a proof of concept example, we use summary data from the GLGC on genetic associations with lipid fractions, and from CARDIoGRAM on associations with CAD risk, and perform the analysis methods discussed in this paper to see whether the causal effect of LDL‐c is preserved by the weighted median and MR‐Egger regression methods, as well as whether the spurious causal effect of HDL‐c is contradicted by the robust methods. We use data provided by Do et al. [2013] on 185 genetic variants for this analysis. The genetic associations with the lipid fractions are in standard deviation units, and with the outcome are log odds ratios, so the causal effects represent log odds ratios per 1 standard deviation increase in the lipid fraction. Further details of the analysis are provided in Supporting Information Appendix 4.
