BackgroundCurrent guidelines provide recommendation for transcatheter aortic-valve replacement (TAVR) in severe aortic stenosis without emphasis on valve systems. The comparative performances of balloon-expandable valves (BEV) and self-expanding valves (SEV) remain unclear. We aim to compare the early (30-day) and midterm (1-year) mortality and cardiovascular outcomes of BEV with SEV. MethodsPubMed, CENTRAL, and EMBASE were searched from inception to February 13, 2020 for randomized controlled trials (RCTs) and propensity-score matched (PSM) studies. Odds ratios (ORs) for binary outcomes and mean differences for continuous outcomes were pooled using random-effect models (DerSimonian–Laird method) with Hartung-Knapp-Sidik-Jonkman variance correction. Primary outcomes were early and midterm all-cause mortality. ResultsWe included 3 RCTs (1418 patients) and 12 PSM studies (36,540 patients). Compared with SEV, BEV was associated with significantly lower mortality at 30 days (OR 0.76, 95% CI 0.67–0.85, p < 0.001, I2 = 0) and 1 year (OR 0.87, 95% CI 0.77–0.99, p = 0.04, I2 = 20.4%) in PSM studies, but not RCTs with insufficient power. Similar findings were found in subgroups analysis based on valve generations and SEV types. The 30-day and 1-year cardiovascular mortality, 30-day incidences of moderate to severe paravalvular leak, procedural contrast agent volume, and procedure time were lower, but transvalvular pressure gradient was higher in BEV than SEV in PSM studies. The 30-day incidences of permanent pacemaker implantation (PPI), acute kidney injury, stroke, major bleeding, major vascular complications, and rehospitalization were not statistically different between BEV and SEV. Early-generation SEV was associated with a higher 30-day PPI risk than corresponding BEV comparators. PPI risk was lower in ACURATE neo (Boston Scientific, Natick, MA) but higher in Evolut R SEV (Medtronic Inc., Minneapolis, MN), both compared with SAPIEN 3 BEV (Edwards Lifesciences, Irvine, CA). ConclusionsPSM studies suggest lower early and midterm mortality in BEV than SEV, but the contribution of unmeasured confounders cannot be excluded. Results from adequately powered RCTs with long-term follow-up are critically needed to confirm these findings.