The Strength of the Healthy Worker Effect Varies According to the Type of Cancer Being Studied

Abstract

PP-29-199 Background/Aims: One type of healthy worker effect is introduced when estimating health risks in a working population by comparing it with the general population. To study how this bias varies according to specific cancer types, we compared mortality and incidence rates in a large cohort of workers being employed in the whole range of occupations and the general population. Methods: The cohort was established using the Norwegian Registry of Employers and Employees, and comprised 366,114 subjects from the general working population in Norway randomly drawn as referents in a cohort study of petroleum workers. The cohort was linked to the Norwegian Cause of Death Registry and Cancer Registry of Norway including all deaths (n = 10,002) and cases of cancer (n = 11,271) reported up to 2003. Results: There was a marked healthy worker effect in the working population for both overall mortality (standardized mortality ratio: 0.73, 95% confidence interval [CI]: 0.72–0.75) and overall cancer (standardized incidence ratios [SIR]: 0.91, 95% CI: 0.89–0.93). The male working population had markedly lower incidence of cancer of lip (SIR: 0.67, 95% CI: 0.47–0.91), pharynx (SIR: 0.80, 95% CI: 0.64–0.99), esophagus (SIR: 0.60, 95% CI: 0.46–0.78), pancreas (SIR: 0.85, 95% CI: 0.73–0.98), larynx (SIR: 0.74, 95% CI: 0.59–0.92), lung and bronchus (SIR: 0.81, 95% CI: 0.76–0.87), prostate (SIR: 0.93, 95% CI: 0.88–0.98), kidney (SIR: 0.83, 95% CI 0.73–0.94), bladder (SIR: 0.77, 95% CI: 0.70–0.85), and leukemia (SIR: 0.80, 95% CI: 0.68–0.92). In contrast, there was an elevated and significant risk of malignant melanoma both for men (SIR: 1.09, 95% CI: 1.02–1.17) and women (SIR: 1.29, 95% CI: 1.14–1.45), in addition to an increased risk of ovary cancer (SIR: 1.32, 95% CI: 1.1–1.5). Conclusion: Depending on the type of cancer and population being studied, there is a marked potential for both under- and overestimation of risk of cancers when using the general population as the reference for studies of specific working populations.