Quantile causal mediation analysis allowing longitudinal data

Area of Science:

• Environmental Epidemiology
• Biostatistics
• Molecular Epidemiology

Background:

• Traditional mediation analysis using mean regression may miss effects on extreme values.
• Individuals with extreme biomarker levels are often more susceptible to environmental exposures.
• Understanding effects on biomarker tails is crucial for public health.

Purpose of the Study:

• To develop and apply a novel mediation analysis framework using quantile regression.
• To investigate the effects of air pollution on fibrinogen levels via interferon-gamma (IFN-γ) methylation.
• To examine effects across the distribution of mediator and outcome variables.

Main Methods:

• Utilized quantile regression within a causal inference framework.
• Modeled direct and indirect effects of particle number on fibrinogen percentiles.
• Incorporated exposure-mediator interactions and random intercepts for longitudinal data.
• Applied methodology to environmental data linking particle number, IFN-γ methylation, and fibrinogen.

Main Results:

• Found a direct effect of particle number on the upper tail of the fibrinogen distribution.
• Observed a suggestive indirect effect of particle number on the upper fibrinogen tail through lower percentiles of IFN-γ methylation.
• Demonstrated the utility of quantile regression for capturing tail-specific effects.

Conclusions:

• Quantile regression provides a more sensitive approach to mediation analysis, especially for extreme values.
• Air pollution may impact cardiovascular health through epigenetic modifications (IFN-γ methylation) and direct effects on fibrinogen.
• This method enhances understanding of environmental health risks at the population level.