2 Neuroscience2.9 Genetics
2.9.8 Mendelian randomisation
2 min read1 figure
Written by Resident Doctor Committee · Published
New in the 2027 syllabus
The College added this topic for exams from January 2027.
Using genetic variants as instruments
Mendelian randomisation uses genetic variants linked to an exposure to test whether that exposure causes an outcome. The variant acts as an "instrument": a stand-in for the exposure that should not be tied to the usual confounders.
The logic rests on timing. Genes are inherited before disease starts, so some reverse causation is avoided. But inheriting a variant is not the same as being randomised in a treatment trial.
The method depends on three assumptions:
Assumption | Requirement | Example threat |
|---|---|---|
Relevance | The variant predicts the exposure | A weak association gives an unreliable instrument |
Independence | The variant is not linked to confounders of the exposure-outcome relationship | Population structure or family effects |
Exclusion restriction | The variant affects the outcome only through the exposure | Horizontal pleiotropy, where the variant acts through another biological pathway |
Worked example: the ratio estimate
These figures are made up. Both measures are continuous, and both associations use the same effect allele.
Change per additional effect allele | Estimate |
|---|---|
Exposure | 0.20 units |
Outcome score | 0.06 points |
Divide the variant-outcome association by the variant-exposure association: 0.06 / 0.20 = 0.30.
Read this as 0.30 outcome points per unit of exposure, provided the instrument assumptions hold.
The ratio alone tells you nothing about precision. You need the standard errors and the uncertainty in both associations.
Interpreting a study
Multiple instruments give more information and allow sensitivity analyses. If the methods agree, that still does not prove every assumption holds.
Two-sample studies take the exposure and outcome associations from separate datasets. Check for ancestry differences between the samples, overlap between them, and correct alignment of alleles.
Horizontal pleiotropy breaks the exclusion restriction. Downstream effects that run through the exposure are different: they fit the proposed causal pathway.
A genetic estimate usually reflects differences in exposure across a lifetime. It may not predict what a short-term clinical intervention will do.
Triangulation means checking the genetic findings against longitudinal, experimental and clinical evidence before concluding that the exposure is causal.