Approved Research

Developing multiverse approaches to Mendelian Randomisation

Lay summary

The best evidence to support whether a risk factor is a true cause of disease comes from experimental trials in which people randomly receive a specific intervention (e.g., to a drug or some sort of lifestyle advice). However, sometimes we can't conduct experiments due to ethical or feasibility considerations. Using a method known as Mendelian Randomisation (MR), genetic predisposition to a proposed risk factor can be used as a substitute for randomising someone to this risk factor or not. However, when the risk factor is not strongly influenced by genetics the results may be biased. Also, more information is needed on the accuracy of this method in different situations. This project has two main focuses. The first is to develop a novel method for using MR when risk factors aren't strongly influenced by genetics. The second is to use several MR approaches side by side to assess the accuracy of different MR methods. This will improve how we conduct MR and our confidence in MR results.