Tobacco smoking is the main cause for lung cancer worldwide with a 20-fold elevated risk linked to this factor.1,2 Evidence on a carcinogenic role of other risk factors should, therefore, be treated with caution because of possible confounding by smoking. Of particular concern are studies of occupational exposures, which result in risk estimates for lung cancer associations of small to moderate magnitude. The best way to estimate the independent effects of other potential risk factors of lung cancer is to restrict the study to never-smokers; however, studies that have examined occupation in relation to lung cancer often do not have sufficient power to restrict analyses to never-smokers.
Within the framework of a large-scale case–control study of lung cancer in 7 European countries, we explored the association between lung cancer and occupation among never-smokers.
Study Design and Population
A multicenter case–control study of lung cancer carried out between March 1998 and January 2002 formed the basis of the present investigation. The study included 16 participating centers from 7 European countries: Czech Republic (Olomouc, Brno, Prague), Hungary (Budapest and counties Borsod-Abauj-Zemplen, Heves, Jasz-Nagykun-Szolnok, Szabolcs-Szatmar), Poland (Warsaw, Lodz), Romania (Bucharest), Russia (Moscow), Slovakia (Bratislava, Banska Bystrica, Nitra), and the United Kingdom (Liverpool).
Incident lung cancer cases (International Classification of Diseases for Oncology 9th revision code 162), diagnosed in clinical and pathologic departments of each participating center, were selected for the study. Initial selection of the cases was restricted to subjects between 20 and 74 years old and to those subjects who had lived in the study area for at least 1 year. For each incident case, final diagnosis was based on clinical evidence, histology, or cytology confirmation. Each case was interviewed within 3 months of diagnostic confirmation.
In all centers except Warsaw and Liverpool, controls were selected from among hospital patients who had one or more of a list of diseases that excluded malignant neoplasms, diseases of the respiratory system, and any other smoking-related disease or disorder (defined to include both active smoking and environmental exposure to tobacco smoke). No single diagnostic groups provided more than 10% in each study center. In Warsaw and Liverpool, controls were selected from among healthy individuals in the general population. All controls were frequency-matched on the lung cancer cases’ sex and age group (±3 years). Eligible controls were interviewed within 3 months of their case interview.
Of 3403 eligible cases, 542 (16%) were not included in the study; 27 of those had been discharged from hospital before the interview, 53 were too ill to be interviewed, 13 had died before the interview, and 449 refused to participate. Of 3670 eligible controls, 550 (15%) were not included in the study; 16 had been discharged from the hospital before the interview, 21 were too ill to be interviewed, 2 had died before interview, and 511 refused to participate. For the present study, we selected the subset of never-smokers defined by subjects who smoked less than 100 cigarettes in their lifetime.
In-person interviews were carried out in the hospital by centrally trained interviewers using structured questionnaires. A lifestyle questionnaire obtained information on education, tobacco smoking and alcohol consumption, diet, lifelong residential history, exposure to secondhand smoke from partner and work, history of lung disease, use of diagnostic or therapeutic x-rays, specific medications, and family history of cancer.
A complete work history of jobs held for at least 1 year was recorded on an occupational questionnaire. The work history included job title, job description, years when jobs started and finished, and type of activity. A general questionnaire was completed for each occupation in the work history focusing on a detailed description of tasks performed and materials used. Specialized questionnaires captured additional information on specific circumstances of exposure for 18 occupations prevalent in the study areas that exposed workers to known or suspected lung carcinogens.
Occupational Exposure Assessment
Occupational questionnaire information was assessed by a group of experts (chemists, industrial hygienists, and occupational physicians) selected in each study center who had a detailed knowledge of the industries of the area and who were centrally trained. Experts were unaware of the case or control status of the study subjects. Information obtained from the questionnaire was evaluated from the expert assessors to derive industry and job codes for each occupation in the job history.3,4 Industry and job title codes were used to create 2 exclusive lists of high-risk occupations for lung cancer list A, comprising occupations with exposures to known carcinogens, and list B, comprising occupations with exposures to suspected carcinogens.5 Based on the tasks performed in each specific job and the detailed information on technologic practices, use of protection equipment, and other determinants of exposure, experts estimated the frequency and level of exposure to 70 known or suspected occupational lung carcinogens as well as their confidence in the presence of exposure (categorized as possible, probable, or certain). Frequency of exposure was scored as low (1–5% of working time), medium (5–30%), and high (greater than 30%). Level of exposure was scored in 3 categories using agent-specific cut points.
We used unconditional logistic regression models to investigate the association between occupational exposures and risk of lung cancer. Because men and women could differ in their exposure patterns even if employed in the same job, we carried out sex-specific analyses to examine increases in risk of lung cancer in association with employment estimated by odds ratios (ORs) and associated 95% confidence intervals (CIs). Therefore, the OR for lung cancer were estimated for ever being employed in list A occupations (ie, disregarding whether employment in list B occupations also occurred), only employed in list B occupations (ie, never employed in list A occupations), and ever employed in occupations on either list. The OR for lung cancer was also estimated in association with duration in years in each list (A or B) and, in either list, based on categories defined by the 50th percentile of the distribution of exposed controls. Subjects never employed in occupations on either list were used as the reference group.
We grouped the 70 agents, as assessed by the experts, based on physical and etiologic similarities as follows: mineral fibers, including asbestos and manmade mineral fibers; nonferrous metal dust and fumes, including inorganic pigment dust, arsenic and compounds, chromium, cadmium, and nickel dust, fumes, or mists; ferrous metal dust and fumes, including mild steel dust, hard alloy dust, stainless steel dust, and iron and compound fumes; combustion fumes, including coal, coke, petroleum oil, and wood combustion fumes; engine emissions, including diesel, gasoline, and kerosene engine emissions; petrol fuels, including diesel, gasoline, kerosene, and mineral spirits; welding fumes, including arc and gas welding fumes; and metal working fluids, including lubricating oil mists, cutting fluid mists, and other mineral oil mists. Individual agents with sufficiently high prevalence of exposure to be retained individually in the analyses were organic solvents, plastic pyrolysis products, polycyclic aromatic hydrocarbons (PAH), and silica. The remaining agents were excluded from the analyses because of low prevalence of exposure.
Analyses that examined the association between lung cancer and agent exposures were carried out among all study subjects. For each agent (or agent group), we calculated exposure duration (the total number of years exposed) and a cumulative exposure index (a combined measure of intensity, frequency, and duration). We estimated the OR of lung cancer in relation to each exposure measure using agent-specific (or agent group-specific) categorical models. Two exposure categories were defined according to the median value for each agent (or agent group) among exposed controls. Subjects unexposed to that agent (or agent group) were used as the reference group.
We explored collinearity between exposure agents by calculating correlation between pairs of agents or groups of agents. Analyses were adjusted for age, sex (when all study subjects were included), and study center. To examine whether country was an important modifier, we carried out additional analyses stratified by country. We also carried out separate analyses for adenocarcinoma, the most common histologic type among never-smokers. We controlled for potential confounding by environmental tobacco smoke (ETS) at work, ETS at home, and a combined measure of these two, all defined as duration in years during the lifetime. Other potential confounders and effect modifiers were also considered: occupational agents with evidence of collinearity; dietary factors (consumption of fruits, vegetables, and dairy products); level of education; duration of residential exposures to gas, wood, and coal as fuels for heating or cooking; family history of lung cancer; and history of lung diseases.
The study included 1262 never-smoking subjects, including 48 male cases, 175 female cases, 534 male controls, and 505 female controls (Table 1). The largest number of cases was enrolled in Russia. The mean age of cases and controls by sex was comparable. Duration of workplace exposure to ETS was higher among male cases than in the other groups. Only a small proportion of subjects in our study had a history of asbestosis, silicosis, tuberculosis, or emphysema with no great difference between cases and controls of either sex. Among all study cases, 96 women and 16 men were diagnosed with adenocarcinoma of the lung.
Only 2 men with lung cancer were ever employed in list A occupations and only 6 women with lung cancer (all painters) (Table 2). The most prevalent occupations in list B included rubber workers among women (5 cases and 5 controls) and transport workers among men (3 cases and 50 controls) (Table 3).
The ORs of lung cancer for employment in list A or B occupations are presented in Table 4. The OR of lung cancer was not increased in either sex for employment in list A occupations. Ever employment in list B occupations resulted in a modest, increased OR among women (OR = 1.4; 95% CI = 0.66–2.8) with the suggestion of a duration–response relationship (OR for employment more than 12 years = 1.8; CI = 0.63–4.9). The OR for employment in list B occupations among men was close to 1.0.
We carried out separate analyses for prevalent occupations among women. Suggestive evidence of positive associations was obtained for painters (list A) (1.8; 0.53–6.0 based on 6 cases and 6 controls) and employment in the rubber industry (list B) (2.6; 0.78–8.4 based on 5 cases and 5 controls).
The OR of lung cancer risk for exposure to groups of agents among all study subjects are presented in Table 5. Increased ORs of lung cancer were found in association with exposures to nonferrous metal dust and fumes (1.7; 1.0–2.9), silica (1.7; 0.97–3.2), and organic solvents (1.5; 0.94–2.2). The results of the analysis on duration of exposure and cumulative exposure to selected groups of agents are presented in Table 6. ORs above 2 were seen for the highest exposure categories of duration and cumulative exposure to nonferrous metal dust and fumes and silica.
We carried out sensitivity analyses in which ETS at work was considered as a specific occupational exposure (defined as total duration in years). The analyses adjusted for other potential occupational exposures (presence of other putative agents) and secondhand smoke at home. Findings of categorical models, with categories defined by the 50th percentile of exposure among exposed controls, showed no risk for occupational secondhand smoke for ≤22 years (0.95; 0.61–1.5) and a 30% excess risk for exposures occurring for >22 years (95% CI = 0.88–2.0). Adjusting for either the other putative agents or secondhand smoke at home did not change this estimate.
When the risk of adenocarcinoma of the lung was analyzed separately, slight differences were seen when compared with all cancer cases. The OR for ever exposed to nonferrous metals was 1.7 (95% CI = 0.81–3.5 based on 12 exposed cases), that for silica was 2.2 (1.0–4.9 based on 11 exposed cases), and that for solvent exposure was 2.0 (1.1–3.5 based on 22 exposed cases).
We found no important correlations between agents or groups of agents. No confounding was seen for exposures to ETS at home or other factors mentioned previously nor were any effect modification factors important.
The present study, based on a large case–control study carried out in 6 central and eastern European countries and the United Kingdom aimed to examine the association between occupational exposures and risk of lung cancer in a never-smoking population. The study findings suggest an increased risk of lung cancer among women employed in suspected high-risk occupations, and in men and women combined, for exposure to nonferrous metal dust and fumes, silica, and organic solvents.
Occupational information in this study was collected through in-person interviews, performed by centrally trained interviewers, using structured questionnaires. The validity and precision of exposure estimates in a case–control study will depend greatly on the methods used for data collection and evaluation. This study used an elaborate system of questionnaires developed specifically for this research after which experts with detailed local knowledge of the relevant industries and agricultural practices evaluated the information gathered blind of case–control status. Exposure assessment was based on detailed task description rather that from self-reported exposures. One important aspect of the assessment was the expert scoring process. To examine reproducibility of the exposure assessment procedure, we evaluated the agreement among expert assessors.6 Of the 12 specific agents or agent groups discussed in our article, one had excellent agreement (welding fumes; group average Cohen's kappa = 0.88), 6 had fair to good agreement (ferrous metals, combustion fumes, engine emissions, diesel/petroleum fuels, metal working fluids, and silica; Cohen's kappa = 0.66), and 5 had poor agreement (mineral fibers, nonferrous metal dust and fumes, organic solvents, plastic pyrolysis products, and PAH; Cohen's kappa = 0.34).
Because of the nondifferential misclassification of exposure from this assessment, one would expect that the true associations are likely to have been underestimated. t'Mannetje and colleagues6 carried out sensitivity analyses to estimate the effect of misclassification of these exposures on hypothetical ORs and found an underestimation of effects; if the prevalence of exposure were 10%, the OR would be reduced from 2.0 to 1.75 for excellent agreement, to 1.51 for fair to good agreement, and to 1.33 for poor agreement. Because, overall, the prevalence of agent exposures in our study was approximately 10% or greater, we would expect the estimates of exposure–lung cancer risk in a large study like ours to be distorted toward the null by an amount not greater than those observed in the simulation study.
Questionnaire information on job description was used to assess the relation of lung cancer risk and employment in certain high-risk occupations. Such exposure definition does not take into consideration the fact that men and women may experience different exposure circumstances while employed in the same job. In contrast, the expert assessment took into account sex differences when assessing exposures to specific agents.
To examine risk of lung cancer by occupation, we applied lists of known and suspected high risk occupations. A small-to-moderate increase in lung cancer risk was found among women employed in list B occupations, but no association with occupation was observed for men. Our findings were partly consistent with a recent study conducted in western Europe between 1988 and 1994 based on the same classification of occupations.7 The earlier study found a moderate increase in risk for employment in list A occupations among men and strong association among women ever employed in list B but not list A occupations. The results from the 2 studies are similar despite their different time periods and geographic locations in Europe with potentially different socioeconomic backgrounds. The consistency in the 2 studies for the observed excess risk in list B occupations among women may suggest that the current knowledge on occupational carcinogens is biased toward risks present typically in male jobs. Occupational risks among women are usually underestimated due to the fact that exposures in industries and occupations prevalent among women are mostly among those listed as “suspected” carcinogens.
We also examined whether the difference in defining never-smoker status between the 2 studies would affect the findings of occupational-related risk of lung cancer. For this, we carried out sensitivity analyses including subjects who had smoked at least one cigarette per day for at least a year (totals about at the same cut point as in the previous European study7) and found no change in the effect estimates for occupation.
Due to the small number of subjects in each occupation, we were unable to examine most of the individual occupations for either sex. However, numbers allowed us to analyze women employed as painters (list A) and in the rubber industry (list B) for which we observed an increased risk, thus corroborating previous evidence. Employment as a painter has been listed by the International Agency for Research on Cancer (IARC) as a human carcinogen (group 1)8 as has employment in the rubber industry.9 The risk of lung cancer in the rubber industry has been linked to the presence of several agents: asbestos,10,11 nitrosamines,12 carbon black,10,13 and solvents.14 More than one agent may contribute to the observed association with employment as painters,8,15 in particular the presence of solvents, metals, and thinners.
Gender differences in the risk of lung cancer reported here and in the earlier European study7 could be explained by differences in either exposure circumstances or in inhalation, deposition, and absorption of the relevant carcinogens. Studies have shown that, for example, different particle deposition in lung occurs for men and women.16,17 However, we do not exclude the possibility that differences in occupational coexposures and other interacting factors such as indoor exposure to air pollutants and dietary factors might have contributed to the pattern of risk observed here.
In the analysis of expert-assessed exposure to specific agents and groups of agents, positive associations were suggested for nonferrous metal dust and fumes comprising cadmium, chromium, nickel, arsenic and their compounds, silica, and organic solvents. A moderately high proportion of subjects were ever exposed to any of the first 3 classes of nonferrous metals and fewer to arsenic dust and fumes. Cadmium and its compounds, nickel compounds, hexavalent chromium (VI), and arsenic and its compounds have been classified by IARC as human carcinogens (group 1),9,18,19 and metallic nickel has been classified as a possible human carcinogen (group 2B).18 Evidence from recent epidemiologic studies has added to previous findings on the carcinogenicity of these agents,20–25 as also summarized in recent reviews.15,26
Our finding for silica corroborated previous evidence for an increased risk of lung cancer in the presence of this exposure.27–29 Crystalline silica has been classified by the IARC as a human carcinogen (group 1).30 Confounding by smoking has been debated as the explanation for the observed association between silica and lung cancer. Our results offer evidence against this hypothesis and suggest a carcinogenic effect of crystalline silica independent from that of tobacco smoking. The estimate was over 2-fold for exposures lasting more than 8 years (OR = 2.4; 95% CI = 1.1–5.2) and also for the highest category of the cumulative exposure index (2.5; 1.2–5.2).
Published epidemiologic evidence on the relation between organic solvent exposure and risk of lung cancer has been mixed.31–34 We found a positive relation between these agents and lung cancer, although low power limits our conclusions. However, such findings were consistent with the positive (if imprecise) findings for employment in the rubber industry and employment as painter considering that both these occupations involve high use of organic solvents.
Although there is considerable evidence from epidemiologic studies for the association of occupations and occupational agents with lung cancer, only a few of those have examined these associations in never-smoking populations. An early study from Sweden35 observed an increased risk of lung cancer due to arsenic exposure among never-smoking men employed in copper smelting. Additional studies have shown elevated risk of lung cancer for occupations listed in our study as high or suspected risk for lung cancer in men and women.7,36–40
There has been little evidence to date of associations between occupational exposures and specific lung cancer histologies. We explored this possibility by examining the risk of adenocarcinoma (the most common morphologic type) for several occupational agents. The observed association of adenocarcinoma with nonferrous metal dust and fumes was slightly lower than for all lung cancer cases, whereas the association with silica and solvent exposures were higher than for all lung cancer cases.
Never-smoking status in this study was defined by self-report. Thus, misclassification can occur from incorrect reporting of either smoking status or amount smoked among light smokers. Other studies have addressed the issue of misclassification of current smokers or former smokers as never-smokers and found it unlikely to represent a major problem.41–44 Given this finding, we would expect little distortion of our study estimates from such misclassification. In addition, because of the conservative definition of never-smokers in this study, we would expect little misclassification of amount smoked among light smokers (1–100 cigarettes in a lifetime).
This study used hospital-based controls. To ensure validity of the study results, controls were specifically selected to exclude all smoking-related or ETS-related diseases. In addition, no single diagnostic groups provided more than 10% of controls in each center. To further assure that such selection did not affect the validity of our results, we ran 2 types of sensitivity analyses excluding diseases potentially linked to occupational exposures. In the first, we excluded all controls ever diagnosed with emphysema, asthma, silicosis, or asbestosis. In the second, we excluded all controls responding “yes” to the question: “Have you ever been diagnosed with any occupational disease?” No important deviations were observed in the estimates when compared with those of the main analyses.
This study did not permit an investigation of relationships for specific occupations and agents due to insufficient number of subjects. Nonetheless, the study does provide evidence of assertions of occupation and occupational exposure agents with lung cancer independent of smoking. We used a stricter definition of never-smoking status than in previous studies to include only those individuals who had never smoked or smoked no more than 100 cigarettes in their lifetime. This should strengthen the validity of the observed associations. The findings for exposure to silica and nonferrous metal dust and fumes corroborated previous evidence for a carcinogenic effect of these agents independent of smoking. Suggestive evidence of increased lung cancer risk was seen for exposures to organic solvents. Suspected high-risk occupations, consistent with earlier studies,7,40 carried higher risk among women. This suggests that more attention should be paid to jobs held predominantly by women with improved strategies for exposure assessment and risk evaluation.
The authors thank Gilles Ferro for assistance in data management and analysis.
1. Tobacco smoking. In: IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans
, vol 38. Lyon: World Health Organization, International Agency for Research on Cancer; 1986.
2. Tobacco smoke and involuntary smoking. In: IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans
, vol 83. Lyon: World Health Organization, International Agency for Research on Cancer; 2004.
3. Eurostat (ES). NACE Rev. 1: Statistical Classification of Economic Activities in the European Community.
Luxembourg: Statistical Office of the European Communities, Classifications; 1996.
4. International Standard Classification of Occupations: ISCO-68
, 2nd ed. Geneva: International Labour Office Publications. 1968.
5. Ahrens W, Merletti F. A standard tool for the analysis of occupational lung cancer in epidemiologic studies. Int J Occup Environ Health
6. 't Mannetje A, Fevotte J, Fletcher T, et al. Assessing exposure misclassification by expert assessment in multicenter occupational studies. Epidemiology.
7. Pohlabeln H, Boffetta P, Ahrens W, et al. Occupational risks for lung cancer among nonsmokers. Epidemiology
8. Some organic solvents, resin monomers and related compounds, pigments and occupational exposures in paint manufacture and painting. In: IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans
, vol 47. Lyon: World Health Organization, International Agency for Research on Cancer; 1989.
9. Overall evaluations of carcinogenicity: an updating of IARC Monographs volumes 1 to 42. In: IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans
, suppl 7. Lyon: World Health Organization, International Agency for Research on Cancer; 1987.
10. Straif K, Chambless L, Weiland SK, et al. Occupational risk factors for mortality from stomach and lung cancer among rubber workers: an analysis using internal controls and refined exposure assessment. Int J Epidemiol
11. Straif K, Keil U, Taeger D, et al. Exposure to nitrosamines, carbon black, asbestos, and talc and mortality from stomach, lung, and laryngeal cancer in a cohort of rubber workers. Am J Epidemiol
12. Li K, Yu S. Economic status, smoking, occupational exposure to rubber, and lung cancer: a case–cohort study. J Environ Sci Health Part C Environ Carcinog Ecotoxicol Rev
13. Sorahan T, Hamilton L, van Tongeren M, et al. A cohort mortality study of UK carbon black workers, 1951–1996. Am J Ind Med
14. Szeszenia-Dabrowska N, Wilczynska U, Kaczmarek T, et al. Cancer mortality among male workers in the Polish rubber industry. Pol J Occup Med Environ Health
15. Siemiatycki J, Richardson L, Straif K, et al. Listing occupational carcinogens. Environ Health Perspect
16. Bennett WD, Zeman KL, Kim C. Variability of fine particle deposition in healthy adults: effect of age and gender. Am J Respir Crit Care Med
17. Kim CS, Hu SC. Regional deposition of inhaled particles in human lungs: comparison between men and women. J Appl Physiol
18. Chromium, nickel and welding. In: IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans
, vol 49. Lyon: World Health Organization, International Agency for Research on Cancer; 1990.
19. Beryllium, cadmium, mercury, and exposures in the glass manufacturing industry. In: IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans
, vol 58. Lyon: World Health Organization, International Agency for Research on Cancer; 1993.
20. Bulbulyan MA, Jourenkova NJ, Boffetta P, et al. Mortality in a cohort of Russian fertilizer workers. Scand J Work Environ Health
21. Sorahan T, Lancashire RJ. Lung cancer mortality in a cohort of workers employed at a cadmium recovery plant in the United States: an analysis with detailed job histories. Occup Environ Med
22. Gibb HJ, Lees PS, Pinsky PF, et al. Lung cancer among workers in chromium chemical production. Am J Ind Med
23. Sorahan T, Harrington JM. Lung cancer in Yorkshire chrome platers, 1972–97. Occup Environ Med
24. Englyst V, Lundstrom NG, Gerhardsson L, et al. Lung cancer risks among lead smelter workers also exposed to arsenic. Sci Total Environ
25. Richiardi L, Boffetta P, Simonato L, et al. Occupational risk factors for lung cancer in men and women: a population-based case–control study in Italy. Cancer Causes Control
26. Hayes RB. The carcinogenicity of metals in humans. Cancer Causes Control
27. Rice FL, Park R, Stayner L, et al. Crystalline silica exposure and lung cancer mortality in diatomaceous earth industry workers: a quantitative risk assessment. Occup Environ Med
28. Steenland K, Mannetje A, Boffetta P, et al.; International Agency for Research on Cancer. Pooled exposure–response analyses and risk assessment for lung cancer in 10 cohorts of silica-exposed workers: an IARC multicenter study. Cancer Causes Control
29. Park R, Rice F, Stayner L, et al. Exposure to crystalline silica, silicosis, and lung disease other than cancer in diatomaceous earth industry workers: a quantitative risk assessment. Occup Environ Med
30. Silica and some silicates. In: IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans
, vol 42. Lyon: World Health Organization, International Agency for Research on Cancer; 1987.
31. Chen R, Seaton A. A meta-analysis of painting exposure and cancer mortality. Cancer Detect Prev
32. Boice JDJr, Marano DE, Fryzek JP, et al. Mortality among aircraft manufacturing workers. Occup Environ Med
33. Wiebelt H, Becker N. Mortality in a cohort of toluene exposed employees (rotogravure printing plant workers). J Occup Environ Med
34. Travier N, Gridley G, De Roos AJ, et al. Cancer incidence of dry cleaning, laundry and ironing workers in Sweden. Scand J Work Environ Health
35. Pershagen G, Wall S, Taube A, et al. On the interaction between occupational arsenic exposure and smoking, and its relationship to lung cancer. Scand J Work Environ Health
36. Brownson RC, Reif JS, Keefe TJ, et al. Risk factors for adenocarcinoma of the lung. Am J Epidemiol
37. Brownson RC, Alavanja MC, Chang JC. Occupational risk factors for lung cancer among nonsmoking women: a case–control study in Missouri (United States). Cancer Causes Control
38. Keller JE, Howe HL. Risk factors for lung cancer among nonsmoking Illinois residents. Environ Res
39. Kreuzer M, Gerken M, Kreienbrock L, et al. Lung cancer in lifetime nonsmoking men—results of a case–control study in Germany. Br J Cancer
40. Kreuzer M, Heinrich J, Kreienbrock L, et al. Risk factors for lung cancer among nonsmoking women. Int J Cancer
41. Riboli E, Haley NJ, Tredaniel J, et al. Misclassification of smoking status among women in relation to exposure to environmental tobacco smoke. Eur Respir J
42. Heller WD, Scherer G, Sennewald E, et al. Misclassification of smoking in a follow-up population study in southern Germany. J Clin Epidemiol
43. Nyberg F, Agudo A, Boffetta P, et al. A European validation study of smoking and environmental tobacco smoke exposure in never smoking lung cancer cases and controls. Cancer Causes Control
44. Brennan P, Buffler PA, Reynolds P, et al. Secondhand smoke exposure in adulthood and risk of lung cancer among never smokers: a pooled analysis of two large studies. Int J Cancer