Meta-analysis of benzene exposure and non-Hodgkin lymphoma: biases could mask an important association.

OBJECTIVES Benzene is a widely recognised cause of leukaemia but its association with non-Hodgkin's lymphoma (NHL) is less well established. The goal of this project is to review the current published literature on this association. METHODS We performed a meta-analysis of cohort and case-control studies of benzene exposure and NHL and a meta-analysis of NHL and refinery work, a potential source of benzene exposure. RESULTS In 22 studies of benzene exposure, the summary relative risk for NHL was 1.22 (95% CI 1.02 to 1.47; one-sided p value = 0.01). When studies that likely included unexposed subjects in the "exposed" group were excluded, the summary relative risk increased to 1.49 (95% CI 1.12 to 1.97, n = 13), and when studies based solely on self-reported work history were excluded, the relative risk rose to 2.12 (95% CI 1.11 to 4.02, n = 6). In refinery workers, the summary relative risk for NHL in all 21 studies was 1.21 (95% CI 1.00 to 1.46; p = 0.02). When adjusted for the healthy worker effect, this relative risk estimate increased to 1.42 (95% CI 1.19 to 1.69). CONCLUSIONS The finding of elevated relative risks in studies of both benzene exposure and refinery work provides further evidence that benzene exposure causes NHL. In addition, the finding of increased relative risks after removing studies that included unexposed or lesser exposed workers in "exposed" cohorts, and increased relative risk estimates after adjusting for the healthy worker effect, suggest that effects of benzene on NHL might be missed in occupational studies if these biases are not accounted for.