Reproductive factors (AFS, LNSP and AFB)
In the MVMR analyses, we arrived at a result of causality between later age at first sexual intercourse and MA: reverse causality, by adjusting LNSP (AFS in model 5, OR=0.831, 95%CI, 0.785-0.880, P=1.64E-10). An increase in LNSP indicated greater likelihood of MA, after controlling/adjusting AFS(model 5), or EA, HI(model 2), or EA, AC, HI(model 3), or EA, SB, HI(model 4) (see Figure 3). Increasing years of schooling are accompanied by increases AFB (effect of EA on AFB, β= 0.377, 95% CI, 0.344 to 0.409, P = 1.56E-114), so EA may indirectly have an impact on the risk of MA through AFB (effect of AFB on MA, β= -0.205, 95% CI, -0.263 to -0.147, P = 5.46E-12) (Supplementary Table 1).By multiplying the estimate of the effect of EA on AFS, AFB with the estimate of the effect of AFB on MA, the mediation effect of AFB in the causal pathway from EA to MA was -0.08, accounting for 57.8% of the total effect (Supplementary Table 2).