摘要：OBJECTIVES: We undertook a meta-analysis of epidemiological studies investigating the relationship between occupational diesel exhaust exposure and lung cancer. METHODS: Thirty of 47 studies initially identified as potentially relevant met specified inclusion criteria. We extracted or calculated 39 independent estimates of relative risk and derived pooled estimates of risk for all studies and for numerous study subsets by using a random-effects model. We also examined interstudy heterogeneity by using linear metaregressions. RESULTS: There was substantial heterogeneity in the pooled risk estimates for all studies combined and for most subsets. Several factors consistent with higher study quality, however, contributed to increased pooled estimates of risk and lower heterogeneity, including (1) adjustment for confounding by cigarette smoking and other covariates, (2) having a lower likelihood of selection bias, and (3) having increased study power. CONCLUSION: This analysis provides quantitative support for prior qualitative reviews that have ascribed an etiologic role to occupational diesel exhaust exposure in lung cancer induction. Among study populations most likely to have had substantial exposure to diesel exhaust, the pooled smoking-adjusted relative risk was 1.47 (95% confidence interval = 1.29, 1.67).
