Mendelian randomization is an epidemiological technique that compares disease risk in groups of individuals defined based on their genetic variants to make causal inferences (1). Rather than comparing individuals with high levels of an exposure versus those with low levels of an exposure as in a conventional epidemiological analysis, the approach compares those with genetic variants predisposing them to increased versus decreased levels of the exposure. This is analogous to the analysis of a randomized controlled trial for a treatment that increases levels of the exposure, which does not compare those with high versus low levels of the exposure after treatment, but rather compares those who were randomly assigned to receive the treatment versus those who were randomly assigned to the control group. The logic is that randomization should be independent of all competing risk factors, meaning that the randomly-assigned groups only differ systematically with respect to their average levels of the treatment and any downstream consequences of the treatment. Hence, an association between randomization and the trial outcome is indicative of a causal effect of the treatment. In the same way, if genetic variants act analogously to randomization by dividing the population into groups that differ systematically only with respect to the exposure and its consequences, then an association between the genetic variants and the outcome is indicative of a causal effect of the exposure (2).