More evidence on the risks of passive smokingBMJ 2005; 330 doi: https://doi.org/10.1136/bmj.330.7486.265 (Published 03 February 2005) Cite this as: BMJ 2005;330:265
- Ichiro Kawachi, professor of social epidemiology ()
- Department of Society, Human Development and Health, Harvard School of Public Health, Boston, MA 02115, USA
But existing evidence is enough to implicate it as a health hazard
Recently the International Agency for Research on Cancer (IARC) reviewed more than 50 studies of passive smoking and lung cancer.1 The pooled excess risk of lung cancer from exposure to spousal smoking was about 20% for women and 30% for men. So what do the two articles featured in this issue add to the evidence on the health hazards of passive smoking, which is already quite extensive (277, 287)?2 3
Compared with the large number of cases of lung cancer (about 6700) available for the IARC meta-analysis, the report from the EPIC study makes a relatively modest contribution (about 100 newly diagnosed cases).2 The adjusted hazard ratio for lung cancer among people exposed to passive smoking was 1.70 (95% confidence interval 1.11 to 2.60). Some methodological limitations have been noted in previous studies of passive smoking, including inadequate control of confounding and misclassification of exposure—for example, the use of spousal smoking status as a proxy.4 The strengths of the EPIC study included its prospective design, control for potential confounding variables such as diet, and the ability to look for effect modification by polymorphisms in genes involved in carcinogenesis.
Some researchers have advocated using the naturally occurring phenomenon of Mendelian randomisation as an approach to get rid of the potential bias caused by omitted variables in studies of passive smoking and cancer.5 Mendelian randomisation occurs naturally during the formation of gametes and offspring and can sometimes provide a study design similar to randomised comparison. With respect to lung cancer and environmental tobacco smoke, the goal is to identify the naturally occurring polymorphisms in genes that are associated with impaired detoxification of carcinogens present in tobacco smoke. In theory, these should be related to increased risk of cancer among people exposed to secondhand smoke, whereas “among people unexposed to the carcinogens in environmental tobacco smoke there is no reason to believe that the detoxification polymorphisms should be related to risk of lung cancer.”5 The problem, however, is that these polymorphisms (including several examined in the EPIC study) are not specific to the carcinogens in tobacco smoke but are also involved in the metabolism of carcinogens derived from sources other than secondhand smoke.
An alternative approach, as yet untried, to identify the causal effect of passive smoking on risk of disease would be to take advantage of “policy randomisation” or natural experiments resulting from, for example, examining variations at the level of the city or state in the introduction of smoke free policies with respect to mortality risks among non-smokers.
A further aspect of the EPIC study deserves comment. Self reports of passive smoking in that study were strongly correlated with serum concentrations of cotinine, but the latter did not predict the risk of lung cancer. The authors say that cotinine is a poor biomarker of carcinogenic risk. Other studies have found that among non-smokers exposed to secondhand smoke, the urinary concentrations of carcinogenic N-nitroso compounds specific to tobacco smoke are 1-5% of those found in active smokers, which is roughly proportional to the excess risk of lung cancer reported in epidemiologic studies.6 In contrast to respiratory cancer, however, a recent study found an association between serum concentrations of cotinine in non-smokers and elevated risk of coronary heart disease (adjusted hazard ratio 1.57, 1.08 to 2.28), comparing the top to bottom fourth of cotinine concentrations.7
The second study in this issue examined the association between home exposure to secondhand smoke and all cause and cause specific mortality (p 287).3 The study, based in Hong Kong, was limited because of its retrospective assessment of passive smoking through proxy respondents and minimal adjustment for potential confounders. On the other hand, the study adds to the small but growing number of studies that implicate passive smoking as a cause of stroke.
The bottom line is, do these two studies change public policy in relation to restrictions on secondhand tobacco smoke? The answer is no. Existing evidence is already sufficient to implicate passive smoking as a cause of lung cancer and coronary heart disease. Moreover, smoke free workplace policies are effective in eliminating secondhand smoke and in encouraging active smokers to quit.8 Eliminating exposure to secondhand smoke is a public health priority not just for European countries but for the rest of the world as well.
Competing interests None declared