Tobacco Documents Online

TR664 REVIEW 397 CONFIDENTIAL Subiect ref.12k "Interactions between smoking and other exposures: Occupation and'Diet" S.D.Stellman Banbury Report 23 : Mechani~sms in Tobacco Carcinogenesis (1986) Cold Spring Harbor Laboratory, 377-395 The question of interaction between smoking an&other exposures is an important one. In some ways, this is a usefuli paper, both because it provides a helpful list of references to papers that have studied the question in theoretical terms and more so because it provides data from quite a range of studies where evidence is available. Thus, evidence in relation to asbestos/shipyard working„ woodworking, underground mining (radon), metal working, occupational exposure to industrial carcinogens, alcohol, green/yellow vegetables, vitamin A supplements and fruit juices are all considered. While the evidence generally seems to indicate that at least in some circumstances a multiplicative model exists, I found the paper somewhat disappointing and superficial (though this may be because of limited'itime for presentationiat a conferenee). One weakness was any real discussion as to the relative implications of data showing,a multiplicative relationship:and data showing an additive relationship. To my mind, one important

-2- inference is that the first implies the two agents operate on different processes, while the second implies they may operate on the same process (and have similar biological effects). A second weakness is that all the comparisons with predictions from the additive and multiplicative model are based on rates and impl~icitly assume there is no random error associated with the estimated risk from exposure to neither agent or the agents individually. One cannot tell for any example whether any apparent superiority of one model over another is statistically significant or not. I would have presented observed numbers of deaths (or onsets) in each of the 4 cells, together with predicted numbers and results of a chisquared test of departure from the 2 models. A third weakness is the inadequate discussion of misclassification. It is noted that miscliassification of just a few subjects can lead to a large bias in the apparent effect of the factor other than smoking - but this is not the whole story and one would have liked to see more. Also - a point alluded to by Tannenbaum in the discussion - the approach is relatively unsophisticated because it fails to take into account time of exposure. A then B may produce a different effect from B then A. Table 8 of the paper is interesting. It is amazing that in this day and age any study of occupation on risk of a tobacco-related disease should fail to record smoking habits. That half still fail to do so in 32 original articles published in the Journal of

-3- Occupational Medicine in 11984 is certainly worthy of mention (though it is interesting to note that "sleep" is included in the tobacco-related diseases!). P.N.Lee list of 9.2.88'