This population-based study of men employed across a diverse range of jobs found that workplace exposure to asbestos was associated with an increased risk of lung cancer. This association persisted after adjusting for cigarette smoking, second hand smoke, and other occupational exposures previously implicated as possible risk factors for lung cancer. The approximate 28% increased risk observed among men ever exposed to asbestos is similar to the finding of Pintos et al.
. In their Montreal based case–control study, those who were exposed to asbestos had an odds ratio of 1.21, (95% CI=0.98-1.49) relative to those with no exposures. The population attributable risk (PAR) percent is often used to provide an estimate of the percentage of cases that be avoided if the putative exposure was eliminated
 . We calculated the PAR in our study using the odds ratio of 1.28 among ever exposed, and an estimated prevalence of exposure of 11.3% (based on our control series). This yielded a PAR of 3.1% which suggests that a relatively small percentage of Canadian male lung cancer cases are due to occupational exposure to asbestos. Based on an estimated 13,300 incident lung cancers among men in Canada in 2012
 this would account for approximately 412 incident cases.
Our study provided support for a dose–response relationship between asbestos exposure and lung cancer as higher risks were observed among those who were ever exposed to ‘medium’ or ‘high’ concentrations of asbestos. Pipefitters accounted for nearly half of these cases and controls (41 of 87). While the limited number of subjects did not allow us to characterize risks for specific types of jobs, our results are consistent with a previously published study of Ontario pipe trade workers
. They reported a 53% increased risk of lung cancer mortality among pipefitters who had been registered trade members for at least 30 years, relative to the Ontario general population. However, their study was somewhat limited due to a lack of data on smoking. Our findings support the hypothesis that asbestos and cigarette smoking affect the risk of lung cancer in a multiplicative fashion.
In many occupational studies, duration of exposure is regarded as valid surrogate measure of cumulative exposure due to the inherent difficulties in retrospective studies to precisely characterize exposure intensity. In their Montreal case–control study, Pintos et al. found a higher risk of lung cancer among those exposed to asbestos for at least 20 years when compared to those exposed for shorter durations
. Duration of exposure was also positively associated with lung cancer risk in other industry-specific cohorts
. In contrast, we found that only intensity but not duration of exposure was associated with statistically significant increased risks of lung cancer. This observation is consistent with recently published findings on a cohort of workers employed in an asbestos reprocessing plant in the Calvados region of France
. In this study, Clin and colleagues observed that the average exposure to asbestos expressed in terms of fibers per ml was associated with pleuro-peritoneal mesothelioma, lung cancer, and colorectal cancer (p<0.05), however, no statistically significant associations were evident with duration of exposure for any of these three cancer sites. Other studies of asbestos workers have also found associations with intensity but not duration of exposure
[12, 30, 31]. Our finding of a stronger positive association between duration of exposure at medium or high levels of asbestos when compared to durations spent at lower levels suggests that time exposed above a threshold level may be a relevant marker of risk. However, this finding should be interpreted cautiously as it based on a very small number of subjects who were exposed to either medium or high intensities.
It is well recognized that there is a lengthy latency period between the time of first exposure to an environmental carcinogen and the development of a solid tumour such as lung cancer. For example, the latency period associated with cigarette smoking and lung cancer has been estimated to be several decades following the initiation of smoking
. By extension, the increased risks of lung cancer due to exposure to asbestos observed in this study are a reflection of workplace exposures many years if not decades earlier. Indeed, among those classified has having ‘medium’ or ‘high’ concentrations to asbestos in the workplace, the start date of employment was after 1980 in only 6% of these jobs.
Participants in our study were asked to provide information for only those jobs that were held for at least one year. The exclusion of these short-term jobs raises the possibility that some exposure misclassification has been introduced. Previous analysis of 27.5 million workers found increased risks of lung cancer among those exposed to high levels of asbestos (20 to 40 fibers per cubic centimeter of air) for only a few months
. Under a classical error model where the possible exposure misclassification error arising from excluding these short term jobs is non-differential to case–control status, our risk estimates would be understated.
An important strength of this study was the availability of other risk factor data obtained through both the questionnaire, as well as expert-based coding of occupational histories. Unlike many other occupational case–control studies, we had extensive data on cigarette smoking, most notably, exposure to second hand smoke. This measure allowed our risk estimates to take into account lifetime exposure to second-hand smoke incurred at both home and workplace settings. In addition, the industrial hygienists also coded each job for possible exposure to other known or suspected lung carcinogens including: crystalline silica, gasoline and engine emissions. We recently found that occupational exposure to diesel but not gasoline engine emissions increased the risk of lung cancer; the risk of lung cancer was also increased among individuals exposed to crystalline silica
. The addition of these two covariates (diesel and silica) strengthened the association for asbestos by approximately 20%.
Approximately 68% of eligible cases and 64% of eligible controls completed a questionnaire. This raises the potential to introduce some bias in our risk estimates, and our results should be interpreted cautiously because of this possibility. However, for several reasons, we do not believe this bias fundamentally changes our results. First, observed associations with known and suspected risk factors such as cigarette smoking, and exposure to second-hand smoke are similar in direction and magnitude to risk estimates reported in other epidemiological studies. Moreover, our published findings for other occupational exposures within the same study population
 are also consistent with the epidemiological literature. Lastly, the distribution of lung cancers by histology in our study is remarkably similar to population-based figures for North America
 and provides some support for the generalizeability of these results to incident lung cancers in Canada. Unfortunately, the NECSS did not collect data from those diagnosed with mesothelioma, and therefore, we were unable to investigate associations with this endpoint.
We were unable to distinguish asbestos on the basis of fiber type. Asbestos fibers can be described according to two broad classes serpentines (phyllosilicates) and amphiboles (inosilicates) that differ substantially with respect to biopersistence and physical and chemical properties. Serpentines include chrysotile asbestos which is the predominant type of asbestos in Canada. The International Agency for Research on Cancer has determined that there is sufficient evidence to conclude that all these forms of asbestos can cause cancer in humans
[4, 6]. There remains considerable uncertainty regarding differences in lung cancer risk resulting from exposure to different types of asbestos fibers. A review of cohort studies where quantitative measurements of asbestos exposure were available demonstrated clearer and consistent associations between exposure and lung cancer for crocidolite or amosite
. On the other hand, associations from cohorts exposed primarily to crysotile asbestos were less consistent
[37, 38]. It is generally accepted that amphibole fibers are more harmful than chrysotile fibers for mesothelioma
[36, 39]. However, it has been argued that these differences are not all that important given that chrysotile is the most commonly used type of asbestos
[40, 41]. In our study, those who were determined to have been exposed to asbestos were believed to have been exposed to chrysotile, however, it is possible that some exposure to less prevalent yet more potent types of fibers occurred and was unaccounted for.
Another limitation of our study was the relatively small number of study subjects who were ever exposed to medium or high levels of asbestos. In total, there were only 39 cases and 24 controls exposed at these levels. These small numbers hindered our ability to characterize the joint relationship between smoking and asbestos exposure on the risk of lung cancer. It also limited our examination of the risks of lung cancer with exposure to asbestos according to different histological subtypes. Several studies have found associations that were most pronounced for adenocarcinoma subtypes
[28, 42–44], however, others did not
[45–47]. The three most common histological types of lung cancer in our study population were squamous cell carcinoma (35%), adenocarcinoma (28%), and small cell carcinoma (15.9%)
. When we restricted analysis to adenocarcinoma, the odds ratio among those exposed to medium or high levels of asbestos increased from 2.16 to 3.14 (95% CI=1.50 – 6.58). However, the latter estimate was based on only 13 incident cases and therefore, our study has very limited statistical power to make inferences by histological type.