Basic MR analysis: Testing for a causal effect.
The first step in any MR approach involves finding genetic polymorphisms to use as proxies, or “instruments,” for a target exposure. Testing for an association between a genetic instruments and the outcome of interest is a valid test of the hypothesis that the exposure causally affects the outcome (9, 11). This requires that all genetic variants satisfy the instrumental variable assumptions (IV1-IV3):
IV1: A genetic variant is associated with the exposure
IV2: A genetic variant is notassociated with any confounders of the exposure and outcome
IV3: The genetic variant notassociated of the outcome conditional on the exposure and all confounders
The IV assumptions are represented by the causal diagram in Figure 1.
Figure 1: Hypothesised relationship between genetic variant Gk, modifiable exposure X and outcome Y in the presence of unobserved confounder U. Dashed lines represent violations of the instrumental variable assumptions.
The causal effect of a unit-change in X on Y is represented by the parameter . If a gene is a valid IV then it should not be associated with Y if = 0. If the two are associated, this points to a non-zero causal effect, but extended methods are required to estimate it.