2.4.2 Heterogeneity and pleiotropy analyses
To evaluate heterogeneity, we used Cochran heterogeneity test. If there
is heterogeneity, random-effects IVW models are applied; otherwise, the
fixed-effect IVW model is applied. To evaluate pleiotropy, we used
MR-Egger regression (assessing whether the MR-Egger intercept was
different from zero). If causality was estimated in the same direction
among the various MR methods, the conclusions will be considered as
relatively robust causal inferences. Analyses were carried out using
MendelianRandomization version 0.9.0, TwoSampleMR version 0.5.7 and MVMR
version 0.4 in the R environment version 3.6.3 (The R Foundation for
Statistical Computing, Vienna, Austria).
Results