Figure 1:

The Three MR Assumptions, where Z is the IV associated with the exposure, X is the exposure, Y is the outcome, and U is confounder. 1) Relevance Assumption: The IV is strongly associated with the exposure of interest. 2) Independence assumption: there are no confounders of the association between the IVs and the outcome, and 3) exclusion restriction assumption: the IV is not related to the outcome other than via the exposure