Retrospective exposure reconstruction using approximate Bayesian computation: A case study on perfluorooctanoic acid...

Area of Science:

• Environmental Epidemiology
• Toxicology
• Biostatistics

Background:

• Exposure biomarkers in environmental epidemiology often represent a single time point, risking misalignment with disease development.
• This temporal mismatch can lead to exposure measurement error and reverse causality in health outcome studies.
• Accurate individual exposure assignment is crucial for understanding toxicant effects on health.

Purpose of the Study:

• To enhance perfluorooctanoic acid (PFOA) exposure estimations using an approximate Bayesian computation (ABC) method.
• To investigate the association between PFOA exposure and preeclampsia within the C8 Studies cohort.
• To improve the characterization of historical toxicant exposures by integrating multiple data sources.

Main Methods:

• Utilized an ABC method to combine measured serum PFOA concentrations with environmentally modeled PFOA data.
• Reconstructed historical PFOA exposures by incorporating lognormal distributions for key pharmacokinetic and exposure model parameters.
• Analyzed serum PFOA levels from blood samples collected in West Virginia and Ohio (2005-2006).

Main Results:

• The ABC method demonstrated similar correlations to traditional methods but reduced mean squared error by over 50%.
• Compared to previous analyses, the ABC approach yielded a similar adjusted odds ratio (AOR) for PFOA and preeclampsia.
• The study confirmed the utility of ABC in reducing exposure measurement error.

Conclusions:

• Combining modeled exposure data with measured biomarkers via Bayesian methods effectively reduces exposure measurement error.
• The ABC approach offers a more robust method for estimating historical toxicant exposures.
• Accurate exposure assessment is vital for reliable epidemiological studies on environmental toxicants and health outcomes.