Mendelian randomisation for mediation analysis: current methods and challenges for implementation
Alice R Carter, Eleanor C. M. Sanderson, Gemma Hammerton, Rebecca Claire Richmond, George Davey Smith, Jon Heron, Amy E. Taylor, Neil M Davies, Laura D Howe
Multivariable and two-step Mendelian randomisation recovered direct and indirect effects with little bias in simulations with confounding or measurement error, conditions that biased phenotypic regression mediation.
European Journal of Epidemiology · May 1, 2021Peer-reviewedGenetics✦ In-depth digest