Exposure to environmental tobacco smoke (ETS) and risk of lung cancer in Montreal: a case–control study

Less than 10% of all lung cancer cases occur among never smokers [1]. Environmental tobacco smoke (ETS) is one of the factors that have been linked to lung malignancies among these subjects. Several epidemiological studies conducted in North America, Asia, and Europe and two meta-analyses showed a slight increase in the risk of lung cancers among never smokers exposed to ETS, particularly among female non-smokers married to smokers [2, 3], and the International Agency for Research on Cancer (IARC) concluded that there is sufficient evidence that ETS exposure causes lung cancer in humans [4, 5]. Studies focused on sources of ETS other than spouses have also found a slight increased risk of lung cancer, such as those that examined the effect of ETS during childhood [6], in the workplace [7, 8], or spouse, workplace and social sources [9]. But the available information is not definitive. Only a few have examined the effect of ETS exposure at different ages [10–12], and very few studies have looked at a host of possible sources of ETS. Most studies have involved very small sample sizes and because it is difficult to assemble large samples of nonsmoking lung cancer cases, it is necessary to increase the number of studies in order to improve the collective ability to discern patterns of risk.

Second-hand tobacco smoke is present in all places where smoking takes place: at home, in the workplace, in bars, restaurants, public buildings, and public transport [3]. The setting that represents the most important source of exposure varies by locale and time period, and it is very unlikely to find completely unexposed populations in urban areas.

In the late 1990s we carried out a population-based case–control study in Montreal, Canada, to examine the possible associations between hundreds of occupational substances and respiratory cancers. In addition to these factors, we set out to assess exposure to ETS among the nonsmoking cases and controls. The purpose was to examine the risk of developing lung cancer among never smokers as a consequence of ETS exposure from all sources combined: inside homes –parents, spouses, and other co-residents, as well from outside homes –workplace, social settings, and public and private transportation. In addition, we examined the effect of exposure at different life stages.

A total of 44 never-smoking cases and 436 never-smoking controls were included in the present analysis. There was a similar distribution by age and sex between cases and controls (Table 1). A higher proportion (27%) of proxy interviews were conducted for cases than for controls (6%). Controls had on average more years of formal education than cases; 29% reached postsecondary school compared to 23% of cases. Median census tract home income was similar for cases and controls. More than half of the cases were diagnosed as adenocarcinomas.

While there were not sufficient numbers of cases of specific histologic subtypes to warrant presentation of detailed analyses by subtype, we did carry out analyses in particular for adenocarcinoma of the lung. The results were not noticeably different from those shown for all histologic types combined. For instance, combining all sources of ETS and comparing the top tertile of pack-years with the lowest tertile (as per Table 5) gave an OR of 0.95 (95%CI 0.43-2.29).

There have been several studies investigating the association between ETS and lung cancer [2–6]. While a consensus has developed that ETS is a risk factor for lung cancer, in fact results from studies conducted to date have been very heterogeneous [2], and further exploration of this hypothesis remains an important one to explore, both overall and in terms of characterizing the possible risks by type of exposure to ETS and type of lung cancer. Because of the difficulty of assembling a large series of nonsmoking lung cancer cases in one locale, except for large international collaborations, most studies have been small, including our own. But it is also a reason for producing and publishing as many valid studies as possible, so as to provide the basis for future syntheses and meta-analyses.

The study on which this analysis was based was designed primarily to examine the risk of lung cancer related to occupational exposures. As in any case–control study of lung cancer, the proportion of cases who were lifelong nonsmokers was very low, and even though the total number of cases in our study was over 1200, the number of never smoking cases was only 44. This is the main limitation of this study, although this is somewhat palliated by the fact that exposure to ETS is quite prevalent. As a result of the low numbers, the estimates of OR from our analysis alone are quite imprecise. But this is typical of most studies of ETS and cancer. We believe that our results are useful, not only because they stand by themselves, but because they may be included in future meta-analyses or pooled analyses, which may accumulate sufficient numbers of subjects to achieve more precise results. In addition, to our knowledge, there is no other published study assessing as many sources of ETS as the present manuscript.

In the present manuscript we present results for ETS exposure from individual sources, namely parents, spouses, other co-residents, and outside home sources. Results for ETS exposure outside homes were associated with higher risk than for exposure inside homes. This is in accordance with the several studies that reported higher intensity and more consistent exposure in the workplace, an exposure that occurs outside homes. From risk assessment perspective, the results presented in Table 5 for TES exposure from all sources are considered the most relevant. Also it is not surprising that combining all sources of ETS, very few subjects reported never have been exposed, especially in a population that was mostly born before 1960 and grew up in a society where smoking was very common.

Strengths of this work include the detailed information collected among never smokers, including ETS exposure from several different sources and during different life stages. In addition, we adjusted the analyses for several potential confounders, including exposure to occupational carcinogens and socioeconomic status. Control for socioeconomic status is particularly important since some studies suggested that subjects with lower socioeconomic status have more intense and longer workplace exposure, which results in higher cumulative lifetime ETS exposure [7, 8]. Our study also looked into the effects of increasing levels of combined household and workplace exposures, which better represents total ETS exposure. Limitations of our study include the already cited low number of cases, and exposure misclassification that inevitably occurred in attempting to retrospectively estimate whether or not each subject was exposed, and the magnitude of this exposure. It is very likely that subjects who reported never exposure to ETS had indeed been exposed. However, it is also likely that their cumulative level of exposure is low, and they served well as a reference group for those subjects with high levels of exposure.

We failed to find a significant association between ETS and lung cancer. Possible explanations include the small sample size and the small ETS exposure levels. Measurement bias resulting from inaccuracies in the assessment of individual exposure to ETS may lead to error in estimation of the risk. Subjects may not recall all exposure sources especially outside the home. For example, it was reported that for 25-50% of women who work outside the home, the spouse’s smoking habit may not accurately reflect their overall exposure to ETS [17]. Similarly, controls who report no ETS exposure may in fact be exposed to other sources in the workplace or the public. If this misclassification of exposure is equal among cases and controls it will lead to underestimation of the risk [18, 19].

Regarding ETS exposure inside homes, which mainly occurs at younger ages, we did not find a significant increased risk of lung cancer. These findings are not in disagreement with previous investigations. Some of the largest studies did not find an association between exposure during childhood and risk of lung cancer [11, 12, 20, 21], including a multi-center study reporting that ETS exposure during childhood was associated with reduced risk of lung cancer: OR = 0.78 (95%CI 0.64-.96) [7]. On the other hand, recent investigations have shown an increased risk of lung cancer associated with ETS exposure during childhood. In a large cohort study conducted in 10 European countries (European Prospective Investigation into Cancer and Nutrition, EPIC), it was estimated that the hazard ratio (HR) for lung cancer risk from ETS exposure during childhood was 2.00 (0.94–4.28); among children with daily exposure for many hours each day the hazard ratio was 3.63 (1.19–11.12) [6].

Other studies have also reported increased of lung cancer associated with ETS exposure at younger ages [10, 22]. Recent investigations that examined ETS during adulthood have also shown an increased risk of lung cancer. A meta-analysis estimated an OR of 1.27 (95%CI: 1.17-1.37) in the risk of lung cancer among non-smoking women exposed to spousal ETS [4]. A second meta-analysis showed a relative risk of lung cancer among nonsmoking women ever exposed to ETS from their husbands of 1.20 (95%CI: 1.12-1.29) [5]. Studies that focused on other sources of ETS have also found a slight increased risk of lung cancer, such as in the workplace [7, 8], or combining spouse, workplace, and social sources [9].

We conducted several exploratory analyses, including comparing risk in groups younger than 65 years of age at diagnosis versus older than 65, males versus females, and risk due to exposure at work only. None of these analyses resulted in increased risk. In a recent meta-analysis of 22 studies, ETS exposure in the workplace was linked to increased risk of lung cancer (OR = 1.24, 95%CI: 1.18-1.29) [23]. In the EPIC cohort, workplace exposure to ETS was linked to lung cancer among women: HR of 2.13 (1.6–3.4) [24]. Despite not detecting risk due to ETS exposure at work (data not shown), our results suggest an increased risk of lung cancer for subjects who were exposed from outside the home.

A large proportion of the cases (52%) in our study were adenocarcinoma. In our analyses, based on small numbers, there was no noticeable difference in ETS ORs for adenocarcinoma and for all tumour types combined. Evidence from previous literature is conflicting. Several studies reporting results on different histological types of lung cancer found that risk of other types was higher than that of adenocarcinoma [7, 12, 20, 21, 25, 26], some found no differences by histologic type [27, 28], while another study reported higher risk for adenocarcinomas [8].

Regardless of the effect on lung cancer, ETS exposure is far from harmless and it should be reduced and controlled. It contributes to significant morbidity and mortality among children, including preterm birth, intrauterine growth retardation, perinatal mortality, respiratory illness, neurobehavioral problems, and decreased performance in school [29].

We did not find a clear association between lifetime ETS exposure from all sources and increased risk of lung cancer. Among the sources of ETS that were examined, only exposure outside the home, in adulthood, showed any indication of excess risks. Small numbers preclude strong inferences from this study on its own.