Active and Passive Smoking in Breast Cancer

To the Editor: Egan et al. 1 have provided a detailed analysis and thoughtful discussion of passive and active smoking and breast cancer. However, at least three aspects of the study—the occupational passive exposure measure, the choice of referent groups, and the cohort studied—are likely to have resulted in substantial passive-smoking exposure misclassification. This may have contributed to their largely negative results that are apparently at odds with the positive results for passive smoking and breast cancer in all 10 studies using breast cancer incidence as an endpoint and one of two studies using breast cancer mortality as the endpoint, as reviewed by Morabia. 2 Furthermore, active-smoking risk was positive for all 10 incidence studies and one of the two mortality studies when the referent group excluded never-smokers who had reported regular exposure to second-hand smoke. The first concern is that the only measure of occupational passive smoke exposure collected was current exposure in 1982. Therefore, the analysis is dependent on a single proxy measure for lifetime occupational exposure over the likely 15 to 40 years of employment before 1982 and employment after 1982. The very limited study of passive-smoke exposure in hospitals 3,4 concurs with what nurses have suggested anecdotally, that many hospital cafeterias, nursing stations and medical staff lounges historically had high levels of second-hand smoke. Second, none of the referent categories for the passive smoking analyses presented excluded all women who reported regular passive-smoking exposure. For the passive-smoking analyses it would be prudent to use as a referent group women who reported no regular exposure to passive smoking, as a child or an adult. Presenting separate analyses of pre- and postmenopausal women might also be helpful as premenopausal risks have been consistently more elevated in other studies. 2 Third, as a cohort defined by being registered nurses in 1976, almost the entire cohort of never–active-smokers may have had regular (unrecorded) exposure to passive smoking for several years during their in-school and/or in-hospital nursing training. Over half the cohort reported having been active smokers and most of those would have smoked during nursing training, as more than 85% of the active smokers reported started smoking by age 22, and two-thirds smoked for at least 20 years. 1 Most nurses probably took their training as young adults before having children, a potentially critical period for exposure. 1 Important exposure misclassification, particularly exposed never-smokers misclassified as nonexposed, can seriously dilute observed risks 5 and may explain the negative results in the other large American cohort study. 6 Furthermore, if there are antiestrogenic effects of tobacco smoke that impact risk, 7 critical periods of exposure, or a risk profile that deviates from a linear dose-response (as suggested by observation of similar magnitudes of risk associated with passive and active exposure in other studies), misclassification may be particularly important. Egan et al. 1 used the positive results of the coronary heart disease-passive-smoking analysis of the same cohort 8 to suggest the validity of their passive exposure measures. Because heart disease is understood to be particularly related to recent acute passive-smoking exposure, 9 the cross-sectional measure of passive smoking in 1982, just before the case detection period began, may have been a reasonably good measure for detecting a coronary heart disease-passive-smoking relation, but a poor proxy for examining earlier occupational exposure, which is likely to be more important in assessing cancer risk. Finally, Egan et al. 1 reported passive-smoking–controlled breast cancer relative risks of 1.15 (95% confidence interval = 0.98–1.34) for current active smokers and 1.17 (1.01–1.34) for ex-smokers, as well as a non–passive-controlled risk of 1.19 (1.03–1.37) for starting smoking before age 17, and 1.31 (1.07–1.61) for preparous smoking initiated before age 17. If the passive exposure misclassification had been reduced through design and analysis it is quite conceivable that a passive-smoking risk could have emerged; in turn, the passive-controlled active-smoking risks would then also have likely been higher. In conclusion, studies of an association of breast cancer with passive and active smoking may require lifetime histories of passive smoking covering the major sources of exposure back to childhood before one can be confident in the results of the risk assessment. Kenneth C. Johnson A. Judson Wells

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.