Edgar Daeter, Maaike M Roefs, Richard A J Post, Jurien Ten Berg, Pim A L Tonino, Dennis Van Veghel, Robert Klautz, Johanna J M Takkenberg, Kevin M Veen, Cardiothoracic Surgery Registration Committee of the Netherlands Heart Registration and DEDICATE author group
In PS-matched analyses, SAVR-registry patients showed lower combined mortality/stroke and stroke rates than SAVR-trial participants, suggesting caution in transporting trial-derived relative effects to this population. For mortality the relative trial-effects could be reproduced, but absolute benefit of TAVR was diminished in the registry population.
OBJECTIVE: To assess the transportability of the randomized DEDICATE trial results, comparing SAVR with TAVR in low- to intermediate-risk patients, to an external Dutch population derived from the Netherlands Heart Registration (NHR).
METHODS: Data sources included the DEDICATE-RCT data (TAVR-trial, SAVR-trial) and the NHR (SAVR-registry). The primary endpoint was composite of all-cause mortality or stroke at one year; secondary endpoints were mortality and stroke. As-treated TAVR-trial and SAVR-trial patients were propensity-score (PS) matched to SAVR-registry patients. Model-based and weighting-based transportability analyses were performed.
RESULTS: After applying trial criteria, 3,389 SAVR-registry patients were eligible and 1,211 trial patients were included (654 TAVR; 557 SAVR). Trial-eligible SAVR-registry patients were younger and had fewer comorbidities. PS-matching of TAVR-trial to SAVR-registry yielded 1,150 patients, and matching of SAVR-trial to SAVR-registry yielded 892 patients. In PS-matched cohorts, SAVR-trial patients experienced more composite events (HR 1.63 [1.02-2.62],p=0.04) and strokes (HR 2.96 [1.37-6.37],p=0.01) than SAVR-registry patients, with comparable mortality (HR 1.15 [0.65-2.02], p=0.63). TAVR-trial (vs. SAVR-registry) patients had lower 1-year mortality (HR 0.46 [0.24-0.87], p=0.02). Transported absolute risk reduction (ARR) of 1-year mortality of TAVR was attenuated when the trial was reweighted to resemble the registry (ARRtransported: 2.2% [0.1-4.2]; ARRtrial: 3.7% [1.4-6.7]).
CONCLUSION: In PS-matched analyses, SAVR-registry patients showed lower combined mortality/stroke and stroke rates than SAVR-trial participants, suggesting caution in transporting trial-derived relative effects to this population. For mortality the relative trial-effects could be reproduced, but absolute benefit of TAVR was diminished in the registry population.