Figure 3.

Causal diagrams representing the assumed causal relationships in the two-sample Mendelian randomization analysis of waist circumference (WC) on blood pressure (BP). The genetic instruments (SNPs) are assumed to influence WC by either affecting overall body adiposity (proxied by body mass index (BMI)) [A] or by specifically changing body fat distribution [B]. The grey solid lines represent the effect of confounders between exposure-outcome (U_{X, Y}), exposure-covariable (U_{X, W}) and covariable-outcome (U_{W, Y}). The dashed lines represent the relationship being tested between WC and BP. IV: instrumental variable; SNP: single nucleotide polymorphism; U: unmeasured confounders