Summary
This methodological review describes how Mendelian randomisation can strengthen causal inference in mediation analysis by mitigating bias from confounding and measurement error that affect traditional approaches. The authors present two implementable frameworks—multivariable and two-step MR—and demonstrate through simulation that both are robust to exposure-outcome confounding and non-differential measurement error, though they require different assumptions than conventional mediation methods. The paper provides practical guidance on when MR assumptions are more plausible than traditional alternatives and thus when these methods can improve causal inference.
UK applicability
These statistical methods are applicable to UK epidemiological and biobank research, particularly leveraging large-scale population cohorts and genetic data now available through resources such as the UK Biobank. The methodological advances are relevant to UK health research seeking to understand causal pathways in chronic disease aetiology.
Key measures
Direct and indirect effects in mediation pathways; bias from confounding, measurement error, and weak instruments; horizontal pleiotropy; interaction effects between exposures and mediators
Outcomes reported
The paper describes and evaluates two approaches for causal mediation analysis using Mendelian randomisation (multivariable MR and two-step MR), addressing methodological challenges in understanding exposure-mediator-outcome pathways. It demonstrates performance characteristics of these methods through simulations and real data examples.
Topic tags
Dig deeper with Pulse AI.
Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.