Results are given in Table 4. The causal effects of LDL‐c and triglycerides are robustly detected by all the analysis methods. In contrast for HDL‐c, both MR‐Egger regression and the weighted median methods suggest that the “causal effect” of HDL‐c on CAD risk detected by the IVW method is suspect. With the exception of the simple median estimates, all other robust analysis methods give estimates compatible with the null. The test for directional pleiotropy in the MR‐Egger regression method suggested pleiotropy for HDL‐c, but not for LDL‐c or triglycerides. Although similar results for HDL‐c can be obtained by filtering out genetic variants that are suspected to have pleiotropic effects, by omitting genetic variants there is a loss of power (e.g., in Holmes et al. a liberal genetic risk score explained 3.8% of the variance in HDL‐c, whereas a conservative score omitting potentially pleiotropic variants explained only 0.3%).
