This paper proposes a Phenome-Wide association Clustering of MR instruments (PWC-MR) approach to address biases in Mendelian Randomisation (MR) studies caused by pleiotropy, confounding, and heterogeneous causal effects. Applying this method to the BMI-educational attainment relationship, the study reveals heterogeneous causal effects across six clusters of BMI-associated genetic instruments. A cluster enriched for socioeconomic indicators shows a large negative effect, while a cluster enriched for body mass traits shows a much smaller effect. Sensitivity analyses using within-sibling MR, childhood BMI, and multivariable MR support these findings, suggesting that the large apparent causal effect of BMI on education is driven by confounding.
Publisher
Nature Communications
Published On
Feb 15, 2024
Authors
Liza Darrous, Gibran Hemani, George Davey Smith, Zoltán Kutalik
Tags
Mendelian Randomisation
Pleiotropy
Causal Effects
BMI
Education
Genetic Instruments
Confounding
Related Publications
Explore these studies to deepen your understanding of the subject.
