METHODS

This was a study of exposure-response relations within a prospective cohort study on new work related cases.

Subjects

The cohort has been described previously.⁴ Briefly, it comprised 300 bakers and millers in the UK. In 1990 they had worked for a maximum of four years and a minimum of one month in a job with exposure to flour dust. At six-monthly intervals between 1990 and 1993 all subject still employed were surveyed—a total of seven visits. Subjects who became employed were also included in the cohort after one month of flour exposed work. Persons with exposure to flour in a previous job were excluded. The median time of follow up from 1986 was 40 months (range 1–91). About 80% of all eligible workers were seen at each time point.

All participants completed a self-administered questionnaire at each survey visit. “Chest symptoms” were defined as at least one positive answer to questions on chest tightness, breathing difficulties, wheezing, or whistling in the chest. “Eyes/nose symptoms” were defined as at least one positive answer to questions on eyes being itchy or runny or the nose being blocked, itchy, runny, or sneezing (without having a cold or flu). Symptoms were regarded as work related if improved during weekends or holidays or if provoked by contact with flour or grain. Symptoms were regarded as “new” if they started after employment at the site. Only new work related symptoms were used in the analyses. Workers were classified as smokers if they had smoked at least one cigarette daily for a year while employed at the site during the follow up—that is, between 1990 and 1993.

Skin prick tests (SPT) were carried out at every survey visit. Positive and negative control solutions, three common inhalant allergens (grass, cat, and Dermatophagoides pteronyssinus), fungal α-amylase (10 mg/ml, Novo Nordisk), and an extract of wheat flours were used. A positive test was defined as a mean weal diameter at least 3 mm greater than that of the negative control. Subjects were classified as atopic when having a positive response to at least one common allergen.

Workers were questioned for work tasks from the time of joining the bakery/mill and at every survey. With this information and results of previous measurements of α-amylase aeroallergen, each worker was assigned an exposure concentration, as described previously.³ If the worker reported any change of work during follow up, the highest exposure ever worked in was used in the analyses. The population was divided into three exposure categories of α-amylase based on the average exposure by exposure group; <5 ng/m³; 5–15 ng/m³, or >15 ng/m³, as described previously.³ Forty persons (13%) were not classified due to missing information (table 1).

Analyses

The associations between exposure and the probability of reporting a respiratory symptom or having a positive SPT to α-amylase during follow up were analysed with a proportional hazards model (Cox regression) using the Stata software package, version 5. The lowest exposure category was used as referent. Age, sex, atopy, and smoking were introduced as potential confounders. Sex and atopy were significant confounders and thus retained in the final model.

RESULTS

There were 36 cases of chest symptoms, 86 of eyes/nose symptoms, 24 of a positive SPT to α-amylase, and 21 of a positive SPT to flour over the study period. Demographic details of the cases have been reported previously.⁴

A significantly increased prevalence ratio (PR) was observed for chest symptoms in the highest exposure category to α-amylase (PR 3.0, 95% CI 1.1 to 8.1) (table 1). There were trends towards increased PRs for eyes/nose symptoms, and for a positive SPT to α-amylase by increasing exposure.

The predictive value of a positive SPT to α-amylase was 29% (7/24) for chest symptoms and 42% (10/24) for eyes/nose symptoms.

DISCUSSION

Our longitudinal analysis indicated that there are clear exposure-response relations between α-amylase allergen exposure and work related respiratory symptoms. To our knowledge, this has not been reported before. In addition, our results support the findings of the two cross-sectional studies on exposure and sensitisation to α-amylase.^2,3 Sensitisation to fungal α-amylase was more common than sensitisation to flour in our cohort, underscoring the allergenic capacity of this enzyme.

The cohort was dynamic—that is, some workers left employment during the study period and even though we tried to survey them before leaving, this was not always possible. If those were more often symptomatic or sensitised, the exposure-response relation might have been underestimated. Newly employed workers also entered the cohort. No systematic pre-employment screening was reported at any of the work sites, and those newly employed could be assigned to any exposure category. One of the limitations of the cohort is its size. Even though it is the largest cohort of flour and amylase workers with good exposure data, the cohort is still fairly small, resulting in imprecise risk estimates.

The categorisation of exposure was based on analysis of work tasks. The participants were not aware of the categorisation and recall bias is improbable. However, there is variability in exposure within the categories and some misclassification of exposure is probable. As such misclassification is based on task and not on outcome it is probably random, leading to an attenuated exposure-response relation.

The highest exposure ever worked in was used in order to reduce misclassification of exposure; it is biologically more plausible that workers develop symptoms and are sensitised at the highest exposure level. Subsequently they may move to a lower exposure group. Nieuwenhuijsen and colleagues³ reported the best exposure-relations using this approach.

Sex, atopy, age, and smoking were evaluated in the model as potential confounders. The first two were of significant importance and were retained in the analyses.

The distribution of α-amylase exposure was highly skewed, with 86% of workers in the low exposure category. This makes the precision of the estimates low in the other exposure categories. There were trends for increased PRs in the second exposure category, with an arithmetic mean of 10.7 ng/m³, and still higher PRs in the highest exposure category with a mean of 46.7 ng/m³. In the discussion on prevention of respiratory allergy to fungal α-amylase, levels in the low ng/m³ range have been considered as possible occupational limit values.^5,6

Conclusion

There were clear trends for exposure-response relations for exposure to fungal α-amylase, and respiratory symptoms and sensitisation. A reduction in α-amylase exposure is likely to reduce the risk for respiratory morbidity in bakery workers.