paperKB
coga / coga-kb
Help
Sign in

Chunk #21 — METHODS — Advantages of multivariable MR‐Egger and comparison with univariable MR‐Egger

Source
Extending the MR-Egger method for multivariable Mendelian randomization to correct for both measured and unmeasured pleiotropy.
Embedded
yes

Text

The expression in Equation 13 follows from the multivariable framework outlined in Equation 5, where the direct effect for univariable MR‐Egger has been decomposed into the residual direct effect αj′ of multivariable MR‐Egger and the indirect effects via each risk factor. The residual direct effect αj′ will be altered with each additional risk factor included in the multivariable MR‐Egger model. If these additional risk factors are causally associated with the outcome (θ k≠0), then αj′ will consist of fewer components. It seems likely that the InSIDE assumption would be easier to satisfy for multivariable MR‐Egger than its univariable counterpart as the direct effect for univariable MR‐Egger consists of unmeasured and measured pleiotropy.