News blog readers know that I am a fan of Mendelian randomisation studies. Much better than observational studies and second only to an experiment of a type that is often hard to do with sufficient precision. Some of my colleagues are less enamoured, worrying that they are overinterpreted. The BMJ has featured many meticulously reported studies where Mendelian randomisation is used to create an ‘instrument’ for cause/effect inferences. Many of these have been reported in our predecessor NIHR CLAHRC West Midlands News Blog (see Table).

It is timely that the BMJ has now published an excellent guide to responsible use of this type of instrumental variable.[7] The paper describes the single and merged study approaches and tests for horizontal pleiotropy (whereby the genetic variant may violate the exclusion assumption because it can bypass the putative explanatory variable). One thing that I learned was the potential for time to interact with genes. For instance, genetic influences may operate only at specific times. For example, Mendelian randomisation involving genetic variants for vitamin D metabolism corroborate studies of an association between low levels of vitamin D and multiple sclerosis. However, vitamin D only provides protection at early ages. This point about life time effects of exposure can make it harder to estimate the magnitude of an effect than the direction of effect.

Mendelian randomisation is an important tool in modern epidemiology and no-one can really call themselves an epidemiologist without nuanced knowledge of the subject. In terms of causal inference it occupies a status intermediate between standard observational methods and RCTs. And like all methods it should be close-coupled with theoretical knowledge, as we argue repeatedly in this News Blog.

Richard Lilford, ARC WM Director

References:

1. Larsson SC, et al. Modifiable pathways in Alzheimer’s disease: Mendelian randomisation analysis. BMJ. 2017; 359: j5375.
2. Do R, et al. Common variants associated with plasma triglycerides and risk for coronary artery disease. Nat Genet. 2013; 45(11): 1345-52.
3. Frikke-Schmidt R, et al. Association of loss-of-function mutations in the ABCA1 gene with high-density lipoprotein cholesterol levels and risk of ischaemic heart disease. JAMA. 2008; 299(21): 2524-32.
4. Tillmann T, et al. Education and coronary heart disease: Mendelian randomisation study. BMJ. 2017; 358: j3542.
5. Luo S, et al. Association of genetically predicted testosterone with thromboembolism, heart failure, and myocardial infarction: mendelian randomisation study in UK Biobank. BMJ 2019; 364: l476.
6. Mountjoy E, et al. Education and myopia: assessing the direction of causality by mendelian randomisation. BMJ. 2018; 361: k2022.