Assessing metal mixture effects on neuropsychological development: A trade-off between complexity and...

Area of Science:

• Environmental Health
• Neuroscience
• Biostatistics

Background:

• Assessing the health impacts of metal mixtures is complex.
• Sophisticated statistical methods like Bayesian Kernel Machine Regression (BKMR) are used but challenging to interpret.
• Understanding metal mixture effects on early childhood neurodevelopment is crucial.

Purpose of the Study:

• To evaluate the feasibility and interpretability of advanced statistical techniques for assessing metal mixture exposures on neuropsychological development.
• To compare the results of BKMR with traditional linear regression models (LRM).

Main Methods:

• Cross-sectional study design.
• Principal Components Analysis (PCA) to identify neurodevelopmental domains.
• Bayesian Kernel Machine Regression (BKMR) and Linear Regression Models (LRM) were employed and compared.
• Analysis focused on lead, molybdenum, and selenium mixture exposures.

Main Results:

• High correlations (0.92-0.95) were found between BKMR and LRM estimations for executive, motor, and visual/verbal functions.
• Differences between models were minimal, primarily at low lead and varying selenium concentrations.
• Findings suggest linear associations rather than high-order interactions between metals.

Conclusions:

• Linear regression models provide comparable results to complex methods like BKMR for assessing metal mixture impacts on neurodevelopment.
• Simpler models like LRM can facilitate a better understanding of individual and interaction effects of metal exposures.
• This supports the use of accessible statistical approaches in environmental health research.