Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption
Authors
Discussion
Loading...
Member takes
Implication
Simulation study demonstrates reduced bias and lower type-I error rates in coronary heart disease, highlighting robust causal inference even when most genetic instruments are invalid.
Key Points
- The mode-based estimate yields less bias and lower type-I error rates under the null, ensuring consistent causal inference even when invalid instruments predominate.
- Simulation in two-sample summary data demonstrates greater statistical power than MR-Egger regression, while requiring feasible genome-wide association study cohorts.
- Highlights potential as a robust sensitivity analysis that relaxes instrumental variable assumptions, strengthening causal evaluations across complex phenotypes.
Cite This Study
Hartwig et al. (2017) studied this question.