Valvular & structural heart diseaseMeta-analysis
Updated 5-year outcomes of transcatheter versus surgical aortic valve replacement in patients with severe aortic stenosis at low- to intermediate-surgical risk
- systematic review and meta-analysis
- Bayesian hierarchical design
- time-to-event data reconstructed
- minimally informative priors
Authors' conclusion
In this meta-analysis of RCTs, TAVI resulted in a clinically relevant increase in all-cause mortality, and a high probability of an increased risk of stroke, at 5 years of follow-up in low-risk to intermediate-risk patients, when compared to SAVR.
Study context
- Population
- Patients with severe aortic stenosis at low- to intermediate-surgical risk, in randomised controlled trials (RCTs) comparing transcatheter aortic valve implantation (TAVI) to surgical aortic valve replacement (SAVR) with at least 5-year follow-upTAVI 3,704 · SAVR 3,545
- Participants
- 7,249 six RCTs
- Comparison
- TAVI versus SAVR
- TAVIn = 3,704
- SAVRn = 3,545reference
- Outcome
- All-cause mortality
- Follow-up
- 5 years
Pooled result
As reported in the abstract
TAVI vs SAVREstimateP value
heterogeneity τ 2 =0; 99.3% posterior probability that SAVR is superior to TAVI
median RR 1.12(95% CrI CI 1.02–1.22)
conventional frequentist random-effects model
RR 1.13(95% CI 1.05–1.22)
P = 0.001conventional frequentist fixed-effects model
RR 1.12(95% CI 1.03–1.23)
P = 0.011Further pooled results
| Outcome | median RR (95% CrI CI) | |
|---|---|---|
| Stroke · 5 years | 1.13 (0.93–1.39) | |
| 0.512Median relative risk (log scale) |
- Stroke: heterogeneity τ 2 =0.04; 88.0% posterior probability that SAVR outperformed TAVI
Interaction effect of surgical risk category
- No evidence for an interaction effect of surgical risk category
- P = 0.470
Original publication
Updated 5-year outcomes of transcatheter versus surgical aortic valve replacement in patients with severe aortic stenosis at low- to intermediate-surgical risk. Heart. Published 11 February 2026. doi:10.1136/heartjnl-2025-327092
Summarised from the published abstract; every number is checked against it