Structural equation models in occupational health: an application to exposure modelling

Area of Science:

• Occupational hygiene
• Environmental health
• Statistical modeling

Background:

• Occupational hygiene surveys often collect simultaneous pollutant data from multiple locations.
• Exposure models must handle complex study designs and facilitate dose-response modeling.
• Accurate exposure assessment is crucial for risk assessment in occupational settings.

Purpose of the Study:

• To explore Structural Equation Models (SEMs) for analyzing occupational pollutant monitoring data.
• To evaluate SEMs as a tool for studies with multiple concurrent sampling locations.
• To compare SEMs with traditional methods for occupational exposure modeling.

Main Methods:

• Utilized exposure data from a diesel exhaust assessment in the US trucking industry.
• Employed Structural Equation Models (SEMs) to analyze elemental carbon exposure data.
• Compared SEMs against two Ordinary Least Squares (OLS) specifications.

Main Results:

• SEMs demonstrated a superior fit to layered exposure data compared to OLS.
• OLS specifications exhibited coefficient bias, including downward bias and overstated smoking effects.
• The SEM identified significant covariates that were not significant in OLS models.

Conclusions:

• Recommends wider adoption of SEMs in occupational health research.
• SEMs provide a more realistic framework for modeling multiple exposure pathways.
• SEMs can reduce exposure misclassification bias and strengthen exposure-disease outcome linkages.