Abstract 4342855: Comparative Effectiveness of Cardiac Rehabilitation After Surgical Aortic Valve Replacement – A Target Trial Emulation

Abstract

Background: Cardiac rehabilitation (CR) after surgical aortic valve replacement (SAVR) is recommended by guidelines, but participation is low. This is partly because the supporting evidence is weak: prior trials were small with short follow-up and few clinical events. Because CR is guideline-recommended, randomized trials of CR vs. no CR would not be ethical, and high-quality observational comparative effectiveness studies are urgently needed. Research Question: Does CR reduce the risk of death and major adverse cardiovascular events (MACE, defined as acute myocardial infarction, stroke, or heart failure) in Medicare beneficiaries undergoing SAVR? Methods: We first specified a target trial – a hypothetical pragmatic randomized trial that would answer the causal question of interest – and then emulated the target trial using 100% Medicare claims ( Fig. 1 ). We included Medicare beneficiaries aged ≥65 years who underwent SAVR between 10/2016 and 12/2022. The intervention (“CR participation”) was defined as receipt of ≥2 CR sessions within 90 days of discharge, compared with receipt of 0 or 1 CR session (“control”). The cloning-censoring-weighting approach was used to align the time for determining eligibility, treatment assignment, and start of follow-up, and accounted for baseline and time-varying covariates. The primary outcome was death from any cause or hospitalization for MACE. Two falsification endpoints (new cancer; acute infection) were used to test for residual confounding. Results: Among 44,136 Medicare beneficiaries, 48.5% participated in ≥2 CR sessions after SAVR ( Fig. 2 ). CR participants were more likely to be male and identify as Non-Hispanic White, and less likely to be dually enrolled in Medicare and Medicaid or reside in neighborhoods with the highest social vulnerability. CR participants had a lower risk of the primary outcome at 3 years compared with the control (17.5% vs 20.1%; adjusted risk difference, -2.6 pp; 95% CI, -3.5 pp to -1.7 pp). Analyses of falsification endpoints were compatible with the absence of strong residual confounding ( Fig. 3 ). Discussion: In the largest and most diverse study of CR after SAVR, patients with CR participation had a lower risk of death or MACE at 3 years. Fewer than half of Medicare beneficiaries participated in CR after SAVR, with marked socioeconomic inequities. Our findings highlight the urgent need for strategies to equitably improve participation in CR after SAVR.