Mendelian randomization with a binary exposure variable: interpretation and presentation of causal estimates.

Area of Science:

• Epidemiology
• Genetic Epidemiology
• Biostatistics

Background:

• Mendelian randomization (MR) is a powerful tool for causal inference in observational studies.
• Genetic variants serve as instrumental variables for modifiable exposures.
• Binary exposures, such as hypertension, pose challenges to standard MR assumptions.

Purpose of the Study:

• To provide methods for causal estimation with binary exposures in Mendelian randomization.
• To address complications arising from dichotomizing continuous risk factors.
• To clarify the interpretation of causal estimates in the context of binary exposures.

Main Methods:

• Utilizing genetic variants as instrumental variables for underlying continuous risk factors.
• Addressing potential violation of the exclusion restriction assumption with binary exposures.
• Applying parametric assumptions for causal estimation with binary exposures.

Main Results:

• Causal inferences for binary exposures are valid when instrumental variable assumptions hold for the underlying continuous risk factor.
• Causal estimates for binary exposures assume a stepwise function at the point of dichotomization.
• The interpretation of causal estimates depends on assumptions like monotonicity and homogeneity, which may not be widely applicable.

Conclusions:

• Mendelian randomization investigations with dichotomized binary exposures should be conceptualized in terms of the underlying continuous variable.
• Standard MR assumptions can be violated with binary exposures, necessitating careful methodological consideration.
• The methods provided offer a framework for causal estimation with binary exposures, acknowledging inherent limitations.