Weighted quantile sum (WQS) mixed-effects model.

Area of Science:

• Environmental Health
• Biostatistics
• Toxicology

Background:

• Environmental exposures often occur as complex mixtures with correlated patterns.
• Individual exposures may be sub-threshold, but their joint action can lead to significant health effects (mixture effect).
• Existing Weighted Quantile Sum (WQS) regression methods assume independence and do not accommodate multiple intra-subject outcomes, limiting their application.

Purpose of the Study:

• To extend WQS regression to handle multiple intra-subject outcome variables or repeated measures, accounting for intra-subject correlation.
• To develop a statistically valid inference method for analyzing mixture effects with correlated outcomes.
• To address a research gap in analyzing complex environmental exposures and their impact on health outcomes.

Main Methods:

• A novel WQS mixed-effects model was developed.
• Data were randomly split, with weights estimated via resampling in the training set.
• Inference was performed using repeated holdout validation sets with a mixed-effects model to manage intra-subject correlation.

Main Results:

• The WQS mixed-effects model successfully accommodates multiple correlated outcome variables within subjects.
• The method allows for statistically valid inference relating weighted environmental exposure indices to health outcomes.
• The model was applied to a pilot study on environmental factors and kidney function in runners.

Conclusions:

• The developed WQS mixed-effects model provides a robust approach for analyzing environmental mixture effects when multiple or repeated outcome measures are present.
• This extension enhances the capability of WQS regression for complex exposure-health outcome relationships.
• The pilot study demonstrated the model's utility in investigating environmental impacts on kidney function, with potential for sex-specific analyses.