Heterogeneity was quantified using between-study standard deviation (I2-statistic in the case of cardiovascular disease). Publication bias was assessed by visual inspection of contour-enhanced funnel plots and Egger’s test if there were at least ten studies [16, 17].