Among patients with LMCAD, in-hospital complications (such as cardiogenic shock, permanent pacemaker implantation, sepsis, myocardial infarction and cerebrovascular events) were not significantly different between those who underwent PCI versus CABG. Similarly, 30-day mortality was not significantly different between the 2 groups (PCI vs CABG 3% vs 2.9%, P = 1.000). However, Kaplan–Meier survival analysis (Fig. 3A) showed that at 3 years of follow-up, the rate of all-cause mortality among patients with LMCAD was significantly higher in those who underwent PCI (26.9%) vs CABG (8.7%; log-rank P-value <0.001 for the overall difference during follow-up). Consistent with the univariable Kaplan–Meier findings, multivariable analysis (Table 3) showed that PCI versus CABG referral was the most powerful predictor of long-term mortality among patients with LMCAD and was associated with a >2-fold mortality hazard increase. Additional predictors of mortality among patients with LMCAD included older age and diabetes mellitus (Table 3). Similar results were obtained after further adjustment for centre volume (Supplementary Material, Table S3).