Fig. 1

Study design overview. A Univariable MR. MR fundamental assumptions: (1) Relevance assumption: The genetic variants (instrumental variables) are strongly correlated with exposure(s) (P < 5E−8; r² < 0.001 and distance > 10 MB, the SNPs are in pairwise linkage disequilibrium). (2) Independence assumption: no unmeasured confounders of the correlations are present between genetic variants and outcome(s). (3) Exclusion restriction assumption: the genetic variants influence the outcome(s) only via exposure(s). B Multivariable MR. MR: Mendelian randomization; PM: Proportion mediated; SNP: single nucleotide polymorphism; BMI: Body mass index; Educational attainment: years of schooling; Smoking: smoking behavior; Watching TV: time spent watching the television