Causal inference in occupational epidemiology: accounting for the healthy worker effect by using structural nested...

Area of Science:

• Epidemiology
• Occupational Health
• Biostatistics

Background:

• The healthy worker effect (HWE) is a well-documented phenomenon in occupational epidemiology.
• Previous research has highlighted the potential for HWE to confound exposure-effect relationships.
• Understanding and quantifying HWE is crucial for accurate risk assessment in various occupations.

Purpose of the Study:

• To provide historical context and a clear definition of the healthy worker effect using causal diagrams.
• To illustrate the application of structural nested models (SNMs) for estimating exposure effects while accounting for the healthy worker survivor effect.
• To offer practical guidance with annotated SAS code for implementing SNMs.

Main Methods:

• Definition of the healthy worker effect through causal diagrams.
• Simulation of data to demonstrate the utility of structural nested models.
• Step-by-step illustration of SNM application for exposure effect estimation.
• Provision of annotated SAS code for reproducible analysis.

Main Results:

• Demonstration that structural nested models can effectively adjust for the healthy worker survivor effect.
• Empirical evidence supporting the potential for HWE to influence occupational health study findings.
• Methodological framework provided for researchers to implement HWE adjustments.

Conclusions:

• Structural nested models offer a robust approach to mitigate bias from the healthy worker effect in epidemiological research.
• Accurate estimation of occupational exposure effects requires explicit consideration and adjustment for the healthy worker survivor effect.
• The provided methodology and code facilitate the application of these advanced statistical techniques in practice.