Document resource
Introduction A recent meta-analysis confirmed that chest X-ray (CXR) has low sensitivity for diagnosing silicosis. We re-estimated previously published dose-response relationships between cumulative respirable crystalline silica (RCS) exposure and silicosis risk, under the assumptions that sensitivity was either fixed or relative to the population proportion of severe silicosis.Methods We combined unpublished logistic regression models from Scottish coal miners with meta-analysis results to model how CXR sensitivity changed according to cumulative RCS exposure. We assumed specificity was 0.95. Among mining cohorts, we calculated the difference in the cumulative risk of silicosis between the unadjusted and fixed and relative scenarios. Finally, we re-estimated a published dose-response meta-analysis and associated absolute risk reductions (ARR).Results The cumulative risk of silicosis was substantially higher in both the fixed and relative sensitivity scenarios compared with the unadjusted estimate in all mining cohorts. This was most pronounced in the relative scenario and when cumulative RCS exposures were below approximately 6 mg/m³-years. A reduction in cumulative RCS exposure from 4 to 2mg/m³-years corresponded to larger ARRs in the fixed and relative scenarios than the unadjusted scenario; 382 (95% CI 361 to 399) and 529 (95% CI 353 to 592) cases per 1000 miners compared with 313 (95% CI 288 to 333) cases per 1000 miners, respectively.Discussion We relied on a single estimate of the proportion of severe disease to link sensitivity and cumulative RCS exposure. Nevertheless, adjusting for the reduced diagnostic accuracy of CXR for silicosis suggests the burden of silicosis is underestimated in published mining cohorts.