Mendelian Randomisation: Past Success and Future Challenges

Morning session specifically aimed at career-young statisticians, although all are welcome to join.

Stephen Burgess (University of Cambridge)

Mendelian randomization is the use of genetic variants to make inferences on causal effects using observational data. The causal effect of a risk factor on a disease outcome can be inferred under the assumption that certain genetic variants are instrumental variables for the risk factor. Mendelian randomization represents a fast and inexpensive technique for prioritizing or de-prioritizing risk factors as targets for clinical intervention.

I will give an introduction to the topic of Mendelian randomization, explaining why causal inference in an observational setting is an important (and a difficult) question, and how and why genetic variants are able to untangle this problem. The talk will be based around the four stages of conducting a Mendelian randomization analysis: 1) specification of the dataset(s) for analysis, 2) search for candidate genetic instrumental variables, 3) validation of the instrumental variable assumptions, and 4) estimation of the causal effect (if appropriate). Particular attention will be given to biological reasons why a genetic variant may or may not be a valid instrumental variable, and the interpretation of an instrumental variable estimate as compared to a clinical or pharmacological intervention on the risk factor under analysis. The talk will be illustrated by an analysis of the causal effect of LDL-cholesterol on coronary heart disease risk.

Annual General Meeting of the IBS-BIR

Vanessa Didelez (University of Bristol)

Instrumental variables provide an approach for consistent inference on causal effects even in the presence of unmeasured confounding. Such methods have for instance been used in the context of Mendelian randomisation, as well as in pharmaco-epidemiological contexts. In these and other applications, it is common that covariates are available, even if deemed insufficient to adjust for all confounding.

As IVs allow inference when there is unobserved confounding, it appears that often the analyst assumes that even observed confounders / covariates do not need to or should not be taken into account. However, this is not generally the case. With view to the role of covariates, we here contrast two-stage least squares estimators, generalized methods of moment estimators and variants thereof with methods more common in biostatistics using G-estimation in so-called structural mean and distribution models.

When using covariates, there are structural aspects to be considered, e.g. whether the covariates are prior to or potentially affected by the instruments. But in addition, one has to worry even more about efficiency versus model misspecification when modelling covariates. We discuss this for the IV procedures mentioned above, especially for linear instrumental variable models. Our results motivate adaptive procedures that guarantee efficiency improvements through covariate adjustment, without the need for covariate selection strategies. Besides theoretical findings, simulation results will be shown to provide numerical insight.

(This is joint work with Stijn Vansteelandt)

John Thompson (University of Leicester)

Mendelian randomization (MR) is a neat way of establishing causation provided that its key assumptions hold. Unfortunately the growing popularity of MR has led to its use in situations where the assumptions are highly questionable. Even if we dismiss these examples as the misuse of MR, we are still left with the majority of applications that are characterised by uncertainty over whether the assumptions hold or not; in these cases, the only honest conclusion is that either the factor is causal or it isn’t. In this talk I want to ask two questions: does a MR estimate have any interpretation when the assumptions do not hold? and when we read a report of a MR, how do we judge whether the assumptions are reasonable or not?