INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Occupational asthma (OA) has been defined as asthma originating from causes and conditions attributable to a particular occupational environment and not to stimuli encountered outside of the workplace (1). Estimates of the proportion of asthma attributable to occupational exposure in adults range from 1.7% to 20% (2). The variation in estimates across countries with different levels of industrialization may reflect a true variation in the prevalence of OA, but could be due to differences in methodology, such as in the source of the study population (hospital- or clinic-based, population-based) or in case definitions (self-attributed OA or working in an occupation with risk for the development of OA). We report here a study of the prevalence, determinants, and characteristics of OA among a large, population-based sample of adult residents in six communities in Canada.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A multicenter prevalence study of asthma and asthmalike conditions was conducted in Canada between March 1993 and November 1994. The protocol followed that of the European Community Respiratory Health Survey (ECRHS) (10). Six communities were studied: five urban (Vancouver, Montreal, Winnipeg, Halifax and Hamilton) and one mixed urban and rural (Prince Edward Island [PEI]). In each location, a randomly selected sample of 3,000 to 4,000 men and women, ranging in age from 20 to 44 yr, were selected. In five centers (all except PEI), the random-digit telephone dialing method was used to identify subjects. In PEI a random sample was selected from the population registry of the provincial health plan. Subjects were sent an initial questionnaire by mail that was a modification of the ECRHS screening questionnaire. A repeat questionnaire was sent to nonresponders. Those who did not respond to two mailings were invited to complete the questionnaire over the telephone. A preselected random subsample of the respondents (without oversampling of symptomatic subjects) was invited to the laboratory.

Questionnaires were administered through interviewers. For additional studies, written informed consent was obtained from all subjects. These subjects completed the second-stage ECRHS questionnaire on symptoms and risk factors for asthma. An additional battery of questions, not part of the ECRHS protocol and developed specifically for the present study, sought detailed information on occupational history (all jobs held for 3 mo or more), on extent of exposure to any of a previously defined list of 39 specific materials known to induce asthma, and on work-related symptoms. The wording of the question about exposure was as follows: "If you have worked directly with these materials or have worked closely to others working with these materials, check under `yes' and indicate whether this occurred on an occasional or regular basis and for how long. Only answer `yes' if you are certain." Subjects with asthma as subsequently defined and who completed laboratory testing were the basis of this report.

Occupational Histories

Occupations and industries were coded according to the 1980 Statistics Canada classification system for occupations and industries (11). Coding of the separate occupational questionnaire was done without knowledge of the subjects' asthma status.

Definitions

Asthma was defined by "yes" answers to the questions "Have you ever had asthma?" and "Was it diagnosed by a doctor?" Adult-onset asthma was defined as asthma with a first attack at age 15 yr or older. Fifteen years of age was chosen as the criterion for this because it was the minimum first-year age of employment for asthmatic individuals in this survey.

OA was defined by using two mutually exclusive criteria, as follows: (1) Probable OA, consisting of a history of asthma in which the first attack occurred during employment in a previously defined combination of occupation and industry considered to carry a high risk for OA (1, 18) (Table 1); and (2) Possible OA, consisting of a history of asthma with the first attack occurring during employment when exposure to a known sensitizing agent or irritant was reported, but not in one of the high-risk combinations of occupation and industry listed in Table 1. The substances in Criterion 2 were defined a priori, and included hardeners/accelerators, paints, resins (all potential source of isocyanates), soldering, biologic enzymes, dyes, formaldehyde, glues, pharmaceutical agents, ammonia, solvents, vegetable gum (printing), grain or flour, cotton dust, cedar dust, other wood dust, fur dust, coffee dust, hay dust, platinum, nickel, chromium, and cobalt.

These definitions of Probable OA and Possible OA are consistent with the surveillance definitions for OA used by the U.S. National Institute for Occupational Safety and Health (NIOSH) in its Sentinel Event Notification System of Occupational Risks (SENSOR) program (12). The remaining subjects with adult-onset asthma did not report exposure to any of the foregoing agents at the onset of their asthma, and were classified as Not OA.

Spirometry

Spirometry was done by trained technicians in all centers, using dry-rolling-seal spirometers (Spirotech Inc.). Calibration and testing procedures for spirometry were those recommended by the American Thoracic Society (13).

Allergy Skin Tests

Skin tests for allergy were done with the skin prick method, using 14 common inhalant allergens together with normal saline as a negative control and histamine (1 mg/ml) as a positive control. The skin tests were read at 15 min, and the largest diameter of a wheal and its perpendicular dimension were measured and averaged. A positive skin reaction was defined as one with an average wheal diameter that was 3 mm greater than that of the negative control. Atopy was defined as a positive skin test reaction to one or more of the common allergens.

Statistical Analysis

We compared the characteristics of the Probable OA, Possible OA, and Not OA groups. Chi-square analysis or Fisher's exact test (if the expected count in one cell was less than five) was used for comparing categorical variables. For continuous variables, two groups were compared through Student's t test, and more than two groups were compared with one-way analysis of variance (14) or Tukey's modified t test. A value of p < 0.05 was considered significant.

Population attributable risk (PAR) (15) was calculated with the formula PAR = (Pe [RR  1]/RR), where Pe is the prevalence of exposure in cases of adult-onset asthma and RR is the relative risk. RR was estimated with the odds ratio (OR) for adult-onset asthma of work before or at the time of commencement of asthma in the high-risk occupation/ industries or exposures. This was calculated by using conditional logistic regression, with each case of adult-onset asthma compared with five randomly selected controls (who never had asthma) of the same age and sex as the case. Occupations, industries, and exposures of the case up to the age of asthma onset were considered. Smoking was controlled for in the analysis. ORs were calculated for the association between each combination of high-risk occupation/industry and high-risk exposure. Occupation/industry combinations and exposures that were responsible for OA cases (Table 2) were used in the calculation of PAR.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The initial questionnaire was sent to 21,617 individuals in the six study communities. Of these individuals, 18,701 (87%) completed and returned the questionnaire. Approximately 28% of the initial questionnaires were completed by telephone interview. Of those individuals who completed the initial questionnaire, 7,701 were invited to the laboratory. Of the 2,974 (39%) who attended, 383 had asthma. All of the individuals who attended the laboratory completed additional questionnaires on symptoms and occupation. Almost all had spirometry and allergy skin testing (96% and 97% for each procedure, respectively).

Table 3 compares the laboratory attendees with individuals who answered the initial questionnaire only. It can be seen that more women than men took part in both the initial and laboratory part of the survey. Laboratory attendees were slightly older, and were more likely to be nonsmokers, to have respiratory symptoms, and to report a diagnosis of asthma. Approximately 70% of all attendees with adult-onset asthma were women.

Among the laboratory attendees were 383 individuals with asthma, of whom 166 had adult-onset asthma. Of the individuals with adult-onset asthma, 27 were working in a previously defined high-risk occupation and industry for OA at the time of onset of their asthma, and therefore fulfilled the criteria for Probable OA. Thirty-three individuals reported occupational exposure to an agent known to cause OA, but did not report being in a defined high-risk occupation and industry group at the time of onset of their asthma (Possible OA). The remaining 106 individuals with adult-onset asthma formed the Not OA group. Twenty-six members of this group were unemployed at the onset of asthma, and 80 were working but did not meet the criteria named earlier for Possible or Probable OA. The percentage (95% CI) of the attendee population with Probable OA was 16.3% (12.6 to 20.0%), the percentage with Possible OA was 19.0% (15.9 to 23.9%), and the percentage with Probable and Possible OA was 36.1% (31.3 to 41.0%) of all cases of adult-onset asthma. The corresponding figures for all cases of asthma were 7.1% (4.5 to 9.6%), 8.6% (5.8 to 11.8%), and 15.7% (12.0 to 19.3%), respectively.

Of the 166 attendees with adult-onset asthma, 94 (56.6%) worked in a combined high-risk occupation/industry or had exposure either before or at the time of commencement of their asthma. The OR for the association of working in a combined high-risk occupation/industry or of exposure with the development of adult-onset asthma was 1.48 (1.05 to 2.09). The PAR was 18.2%.

The occupation, industries, and exposures at work reported by subjects with Probable OA and Possible OA are presented in Table 2. Nursing, baking, and hairdressing, and chemical, rubber, or plastics processing were common occupations in the Probable OA group. Of the 27 attendees with Probable OA, 20 (74%) also reported exposure to an agent known to cause OA. In the Possible OA group, common industries were healthcare, civil service, and food, and common occupations were clerical work and food preparation. The most common exposures of any kind reported at the time of commencement of asthma by attendees with Possible OA and Probable OA were similar. These included ammonia, paints, acids, grain or flour, glues, insecticides/herbicides, enzymes, and solvents. Attendees with Possible OA reported exposure to more agents than did those with Probable OA, at a mean of 2.9 agents per case as compared with 1.7, respectively (p = 0.02). On the other hand, more of the exposures in the Probable OA group were reported as regular rather than occasional (53% versus 45% in the Possible OA group), although this was not statistically significant.

The characteristics of the Probable OA, Possible OA, and Not OA groups are given in Table 4. There were no significant differences in sex, age, FEV₁ or smoking habit among the groups. All groups had a high prevalence of symptoms consistent with asthma over the 12 mo previous to the survey. The proportion with current wheezing or dyspnea at or after work was similar in the Possible OA and Probable OA groups, and differed significantly from that in the Not OA group. Subjects with Probable OA were significantly more likely to have changed jobs since their first asthma attack, and the median interval between commencement of the high-risk job and the first asthma attack was 1.0 yr (range: 0 to 18 yr), and the median interval between the first asthma attack and leaving the high-risk job was 1.5 yr (range: 0 to 9 yr). The median interval between the first asthma attack and the study date was 6.3 yr (range: 2 to 24 yr). By contrast, the prevalence of atopy was lower in the Probable and Possible OA groups than in the Not OA group.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We examined the prevalence of OA among individuals with adult-onset asthma in a large, population-based cross-sectional study of adults in six communities in Canada. Our criteria for this were: (1) work in a high-risk occupation and industry; or (2) exposure to a known cause of OA. Of subjects with adult-onset asthma, 36.1% (95% CI: 31.3 to 41.0%) and of all subjects with asthma, 15.7% (95% CI: 12.0 to 19.3%) satisfied one of the case definitions. The PAR of these occupation/industry combinations and exposures for the development of adult onset asthma was 18.2%. These figures are similar to those reported elsewhere (2, 3), and indicate that an estimated 18.2% of cases of adult-onset asthma in our Canadian urban population would have been prevented if exposure to the combined high-risk occupation/industries and exposures that we investigated were eliminated.

The exposures commonly reported were similar in cases of both Probable and Possible OA. In the majority of cases, the suspected etiologic agent of OA was a high-molecular-weight substance. These substances included latex, flour, grain, biologic enzymes, vegetable gum (printing), and laboratory animal antigens. Common low-molecular-weight etiologic agents included isocyanates and western red cedar dust. In both the Probable and Possible OA groups, the majority of subjects had changed jobs since their asthma commenced, and most still had respiratory symptoms in their current job. Since subjects with the two categories of OA were similar in all respects, they were considered together.

A limitation of the present study was its reliance on questionnaire data, which depend on subject recall. Reported occupations, industries, and exposures were not validated. Questionnaire data are also problematic in the diagnosis of asthma, owing to the absence of a "gold standard". Assessment of physician-diagnosed asthma by questionnaire has a low sensitivity but high specificity (99%) when validated with nonspecific bronchial hyperresponsiveness (16). This may result in an underestimation of the prevalence of asthma. However, the prevalence of asthma in our study was similar to previously published prevalences in adult populations (17). Laboratory attendees were more likely to have asthma and respiratory symptoms than individuals answering the initial questionnaire only. Again, this would not have biased our estimates of OA unless asthmatic individuals with significant occupational exposures were more likely to have attended the laboratory. The latter scenario is highly unlikely, since detection of cases of OA was not an original aim of our study, and questions on asthma symptoms and occupational exposures were contained in separate questionnaires.

The estimated prevalence of OA in adult asthmatic populations has varied in published studies, depending on the case definitions of OA used and the population studied. Blanc and colleagues (3) analyzed data from the 1978 U.S. social security survey, and found that 15.4% of asthmatic individuals attributed their asthma to workplace exposure. In another study, the same investigators reported the prevalence of OA among adult asthmatic individuals as being 6% to 17%, depending on the definition used (2). In the latter study, adult asthmatic individuals (aged 18 to 50 yr) were recruited from patient-visit logs maintained by pulmonary or allergy specialists in northern California who were sampled at random. The definition of OA was based on subjects' reports of exposures at the time of asthma onset. An asthma-specific job-exposure matrix for occupations carrying a risk for OA was also used. Timmer and Rosenman (4) studied 94 patients with asthma, aged 20 to 65 yr, who were discharged from Michigan hospitals. They found that 3% of adult asthma patients satisfied the NIOSH criteria for probable OA, and that up to 20.2% satisfied the criteria for possible and probable OA. Probable OA was defined as patients with: (1) asthma; (2) an association between symptoms of asthma and work; and (3) workplace exposure to an agent or process associated with OA. Cases of possible OA satisfied criteria (1) and (2) only. The definition of an association between asthma and work (improvement in symptoms on days away from work, or worsened symptoms during the working day and working week) was more restricted than in our study (asthma commencing during employment in a specific job). Of 580 adult asthmatic individuals seen at a clinic in Zambia, 5.9% had OA diagnosed according to a case definition that included a change of employment because of persistence of asthma symptoms (5). Measures of the incidence of OA are also available. In a report based on national registries of asthma and OA cases between 1986 and 1993 in Finland, the proportion of newly diagnosed OA was 4.8% of all new cases of asthma (6).

One of the major strengths of our study was that it was population-based, unlike the other studies just described. Three other population-based studies (in New Zealand [7], Spain [8], and Europe and other industrialized countries not including Canada [9]) reported the attributable risk of occupational exposures in adult asthma. The estimates were lower than the results in the present study, and varied from 1.9% to 10%, depending on the asthma definition used. This may have resulted from the more detailed occupational information we had available in the present study. In these other studies, only current occupation or occupation at the time of a change of occupation because of health problems was available. The varying results among these other studies and their differences from our results may also reflect differences in the industrial exposures in the different regions.

Some industries in which cases of Possible OA in our study had worked are known to carry a high risk for OA, such as baking and healthcare, but the cases were not included in the Probable OA group because their occupations were clerical. Other industries, such as building and telecommunications, are not generally recognized as high-risk industries for OA. Our results indicate that obtaining exposure information is useful in detecting OA, since cases identified by the criteria for both Probable and Possible OA were similar clinically and reported similar exposures. We believe that this is the most likely reason for the higher estimates of OA in our population-based study, since the other two population-based studies of OA (7, 8) did not have detailed information on exposure. The results of our study underscore the importance of taking a history not only of occupation but also of exposures in all patients with new adult-onset asthma.

We did not rely on subjects ascribing symptoms to work, but decided a priori about at-risk groups. Different questionnaires were used for symptoms and for occupation. It is unlikely that subjects with asthma are more careful in recalling all of their jobs. It is possible that cases of asthma arising in the predefined risk groups in our study were unrelated to occupational exposure, as objective evidence of work-relationship was not available. This criticism can also be applied to several published studies and to all community-based studies (2).

In conclusion, in a population-based cross-sectional study of adults in six communities in Canada, we found that approximately one in three cases of adult-onset asthma may have been caused by occupational exposures, and that 18% of adult-onset asthma may be prevented by eliminating exposure to known occupational inducing/sensitizing agents. Occupational exposure as a cause of adult-onset asthma may be more common than is realized.