Seventeen studies (N = 565) were included in the review. All were non-randomised and retrospective.
There was statistically significant lower mortality in patients treated with EVR (8.1 per cent) compared to OR (20.8 per cent), pooled OR 0.43 (95% CI: 0.25, 0.76, p < 0.01, 17 RCTs). There was no evidence of statistical heterogeneity. Sensitivity analyses did not significantly alter the findings; this was also the case with a sub-group analysis of studies with similar pre-operative variables. No significant publication bias was detected.