Physical examination and a questionnaire, including previous and current job history and environmental factors, were completed at the time of visiting a physician. Occupational exposures were estimated for the current or most recent job code (24) and assessed by trained interviewers. Government occupational hygienists performed the expert step according to the published method (17). Our study used the asthma-specific JEM developed by Kennedy and colleagues (17). The JEM has 22 exposure groups, including 18 high-risk groups, based on known risk factors (referred to here as “asthmogens”) for occupational asthma, grouped into high-MW agents, low-MW agents, and mixed environments (see Table E1). The “no-risk group” means that the study subjects are unlikely to be exposed to substances associated with a risk of asthma or to other irritating chemicals. The “low-risk group” consists of patients with low levels of exposure to irritant chemicals (no high peak exposures), exhaust fumes, and environmental tobacco smoke (17, 18). The tables display the results for a specific exposure if there were at least five patients and four control subjects for that exposure, and the numbers of exposed subjects in both case and control groups are shown in Table E1.

Blood samples were collected in heparin-containing venipuncture tubes for the Phadiatop test and determination of total plasma IgE (IMMULITE 2000 analyzer; Diagnostic Products Corporation, Los Angeles, CA). The Phadiatop test was analyzed by immunoassay system (Pharmacia ImmunoCAP; Pharmacia, Uppsala, Sweden). This test is an in vitro assay for specific IgE production in response to a mixture of commonly inhaled allergens (>20 items). Results are expressed as Pharmacia arbitrary units per liter (PAU/L), indicating the degree of sensitization. Phadiatop values equal to or greater than 0.35 PAU/L are considered indicative of sensitization, and this was the threshold for a positive or negative response in all our dichotomous analyses.

Total plasma IgE was measured in a solid-phase, chemiluminescence immunometric assay (IMMULITE 2000 analyzer; Diagnostic Products Corporation, Los Angeles, CA). We defined high responsiveness of total IgE in this study as greater than 100 U/ml (25). Atopy was defined as an increase in total IgE (≥100 U/ml) or a positive Phadiatop test result (≥0.35 PAU/L). Asthma with atopy was defined as having asthma in combination with an increase in total IgE (≥100 U/ml) or a positive Phadiatop test result (≥0.35 PAU/L).

The χ² test was performed to evaluate percentage differences in risk factors between the case and control groups. Analysis of variance was used to assess the mean differences between the three groups. The adjusted odds ratio (AOR) and 95% confidence interval (CI) for each asthmogen and environmental factor was calculated by a multiple logistic regression. The multiple logistic regression included several factors that are associated with asthma, including alcohol use; smoking; body mass index (BMI); and exposure to pets, cockroaches, and indoor incense burning. Lung function was compared between the three groups after adjusting for smoking, age, and sex by using multiple linear regressions. The population-attributable fraction was calculated according to the following formula: frequency of any asthmogen in the case group × [(odds ratio – 1)/odds ratio], where the odds ratio for the effect of an occupational asthmogen on current asthma was estimated on the basis of the multiple logistic regression model. The statistical analyses were performed with SPSS, release 14.0 (SPSS Inc., Chicago, IL).

We classified asthma into two groups, asthma with or without atopy, and found that high-MW and certain low-MW occupational asthmogens have a significant effect on atopic asthma. We also observed that the effect of low-MW agents on nonatopic asthma was higher than on atopic asthma. Our results support the hypotheses of Bardana, who proposed that atopic individuals are more likely to become sensitized and to have allergic OA when exposed to high-MW allergens (10). On the other hand, the relationship between atopic OA and exposure to low-MW agents remains inconsistent (4). Atopic OA is affected only by some specific low-MW agents, such as salts of platinum (10). In contrast, atopic individuals do not have a higher susceptibility to asthma induced by low-MW agents, such as western red cedar (26) or diisocyanates (27). Nonatopic OA can also be induced by exposure to high-level irritants (8).

We found that exposure to specific asthmogens, such as latex, is associated with atopic asthma. Highly reactive chemicals are associated with nonatopic asthma, whereas agriculture, industrial cleaning agents, and metal sensitizers (Tables E2 and E3) are associated with both types of asthma. In many countries, powdered natural rubber latex (NRL) gloves have been largely replaced by NRL-free gloves, and their use has declined from 80.9% in 1989 to 17.9% in 2004 (28). Thus, the incidence of NRL-induced OA has significantly decreased since the late 1990s in western countries (28). In Taiwan, however, disposable NRL gloves are still predominantly used. Few hospitals have latex allergy interventions or a named consultant responsible for latex allergy provisions (29). Our findings indicate a high risk (3.8- to 6.5-fold) of exposure to latex on atopic asthma compared with community control subjects.

Several prospective studies have provided evidence of an increased risk for asthma in those exposed to domestic cleaning agents either professionally (5) or privately (30). Specific cleaning products, including chlorine bleach and frequent use of cleaning sprays, are associated with asthma symptoms (30, 31). Cleaning agents have both irritant and sensitizing properties. However, it remains unclear which mechanisms play a predominant role (32). Various metals, including Pt salts, Zn, Ni, Cr, and Co, have been recognized as possible causes of occupational asthma (33). Studies also suggest that exposure in the aluminum industry is associated with asthma symptoms (34) and exhaled nitric oxide (35). Farmers and other agricultural workers may develop occupational asthma as a result of exposure to grains, animals, dust, or pesticides (36, 37). High pesticide exposure may contribute to both atopic and nonatopic asthma in male farmers (37), but it was associated only with atopic asthma in female farmers (38). In our study, we found that agricultural workers are at an increased risk for atopic (AOR, 7.8) and nonatopic asthma (AOR, 4.1) compared with community control subjects.

Our study has several strengths. First, we used a case–control study to minimize healthy worker effects and to assess the effects of occupational exposure across the entire spectrum of jobs. Cross-sectional workplace surveys may be better at assessing the current frequency, duration, and intensity of exposure to suspected agents in the workplace. However, such studies can be done only in selected factories or workplaces. Moreover, workers who develop symptoms often leave their jobs or transfer to positions where less exposure is required. This healthy worker bias leads to underestimates of relative risks (39). By contrast, case–control studies are likely to recruit all patients ever exposed in at-risk workplaces, rather than those currently exposed. Therefore, case–control studies are less biased by the healthy worker survivor effect. Second, we used a JEM based on the current knowledge of risk factors for asthma and this JEM has been shown to give reliable estimates of asthmogen exposure (17, 18). The questions regarding current job and industry title used in JEM are less subject to recall bias. Sun and colleagues compared surrogate indicators of exposure, which were defined by “only,” “ever,” “longest,” “last,” and “current” employment history, in a hypothetical cohort with those in an actual cohort. Except for workers in the “only” employed group, all other definitions overestimated the risk in the low-risk jobs and diminished dose–response effects (40). Third, our multiple logistic regression analysis allowed us to control for potential confounding factors. As expected, several factors, including alcohol use, smoking, BMI, and exposure to pets, cockroaches, and indoor incense burning, were associated with asthma in the present study. These potential confounding factors were included and adjusted for in the multiple logistic models.

The limitations of this study should also be discussed. The first limitation is its small sample size for comparisons of some job categories, thus leading to wide confidence intervals for OR. Second, the mean age of the study group was rather high (average, 50 yr old). Chronic obstructive pulmonary disease (COPD) and asthma are major chronic obstructive airway diseases that involve airway inflammation. Unlike in asthma, airflow limitation in COPD is not fully reversible and is often progressive. COPD is usually associated with an abnormal inflammatory response of the airway to noxious particles or gases. In general, COPD is defined by fixed airway obstruction as measured by postbronchodilator FEV₁/FVC < 0.70 (41). We have done our utmost to exclude those with overlapping diagnoses. However, even though asthma can usually be distinguished from COPD, in some individuals it is difficult to differentiate the two diseases. Despite the availability of consensus guideline diagnostic recommendations, confusion between COPD and asthma appears to be common in patients 40 years of age and over (42). Our pulmonary function testing followed the guidelines of the American Thoracic Society (ATS). The ratios of FEV₁ to FVC may be higher but were all consistent between the three risk groups of patients (Table 5). Third, it is difficult to obtain direct information about past work exposure from workplaces after subjects have quit or retired from their jobs. If the subjects had worked in more than one place, we evaluated the occupational exposure for the current or most recent job to reduce recall problems. However, when we restricted our analyses to currently employed subjects with asthma (Table 4 and Table E5), the AORs were consistent with, but higher than, those found for all subjects with current asthma (Table 3 and Table E4). On the basis of these findings, recall problems did not appear to confound the results significantly. Fourth, in our present study, the 1,560 hospital-based control subjects (994 males and 566 females) were recruited irrespective of job title and exposure. Many female control subjects (50.8%, 288 of 566) were excluded because they had never worked or they had missing answers for questions concerning job title and work exposure. So, we were unable to fully match hospital control subjects to asthma for age and sex by 1:1 because insufficient suitable hospital control subjects were available to match some female subjects with asthma. However, the confounding effects of age and sex were adjusted for by running multiple logistic regressions (Tables E2–E4).

Total serum IgE, Phadiatop, and skin prick tests (SPTs) are usually used to diagnose atopic diseases, but in the present study we did not perform SPTs. A randomized study of 8,329 adults compared the diagnostic value of these three atopic markers in adults with current allergic asthma and rhinitis. The sensitivity of the Phadiatop test was significantly higher than that of the SPT and IgE. Nevertheless, the SPT has the best positive predictive value and is the most efficient test (43). The validity of the Phadiatop test was compared with the SPT in schoolchildren and an adult population. The Phadiatop test was 85% accurate, qualifying it as a valid alternative to the SPT for diagnosing atopy (44, 45).

To improve the health status of adults with asthma, a public health approach is required. Smith and Bernstein proposed a three-stage surveillance program in a disease prevention initiative (3). Primary prevention consists of assessing workplace risks and instituting appropriate exposure control measures, whereas secondary prevention involves early identification of workers with occupational sensitization. Tertiary prevention involves removing a worker with established OA from exposure to a recognized occupational sensitizer and preventing this employee from developing more severe asthma (3). The detergent-manufacturing industry developed an exemplary medical surveillance program and used medical questionnaires combined with skin testing to identify at-risk employees. This program successfully maintained the incidence of newly sensitized workers below 3% and prevented new cases of OA (46).

In conclusion, we found that high-MW asthmogens and low-MW asthmogens may pose different risks for inducing atopic and nonatopic asthma. Our results suggest that the population-attributable fraction due to occupational asthmogens for current asthma is 16.3%. Protective standards will have to be established that involve improving the social, occupational, and environmental conditions of workers to minimize OA.