Methods: We did a systematic review up to July 19, 2017, to identify randomised clinical trials comparing CABG with PCI using stents. Eligible studies included patients with multivessel or left main coronary artery disease who did not present with acute myocardial infarction, did PCI with stents (bare-metal or drug-eluting), and had more than 1 year of follow-up for all-cause mortality. In a collaborative, pooled analysis of individual patient data from the identified trials, we estimated all-cause mortality up to 5 years using Kaplan-Meier analyses and compared PCI with CABG using a random-effects Cox proportional-hazards model stratified by trial. Consistency of treatment effect was explored in subgroup analyses, with subgroups defined according to baseline clinical and anatomical characteristics.
Findings: We included 11 randomised trials involving 11 518 patients selected by heart teams who were assigned to PCI (n=5753) or to CABG (n=5765). 976 patients died over a mean follow-up of 3·8 years (SD 1·4). Mean Synergy between PCI with Taxus and Cardiac Surgery (SYNTAX) score was 26·0 (SD 9·5), with 1798 (22·1%) of 8138 patients having a SYNTAX score of 33 or higher. 5 year all-cause mortality was 11·2% after PCI and 9·2% after CABG (hazard ratio [HR] 1·20, 95% CI 1·06–1·37; p=0·0038). 5 year all-cause mortality was significantly different between the interventions in patients with multivessel disease (11·5% after PCI vs 8·9% after CABG; HR 1·28, 95% CI 1·09–1·49; p=0·0019), including in those with diabetes (15·5% vs 10·0%; 1·48, 1·19–1·84; p=0·0004), but not in those without diabetes (8·7% vs 8·0%; 1·08, 0·86–1·36; p=0·49). SYNTAX score had a significant effect on the difference between the interventions in multivessel disease. 5 year all-cause mortality was similar between the interventions in patients with left main disease (10·7% after PCI vs 10·5% after CABG; 1·07, 0·87–1·33; p=0·52), regardless of diabetes status and SYNTAX score.
Interpretation: CABG had a mortality benefit over PCI in patients with multivessel disease, particularly those with diabetes and higher coronary complexity. No benefit for CABG over PCI was seen in patients with left main disease. Longer follow-up is needed to better define mortality differences between the revascularisation strategies.