Multiple-index interaction models to accommodate exposure grouping in environmental mixtures.

Area of Science:

• Environmental health
• Toxicology
• Biostatistics

Background:

• Assessing risks from environmental exposure mixtures is crucial in environmental health research.
• Existing methods often fail to utilize biological grouping of exposures, limiting population-based analyses.
• Lack of appropriate statistical tools hinders the integration of biological insights into mixture analyses.

Purpose of the Study:

• To propose a novel semiparametric multiple-index interaction model (MIIM) for analyzing environmental mixtures.
• To explore the impact of persistent organic pollutants (POPs) groups on leukocyte telomere length (LTL) using NHANES data.
• To provide a flexible statistical framework for high-dimensional exposure mixtures and various health outcomes.

Main Methods:

• Developed a semiparametric multiple-index interaction model (MIIM) to summarize high-dimensional exposures into group-level indices.
• Allowed for nonlinear effects and interactions among exposures within and between groups.
• Utilized Monte Carlo simulations to evaluate MIIM performance under diverse mixture scenarios.
• Applied MIIM to National Health and Nutrition Examination Survey (NHANES) data.

Main Results:

• MIIM effectively handles high dimensionality and correlation in exposure mixtures.
• The model provides interpretable insights into overall group effects and between-group interactions.
• MIIM can identify key contributors within exposure groups.
• Demonstrated application to NHANES data for POPs and LTL analysis.

Conclusions:

• MIIM is a valuable translational tool for environmental health research.
• It bridges biological insights with epidemiological data for mixture analysis.
• The model enhances understanding of environmental mixtures' effects on health outcomes, including LTL.