INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Chronic obstructive pulmonary disease (COPD) remains a considerable cause of increased respiratory morbidity and mortality (1) and although it is assumed that the majority of cases of this condition are related to cigarette smoking (2), other causes are recognized (1). The potential link with occupational exposures, and particularly those associated with dust exposure, has long been noted (3), both from cross-sectional and from longitudinal studies (4, 5).

Population-based studies, used to assess the effects of occupational and other exposures on symptoms of respiratory diseases and abnormalities of pulmonary function, are less commonly reported than are specific industry-based, cross-sectional studies. Recent such studies, however, have suggested new effects of occupational exposures on the development of asthma (6, 7) and chronic respiratory disease (8).

Furthermore, as the rate of adult smoking continues to decline in many industrialized societies, it may well be that other important exposures (12), including atmospheric pollution (13) and workplace exposures (5), will become more important etiologic factors in the production of this condition. Most previously reported occupational-based studies have relied on specific exposures or descriptions of exposures such as "dust" or "fumes." We present the results of an analysis of questionnaire data and pulmonary function performed during Phase 2 in New Zealand of the European Community Respiratory Health Survey (ECRHS) (14) in order to assess the effect of occupation on the presence of symptoms suggestive of COPD (chronic bronchitis and shortness of breath), either alone or in combination with evidence of mild airway obstruction. This study uses job title to categorize exposure. This is identical to the classification used previously in a New Zealand population-based study of occupational asthma conducted by our group (7) and a similar study based in Spain (6). This study used the same random population-based sample for analysis, but rather than focusing on asthma symptoms (7), we investigated the relationship between occupation and four different symptom categories suggestive of COPD in nonasthmatics.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The study methodology has been described in a previous publication (7). Briefly, a random sample (n = 14,318) of the population 20 to 44 yr of age living in four geographic areas were contacted and asked to complete a short screening questionnaire on respiratory symptoms as part of the Phase 1 sampling in the New Zealand part (15) of the ECRHS; 11,978 of them (83.7%) responded to this initial phase.

We contacted 2,004 subjects from three of the four original areas in the second phase of the study (the fourth area did not take part in the second phase). This represented a 22% random subsample of the first phase participants in addition to an "enriched" subsample of 515 (from the same three areas) reporting respiratory symptoms in the screening questionnaire but who had not already been selected in the random sample. The random selection of this group was made from all participants answering yes to at least one of a triad of questions relating to asthma symptoms, attacks, and medication use (corresponding to questions 3, 5, and 6 from Phase 1 of the ECRHS).

A detailed questionnaire (including questions on occupational history) was administered in addition to pulmonary function testing; 1,609 (63.9%) subjects completed the more detailed questionnaire from the 2,519 invited to do so. The response rates for the three individual areas of New Zealand were 58, 66.6, and 65.2%. The questionnaire included a comprehensive enquiry on respiratory symptoms, smoking, and occupation. Previous occupations were also recorded if the workers had changed employment at any time in the past because of respiratory symptoms.

Lung Function Assessment

There were 1,132 (70.4%) subjects who satisfactorily completed baseline lung function assessment consisting of FEV₁ and FVC. This consisted of five technically satisfactory forced expiratory maneuvers performed in the standing position. The best of these attempts for both FEV₁ and FVC was used to calculate the ratio of (FEV₁/FVC); this value is subsequently referred to as the FEV%.

Definitions of Chronic Bronchitis, Shortness of Breath, and Airway Obstruction

Four definitions were chosen to investigate the relationship between symptoms suggestive of COPD and occupation. In all cases, subjects who had reported a diagnosis of asthma confirmed by a doctor were excluded. This was done to exclude any participants who may have had current asthma and symptoms as a consequence of this.

1. Chronic bronchitis. Cases of chronic bronchitis were defined as all subjects answering positively to "usually bringing up phlegm from the chest first thing in the morning in winter or at any other time of the day or night in winter."
2. Chronic bronchitis and airway obstruction. This category included all subjects with symptoms of chronic bronchitis and a FEV% less than or equal to 0.75. This value was chosen so as not to censor a large proportion of a generally fit working population while acting as a threshold for mild airway obstruction, and it was done essentially to be able to identify those in the population with early airway obstruction.
3. Shortness of breath. Cases of "grade 1" shortness of breath were defined as all subjects answering positively to the question "are you troubled by shortness of breath when hurrying on level ground or walking up a slight hill?"
4. Shortness of breath and airway obstruction. Similarly, this category was defined as all subjects with shortness of breath and a FEV% less than or equal to 0.75.

All participants, including those with asthma and other respiratory symptoms, who did not meet the criteria to be included as cases of these symptom categories were included in the analysis as noncases.

Occupational History

The question "what is or was your current or most recent job?" was used to code current occupation. If subjects reported a change of job related to breathing problems, the job at that time was also noted and was used in the analysis rather than the most recent job, in order to strengthen any possible association with occupation and diagnostic category. The question used to investigate the change in job was "have you ever had to change or leave your job because it affected your breathing?", which was followed by "what was this job?" This approach has been used in a recent population analysis relating to occupational asthma by this group (7) and by others (6). Each of these occupations was initially coded into one of the 350 Office of Population Censuses and Surveys (1980) codes (16). Subsequently, these were converted into one of 21 combined groups as used in the surveillance of work-related respiratory diseases (SWORD) project to classify occupations (17), and these 21 groups formed the basis of the data analysis, with professional and administrative workers being used as the referent occupation for each analysis.

In order to investigate further any possible connection between occupational exposures and the diagnostic categories used in the main analysis, we also did an additional analysis of persons who reported having worked in a job that exposed them to vapors, gas, dust, or fumes.

Data Analysis

The data were analyzed as prevalence odds ratios for various combinations of symptoms and/or evidence of airway obstruction and also for comparison between those with and without specific occupational exposure. All adjusted odds ratios cited are adjusted for age, sex, and current, ex- or never smoking of cigarettes. All categories of COPD definition excluded those who reported ever having asthma diagnosed by a doctor.

Prevalence odds ratios were then calculated for each of the 21 occupational groups for each definition of "COPD" using SAS (18), and adjusted for age (in the categories shown in Table 1), sex, and tobacco smoking (defined as current smoker, ex-smoker, or never smoker) using the Mantel-Haenszel method (19). The group of "remainder professional, clerical, and administrative workers" was taken as the reference category for all analyses.

To assess the proportion of COPD prevalence within the population that could be attributed to dust, gas, fume, or vapor exposure, the population attributable risk was estimated for those subjects who reported such exposure and for those who did not, using the combination of chronic bronchitis and airway obstruction as the definition of COPD.

Because the study included both a random sample and an "enriched" sample of persons chosen on the basis of reported asthma symptoms, initial analyses were done adjusting for the selection factor (i.e., the presence or absence of the selected asthma symptoms). However, this adjustment made little difference in the findings reported here for chronic bronchitis; accordingly, for simplicity, it was decided not to adjust for the selection factor in the final analyses presented.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The limited data available from the initial Phase I questionnaire suggested that a higher proportion of nonresponders (n = 910) were male (52.6%) in comparison with responders (45.9%). Nonresponders were generally less symptomatic (as judged by the initial Phase 1 questionnaire) than were responders. For example, the prevalence of wheeze or whistle in the chest in the last 12 mo was 32.5% in the nonresponders in comparison with 37.7% in the responders.

The study population according to demographic factors, smoking status, and numbers satisfactorily completing pulmonary function testing is described in Table 1. Only 45% (1,132 of the 2,519 invited) underwent pulmonary function testing.

The mean FEV% (SD) for the total population of 1,132 participants with satisfactory pulmonary function assessment was 0.81 (0.08). Two hundred twenty-four subjects (19.8%) had a FEV% less than or equal to 0.75, and 42 (18.8%) of these had a mean percent predicted FEV₁ less than 80%. Similarly, 103 subjects (9.1%) recorded a FEV% less than or equal to 0.70, and 32 of these (30.1%) had a mean percent predicted FEV₁ less than 80%.

The 477 subjects who attended for the second phase of the study did not undergo pulmonary function testing. The prevalence of chronic bronchitis in these subjects was 12.4% in comparison with a prevalence of 12.9% in those with measured pulmonary function (OR = 0.95; 95% CI, 0.69 to 1.32).

The 143 subjects with symptoms of chronic bronchitis (41.1% of those with chronic bronchitis) were excluded as cases as they reported that a diagnosis of asthma had been made by a doctor at some stage, leaving 205 subjects with chronic bronchitis for analysis. Similarly, 200 subjects with shortness of breath (46.8% of the total with shortness of breath) were excluded as cases for the same reason, leaving 227 subjects for analysis. The analysis was repeated (results not shown), excluding all subjects with doctor-diagnosed asthma from both cases and noncases. The results generally showed a slightly stronger effect of individual jobs as risk factors for chronic bronchitis.

Again, the analysis used predominantly the current occupation reported by the subject. However, 230 subjects (14.3%) reported a change of job because of breathing problems. There was an increase in the prevalence of chronic bronchitis in this group (15.2%) in comparison with those not reporting a change of job for this reason (12.2%); OR = 1.3; 95% CI, 0.9 to 1.9. However, the FEV% did not differ significantly between groups.

The adjusted odds ratios and their associated 95% confidence intervals for the prevalence of chronic bronchitis (with or without airway obstruction) within each occupational group in comparison with the referent group are shown in Table 2. The identical analysis for shortness of breath is presented in Table 3.

The numbers of subjects in some occupational groups were relatively small; nevertheless, two occupational groups showed a significant association with chronic bronchitis; these were for food processors, other than bakers, (OR = 2.83; 95% CI, 1.27 to 6.29) and chemical processors (OR = 18.84; 95% CI, 3.71 to 95.64). When the definition of chronic bronchitis in combination with airway obstruction was used, bakers (OR = 25.50; 3.86 to 168.53) and spray painters (OR = 16.62; 2.32 to 119.34) showed significant associations with this diagnosis. The association with bakers and spray painters was also seen with the combination of shortness of breath and airway obstruction, as shown in Table 3.

Some occupations also showed elevated odds ratios, which were not statistically significant. In particular, chronic bronchitis was more common in spray painters (OR = 3.33; 0.75 to 14.83) and construction/mining workers (OR = 2.67; 0.70 to 10.20). Shortness of breath was high in bakers (OR = 6.72; 0.57 to 79.66) and in hairdressers (OR = 2.75; 0.80 to 9.42). Nursing was associated with lower levels of shortness of breath in comparison with the referent occupation (OR = 0.42; 0.16 to 1.15).

A further analysis, (not shown in the tables), was performed to investigate the relationship between measured airway obstruction (FEV% less than equal to 0.75), regardless of the presence of symptoms. This showed an association with spray painters (OR = 6.71; 1.29 to 35.01) and nonsignificantly elevated odds ratios for work as laboratory technicians (2.80; 0.55 to 14.42), bakers (2.68; 0.32 to 22.51), and plastics and rubber workers (3.33; 0.55 to 20.14).

When the identical analysis was performed on a more restricted definition of airway obstruction (including only those 32 subjects [2.8% of the population with pulmonary function testing] with a FEV% less than 0.7 in addition to a percent predicted FEV₁ of less than 80% [termed subsequently as severe airway obstruction]), 21 of these subjects were contained within the referent class. The remaining 11 were distributed thorough various occupational groups, and only construction and mining showed a significant relationship to this definition (12.7; 2.4 to 66.7). The small numbers within each group obviously make the interpretation of these data more difficult.

The analysis of chronic bronchitis by occupation was repeated using current job only (data not shown) in order to estimate the effect of using current job or job causing breathing problems. There were generally small changes in the odds ratios for chronic bronchitis relating to each job. For example, the odds ratio for chronic bronchitis in chemical processors reduced from 18.8 to 8.1 (0.8 to 88.2) when change of job was ignored in the analysis.

The significant association between the group ever exposed to dust, gas, vapors, or fumes (817 subjects; 50.8%) and both the presence of shortness of breath and the combination of chronic bronchitis and airway obstruction are illustrated in Table 4. These associations remain after adjustment for current smoking habit (in addition to adjustment for age and sex), despite the relatively small number of subjects in the study population with the degree of airway obstruction incorporated into certain diagnostic categories.

Of the 24 subjects with the combination of chronic bronchitis and mild airway obstruction, 20 had reported previous occupational exposure to dust, gas, vapor, or fumes. This exposure accounted for an attributable risk of developing COPD (using this definition) of 19.3%.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This population-based study, using a random sample of the New Zealand adult population between the ages of 20 and 44, has identified certain occupation groups to be associated both with symptoms of COPD and also diagnostic categories incorporating degrees of airway obstruction. The overall moderate response rate must inevitably lessen the certainty with which the results are interpreted. Furthermore, the analyses presented here involved multiple comparisons, but the fact that the significant associations between job category and definitions of COPD reported here made clinical sense supports the hypothesis that these were not entirely due to chance.

There is now some evidence from cross-sectional, case control, and prospective cohort studies (4, 5), based both in the community and the workplace, suggesting that occupational exposures, particularly to dust and (to a lesser extent) gases, are important in both the development of respiratory symptoms and the presence of airway obstruction. These problems apparently occur in addition to the effect of other exposures such as cigarette smoking.

This study was able in part to overcome some of the potential problems associated with the healthy worker effect (20), namely, that susceptible persons may leave dusty work areas early in the course of the job, leaving behind a relatively less susceptible group of workers who are potentially more able to tolerate the work environment. Our questionnaire and method of analysis allowed us to include those who had changed jobs specifically for this reason, and when this occurred, the job causing this change was used in the analysis.

The information collected in this study on occupational exposures was relatively limited, and it relied primarily on the job title to categorize exposure. Nevertheless, the analysis did show a significant association between these various job titles grouped together into categories and the diagnostic categories used for COPD. It is difficult to comment further, however, on potential differences in exposure between persons within the same job title grouping, and indeed different workers within a given job title may well experience widely differing exposures (21). Furthermore, the numbers of workers in certain categories of occupation were very small, and this must inevitably imply caution when interpreting the data.

It is possible that certain workers will have had their exposures misclassified according to their job title or category. However, this is likely to have occurred at random and is therefore likely to produce bias towards the null and to produce "false negative" rather than "false positive" findings (20). Similarly, it is possible that reporting bias may have influenced the findings, with overreporting of symptoms by subjects who feel that their work has contributed to these, or by those previously employed in dusty occupations. For example, misclassification may have occurred because some workers were recalling jobs from many years prior to the questionnaire, whereas others were reporting the current job. In addition, the present study unfortunately was not able to comment on any specific previous occupational exposures that might be etiologically important in chronic obstructive lung disease.

We defined four symptom categories suggestive of COPD. In all cases, an attempt was made to exclude asthmatic subjects by excluding those who reported ever having asthma diagnosed by a doctor, as there is obviously potential for considerable diagnostic overlap between these two conditions. However, it is possible that some subjects with previously undiagnosed "chronic" asthma, with a degree of fixed airway obstruction, would not have been excluded. This problem may well be common to all such studies attempting to categorize according to relatively limited information about symptoms and pulmonary function, although symptoms of regular phlegm production and Grade 1 shortness of breath are likely to identify most persons with symptomatic COPD.

This study noted significant associations between either chronic bronchitis or Grade 1 shortness of breath and work as a food processor, baker, chemical processor, or spray painter. There was an association also with work in the construction and mining category. In addition, work as a baker or spray painter was associated with airway obstruction alone and with the combination of symptoms and airway obstruction. Both of these occupations are well known to be associated with a risk of occupational asthma. Bakers may become sensitized to many of the individual components of flour used for the baking process (22) and subsequently develop asthma. As the numbers of workers within the bakers category was so low it is impossible to comment if this relates to a diagnosis of long-standing asthma or to an independent effect of dust exposure causing COPD. Similarly, spray painters are known to be exposed potentially to chemicals within paints that can be potent causes of occupational asthma (22, 25), and again it is not possible to dissect with certainty the cause of the effect seen in this case.

We chose to define airway obstruction as a FEV% less than or equal to 0.75, although other analyses were performed and have been summarized. This level was chosen as it was thought likely to represent early development of airway obstruction. Other studies have used lower threshold values for the diagnosis of significant airway obstruction (10, 11). Although both perhaps have their merits, this analysis was designed to identify symptomatic predominantly working young adults who may be developing early degrees of airway obstruction.

The combination of symptoms and airway obstruction in this study was again related to current or most recent occupation as a baker or spray painter, and the odds ratio was elevated, implying an association with work as a chemical processor.

In the subsequent analysis, the presence of airway obstruction alone, regardless of symptoms, was related significantly again to work as a spray painter. Although not significantly associated, odds ratios were also elevated for work as a baker, laboratory technician, and plastic and rubber worker. Previous work has noted a relationship between these occupations and the development of respiratory symptoms with or without evidence of COPD (22, 26), although the classic relationship between work as a laboratory technician relates to a diagnosis of asthma. The limited information in our questionnaire does not, however, permit further insight into relevant exposures within these groups.

These findings concord interestingly with our analysis for differing definitions of asthma by occupation (7). The presence of bronchial hyperresponsiveness, regardless of symptoms, was significantly elevated in laboratory technicians (OR = 4.82) and was associated with work as a plastic and rubber worker, although in the current analysis, it was the intent to specifically exclude those with a previous diagnosis of asthma. It may well be that this analysis has identified certain laboratory technicians with airway obstruction related to long-term, undiagnosed, work-related asthma.

A large population-based study (10) has previously noted a significant relationship between work with dusts or fumes/ gases and the presence of chronic respiratory symptoms. In addition, there appeared to be a dose-response relationship between dust exposure and phlegm production, and no significant interaction between the two broad categories of exposure. In particular, those with previous exposure to dust and fumes had significantly greater phlegm production (OR = 1.59), breathlessness (OR = 1.96), and "COPD" (OR = 1.57), the latter defined as a FEV₁/FVC ratio of less than 0.6. A case control study from Scandinavia (29) noted that the presence of emphysema was related to regular employment exposed to dusts, vapors, or aerosols, after adjusting for age, sex, and tobacco smoking. A further study based on an older population than the current study (30), also noted a significant relationship between self-reported exposure to dust, gases, or fumes and respiratory symptoms.

Again, when subjects from a large Norwegian population-based study (11) were grouped into categories of "potentially harmful" airborne exposures, obstructive lung disease was more common (OR = 3.6) in jobs with a high degree of airborne exposure in comparison with the low exposure group. Interestingly, there was no relationship between airway obstruction alone, as defined by a FEV₁/FVC less than 0.7, a percent predicted FEV₁ less than 80%, and exposure category.

Further population-based studies have confirmed this association (9), and other studies with a longitudinal design based on random population samples (28) and specific occupational groups (32) have noted accelerated loss of FEV₁ to be associated with work in dusty jobs. The above studies relating to specific exposures of grain (32), wood dust (33), and cotton dust (34) are examples of the emerging number of studies documenting work-related FEV₁ loss, independent of cigarette smoking. This concept has been referred to as the COPD effect (35).

These studies, when considered together, provide evidence that occupational exposure to dusts, gases, fumes, and vapors are likely to be important determinants of respiratory disease, particularly as this finding has been consistent in many studies with different population ages and study methods.

In summary, this study has found significant associations, independent of tobacco smoking, between job category and the risk of respiratory symptoms and physiologic abnormalities suggestive of early COPD. These include food processors, bakers, chemical processors, spray painters, and workers who reported to be ever exposed occupationally to vapors, gases, dust, or fumes. It is likely that these occupational environments are risk factors for COPD, although the relative importance of the occupation weighed against other risk factors (such as pollution and cigarette smoke) will vary from worker to worker and industry to industry.