Mendelian randomization analysis of a time-varying exposure for binary disease outcomes using functional data...

Area of Science:

• Epidemiology
• Biostatistics
• Genetics

Background:

• Mendelian randomization (MR) is a popular method for causal inference in observational studies.
• Current MR studies often use single measurements for time-varying exposures, failing to capture long-term effects.
• Longitudinal data offers rich information on exposure trajectories.

Purpose of the Study:

• To develop and evaluate novel MR methods that incorporate longitudinal data for time-varying exposures.
• To improve the accuracy and power of causal effect estimation in epidemiological studies.
• To address limitations of standard MR analysis in handling dynamic exposure variables.

Main Methods:

• Functional principal component analysis (FPCA) to model individual exposure trajectories from sparse, longitudinal data.
• Two novel MR analysis frameworks: one for cumulative exposure effects and another for time-varying genetic effects using functional regression.
• Statistical testing for causal effects using the proposed functional MR methods.

Main Results:

• Simulation studies demonstrated substantial power gains for the proposed functional MR methods over standard MR.
• The new methods effectively utilize longitudinal data to capture long-term exposure dynamics.
• Application to Framingham Heart Study data highlighted improved performance and identified inconsistencies with single-measurement MR.

Conclusions:

• Functional data analysis-based MR methods offer significant advantages for studying time-varying exposures.
• These advanced methods provide more robust causal inference by leveraging longitudinal data.
• The proposed techniques enhance the ability to understand the long-term causal impact of dynamic risk factors on disease outcomes.