Aernoud Fiolet

177 The efficacy and safety of low-dose colchicine in patients with coronary disease Study endpoints For this meta-analysis, the pre-specified primary endpoint was the composite of myocardial infarction, stroke, or cardiovascular death and the key secondary endpoint was as above with the addition of coronary revascularization. In the definition of these composite endpoints, we chose the composition, definition, and tallies of the composite endpoints as reported in the original and subsequent publications [Low-dose Colchicine for secondary prevention of cardiovascular disease 2 (LoDoCo2) and the Colchicine CardiovascularOutcomeTrial (COLCOT)]. For the Colchicine in Patients with Acute Coronary Syndrome (COPS) trial and for the LoDoCo trial, we used LoDoCo2 definitions and supplemented the published data with additional data obtained from the principal investigators (J.L. and S.M.N.). In addition, all endpoints were also analysed using the original definitions and published data. The authors did not have access to the composite endpoints of the trial by Deftereos et al. (Supplementary material, Table S1A). The component-oriented endpoints were myocardial infarction, stroke, coronary revascularization, and cardiovascular death. We used the most inclusive definition as reported in the original main trial paper including supplementary materials, ancillary papers, or by personal communications. Safety endpoints were hospitalization for infection, hospitalization for pneumonia, hospitalization for gastro-intestinal disorders and newly diagnosed cancer, all-cause death, and non- cardiovascular death (Supplementary material, Table SA1B). Endpoint tallies were extracted into a structured data set by two reviewers (T.S.J.O. and A.T.L.F.). Data synthesis and analysis To estimate the pooled treatment effect, pooled relative risks (RR) were calculated using the cumulative incidence rates as reported, by applying inverse-variance weighting combined with a random-effect model with a DerSimonian–Laird estimator. Overall treatment effect was formally tested at a two-sided alpha level of 0.05 without adjustment for multiplicity. Treatment effect modification by subgroups for the primary and secondary endpoint was tested using random- effects models, applying the method of restricted maximum likelihood estimation. In order to obtain absolute risk reductions and number needed to treat, risk estimates of the 1- and 3-year cumulative incidences were calculated from the original data or estimated from the published Kaplan–Meier estimator curves. Weighted-average estimates were calculated with the use of the weights from the overall meta-analysis of reported endpoints.