Background Evidence that painters may be at risk for lung cancer comes mainly from analyses on job titles rather than on specific exposures found in the environments of painters.
Methods In the context of two large population-based case–control studies of lung cancer carried out in Montreal, we were able to assess possible relationships between lung cancer and the occupation of painter as well as exposure to paints, varnishes and stains. Interviews for study I were conducted in 1979–1986 (857 cases, 533 population controls, 1349 cancer controls) and interviews for study II were conducted in 1996–2001 (765 cases and 899 controls). Detailed lifetime job histories were elicited; a team of hygienists and chemists evaluated the evidence of exposure to many occupational substances including paint-related substances. The relative risk of lung cancer was estimated, adjusting for several potential confounders, including smoking, in a three-variable parameterisation.
Results In analyses pooling the two studies, painters had an OR of lung cancer of 1.3 (95% CI 0.9 to 2.2). Regarding exposures, ORs were: for wood varnishes and stains, 1.6 (95% CI 1.0 to 2.3); for wood and gypsum paints, 1.3 (95% CI 0.9 to 1.7); and for metal coatings, 1.1 (95% CI 0.8 to 1.6). Small numbers hampered evaluation of dose–response relationships.
Conclusions While our results cannot exclude chance or residual confounding by smoking or concomitant occupational exposures, they provide further evidence that some exposures in paint-related occupations, most notably wood varnishes and stains, increase the risk of lung cancer.
- Lung cancer
- case-control studies
- occupational risks
- chemical hazards
Statistics from Altmetric.com
- Lung cancer
- case-control studies
- occupational risks
- chemical hazards
What this paper adds
Large numbers of workers are engaged in paint-related occupations.
Evidence that painters may be at risk for lung cancer comes mainly from analyses on job titles rather than on specific exposures found in the environments of painters.
We assessed the possible relationships between lung cancer and exposure to paints, varnishes and stains.
The results suggest that exposure to coatings for wood, plaster or gypsum may carry excess risk; results for metal coatings were more equivocal.
The evidence of a lung cancer risk was stronger for solvent-based coatings than for water-based coatings.
Large numbers of workers are engaged in paint-related occupations. There has been increasing concern regarding health effects from paint-related exposures. Among diseases that have been linked to painting occupations are allergic and non-allergic dermatitis, chronic bronchitis, asthma, and complications of the nervous system, liver, kidney and blood-forming organs.1
Further, there have been indications that painters are at excess risk for cancer, and in particular lung cancer.1 2 Most previous evidence comes from epidemiological studies in which the exposure variable under analysis was simply the fact of having the job title of painter. However, exposure to paints is much more widespread than the occupation of painter.
A paint is a dispersion of a finely divided pigment in a liquid composed of a resin or binder and a volatile solvent; it is most often used as a coating for metal, wood, gypsum or plaster. Varnishes are light-bodied quick drying products that form a glossy or matt finish on application. Stains are varnishes containing enough pigment or dye to alter the appearance of a wood surface. Paint, varnish and stain products contain thousands of chemical components that are used as pigments, extenders, binders, solvents and additives. The chemistry of paint products has evolved over time.
We provide additional evidence on paint-related agents and lung cancer using data from two large population-based case–control studies conducted in Montreal. The primary objective of these two studies was to evaluate the possible role of occupational exposures in cancer aetiology. For study I, subjects were interviewed from 1979 to 1986 and for study II subjects were interviewed from 1996 to 2001. Study I included 14 major sites of cancer including the lung, while study II included only lung cancer. Both studies collected detailed information on job histories and on a large set of agents and mixtures, including paints, varnishes and stains. In addition to detailed occupational histories, the questionnaires sought data on several sociodemographic and lifestyle items, notably lifetime smoking history. This report describes possible associations between lung cancer and exposure to paints, varnishes and stains. As it is important to place such results in the context of risks among painters, associations between the job title of painter and lung cancer are presented.
Materials and methods
Metropolitan Montreal had a population of 2.7 million in 1979 and 3.1 million in 1996. Study I included men aged 35–70 years and study II included both men and women aged 35–75. In both studies, cases and controls were restricted to Canadian citizens who were resident in the Montreal area. Details of subject ascertainment and data collection have been presented previously.3–5 In both studies, cases were ascertained in the 18 largest hospitals located in the metropolitan area. In study I, 857 lung cancer cases were interviewed, representing 79% of eligible subjects. In study II, 1236 cases (471 females; 765 males) were interviewed, representing an overall response rate of 86%. In both studies, population controls were randomly sampled from population-based electoral lists, stratified by sex and age to the distribution of cases. In study I, 533 population controls were interviewed (response rate 70%),while in study II, 1512 population controls (613 females; 899 males) were interviewed (response rate 70%). In study I, an additional control group for the lung cancer series was created by sampling among other cancer patients; we refer to this series of 1349 subjects as ‘cancer controls’. These cancer controls included 14 major sites of cancer, none of which comprised more than 20% of the total. The histology of lung cancer was coded according to WHO/IARC technical report 31.6 Smoking history included information on age at beginning smoking, age at ending and average amount smoked. Due to very low numbers of females in painting occupations, the current analysis was limited to males.
The methodology of exposure assessment has been presented previously.4 5. After obtaining informed consent, interviews were conducted using structured questionnaires to obtain detailed information on sociodemographic and lifestyle factors. A semi-structured questionnaire was used to obtain details of each job the subject had held during his working lifetime. For each job, the subject was asked about the company, its products, the nature of the worksite, main and subsidiary tasks, use of protective devices and equipment maintenance. For some occupations, supplementary questionnaires were developed to assist the interviewers with detailed technical probing. A team of chemists and hygienists examined each subject's description of his jobs and translated each job into potential exposures from a list of 294 substances, without knowledge of case or control status. For each substance that the chemists identified as being present in a job, duration of exposure was estimated as the duration of the job. The experts also coded the exposure according to three dimensions: (i) their confidence that the exposure actually occurred (possible, probable, definite); (ii) frequency of exposure during a normal working week (<5%, 5–30% or >30% of the time); and (iii) relative concentration of the substance (low, medium or high). Concentration levels were established with reference to certain benchmark occupations in which the substance is found. That is, we identified a priori some hypothetical workplace situations which would correspond to low, medium and high exposure for each substance, and the experts rated each real job against these benchmarks. The exposure assessment was based not only on the worker's occupation and industry, but also on individual characteristics of the workplace and tasks as reported by the subject; an illustrative example is given in the appendix of Parent et al.7
The exposures of interest
The three of agents on the checklist of exposures evaluated by the experts were: (i) metal coatings, (ii) wood varnishes and stains, (iii) wood, gypsum and plaster paints which, for simplicity, we will henceforth refer to as wood and gypsum paints. These agents are characterised as follows. (i) Metal coatings consist of combinations of oxidising alkyds, epoxies, formaldehyde resins, thermosetting acrylics, polyesters and others. They are most commonly applied to motor vehicles, structural steel, ships, home appliances and metal furniture. In the past metal coatings often contained lead and zinc chromate, a combination of zinc, potassium and chromic acid; recently manufactured coatings are mainly composed of zinc metal, zinc oxide, molybdates and phosphates.1 8 9 (ii) Wood varnishes and stains are made of resins in drying oils, mixed with driers and thinning agents such as alcohols, ethers and naphthas. Varnishes are now predominantly based on alkyds and urethanes. (iii) Wood and gypsum paints include alkyds, acrylic latexes and (previously) caseins. In the past, wood paints contained petroleum solvents, and may have contained arsenic, benzene, titanium dioxide, chromium compounds, lead and other suspected carcinogens.1 10 Today they contain fewer solvents.11
Exposure to these agents occurs in paint manufacturing operations, but the vast majority of those exposed to paints are in user industries, such as construction, furniture and automobile production and the service sector. There, the exposures may occur in mixing paints, in preparing surfaces for painting, in applying paints, in spray painting operations, in cleaning equipment, and from activities being carried out by work colleagues. As well as the coating, the worker may be exposed to components of the surface being painted. Exposure to paint components can occur via inhalation or dermal contact.
In study II the exposure checklist also included two exposures not on the study I checklist and that cut across the other paint exposures: water-based coatings and solvent-based coatings. These two families of exposures largely, but not entirely, represent a different way of ‘slicing up the pie’ of paint-related coatings.
In addition to carrying out a number of descriptive analyses of the study populations and exposure profiles, the main purpose was to estimate the relative risk of lung cancer in relation to the occupation of painter and in relation to paint exposures. This was done using unconditional logistic regression analyses.12
To assess the relative risks of lung cancer for painters, we combined three categories in the Canadian Occupational Classification, namely: painting and decorating occupations, except construction; construction painters, paper hangers and related occupations; and painters, sculptors and commercial artists. Occupational chemicals such as asbestos were not included as covariates in these analyses because, in these analyses, we want to estimate the risks for painters irrespective of what the responsible agent might be.
For each job in which the subject was exposed to an agent, we had the duration in years and a set of ordinal values for confidence, frequency and concentration. If a subject was exposed to the agent in two or more jobs, then lifetime values of confidence, frequency and concentration were calculated by taking averages, weighted by the durations of the various jobs in which exposure had occurred. The combination of duration, confidence, frequency and concentration was used to categorise the lifetime exposure into categories as follows: unexposed, exposed at a non-substantial level, and exposed at a substantial level. Exposures occurring within 5 years of diagnosis or interview were excluded. In order to be classified as exposed at the substantial level, a subject had to have been exposed at confidence of probable or definite, concentration and frequency of medium or high, and for duration greater than 5 years. All other exposed subjects were then classified in the non-substantial category. We consider this non-substantial/substantial dichotomy to be a simple proxy for cumulative exposure. Unexposed subjects were defined as those not exposed professionally to any of the three agents. Admittedly, many of these unexposed may have exposure to paints for personal or hobby use.
The exposure variables were analysed as never/any and the ‘any’ exposed were further subdivided by non-substantial/substantial, by duration and by era of exposure. In analysing associations between exposure to a substance and risk of lung cancer, we included the following variables as potential confounders (with indication of parametrisation): age (linear), ethnicity (French, Anglo, other), years of school attendance (tertiles), median family income of census tract of residence (tertiles), type of respondent (self, surrogate), smoking history (see below), and exposure to three other known occupational lung carcinogens – asbestos, silica and cadmium compounds (each entered as binary exposed/unexposed). The latter three were selected for inclusion because they are on the IARC group 1 list of carcinogens, and because they had over 3% prevalence of exposure in this population. Also, each of these showed associations with lung cancer in our data.13–15 A smoker was defined as someone who had smoked at least 100 cigarettes in his lifetime; a former smoker was defined as someone who had stopped smoking at least 2 years before the interview. Following the recommendations of Leffondré et al,16 smoking history was represented by three variables: a binary variable indicating whether the person was ever a smoker, a categorical variable for the number of years since quitting (0–2, 3–5, 6–10, 11–15, 16+ years) and cumulative amount smoked, as measured by pack-years. The association between the exposure of interest and the three variables and the most prevalent histological types – small cell, squamous cell and adenocarcinoma – were also evaluated.
The main analyses were carried out with each of the paint-related variables in a separate model with the covariates. A complementary set of analyses was conducted with all of the paint variables in a model with the covariates.
The availability of two studies and two control groups in study I provided various opportunities. There are advantages and disadvantages with cancer controls and population controls and we cannot state that one is better than the other.17 We first carried out analyses of the study I data by comparing the cases separately with population controls and with cancer controls. We analysed study II separately. In order to maximise the precision of estimates, we also conducted two separate analyses pooling the study I and study II samples. In one pooled analysis, we used only the population controls from study I, combined with the population controls from study II. In the second pooled analysis, we included both cancer controls and population controls from study I, but weighted them equally, due to our belief that the two control groups were equally likely to represent the study base. Operationally, this was done by randomly sampling 533 of the cancer controls to be pooled with the 533 population controls. When combining the two studies in a single model, we included an indicator of study, to avoid any confounding that could derive from different case/control ratios in the two studies. The pooled results represent a kind of weighted average of the various control groups and the two studies; it cannot create an artificial excess risk where there is none in the component parts. We thus present risk estimates separately in five distinct sets of case–control combinations: (i) study I using population controls, (ii) study I using cancer controls, (iii) study II using population controls, (iv) pooled cases and pooled population controls from both studies, and (v) pooled cases and pooled controls from both studies, with cancer controls in study I given equal weight to population controls. In addition to presenting ORs for subcategories of cumulative exposure and duration, we conducted trend tests by duration of exposure and by cumulative exposure. The cumulative exposure index was created by multiplying duration of exposure by concentration (on a three-point scale) squared and by frequency of exposure (on a three-point scale) squared. These indices were entered as linear terms in logistic regression models.
The sociodemographic characteristics of study subjects are presented in table 1. Because of different age cut-offs in the two studies, subjects were relatively older in study II than in study I. Over 75% of cases were more than 55 years old when they were diagnosed with lung cancer. Fewer than 3% of cases and about 15% of controls were lifelong non-smokers. As compared with controls, cases had a lower socio-economic status and were more likely to be French Canadian. There were more surrogate respondents for cases than for controls. All of the variables in this table were included as covariates in the analyses.
When pooling the two studies, the OR between lung cancer and the occupation of painter was 1.6 (95% CI 1.0 to 2.4) when cancer controls from study I were excluded and 1.3 (95% CI 0.9 to 2.2) when they were included (table 2). There was no systematic difference in OR from study I to study II and there was no clear indication of a duration–response trend. To determine whether the apparent excess risks among painters were due to co-exposure to asbestos and silica, we also conducted an analysis of painters including asbestos, cadmium and silica as covariates. When including cancer controls from study I in a pooled analysis, the resulting OR was 1.3 (95% CI 0.8 to 2.3). We carried out an additional analysis subdividing painters according to the industry in which they worked. When including cancer controls from study I in a pooled analysis, painters in the construction industry had an OR of 1.6 (95% CI 1.0 to 3.1; 39 painter cases), whereas those in other combined industries had an OR of 1.1 (95% CI 0.5 to 2.4; 28 painter cases).
Exposure to paints and coatings occurred in many occupations and industries, such as construction, renovation, furniture manufacture and repair, automobile production, motor-vehicle maintenance and repair, metal working occupations, janitorial services, artistic painting and many others. For the two studies combined, the lifetime prevalences of exposure to the three agents were: 9% for metal coatings, 6% for wood varnishes and stains, and 13% for wood or gypsum paints. Over 90% of individuals classified as exposed were considered as probably or definitely exposed by the chemists. Over 80% of exposed subjects were exposed for more than 5% of the working week. Very few subjects were exposed at high concentration levels. A majority of exposed subjects were exposed for more than 10 years.
Among all 870 workers in both studies combined who were exposed to any of these three agents, 67% were considered exposed to only one of them, 24% to two of them (most frequently wood and gypsum paints and wood varnishes and stains), and only 9% to all three of them. The exposure ‘wood varnishes and stains’ was rarely coded alone; the other two agents were quite often coded alone.
ORs for the associations between exposure to the three agents and lung cancer are presented in table 3. For each of the three agents, we carried out analyses separately in each of the five sets of case–control combinations, stratified by non-substantial/substantial cumulative exposure, by duration of exposure and by start date of exposure. For the three agents and the five case/control combinations, every OR in this table in the ‘any exposure’ category or in the ‘substantial exposure’ category was greater than 1.0, with a few being significantly elevated. Based on our prior plans, we consider the pooled analyses with population and cancer controls (the last column of table 3) to provide the most importance results. The results for wood varnishes and stains were perhaps the most noteworthy, with significantly elevated ORs for any exposure and for substantial exposure, although workers with over 15 years of exposure did not show excess risks. To a slightly lesser degree, the results for wood and gypsum paints also showed some evidence of excess risks, although not significantly so. The results for metal coatings were close to the null. Trend tests were carried out by modelling duration of exposure and cumulative exposure as continuous linear variables; none had p values below 0.05 (results not shown). The results by start date of exposure do not show clear trends, except for wood varnishes and stains, which had a higher OR in the earlier period.
Smoking did not act as an empirical confounding variable; its inclusion had almost no impact on risk estimates. When we carried out pooled analyses (all controls) for any exposure to the three agents with all covariates except smoking, the ORs were: metal coatings, 1.2 (95% CI 0.8 to 1.6); wood varnishes and stains, 1.6 (95% CI 1.1 to 2.4); and wood and gypsum paints, 1.4 (95% CI 0.7 to 1.6). These are almost identical to the smoking-adjusted estimates in table 3. Similarly, the exclusion of the income variable from the models had almost no impact on point estimates.
The results in table 3 involved models in which only one of the three agents was included at a time. We also implemented a model including all three of these agents, as well as the other covariates. In general, the resulting ORs were lower and less stable than those shown in table 3. For instance, in the pooled analysis with all controls, for ‘any exposure’, the ORs were: metal coatings, 1.1 (95% CI 0.8 to 1.6); wood varnishes and stains, 1.6 (95% CI 1.0 to 2.3); and wood and gypsum paints, 1.3 (95% CI 0.9 to 1.7).
Since there were so many proxy respondents (about 35% of cases and 10% of controls), we explored whether this could have distorted the results. There was hardly any difference in prevalence of exposure between self- and proxy respondents; for both studies combined, 22.4% of self-respondent cases were considered exposed to any of the three agents, while the corresponding fraction among proxy respondents was 23.2%. When we restricted the logistic regression analyses to self-respondents only, the results were almost identical to those in table 3 (results are available from authors).
Table 4 shows the associations from the pooled sample between each exposure and each of the three most prevalent histological types of lung cancer: small cell, squamous cell and adenocarcinoma. The associations were slightly stronger with squamous cell and small cell tumours than with adenocarcinoma. This pattern held in study I and in study II (not shown).
Table 5 shows results of the analyses for the two cross-cutting exposure families coded only in study II. These represent essentially, but not entirely, the same exposures as those presented in table 3, but are distinguished by a basic constituent rather than the use category. There were indications of excess risk of lung cancer for both water-based and solvent-based coatings; for solvent-based paints, there was evidence of a dose–response relationship and for water-based paints the numbers in the substantial exposure category were too small to provide reliable estimates.
Several additional tables of results were produced and can be obtained on request from the authors.
It is legitimate to enquire as to whether painters experience an excess risk of lung cancer, and it is also legitimate to enquire as to whether exposure to paints and related products confers a risk. These are overlapping but not identical questions. Our study has relevant findings pertaining to both of these questions. Most previous studies have evaluated risks related to job titles rather than exposures.
The interpretation of epidemiological studies of cancer risks in painters and the carcinogenicity of paint-related exposures is complicated by the variability and complexity of paint products and painting environments, and by the changes that have occurred in the composition of paint products.
Risk of lung cancer among painters has been studied using national standardised mortality ratio-type analyses of routine records, retrospective cohort studies and case–control studies. Most analyses of national mortality or cancer incidence registries have indicated excess lung cancer risks for painters with relative risks in the range of 1.2–1.6.18–27 In cohort studies of painters, standardised mortality or incidence ratios of lung cancer were generally elevated and in the same range as in the national registry studies.28–33 These studies did not typically control for smoking status, or else did so with limited smoking information.
Several case–control studies have provided relative risk estimates for painters, mostly adjusting for smoking. Irrespective of the design characteristics, nearly all of these studies showed relative risks in the range of 1.3–1.934–42 while only a few were lower43 44 or higher.45 In most of these studies no distinction was made according to type of painting activities, although one of the null results was among spray painters in the automobile industry.44
Although there was no clear indication of a duration–response relationship, our job title analysis of painters, with rigorous control for smoking history, supported the hypothesis of an excess lung cancer risk particularly for those in the construction industry. In the construction industry, painters may have been exposed to such known lung carcinogens as asbestos and silica. Even if exposure to these carcinogens is the reason for excess risk among painters, this does not detract from the observation of excess risk in this profession. Indeed, we also conducted an analysis of painters including asbestos, cadmium and silica as covariates, and the resulting ORs were very similar to those in table 2.
In our study, there was little indication of excess risk following exposure to metal paints. However, there were indications, although not strong ones, that exposure to paints, varnishes and stains on wood, gypsum or plaster surfaces did lead to increased risks of lung cancer, although these increases were not generally statistically significant and there was no clear dose–response pattern.
When we analysed the exposures defined by whether they were water or solvent based, both categories showed some evidence of associations with lung cancer. For solvent-based coatings, but not for water-based coatings, there was an indication that risk was high among those substantially exposed.
Several methodological issues may influence the interpretation of our findings. Two-thirds of the study population was of French Canadian origin. Compared with the typically diverse North American populations, this considerably diminishes the risk of confounding by genetic and environmental factors related to ethnicity. All cancer cases were incident and histologically confirmed. Response rates were quite high, being over 80% for case groups and over 70% for population control groups. We had extensive information on potential confounders covering sociodemographic and lifestyle factors including smoking history, as well as other occupational exposures such as asbestos. In contrast to most occupational cohort studies, we had information on the worker's complete lifetime work history, not just the history with one employer. The assessment and control of smoking history is an obvious concern. For the parameterisation of smoking history, we used an approach based on a risk model derived from our study subjects.16 While this is an optimal approach, it may be that some degree of residual confounding remains. In our data, the paint-related exposures were more strongly associated with small cell and squamous cell tumours than with adenocarcinoma. This pattern is compatible with the possibility of residual confounding, since squamous cell tumours are those with the strongest relationship with cigarette smoking. But this pattern is also compatible with the possibility that paint-related exposures are causally related to the same histological types of lung cancer as smoking.
While there were quite high proportions of proxy response, it was reassuring that prevalence of exposure was similar between proxy and self-respondent subjects, and that the overall results were similar when we excluded proxies from the analyses. Occupational exposure was attributed to subjects on the basis of their detailed lifetime job history reported at the interview. We considered the specific tasks they carried out and their use of protective equipment. We have previously demonstrated that the subjects' reports of occupational history were valid.46 Our team of chemists and industrial hygienists attributed exposure to the subjects using a method for which reasonable reliability47 48 and validity49 has been shown. However, as our exposure assessment protocol was based on expert opinion rather than direct measurement or biomarkers, it inevitably entailed some degree of measurement error. Since this work was carried out blind with respect to disease status, we assume that any misclassification of the exposure variables under consideration would have occurred at random with respect to the outcome and thus would lead to an attenuation of estimates of association. Our assessment of exposure frequency and concentration was semi-quantitative, based on descriptions provided by the subjects and established by expert chemists and industrial hygienists. While errors in exposure assessment for the agents under consideration would likely lead to attenuation of risk estimates, errors in exposure assessment for occupational confounders or failure to include true occupational confounders could lead to bias in any direction. We cannot be certain that the associations observed are not attributable to some other concomitant exposures in the same environments as the agents that were analysed.
The choice of control group is problematic in any case–control study.50 In study II, we had only a population control group, but in study I, we had two options and we exercised both. While a population control group is often considered to be more representative of the base population, cancer controls are often less susceptible to non-participation bias and information bias.4 There are advantages and disadvantages to both options and we cannot affirm that one is necessarily better than the other. We exercised two different tactics in pooling the two studies, and they led to broadly similar results.
In conclusion, the results of our two community-based studies are compatible with the growing evidence that painters are at excess risk of lung cancer. The results further suggest that exposure to coatings for wood, plaster or gypsum may carry excess risk; results for metal coatings were more equivocal. There are indications of excess risk in relation to both water-based and solvent-based coatings, although the evidence was stronger in relation to the latter.
Exposure assessment methods were expertly developed and implemented mainly by Michel Gérin, Louise Nadon, Ramzan Lakhani, Denis Bégin and Benoit Latreille. We thank the many research assistants and interviewers who participated, including Marie-Claire Goulet and Jerome Asselin.
Funding This study was funded by a number of agencies, including the Fonds de recherche en santé du Québec, the Institut de recherche en santé et sécurité au travail du Québec, the Canadian Institutes for Health Research and the Guzzo-Cancer Research Society Chair in Environment and Cancer.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was approved by ethics committees at the Institut Armand-Frappier (University of Quebec), McGill University and each of the 18 hospitals in which cases were ascertained.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.