Background: Job titles or work areas are often used as surrogate indicators of exposure in occupational epidemiological studies. In this article, we assess the validity and comparability of commonly used surrogate indicators.
Methods: We analyzed lung cancer mortality among a hypothetical and an actual cohort of rubber workers. Surrogate indicators of exposure were defined according to jobs in which workers were "only," "ever," "longest" or "last" employed, or in which they were employed at the "census" of the study. Occupational risks were estimated using standardized mortality ratios. Validity of surrogate indicators was assessed in the simulated data by comparison between estimated effects and the known underlying associations. Comparisons of surrogate indicators were conducted in both simulated and empirical data.
Results: Use of the definition "only" as the surrogate indicator gave valid but imprecise results. For all other definitions, we observed a moderate overestimation of risks in no-risk or low-risk jobs and attenuation of underlying dose-response relationships, without substantial differences among the applied definitions.
Conclusions: Our results demonstrate a limitation of using surrogate indicators of exposure in occupational epidemiological studies. However, they suggest that the inconsistencies of published study findings in the rubber industry are unlikely to be attributable to the use of different surrogate indicators.