For SBP, the IVW causal estimate of 0.054 (log odds ratio per 1 mmHg change in blood pressure) is far from the null (P = 4 × 10−6, odds ratio 1.055). In contrast, the MR-Egger test for the intercept gives a P-value of 0.21 and a causal estimate closer to the null (bias-corrected estimate: 0.015, P = 0.64, odds ratio 1.015). For DBP, the IVW causal estimate of 0.083 is again far from the null (P = 1 × 10−5, odds ratio 1.087). The MR-Egger test gives a P-value of 0.054 and a negative causal estimate (bias-corrected estimate: −0.024, P = 0.67, odds ratio 0.976). Additionally, Cochran’s Q test indicated strong evidence of heterogeneity between IV estimates based on the individual variants (P < 0.001 for both SBP and DBP). Given the presence of apparent asymmetry in the funnel plots (Figure 4), the application of the MR-Egger test casts some doubt on the robustness of the original claims that these data allow generation of Mendelian randomization style analyses that provide strong support for the (well established) notion that blood pressure is causally related to coronary heart disease (CHD) risk.
