Meta-analysis of studies of lung cancer among silicotics.

The relation between exposure to crystalline silica and lung cancer has been a controversial topic, and findings have appeared inconsistent. In this paper, we focus on lung cancer risks in epidemiologic studies of silicotics. We abstracted data from 29 studies for quantitative evaluation. We identified several studies that suffered from biases due to competing risks of different causes of death--in particular, death due to silicosis itself. After adjustment for competing risks, all 29 studies demonstrated lung cancer relative risk (RR) estimates greater than one. The pooled RR estimate for the 23 studies that could be combined was 2.2, with a 95% confidence-interval (CI) of 2.1-2.4. The pooled estimates by study design were 2.0 (95% CI = 1.8-2.3) for cohort studies and 2.5 (95% CI = 1.8-3.3) for case-control studies. The proportional mortality studies combined gave a summary RR of 2.0 (95% CI = 1.7-2.4), whereas the studies of cancer incidence gave a summary RR of 2.7 (95% CI = 2.3-3.2). Although statistical tests demonstrated heterogeneity between studies, and the confidence intervals given above may therefore be a little too narrow, the overall findings could not be attributed to chance, confounding by smoking, or other sources of bias. We conclude that the association between silicosis and lung cancer is causal, either due to silicosis itself, or due to a direct effect of the underlying exposure to silica.