In this issue of the journal (see p 789), Rushton and colleagues1 report the first detailed estimate of the burden of occupational cancer in the UK. Estimates for all but leukaemia are greater than those currently used in the UK. During the last 25 years a variety of methods and assumptions have been employed to estimate the local, national and global burdens of occupational cancer resulting in attributable fractions ranging from less than 1% to about 40%.

In 1981, Doll and Peto2 estimated that about 4% of all cancer deaths and 12.5% of lung cancer deaths (15% in men) in the USA were attributable to occupational exposures. Lung cancer accounted for almost 70% of occupational cancers and at least 1%–2% of all cancer deaths were ascribed to asbestos (see Box 1). These estimates have been widely quoted although they were criticised for various methodological reasons, for example for accepting only “definite” occupational carcinogens but “highly speculative” evidence for dietary risk factors.

The World Health Organization’s (WHO’s) Comparative Risk Assessment (CRA)3 provided the most comprehensive and coherent estimates of mortality for more than 135 causes of disease and injury. Due to stringent data requirements only selected occupational carcinogens for lung cancer, mesothelioma and leukaemia were included. The selected carcinogens accounted for 9% of the global lung cancer deaths. The percentages for Western Europe and the USA were 6% and 5%, respectively. Important limitations of the CRA estimates stem from the lack of reliable exposure data for most of the countries, and uncertainties about the strength of the exposure-disease associations.

Several other estimates using reliable national data have been published and can be informative. Steenland et al4 relied on US exposure data (primarily from the National Occupational Exposure Survey, 1980–1983) to estimate the burden of occupational disease in …