Contrasting Causal Effects of Workplace Interventions

Area of Science:

• Occupational health
• Epidemiology
• Biostatistics

Background:

• Occupational exposure guidelines often rely on static intervention models assuming constant exposure.
• Estimating static effects is challenging with long-term follow-up beyond employment, requiring additional assumptions.
• Dynamic interventions, where exposure ceases upon leaving work, are more identifiable but produce distinct estimates.

Purpose of the Study:

• To investigate the factors driving discrepancies between static and dynamic occupational exposure effect estimates.
• To analyze how intermediate adverse health events, acting as time-varying confounders, influence these estimates.

Main Methods:

• A simulation study was conducted to model worker employment termination due to intermediate adverse health events.
• The study explored the relationship between health events, employment termination, and outcome estimates under different intervention scenarios.

Main Results:

• Effect estimates from static and dynamic interventions converged as the strength of the health event-outcome relationship increased.
• Estimates became more similar with longer latency between the health event and employment termination.
• Discrepancies were largest when the intermediate health event occurred early in employment, acting as a screening mechanism.

Conclusions:

• Intermediate adverse health events can significantly influence the identifiability and estimation of occupational exposure effects.
• The timing of health events relative to employment termination is crucial in understanding differences between static and dynamic exposure models.
• Findings have implications for setting accurate and practical occupational exposure limits.