Mendelian randomization

Mendelian randomization (MR) is an instrumental-variable method that exploits genetic variants associated with a modifiable exposure to probe whether the exposure causally affects an outcome. Because variants are allocated at conception and generally unrelated to later behaviors, they mimic the random assignment of a trial and are less prone to confounding and reverse causation than standard observational analyses. Validity rests on three assumptions: the variant is genuinely linked to the exposure, affects the outcome only through that exposure (no pleiotropy), and is not itself confounded by a common cause it shares with the outcome. When reading an MR study, check how strongly the instrument predicts the exposure and what sensitivity analyses were used to detect pleiotropy, since violations can produce misleading causal claims. Example: variants linked to lifelong higher LDL cholesterol are associated with more heart disease, supporting a causal role for LDL, a conclusion later reinforced by randomized cholesterol-lowering trials.