Hierarchical testing of composite endpoints: applying the win ratio to percutaneous coronary intervention versus coronary artery bypass grafting in the SYNTAX trial

The goal of the study was to compare long-term outcomes of percutaneous coronary intervention (PCI) versus coronary artery bypass grafting (CABG), accounting for the clinical impact of individual components in the composite endpoints and prioritising these using the win ratio (Rw).The win ratio was compared with conventional methods of analyses (hazard ratio [HR] and relative risk) in the SYNTAX trial (n=1,800). For the composite of death/stroke/myocardial infarction (MI), the win ratio favoured CABG and was 1.37 (95% CI: 1.10-1.77) for matched analysis, 1.28 (95% CI: 1.11-1.53) for unmatched analysis, while the conventional HR was 1.29 (95% CI: 1.11-1.53). The largest number of winners in favour of CABG over PCI were based on MI (n=39 vs. n=19, respectively). Death was significantly reduced with CABG in matched (Rw=1.39, 95% CI: 1.04-1.86) and unmatched win ratio analyses (Rw=1.27, 95% CI: 1.01-1.42) as compared with non-significant conventional analysis (HR 1.19, 95% CI: 0.92-1.56). In subgroups, matched win ratio analyses had a larger treatment effect in favour of CABG compared with conventional analyses, especially in patients with three-vessel disease and intermediate SYNTAX scores, while unmatched win ratios had a smaller point estimate, but with narrower confidence intervals than matched analyses findings.This re-analysis of the SYNTAX trial using the win ratio shows that the most important benefit of CABG treatment is the reduction of hard clinical endpoints such as mortality and MI. Future trials using this approach can expect to maintain similar statistical power with smaller sample sizes, and thereby reduce the cost of a trial.