Beta-blockers are clinically known to lower HR, therefore the phenotype measurements of beta-blocker users may be under-estimated, and hence the inclusion of beta-blocker users in our analysis may potentially bias our analysis results. We therefore performed a sensitivity analysis by also meta-analysing a subgroup of cohorts that provided beta-blocker data (N = 48 347; 17 cohorts). Results including or excluding beta-blocker users were highly correlated (r2 of the betas = 0.97; r2 of the P-values = 0.74; Supplementary Material, Fig. S4), suggesting there is little or no bias from including beta-blocker users in the analysis. Therefore we report the meta-analysis results from the full dataset for the RR-interval, to maximize sample size and power.
