A weighted quantile sum regression with penalized weights and two indices.

Area of Science:

• Environmental epidemiology
• Biostatistics
• Toxicology

Background:

• Estimating the health effects of multiple chemical exposures presents statistical challenges.
• Weighted Quantile Sum (WQS) regression is a method for assessing mixture effects and identifying key components.

Purpose of the Study:

• To extend WQS regression for estimating dual mixture effects (positive and negative) in a single model.
• To introduce a penalization term to improve the estimation of component weights.

Main Methods:

• Developed an extended WQS regression model incorporating two directional indices and a penalization term.
• Evaluated the model's performance using simulation studies.
• Applied the model to analyze nutrient-obesity associations in the National Health and Nutrition Examination Survey (NHANES) data.

Main Results:

• The extended WQS model demonstrated improved accuracy in parameter estimation compared to the standard WQS regression.
• The penalization term enhanced the estimation of weights for individual chemicals.
• In the NHANES data, the model identified significant positive and negative associations between nutrient mixtures and obesity, highlighting sodium and magnesium.

Conclusions:

• The enhanced WQS regression provides a more accurate method for assessing chemical mixtures with both beneficial and detrimental health effects.
• The inclusion of a penalization term is advantageous for estimating component weights in mixture analysis.