Abstract

OBJECTIVE: To evaluate crystalline silica as a human carcinogen. METHODS: A cohort of 5115 men, born 1916-45 and employed in the pottery, refractory, and sandstone industries of Stoke-on-Trent was identified from occupations subject to health surveillance by the local Silicosis Medical Board (now the Department of Social Security). Detailed occupational and smoking histories, and records of small parenchymal opacities on periodic radiographs were extracted from medical records. An exposure matrix was derived from some 1400 personal or static dust samples and tested against the presence of small parenchymal opacities in a subcohort of 1080 men employed for at least 10 years, who had started working in the industry before 1960. RESULTS: Standardised mortality ratios (SMRs) calculated against mortalities for Stoke-on-Trent, with 95% confidence intervals (95% CIs), were raised for all causes (1.15 (1.05 to 1.26)), lung cancer (1.28 (0.99 to 1.62)) and non-malignant respiratory disease (2.04 (1.55 to 2.65)). Average concentration and duration of exposure to silica were, taken together, significantly related to the presence of small opacities (> or = 1/0). In a nested case-referent analysis of 52 cases of lung cancer and 197 matched referents, conditional logistic regression gave a significantly increased odds ratio (OR) for average silica concentration (micrograms/m3.100), after adjustment for smoking, of 1.66 (1.14 to 2.41) but not for duration of exposure nor, in consequence, for cumulative exposure. CONCLUSION: The association between risk of lung cancer and quantitative estimates of silica exposure supports the SMR analysis and implies that crystalline silica may well be a human carcinogen.