Methods for combining mediation analysis with Mendelian randomization
This workshop will give participants an understanding of Mendelian randomization (MR) methods available to carry out mediation analysis and the opportunity to discuss persistent challenges. A number of MR methods, such as multivariable MR and Network MR, have been developed and applied to understanding complex effects of multiple exposures on outcomes, including unpicking mediating mechanisms. As with any analysis, given different data, some methods are more appropriate than others and it is not currently fully understood how mediation analysis can be conducted within an MR framework in many scenarios. In this workshop we will discuss which methods are currently available for combining MR and mediation analysis and when these methods are most appropriate. We will then discuss the challenges still present in these analyses and how we might be able to address these in the future. The workshop will not cover the specific details of how to conduct the analysis.
This course is organised by tutors based at the MRC Integrative Epidemiology Unit who use these methods in their research along with contributions from external experts.
- Alice Carter (MRC-IEU PhD Student)
- Eleanor Sanderson (Research Associate)
- Rebecca Richmond (Research Fellow)
- Jon Heron (Senior Research Fellow)
- Tom Richardson (Senior Research Associate)
- Others to be confirmed
School of Chemistry, Cantocks Close, Bristol, BS8 1TS
Date and Time
The workshop will be held on Friday 19th July, from 13:30 to 17:00. Refreshments will be provided.
The fee for attending the workshop is £20.
Aims and objectives
The aim of the workshop is to inform participants of the methods available for combining MR with mediation analysis and when different methods should be used. By the end of the session participants should be able to:
- Understand methods for mediation analysis using observed phenotypic data
- Understand what multivariable MR is
- Understand what network MR is
- Understand which method is appropriate for different sources of data (such as individual level or summary data) and different types of variables
- Appreciate persisting methodological challenges and where methods still need to be developed
The workshop will last three and a half hours and split into two sessions. The first session will feature short talks on mediation and advanced MR methods providing participants with an understanding of the different methods available. The talks will be supplemented with applied examples and demonstrations of when each method should be used with different sources of data (such as summary and individual level data) and types of variables (such as binary or continuous measures). Analyses using different types of phenotypes, including molecular phenotypes, will be introduced. Additionally we will introduce potential future applications of the methods and what challenges we still encounter in our analyses.
The second session will be a discussion driven by workshop participants. Workshop tutors will facilitate the discussion and answer questions. Participants should be willing to discuss current challenges in this area of research.
Who is the course intended for?
The workshop will be of interest to researchers familiar with MR methods, although not necessarily with mediation analyses or the particular MR methods potentially applicable to mediation analysis. Tutors will assume a knowledge of MR principles. This workshop will suit those who have previously used mediation methods and those considering adding these methods to their research. Although these methods are not exclusively for mediation analysis, this will be the focus of the workshop.