurement problem, not a risk marker: the EOA source and VARC-3 threshold should always be stated.