A framework for the investigation of pleiotropy in two‐sample summary data Mendelian randomization

• DOI: 10.1002/sim.7221
• Publication Date: 2017-01-23T23:03:00Z
• Authors: Jack Bowden; Fabiola Del Greco M; Cosetta Minelli; George Davey Smith; Nuala Sheehan; John Thompson
• Keywords: Mendelian Randomization; Pleiotropy; Instrumental variable

Mendelian randomization (MR) uses genetic data to probe questions of causality in epidemiological research, by invoking the Instrumental Variable (IV) assumptions. In recent years, it has become commonplace attempt MR analyses synthesising summary estimates association gleaned from large and independent study populations. This is referred as two‐sample MR. Unfortunately, due sheer number variants that can be easily included into analyses, increasingly likely some do not meet IV assumptions pleiotropy. There a pressing need develop methods both detect correct for pleiotropy, order preserve validity approach this context. paper, we aim clarify how established meta‐regression random effects modelling mainstream meta‐analysis are being adapted perform task. Specifically, focus on two contrastin g approaches: Inverse Variance Weighted (IVW) method which assumes its simplest form all valid IVs, MR‐Egger regression allows violate assumptions, albeit specific way. We investigate ability popular models provide robustness pleiotropy under IVW approach, propose statistics quantify relative goodness‐of‐fit over regression. © 2017 The Authors. Statistics Medicine Published JohnWiley & Sons Ltd