Subjects and methods

SUBJECTS

Subjects comprised all current employees of three alumina refineries in Western Australia. A total of 2964 employees were invited to take part in the study, of whom 2639 (89%) participated. The study was approved by the ethics committees of the University of Western Australia and Monash University.

DATA COLLECTION

The study was cross sectional and was conducted between September 1995 and May 1996, avoiding the winter period. All subjects were given a modified British Medical Research Council questionnaire on respiratory symptoms and smoking by the same team of trained interviewers.5 Rhinitis was taken as frequent sneezing and an itchy running nose. Chronic bronchitis was defined as cough or sputum (or both) on most days for at least 3 months. Current smokers had smoked at least one cigarette daily for at least 1 year, and ex-smokers had stopped smoking for a period of at least 3 months.

Participants were asked additional questions about the temporal relation between any reported symptoms and their work. Subjects who answered that a symptom was better at the beginning of the working week, or on days when away from work, or when on holidays, were considered to have a work related symptom.

EXPOSURE ASSESSMENT

A detailed occupational history was taken from each employee to determine work site, operating centre, department, job title, and tasks performed. With pre-existing occupational hygiene data, jobs and job tasks were divided into process groups that defined broad groupings of similar exposures to the dusts and potential irritants of interest. These consisted principally of exposure to bauxite, alumina dusts and caustic mist (table 1). There were no measurements available for the administration and powerhouse process groups because the local occupational hygienists considered these jobs to be unexposed to the contaminants of interest to this study.

MEASUREMENT OF LUNG FUNCTION

Forced expiratory volume in 1 second (FEV₁) and forced vital capacity (FVC) were measured during work hours with a rolling seal spirometer (Graseby Anderson, USA) according to the recommendations of the American Thoracic Society.6 The calibration of the spirometers was checked at the beginning and end of each testing session with a 3 litre syringe to ensure that they remained within the limits of accuracy recommended in the ATS guidelines. The spirometers were also regularly tested for leaks (at least weekly and whenever the face plate was removed for cleaning).

ASSESSMENT OF ATOPY

Atopy was measured with prick skin tests for house dust mite, mixed grasses, mould mix, and cat dander allergens (Bayer, Sydney, Australia). Subjects with one or more wheals with a mean diameter ⩾3 mm greater than the negative control were considered to be atopic.7

STATISTICAL METHODS

Prevalence ratios were used to examine the associations between each of the four work related symptoms and explanatory covariates and were estimated with Cox's regression as suggested by Thompsonet al.8 The models included terms for process group, refinery site, age, smoking, and atopy. Overall differences between refineries and process groups were tested with these model terms. The interaction of refinery or process group with other variables was used to test if associations were consistent in the different locations, and the interaction of refinery and process group was used to examine whether the associations of process group with the various response variables were consistent in the different refineries. Smoking was included as one categorical variable for current smokers, exsmokers, and never smokers. Two work exposure variables were examined: total years employment with the company and current process group. All analyses were repeated for non-work related respiratory symptoms—namely, chronic bronchitis, wheeze in the past 12 months, wheeze with shortness of breath, seasonal rhinitis, perennial rhinitis, chest tightness in the past 12 months, and dyspnoea grade 2 or worse. This was done to investigate consistency of findings for both work related and non-work related symptoms, especially for those where a temporal relation to work (such as symptoms of chronic bronchitis) would not be expected.

Predicted values for lung function were obtained from published regression equations, which include terms for the decline in lung function with age and its increase with height.9 The associations of the same explanatory covariates with FEV₁, FVC, and the FEV₁/FVC ratio were examined with multiple linear regression after a similar analysis strategy to that used for symptoms, but with additional adjustment for height, a major determinant of lung function. Differences in lung function associated with work related symptoms were similarly estimated.

All statistical analysis was performed with SPSS for Windows.10

Results

Of 2964 employees at the refineries, 2388 male and 192 female subjects provided complete data sets representing an 87% participation rate. One hundred and thirty eight (72%) of the women worked in the administration process group. It was therefore decided to confine further analysis to men only. The median duration of employment was 10 years (range 0–33 years).

There were significant differences between refineries in mean age and the prevalence of atopy, and between process groups in the prevalence of current smoking, mean age, and years of schooling (table2).

Overall levels of FEV₁ and FVC were >100% predicted in all refineries (table 3). After internal adjustment for age, height, smoking, and atopy, refinery 1 had the lowest FEV₁ and FVC, and refinery 2 the highest. Mean FVC was 270 ml higher and mean FEV₁ 109 ml higher in refinery 2 than refinery 1 (table 3). The FEV₁/FVC ratio order was reversed. There were no significant differences between process groups for adjusted FEV₁, but there were for FVC and FEV₁/FVC ratio: employees in the clarification process had the highest adjusted FVC and the lowest FEV₁/FVC ratio (table 3). There was no significant effect of duration of employment on any measure of lung function.

Examination of interaction effects of process group and refinery indicated that there were no significant differences between refineries in the associations between process groups and lung function (p>0.2 for all analyses).

Work related wheeze and work related rhinitis were significantly more prevalent in refinery 2 than in the other two refineries, and work related chest tightness was also more commonly reported there, although this was not significant (table 4). Rhinitis was the most commonly reported work related symptom and was more prevalent in all production employees apart from the residue group.

The presence of atopy on prick skin tests was significantly associated with reported work related wheeze, with a prevalence ratio (PR) of 1.8 (95% confidence interval (95% CI) 1.2 to 2.6), and atopy was also more common in subjects reporting other work related symptoms (table 5). Current smoking was associated with a large increase in the prevalence of reported work related wheeze (PR 3.1, 95% CI 1.9 to 5.0) and with lesser increases in the other symptoms.

After adjustment for age, smoking, atopy, and process group, work related symptoms were reported significantly more commonly in refinery 2, particularly wheeze, but also chest tightness and rhinitis (table5). Precipitation employees had significantly increased reports of work related shortness of breath, wheeze, and rhinitis compared with administration employees. Work related wheeze was also more commonly reported by digestion and calcination or shipping employees. Work related rhinitis was widely reported by production employees. There was no significant or appreciable effect of duration of employment on the reporting of any work related symptoms.

The presence of these work related symptoms was significantly associated with levels of lung function: work related shortness of breath was associated with lower levels of FEV₁ by 272 ml, of FVC by 207 ml, and of FEV₁/FVC ratio by 2.6% (table 6). Similarly, work related wheeze was associated with an FEV₁which was 186 ml lower and an FEV₁/FVC ratio which was 3% lower. There were similar changes associated with chest tightness (table 6). Work related rhinitis was associated with only relatively small changes in mean lung function.

Results from analyses of all reported symptoms were broadly similar to those obtained when looking at work related symptoms for wheeze, chest tightness, and dyspnoea. Chronic bronchitis was reported more commonly at refinery 2 (PR 1.4, 95% CI 1.04 to 1.80) and was associated with decreased lung function although this association was not significant (FEV₁ lower by 32 ml, 95% CI −97 to 34). Seasonal and perennial rhinitis indicated different associations. For perennial rhinitis there were no significant differences between refineries but large differences between process groups, whereas for seasonal rhinitis there were significant differences between refineries (with the prevalence of refinery 2 significantly increased (PR 1.3, 95% CI 1.07 to 1.59)) but no differences between process groups.

As well as effects on work related symptoms, atopy and smoking also were significantly associated with lung function. Atopic subjects had a lower mean FEV₁/FVC ratio, (0.8, 95% CI 0.2 to 1.3) but were not significantly different in either mean FEV₁ or FVC. Current smoking was associated with a significantly lower mean FEV₁ (194 ml), mean FVC (42 ml), and mean FEV₁/FVC (2.3%). There were lesser deficits for ex-smokers, which were only significant for a 1% reduction in FEV₁/FVC.

Discussion

Subjects from the three alumina refineries had good lung function compared with published predicted normal values.9

There was a significant increase in rhinitis among most groups of production employees compared with administration workers which is consistent with upper airway irritation caused by either dust or caustic mist. There were no significant deficits in lung function associated with this symptom.

There were also significantly higher frequencies of work related symptoms among precipitation employees. These symptoms were not associated with any significant deficits in lung function. Also, precipitation employees did not have the highest exposures to caustic mist (table 1), and available exposure data do not provide an explanation for why they would be more likely to report symptoms as being work related than other process groups.

Employees at refinery 2 reported a higher prevalence of work related symptoms than those at the other refineries, but employees at that refinery also had the highest levels of FEV₁ and FVC. Average exposures at this refinery were relatively low compared with the other two refineries (table 1). Employees at refinery 2 were more likely to be atopic than employees at the other refineries, but this could not explain the differences in prevalence of symptoms as the increases in reported symptoms were not restricted to atopic subjects (the interaction term of refinery site and atopy was far from significant, p=0.75), and remained similar after adjustment for atopy (table 5). It may be that these reflect differences in source populations, in selection procedures, or in unmeasured exposures at the different refineries. Seasonal factors were unlikely to account for these differences, as our data relate to work related symptoms.

Workplace exposures within some of the process groups in these three refineries were associated with work related upper and lower respiratory symptoms, consistent with known exposure to irritants such as caustic mist and particulates. These symptoms were associated with deficits in lung function with an obstructive pattern in those subjects with work related shortness of breath, wheeze, and chest tightness. These findings suggest that exposures at work are responsible for both the symptoms and impaired lung function, or alternatively, that subjects with lower levels of lung function are more likely to experience symptoms.

Smoking was associated with an increased prevalence of work related symptoms, and a deficit in the level of lung function among all employees. This is consistent with the known effects of smoking, and suggests that the other findings were real effects. Atopic subjects were also more likely to experience work related symptoms than non-atopic subjects and had a lower FEV₁/FVC ratio, consistent with previous studies.11 The distribution of atopy between the various process groups and between refineries was uneven, raising the issue that selection factors before employment may account for some of the differences in symptom prevalence between groups. From these cross sectional data this possible selection bias could not be disentangled. Should there have been such a selection bias, however, it is likely to have operated in such a way as to increase the prevalence of symptoms among the office or administration process group rather than among the production workers as we found.

Although there was some evidence of airflow obstruction with reduced FEV₁/FVC in some process groups, this reduction was not related to any one of the particular exposures that had been estimated for the process groups. The estimated differences in lung function between process groups were also small and unlikely to be of clinical significance.

This study is one of the only two studies of the respiratory health of alumina refinery employees. A previous study in an American refinery found a pattern of decreasing lung function with increasing total dust exposure.2 In our study, quantitative exposure indices could not be estimated at this stage, but there were no important decreases in lung function associated with work in any particular process groups, albeit these process groups represent a relatively crude estimate of exposure. Precise exposure estimates, based on specific exposures in each job and task in the refineries, are to be developed in the next phase of this study to further investigate the findings of these analyses based on process group only. However, the analyses in this paper do not suggest that work in an alumina refinery, in conditions similar to those pertaining at the time of this study, has any major adverse effects on respiratory health.