Towards causal inference in occupational cancer epidemiology--II. Getting the count right.

This second paper in the series assessing causal inference in occupational cancer epidemiology highlights the importance of using the right data sources when examining the association between exposure and tumour incidence in groups of workers. Using the wrong ones can lead to omissions which become insidious and cumulative during a long-term follow-up study, often extending over several decades. The main cause is the changing survival pattern of some cancers and it is a new and important factor to be taken into account. This was illustrated during an in-depth study of an excess of bladder cancer at a large tire factory, where the workforce was exposed to beta-naphthylamine-contaminated antioxidants used in processing. At first the study was confined to a mortality analysis using death certificates to identify those whose underlying cause of death was bladder cancer [International Classification of Diseases (8th Rev.), No. 188]. It soon became apparent, however, that several cases had outlived the study end-date, or had died of an unrelated cause, often with bladder cancer not being mentioned anywhere on the death certificate. The study was then changed also to include morbidity (incidence) data using cancer registration. Between 1946 and 1949 some 2090 men were exposed to two antioxidants contaminated with about 0.2% of beta-naphthylamine. Altogether there were 45 bladder tumours relevant to this 'at risk' group, whereas only 24.8 were expected at the national incidence rates (P < 0.001), which clearly reflects the group's exposure to a recognized bladder carcinogen.(ABSTRACT TRUNCATED AT 250 WORDS)