We investigate the causal effect of HDL‐C on CHD risk further using the multivariable MR‐Egger method. We consider the 185 genetic variants having known association with at least one of HDL‐C, LDL‐C, and triglycerides at GWAS significance in 188 578 participants reported by the Global Lipids Genetics Consortium.23 The point estimates for the associations between these genetic variants and lipids were taken from Do et al.24 The CARDIoGRAMplusC4D consortium consisting of 60 801 cases and 123 504 controls was used to obtain the estimates of the association between the variants and CHD risk.25 The IVW and MR‐Egger methods were applied to the data under univariable and multivariable frameworks as described in Section 2. For the univariable IVW and MR‐Egger methods, the models were fitted using 2 sets of variants: firstly using all 185 variants; and secondly using all variants associated with HDL‐C at GWAS level of significance. The genetic variants were orientated with respect to the risk increasing allele for HDL‐C. These analyses differ from those provided in Burgess et al and Do et al as they use summarized data from different versions of the CARDIoGRAMplusC4D study22, 24; here, we use associations from the 2015 data release.25
