Meta-analyses of published epidemiological studies, 1979-2006, point to open causal questions in silica-silicosis-lung cancer research

Abstract

Background and objectives: Following up on a previous meta-analysis of lung cancer risk in individuals without silicosis, we provide more detailed results of silica associated lung cancer risk in both silicotics and non-silicotics. The objective was to examine in depth whether current data allows to answer the pressing question "does silica cause lung cancer in the absence of silicosis"?

Methods: We updated earlier meta-analyses of silicosis and lung cancer and compared the results with our 2009 meta-analysis of risks in individuals without silicosis. We performed fixed (FE) and random (RE) effects meta-analyses, calculated heterogeneity statistics, stratified the study material, performed sensitivity analyses with modified study results and meta-regressions to detect effect modification.

Results: In silicotics, lung cancer risks were found to be doubled in 38 studies (FE: RR = 2.1; 95% CI = 2.0-2.3). In non-silicotics, eight studies without smoking adjustment suggested marginally elevated risks (FE: RR = 1.2; 95% CI = 1.1-1.3; RE: RR = 1.2; 95% CI =1.0-1.4) but three studies which were controlled for smoking showed null results (FE and RE: RR = 1.0; 95% CI = 0.8-1.3). Heterogeneity was substantial but could be linked to study characteristics, like sector of industry, and other second-level data in meta-regression. As no excess was observe dfor other smoking-related effects in studies ofllung cancer among non-silicotics, smoking was not considered to be an important confounder or modifier. CONCLUSIONn: Our meta-analyses further substantiate evidence of a strong association between silicosis and lung cancer. However, questions remain regarding lung cancer caused by silica in non-silicotics. Ideally, future investigations should consider the entire exposure-response range between silica exposure, silicosis development and lung cancer occurrence, and analyze data in terms of processes taking intermediate confounding into account.