Journal of Exposure Analysis and Environmental Epidemiology (2004) 14, 397–403 r 2004 Nature Publishing Group All rights reserved 1053-4245/04/$30.00
www.nature.com/jea
Evaluation of specific occupational asthma risks in a community-based study with special reference to single and multiple exposures JAN-PAUL ZOCK,a NU´RIA CAVALLE´,b HANS KROMHOUT,c SUSAN M. KENNEDY,d JORDI SUNYER,a, e A´NGELES JAE´N,a NEREA MUNIOZGUREN,f FE´LIX PAYO,g ENRIQUE ALMAR,h JOSE´ L. SA´NCHEZ,i JOSEP M. ANTO´a, e AND MANOLIS KOGEVINASa a
Respiratory and Environmental Health Research Unit, Municipal Institute of Medical Research, Barcelona, Spain National Institute of Occupational Safety and Hygiene, Barcelona, Spain c Institute for Risk Assessment Sciences, University of Utrecht, Utrecht, The Netherlands d School of Occupational and Environmental Hygiene, University of British Columbia, Vancouver, Canada e Department of Experimental and Health Sciences, Pompeu Fabra University, Barcelona, Spain f Epidemiology Unit, Public Health Department of Bizkaia, Basque Country, Spain g National Institute of Silicosis, Oviedo, Spain h Epidemiology Unit, Autonomous Health Council of Castilla-La Mancha, Albacete, Spain i Pulmonology Unit, General Hospital ‘‘Juan Ramo´n Jime´nez’’, Huelva, Spain b
Objectives: To compare two job exposure matrices (JEMs) for the evaluation of asthma risks related to specific occupational exposures in a communitybased study. Methods: A questionnaire on self-reported asthma, respiratory symptoms and current occupation was sent to the participants of the European Community Respiratory Health Survey in five areas in Spain. Both an asthma-specific JEM, including expert judgment steps and a general JEM, were applied to occupational codes. Risks of current asthma symptoms and wheeze in the last year associated with the obtained exposure estimates were evaluated. Correlations between specific exposures were investigated using explanatory factor analysis. Results: Occupational exposures to the highmolecular-weight (MW) agents flour dust, enzymes, mites and animal-derived proteins as obtained by the asthma-specific JEM were positively associated with asthma outcomes. The effect of additional expert judgment steps was limited. High exposures to biological dust assessed by the general JEM without expert judgment was also associated with asthma. Many of the exposed individuals worked in environments with multiple exposures. Conclusions: Asthma risks associated with occupational exposures to specific high-MWagents could be identified from a population-based study using an asthma-specific JEM. Application of JEMs can be a useful tool to estimate asthma risks attributable to specific occupational exposures in the general population. However, these specific exposure risks should be interpreted in connection with the whole of concomitant exposures constituting the work environment. Journal of Exposure Analysis and Environmental Epidemiology (2004) 14, 397–403. doi:10.1038/sj.jea.7500337
Keywords: exposure assessment, job exposure matrix, multiple exposures, ECRHS, respiratory disease.
Introduction Asthma is the most common occupational lung disease in industrialized countries (Venables and Chan-Yeung, 1997), and it has been estimated that 5–20% of all adult asthma cases have an occupational origin (Blanc and Tore´n, 1999). More than 250 different workplace exposures have been associated with asthma (Chan-Yeung and Malo, 1999), among which isocyanates and other reactive chemicals, flour dust, animal-derived proteins, wood dust, metals and natural rubber latex being major causal agents. Most of these agents have been identified through case studies and asthma surveillance. Few community-based epidemiological studies
1. Address all correspondence to: Dr. Jan-Paul Zock, IMIM, Dr. Aiguader 80, E-08003 Barcelona, Spain. Tel.: þ 34-93-221-10-09 (Extension 2213); Fax: þ 34-93-221-64-48. E-mail:
[email protected] Received 20 March 2003; accepted 11 November 2003
have examined asthma risks related to occupational exposures. Certain occupations involving exposure to asthmarelated agents like bakers (flour dust) and spray painters (isocyanates) show an excess risk for asthma in communitybased analyses (Meredith, 1993; Kogevinas et al., 1999; Karjalainen et al., 2000). However, for other occupations such as nurses (potentially exposed to asthmagens like latex, sensitizing drugs and disinfectants), an elevated asthma risk has generally not been observed (Meredith, 1993; Kogevinas et al., 1999; Karjalainen et al., 2001). Overall, there is a lack of general population-based epidemiological studies on workrelated asthma that include assessment of specific asthmarelated agents. The European Community Respiratory Health Survey (ECRHS) is an ongoing multicenter community-based epidemiological study aiming at investigating known or suspected risk factors for asthma. In analyses of the ECRHS, an excess risk of self-reported symptoms suggestive of asthma was found in specific job groups like farmers, cleaners and
Zock et al.
spray painters (Kogevinas et al., 1999). In addition, the prevalence of asthma was associated with exposures to dusts, gases or fumes at work, assessed by either self-report or by an external job exposure matrix (JEM). A follow-up analysis showed that occupational exposures assessed by this JEM were also probably related to the incidence of asthma (Basagan˜a et al., 2001). However, it remained unclear as to which specific occupational exposures were related to asthma. An asthma-specific JEM was developed for the French Epidemiological study of Genetics and Environment in Asthma (EGEA) (Kennedy et al., 2000). The application of this JEM in case–control analysis showed that, in combination with expert judgment, exposures to certain low-molecular-weight (MW) agents such as highly reactive chemicals, industrial cleaning agents and metal sensitizers were associated with asthma at the time of onset. The aim of the present analysis was to evaluate the performance of the new asthma-specific JEM in the Spanish population of the ECRHS. In order to study differences with earlier phases of the ECRHS, the performance of the asthma JEM was compared with that of a general JEM, originally designed for studies on obstructive lung diseases. Finally, to better interpret the entire findings for different exposures according to the JEMs, we aimed to identify groups of workers in environments with exposure to multiple agents using explanatory factor analysis.
Methods Population and Questionnaire The population sampling methodology for the ECRHS has been described elsewhere (Burney et al., 1994). Briefly, in 1992, a community-based epidemiological study on asthma was performed in five areas in Spain. Subjects were derived from a random population sample and from an additional sample of subjects reporting symptoms highly suggestive of asthma in a screening questionnaire (Kogevinas et al., 1996). A repeated survey was conducted in 1998 (Basagan˜a et al., 2001). All 2661 subjects from both samples seen in the baseline study were sent a short questionnaire by mail, and telephone follow-up was performed. The ethics committees of the participating centers approved the study protocol, and subjects gave written, informed consent. The questionnaire included the following respiratory symptoms in the last 12 months: (a) wheezing or whistling in the chest at any time; (b) being woken by an attack of shortness of breath; (c) asthma attack; and (d) use of current medication for asthma. Current asthma was defined as the presence of at least one of factors (b), (c) or (d) listed above (Kogevinas et al., 1999). It was also asked whether subjects had ever had asthma and, if affirmative, whether this had been diagnosed by a doctor. Those who had smoked at least one cigarette daily during the last month were regarded as being current smokers. 398
Asthma risks of specific occupational exposures
Occupational Exposure Assessment Work or principal activity in the last 12 months was recorded in the questionnaire. Occupations were coded using both the British Office of Populations Censuses and Surveys (OPCS)1980 classification (used in ECRHS-I) (OPCS, 1980) and the International Standard Classification of Occupations (ISCO)-88 system (to be used in ECRHS-II) (International Labor Office, 1991) by one experienced coder (NC). The OPCS codes were linked to a general JEM developed for analyses on chronic bronchitis and chronic airflow limitation within the ECRHS (Sunyer et al., 1998). This yielded a semiquantitative estimation (none, low and high) of exposure to biological dusts, mineral dusts and gases or fumes for all individuals in the study. ISCO codes were linked to an asthma-specific JEM (Kennedy et al., 2000). This is a method for estimating exposure risks in population studies of asthma. It combines JEM methodology with selected expert judgment. The method first classifies jobs into more or less specific exposure categories representing known high asthma risks (Table 1). Exposures are coded ‘‘1’’only for jobs with a strong likelihood of exposure relevant to occupational asthma, that is, a high probability of most workers in that job being exposed to agents that would put them at high risk for the development of occupational asthma. The additional expert judgment being part of the JEM methodology involving individual re-evaluation in two steps was performed by two occupational hygienists included in the authors list (NC, JPZ). First, the description of occupation was reviewed for all subjects that needed ‘‘individual re-evaluation’’ according to the JEM, and if necessary exposure categories were changed according to the comments provided in the JEM. For example, all those in the category ‘‘medical doctors’’ needed re-evaluation. When the job title said ‘‘surgeon’’, exposure to latex was assigned. Or, when reviewing bakers, exposure to flour and biological enzymes was removed in all cases where the job title said ‘‘confectioner’’. Second, the description of occupation was reviewed for all exposed according to each risk category, and exposure risk classifications were revised wherever applicable. For instance, when reviewing those with assigned latex exposure, when the job title said ‘‘assistant psychiatrist’’ this exposure was removed. The main aim of this strategy was to enhance the specificity of the exposure assessment. Analysis Questionnaire data were obtained for 2279 subjects (response rate 86%). Excluded from analyses were 804 subjects with missing or incomplete occupational codes, mainly consisting of housewives for whom no ISCO code exists (n ¼ 441), those working in jobs with an insufficient level of detail to code into the four-digit ISCO unit group level (n ¼ 135) and unemployed (n ¼ 58). The group of excluded subjects comprised relatively more women and more nonsmokers. Journal of Exposure Analysis and Environmental Epidemiology (2004) 14(5)
Asthma risks of specific occupational exposures
Zock et al.
Table 1. Selection of the asthma-specific JEMa with examples of job titles with specific exposuresb. ISCO category
Animal-derived HMW agents
Flour
Enzymes Arthropods, Bio-aerosols Isocyanates Sensitizing Checkc mites drugs
General managers in wholesale and retail trade Veterinarians Pharmaceutical assistants Receptionists and information clerks Dairy and livestock producers Varnishers and related painters Bakers, pastry-cooks and confectionery makers Grain- and spice-milling-machine operators Domestic helpers and cleaners
0 1 0 0 1 0 0 0 0
0 0 0 0 0 0 1 1 0
0 0 0 0 0 0 1 0 0
0 0 0 0 0 0 0 1 1
0 0 0 0 1 0 0 0 0
0 0 0 0 0 1 0 0 0
0 0 1 0 0 0 0 0 0
0 0 0 0 0 0 1 0 1
a
Kennedy et al. (2000). Exposures are coded ‘‘1’’only for jobs with a high probability of most workers being exposed to agents that would put them at high risk for the development of occupational asthma. c Individual re-evaluation (i.e., expert judgment) needed; for selected examples exposure to flour and enzymes can be removed if confectioner, and exposure to cleaning agents can be assigned if domestic cleaner. b
The prevalence rates of respiratory symptoms, however, did not differ between the study population and the excluded group. Finally, 20 subjects were excluded from epidemiological analyses because data on asthma symptoms or smoking were missing. Thus, the final population comprised 1455 subjects. Prevalence ratios (PRs) were used to evaluate associations between exposure risk categories derived from the JEMs and symptoms of asthma. PRs adjusted for sex, age and current smoking were obtained using log-binomial models (Skov et al., 1998). The reference group in all models comprised subjects who were unexposed according to all categories of the applied JEM. All assigned exposure categories from both the asthmaspecific JEM (after expert judgment) and the general JEM were joined, and a factor structure for the 1455 subjects was built up using explanatory factor analysis (Bollen, 1989). This is a statistical technique used to identify whether the correlations between a set of observed variables (i.e., exposures) can be explained by a few latent, unobserved variables (i.e., factors). Only the principal factors with Eigenvalues greater than 1.25 were considered significant (i.e., explaining more multiple correlation than a single exposure). Factors were defined by exposures with a factor loading higher than 0.4. Exposures with lower loadings according to all factors identified from the structure were regarded unique and treated as single exposure. The subjects could be assigned only one factor or one single exposure (orthogonal structure). Associations between the obtained factors, single exposures and asthma were evaluated as described above.
Results The study population comprised slightly more men than women (Table 2). Almost half of the subjects were smokers. Symptoms consistent with asthma were reported by 19% of Journal of Exposure Analysis and Environmental Epidemiology (2004) 14(5)
Table 2. General and respiratory health characteristics of the study population (n ¼ 1455). Age in years: Mean (range) Women Current smokers Current wheezea Current asthmab Doctor-diagnosed asthmac
38.9 637 688 400 272 170
(27–53) (43.8%) (47.3%) (27.5%) (18.7%) (11.7%)
Numbers (%) are given, unless otherwise stated. a Wheezing or whistling in the chest at any time in the last 12 months. b Attack of asthma in the last 12 months, and/or being woken by an attack of shortness of breath in the last 12 months and/or current use of asthma medication. c Asthma confirmed by a physician (n ¼ 1449).
the study population, while 12% of the subjects reported asthma diagnosed by a physician. These figures reflect the oversampling of asthmatics in the population. Application of the asthma-specific JEM grouped 20% of the subjects into one of the high-risk categories (Table 3). The first step of the expert judgment procedure was reviewing the description of the job title for all 246 subjects (17%) who needed individual re-evaluation. In total, 23 changes were made in 17 subjects. An exposure was removed on 12 occasions (concerning flour, enzymes, metal sensitizers, metal working fluids, or low-level exposure to chemicals); and an exposure was assigned on 11 occasions (concerning latex, highly reactive chemicals, isocyanates, textile production, irritants, or exhaust fumes and ETS). In the second expert judgment step, where job title was reviewed for all 21 exposure categories, 64 exposures corresponding to 58 subjects were removed. The most common changes were related to the removal of exposure to highly reactive chemicals (13 ‘‘helpers and cleaners in offices, hotels and other establishments’’; four ‘‘institution-based personal care workers’’; and two ‘‘agronomists and related professionals’’), animal-derived HMW agents (16 farmers who were appar399
Asthma risks of specific occupational exposures
Zock et al.
Table 3. Specific exposure groups according to an asthma-specific JEM before and after expert judgment steps, and asthmaa risk estimates (n ¼ 1455). Before expert judgment
After expert judgment
n
PR (95% CI)
n
PR (95% CI)
Unlikely to be exposed
910
1.0 (referent)
918
1.0 (referent)
High-risk exposures (combined) High-MW agentsb Animal-derived Flour Arthropods, mites Enzymes Latex Bio-aerosols
291 138 20 12 19 12 78 12
1.2 1.6 2.3 3.4 2.2 3.4 1.0 1.3
(0.91.5) (1.22.2) (1.34.0) (2.15.3) (1.24.1) (2.15.3) (0.61.7) (0.53.6)
273 118 4 10 19 10 76 12
1.1 1.4 3.9 3.6 2.1 3.6 0.9 1.3
(0.91.5) (1.02.0) (2.07.7) (2.25.8) (1.13.9) (2.25.8) (0.51.5) (0.53.6)
Low-MW agents Sensitizing drugs Highly reactive chemicals Isocyanates Wood dusts Metal sensitizers, fumes Industrial cleaning agents
167 6 88 6 11 58 67
1.1 3.7 1.0 0.0 1.4 0.9 1.0
(0.81.5) (2.06.9) (0.61.6) () (0.53.6) (0.51.6) (0.61.7)
154 4 70 7 11 50 66
1.0 2.5 0.9 0.0 1.4 0.8 1.0
(0.71.5) (0.96.9) (0.61.6) () (0.53.7) (0.41.6) (0.61.7)
99 18 27 16 38
0.9 1.4 1.5 0.4 0.4
(0.51.4) (0.63.0) (0.83.0) (0.12.4) (0.11.2)
87 16 27 16 28
1.0 1.6 1.5 0.3 0.6
(0.61.6) (0.83.4) (0.83.0) (0.12.3) (0.21.7)
254 161 51 97
0.9 1.0 1.0 0.6
(0.71.3) (0.71.4) (0.61.9) (0.31.1)
264 165 53 111
0.9 1.0 1.0 0.6
(0.71.3) (0.71.4) (0.61.8) (0.31.0)
Exposure group
Mixed environments Metal working fluids Agriculture Textile production Irritant peaks Possible exposure other substances Low-level exposure to chemicals Exp. to irritants, not high peaks Exp. to exhaust fumes and ETS
PR (95% CI) for current asthma symptoms or medication, adjusted for sex, age, and current smoking. Attack of asthma in the last 12 months, and/or being woken by an attack of shortness of breath in the last 12 months, and/or current use of asthma medication. b No exposed subjects in categories ‘‘fish/shellfish’’ and ‘‘other plant antigens’’. a
ently crop growers) and irritant peaks (10 police officers). Taking both expert steps together, major changes had only been made for a few exposure categories (Table 3). Only eight of the potentially exposed subjects moved from an exposed to the unexposed group as a result of the expert judgment steps. Assigned exposure to high-MW agents was positively associated with current asthma symptoms or medication, both before and after expert judgment (Table 3). Consistently elevated PRs for asthma were found for exposures to certain specific high-MW agents; arthropods and mites, flour dust and biological enzymes. After expert judgment, only four subjects were assigned exposure to animal-derived proteins, and this exposure was positively associated with asthma. This exposure occurred in animal producers and veterinarians. When analyses were performed for current wheeze as outcome, the results were consistent with those shown for current asthma (results not shown). The overall assigned exposure to low-MW agents was not associated with asthma (Table 3). Exposure to sensitizing 400
drugs was related to asthma, but risk estimates decreased and lost statistical significance after the expert judgment steps. There were no asthmatics among those exposed to isocyanates, and no increased risk for wheeze was found in this group (PR 1.0 and 0.9 before and after expert judgment, respectively; results not shown). Exposure in any of the mixed environments was not related to asthma, although exposure to metal working fluids and agricultural work showed nonsignificantly increased risks for both asthma and wheeze. Possible exposure to other substances was also not associated with asthma, although there was a tendency towards a lower asthma risk for the category of exposure to exhaust fumes and environmental tobacco smoke (PR ¼ 0.6; CI 0.3–1.0). Asthma prevalence was associated with high exposure to biological dusts assessed by a general JEM (Table 4). Exposure to mineral dusts, or to gases or fumes was not related to asthma. Consistent results were obtained for current wheeze (results not shown). Journal of Exposure Analysis and Environmental Epidemiology (2004) 14(5)
Asthma risks of specific occupational exposures
Zock et al.
When the 19 specific exposures of the asthma-specific JEM (after expert judgment) and the six specific exposures of the general JEM were combined, 837 subjects (57.5%) were unexposed according to all categories. Among the 618 exposed subjects, the vast majority (526) was exposed to two or more agents. Simultaneous exposure to three or more agents was found in 342 subjects, while 156 subjects were exposed to at least four agents. Explanatory factor analysis yielded five principal factors with an Eigenvalue of 1.25 or higher (Table 5). Specific exposures with loadings 40.4 were placed into one or more of the five factors accordingly. Eight exposures with uniqueness 40.9 were not included in one of the five factors and were regarded as separate groups. Among the five principal factors, only the category comprising flour, enzymes and high exposure to biological dusts showed a positive association with asthma (Table 6). The PRs for the eight single exposures were comparable to those presented in Tables 2 and 3. In our study population, exposure to sensitizing drugs occurred in pharmaceutical jobs and not in healthcare workers.
Table 4. Exposure categories according to a general job exposure matrix, and asthmaa risk estimates (n ¼ 1455). Exposure group
n
PR (95% CI)
No exposure to dusts/gases/fumes Low exposure to biological dusts High exposure to biological dusts Low exposure to mineral dusts High exposure to mineral dusts Low exposure to gases or fumes High exposure to gases or fumes Low exposure to dusts/gases/fumes High exposure to dusts/gases/fumes
901 131 59 187 82 327 140 358 196
1.0 0.9 1.6 1.0 1.0 0.8 1.1 0.9 1.2
(referent) (0.61.3) (1.12.5) (0.71.4) (0.71.7) (0.61.1) (0.81.6) (0.71.1) (0.91.6)
PR (95% CI) for current asthma symptoms or medication, adjusted for sex, age and current smoking. a Attack of asthma in the last 12 months, and/or being woken by an attack of shortness of breath in the last 12 months and/or current use of asthma medication.
Discussion The application of an asthma-specific JEM in this crosssectional community-based study showed that occupational exposures to certain high-MW agents were related to selfreported asthma. The effect of additional expert judgment steps on specific associations was limited. The application of a general JEM confirmed an increased asthma risk associated with high exposure to biological dusts. Jobs with multiple exposures were common, and therefore specific exposure risks derived from JEMs in community-based studies should in many cases be interpreted as being dependent on concomitant exposures constituting the work environment. In this analysis, asthma was associated with recognized asthma-related workplace agents, including flour dust, biological enzymes, mites and animal-derived proteins. This was in agreement with asthma risks associated with exposure to biological dusts, estimated using a general JEM. Exposure to flour and enzymes mainly occurs in bakery workers and grain-processing operators. Flour dust has frequently been associated with occupational asthma in epidemiological studies based on general population samples or asthma registries (Meredith, 1993; Tore´n et al., 1999), and both flour and biological enzymes in studies among bakers (Houba et al., 1998). To date, asthma caused by exposure to mites in the workplace has only been reported for storage mites (Chan-Yeung and Malo, 1999). It can be speculated that exposure to house dust mites might be an occupational cause of asthma in domestic cleaners (Zock et al., 2001). The most common occupational sources of animal-derived allergens are laboratory animals and livestock (Chan-Yeung and Malo, 1999). Asthma symptoms were not associated with other recognized asthma-related agents, including isocyanates or other sensitizing chemicals, latex, wood dust or metal sensitizers. Although sensitization to natural rubber latex is a well-recognized cause of occupational asthma due to the use of powdered latex gloves (Charous et al., 2002), community-based studies generally failed to show a high asthma risk in healthcare workers commonly using gloves (Meredith, 1993; Kogevinas et al., 1999; Karjalainen et al.,
Table 5. Exposure categories resulting from explanatory factor analysis. Factor
Specific exposures from two JEMs
Working name of category
1
(1) (4) (1) (1) (1) (1) (3)
High exposures in agricultural and other settings Bakery workers Healthcare workers Metal workers Miscellaneous chemical
2 3 4 5
Agricultural environment; (2) high exposure to biological dusts; (3) high exposure to mineral dusts; high exposure to gases or fumes Flour; (2) enzymes; (3) high exposure to biological dusts Latex; (2) highly reactive chemicals; (3) industrial cleaning agents; (4) low exposure to gases or fumes Bio-aerosols; (2) metal sensitisers, fumes; (3) metal working fluids; (4) irritant peaks Low-level exposure to chemicals; (2) exposure to exhaust fumes and environmental tobacco smoke; low exposure to mineral dusts; (4) low exposure to gases or fumes
Single exposures not included in one of the five principal factors (uniqueness 40.9): (1) Animal-derived HMW agents; (2) arthropods or mites; (3) sensitizing drugs; (4) isocyanates; (5) wood dusts; (6) textile production; (7) exposure to irritants, but not high peaks; (8) low exposure to biological dusts.
Journal of Exposure Analysis and Environmental Epidemiology (2004) 14(5)
401
Asthma risks of specific occupational exposures
Zock et al.
Table 6. Exposure categories resulting from factor analysis and asthmaa risk estimates. Exposure category
n
PR (95% CI)
No exposure
837
1.0 (referent)
Factors (multiple exposures) 1. High exposures in agricultural and other settings 2. Bakery workers 3. Healthcare workers 4. Metal workers 5. Miscellaneous chemical
10 114 52 194
3.0 1.0 0.8 0.9
(2.04.6) (0.71.5) (0.51.5) (0.71.3)
Single exposures 1. Animal-derived HMW agents 2. Arthropods, mites 3. Sensitizing drugs 4. Isocyanates 5. Wood dusts 6. Textile production 7. Exposure to irritants, but not high peaks 8. Low exposure to biological dusts
4 19 4 7 11 16 53 101
3.9 1.9 2.3 0.0 1.3 0.4 1.0 0.8
(2.17.5) (1.13.5) (0.96.1) () (0.53.4) (0.12.1) (0.61.7) (0.51.3)
33
1.0 (0.52.1)
PR (95% CI) for current asthma symptoms or medication, adjusted for sex, age and current smoking. a Attack of asthma in the last 12 months, and/or being woken by an attack of shortness of breath in the last 12 months and/or current use of asthma medication.
2001). This may be due to the fact that many hospitals have replaced powdered by nonpowdered gloves during the 1990s. In addition, the job title description is probably not specific enough to predict relevant exposure to airborne latex antigens at work, neither by job grouping nor by a specific JEM. Due to large variation of exposures within jobs not accounted for, the specificity of the categories wood dust and metal sensitizers in the asthma-specific JEM in our study was probably also relatively low, which might explain the lack of association between occupational exposure to these agents and asthma. In addition, occupational asthma due to wood dust exposure has been well described for Western Red Cedar, which is common in North America but not in Spain (Chan-Yeung, 1999). Other wood species contain less plicatic acid, the main sensitizing agent in Western Red Cedar dust. Assessment of occupational exposures in communitybased studies on asthma is commonly done using self-reports (Tore´n et al., 1999; Bakke et al., 2001). This constitutes a simple low-cost method of obtaining information on different occupational exposures and allows for variations in exposure within job titles (Kennedy et al., 2000). However, self-reports are particularly susceptible to reporting or recall bias, which might lead to differential misclassification, with asthma cases being more likely to report or recall exposures than nonasthmatic workers (McGuire et al., 1998; Benke et al., 2001). The application of JEMs does not introduce reporting 402
or recall bias, but JEMs have the general limitation that variations in exposure within job titles are not taken into account. In the approach presented in our paper, expert judgment steps allowed the possibility to revise the specific exposures generated by the JEM on the basis of job description. This resulted in different exposure classifications within subjects with the same job code, and probably increased specificity, which is favorable in this type of community-based studies (Kromhout et al., 1992). Although the labor-intensive expert judgment steps in our study resulted in relevant modifications, substantial changes were only made in a few categories. The specific asthma risks associated with these exposures did not change to a large extent, which was also observed in the French EGEA study (Kennedy et al., 2000). The postal questionnaire used in our study yielded limited job descriptions, and type of industry was not available. It is therefore possible that relevant modifications were not made due to the limited amount of information. In the follow-up of the ECRHS, more detailed occupational history information obtained by face-to-face interviews will be available. Another drawback of the use of JEMs is that assigned exposure categories may not be independent, hampering a clear interpretation of the associated risks. We followed an unconventional strategy to analyze exposure categories obtained from JEMs. In the classical approach, assigned exposures from a JEM are treated as if they were single, noncorrelated exposures when linked to health outcome. However, our study clearly showed that a large number of subjects had a multiple exposure pattern; 25% of individuals were assigned four or more exposures according to both JEMs. We therefore applied a statistical technique to determine underlying factors based on the exposure pattern. The most obvious example that resulted from this strategy was that all those exposed to flour dust were also assigned exposure to biological enzymes, and high exposure to biological dusts (i.e., bakery workers). Another factor mainly consisted of exposures concurrently occurring in healthcare work. Nevertheless, seven categories from the asthma-specific JEM could be considered as more or less unique exposures. Thus, for seven single exposures no underlying factors consisting of concurrent exposures could be identified, however only accounting for about one-third (215 out of 618) of exposed subjects. In addition to the classical approach using only single-exposure categories, it can be recommended to take into account multiple-exposure patterns when evaluating associations with asthma. An alternative for factor analysis to identify independent associations of specific exposures with the outcome could be the application of multiple regression techniques adjusting for concomitant exposures. A drawback of this approach is that statistical problems may occur due to high mutual correlations between exposures in the same model. In addition, multiple regression does not evaluate the correlations between Journal of Exposure Analysis and Environmental Epidemiology (2004) 14(5)
Asthma risks of specific occupational exposures
all the exposures and does not identify the ‘‘exposure cocktails’’. A potential limitation of our analysis arose from the crosssectional analysis of current occupational exposures and current asthma. A considerable selection effect was shown in analyses using previously collected data from this population (Kogevinas et al., 1996). This has probably resulted in biased risk estimates towards the null, and precluded the identification of certain occupational exposures in former jobs that had caused asthma, such as isocyanates. In conclusion, asthma risks associated with occupational exposures to specific high-MW agents could be identified from a community-based study using an asthma-specific JEM, with or without additional expert judgment. Many of the exposed individuals worked in environments with multiple exposures. This indicates that the common interpretation of specific exposures derived from JEMs should be done with caution. It is often difficult to clarify whether specific exposures or a cluster of mixed exposures are related to asthma. Factor analysis can be a helpful tool to explore correlations between concomitant exposures and to identify multiple-exposure patterns.
Acknowledgments This study was supported by the National Institutes of Health, USA (NORA Grant Number 1R01HL62633-01); the Spanish Ministry of Health (FIS-97/0035/01), the Catalan Government (CIRIT-1999SGR 00241); the Health Institute Carlos III, Ministry of Health (Grant 00/0316); and the Spanish Foundation of Respiratory Pathology (FEPAR Grant, 1998). We thank Xavier Basagan˜a for statistical assistance.
References Bakke P.S., Hanoa R., and Gulsvik A. Relation of occupational exposure to respiratory symptoms and asthma in a general population sample: Self-reported versus interview-based exposure data. Am J Epidemiol 2001: 154: 477–483. Basagan˜a X., Sunyer J., Zock J.P., et al. Incidence of asthma and its determinants among adults in Spain. Am J Respir Crit Care Med 2001: 164: 1133–1137. Benke G., Sim M., Fritschi L., et al. Comparison of occupational exposure using three different methods: hygiene panel, job exposure matrix (JEM), and self reports. Appl Occup Environ Hyg 2001: 16: 84–91. Blanc P.D., and Tore´n K. How much adult asthma can be attributed to occupational factors? Am J Med 1999: 107: 580–587. Bollen K.A. Structural Equations with Latent Variables. John Wiley & Sons, New York, NY, 1989.
Journal of Exposure Analysis and Environmental Epidemiology (2004) 14(5)
Zock et al.
Burney P.G.J., Luczynska C., Chinn S., et al. The European Community Respiratory Health Survey. Eur Respir J 1994: 7: 954–960. Chan-Yeung M. Western Red Cedar (Thuja plicata) and other wood dusts. In: Bernstein I.L., Chan-Yeung M., Malo J.L., and Bernstein D.I. (Eds.) Asthma in the Workplace. Marcel Dekker Inc., New York, NY, 1999, pp. 543–563. Chan-Yeung M., and Malo J.L. Tables of major inducers of occupational asthma. In: Bernstein I.L., Chan-Yeung M., Malo J.L., and Bernstein D.I. (Eds.) Asthma in the Workplace. Marcel Dekker Inc., New York, NY, 1999, pp. 683–720. Charous B.L., Blanco C., Tarlo S., et al. Natural rubber latex allergy after 12 years: Recommendations and perspectives. J Allergy Clin Immunol 2002: 109: 31–34. Houba R., Doekes G., and Heederik D. Occupational respiratory allergy in bakery workers: A review of the literature. Am J Ind Med 1998: 34: 529–546. International Labor Office. International Standard Classification of Occupations: ISCO 88. ILO, Geneva, 1991. Karjalainen A., Kurppa K., Martikainen R., Klaukka T., and Karjalainen J. Work is related to a substantial portion of adultonset asthma incidence in the Finnish population. Am J Respir Crit Care Med 2001: 164: 565–568. Karjalainen A., Kurppa K., Virtanen S., et al. Incidence of occupational asthma by occupation and industry in Finland. Am J Ind Med 2000: 37: 451–458. Kennedy S.M., Le Moual N., Choudat D., et al. Development of an asthma specific job exposure matrix and its application in the epidemiological study of genetics and environment in asthma (EGEA). Occup Environ Med 2000: 57: 635–641. Kogevinas M., Anto´ J.M., Soriano J.B., et al. The risk of asthma attributable to occupational exposures. A population-based study in Spain. Am J Respir Crit Care Med 1996: 154: 137–143. Kogevinas M., Anto´ J.M., Sunyer J., et al. A population based study on occupational asthma in Europe and other industrialized countries. Lancet 1999: 353: 1750–1754. Kromhout H., Heederik D., Dalderup L.M., et al. Performance of two general job-exposure matrices in a study of lung cancer morbidity in the Zutphen cohort. Am J Epidemiol 1992: 136: 698–711. McGuire V., Nelson L.M., Koepsell T.D., et al. Assessment of occupational exposures in community-based case-control studies. Annu Rev Public Health 1998: 19: 35–53. Meredith S. Reported incidence of occupational asthma in the United Kingdom, 1989–90. J Epidemiol Commun Health 1993: 47: 459–463. Office of Populations Censuses Surveys. Classification of Occupations 1980 (OPCS-CO). Her Majesty’s Stationery Office, London, 1980. Skov T., Deddens J., Petersen M.R., et al. Prevalence proportion ratios: estimation and hypothesis testing. Int J Epidemiol 1998: 27: 91–95. Sunyer J., Kogevinas M., Kromhout H., et al. Pulmonary ventilatory defects and occupational exposures in a population-based study in Spain. Am J Respir Crit Care Med 1998: 157: 512–517. Tore´n K., Ja¨rvholm B., Brisman J., et al. Adult-onset asthma and occupational exposures. Scand J Work Environ Health 1999: 25: 430–435. Venables K.M., and Chang-Yeung M. Occupational Asthma. Lancet 1997: 349: 1465–1469. Zock J.P., Kogevinas M., Sunyer J., et al. Asthma risk, cleaning activities and use of specific cleaning products in Spanish indoor cleaners. Scand J Work Environ Health 2001: 27: 76–81.
403
