Instrumental variable estimation of causal risk ratios and causal odds ratios in Mendelian randomization analyses

Area of Science:

• Epidemiology
• Biostatistics
• Genetic Epidemiology

Background:

• Causal inference methods are crucial for understanding exposure-outcome relationships.
• Instrumental variable (IV) methods are used to estimate causal effects when confounding is present.
• Continuous exposures present unique challenges for traditional IV estimators.

Purpose of the Study:

• To compare different instrumental variable (IV) estimators for causal risk and odds ratios, focusing on continuously measured exposures.
• To apply these methods to Mendelian randomization analysis investigating the effect of body mass index (BMI) on asthma risk.
• To evaluate the impact of potential interactions on IV estimates.

Main Methods:

• Description and application of various IV estimators, including multiplicative structural mean model (MSMM) and multiplicative generalized method of moments (MGMM).
• Mendelian randomization analysis using data from the Avon Longitudinal Study of Parents and Children.
• Simulation study to assess estimator performance under interaction scenarios.

Main Results:

• MSMM and MGMM estimators produced identical results, suggesting an inverse relationship between BMI and asthma risk.
• Other IV estimators indicated a positive association, though all confidence intervals were wide.
• An interaction between BMI, FTO genotype, and asthma risk explained the divergent estimates.

Conclusions:

• Point estimates from different IV methods can vary significantly in practice.
• Structural mean models (SMMs) make weaker assumptions than other IV estimators, potentially leading to more consistent results.
• Understanding interactions is critical for interpreting IV estimates in complex causal pathways.