School of Clinical Medicine Distinguished Lecture 2025

Evidence triangulation: moving from correlations to causes

Understanding the causal nature of modifiable risk factors is key to much biomedical science. This talk will present the outlines of a "triangulation of evidence" approach, in which the focus is on the variety of evidence, rather than the volume. Estimates from different study types - e.g. conventional observational studies, randomised controlled trials, Mendelian randomization, negative control investigations, cross-context comparisons, non-genetic instrumental invariable analyses, between-twin analyses, among others - are brought together. The key notion is that whilst all study designs can lead to biased findings, if the biases in one study type are independent of the biases in another study type, greater confidence in the causal inference from these studies can be made. The importance of prospective registration of study protocol and full data and code access will be emphasised.