Difficulties in Testing the Instrument Strength Independent of Direct Effect Assumption in Mendelian Randomization

To the Editor Mendelian randomization (MR)¹ is becoming an increasingly popular framework for assessing causality within epidemiological research. A recent example of such an analysis in JAMA Cardiology by White et al² looked specifically at the causal effect of several lipid fractions on the risk of coronary artery disease (CAD) and diabetes, using single nucleotide polymorphisms (SNPs) that predict variation in lipids. We would like to draw attention to a minor mistake in their interpretation of a crucial assumption required by some of the statistical methods used, so that it is not subsequently adopted by others.