INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Chlorine gas is one of the most common irritant encountered in the workplace that can cause irritant-induced asthma and reactive airways dysfunction syndrome (RADS), the latter being a condition that has been defined as occurring after one single inhalational accident (1, 2). Both the general and the working environments can be responsible for these accidents. Subjects affected by irritant-induced asthma report asthmalike symptoms and are left with airway obstruction and/or hyperresponsiveness. Lemière and coworkers (3) have summarized studies performed in working groups with chronic exposure to low levels of chlorine with or without repeated exposures to high concentrations and at risk of developing airway symptoms consistent with RADS, airflow obstruction, and hyperresponsiveness. These epidemiologic studies were performed primarily in groups of workers from pulp and paper mills where accidental chlorine spills occurred frequently (4).

We undertook a study in a modern metal production plant of workers at risk of accidental chlorine exposure: an initial cross-sectional assessment of 239 workers performed in 1992 showed a slight but significant reduction in expiratory flow rates, as well as an increased airway reponsiveness in workers who reported accidental inhalation of chlorine (8). The cross-sectional design of our previous study precluded us from attributing those lung function results solely to chlorine exposure: airway hyperresponsiveness and/or airway obstruction might have preexisted as an underlying condition. To address these questions, the cohort of 239 workers was followed for 2 yr to assess changes in respiratory symptoms, spirometry, and airway responsiveness and to obtain an update on occupational exposure to chlorine puffs, on accident reports to the first-aid unit as well as data on other risk factors associated with decline in lung function. We herein report the results of this longitudinal assessment.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects and Study Design

All 239 employees who took part in the initial study in 1992 (8) were eligible for the follow-up assessment in 1994 and were invited to participate by the nurse of the company. Both surveys took place in the autumn, and at the follow-up visit, workers were assessed in the same month of the year as in the initial survey. The working areas of the ore smelting plant where this study took place have been described earlier (8). The study protocol had been approved by the Hôpital du Sacré- Coeur Ethics Committee. When workers gave their written consent in 1992, they were informed that the series of tests in which they agreed to participate would be repeated in 1994. Data were collected at the workplace by the same nurse and same two medical technologists who had previously done so.

Questionnaires

A French version of the symptom questionnaire developed by the International Union against Tuberculosis and Lung Diseases (9) and administered in 1992 was adapted to focus on the presence of chronic respiratory symptoms and smoking history in the preceding 2 yr. An occupational history questionnaire was used to collect information on accidental exposure to chlorine occurring during the 2-yr period between the surveys (exposures to chlorine alone or with one or more other gases are referred to as "puffs"). An accidental chlorine exposure was considered to have taken place when a worker gave a positive answer to the question: "Have you ever had an episode of short high level chlorine exposure ("a puff")?" Questions were designed to document the working area where the event took place, the year of the event, the number of events per year by predetermined categories in order to make recall easier, i.e., 0, 1 to 5, and > 5, and the presence and intensity of symptoms after inhaling a puff (expressed as no symptom, or as mild or significant symptoms).

First-aid Reports

For each worker seeking care in the first-aid unit after an accidental inhalation, the factory medical unit continued to describe the work area where the event occurred, document the date, the nature and duration of symptoms that followed the event, the treatment provided, and the need for further consultation. These data were used as an objective assessment of accidental chlorine inhalation causing significant symptoms, as compared with the information reported in the questionnaires.

Pulmonary Function Testing

Spirometric measurements were performed according to the criteria of the American Thoracic Society using a Collins-type spirometer (Warren E. Collins, Braintree, MA) (10), the apparatus used in the 1992 survey. The same measurements were performed and derived: FEV₁, FVC, and the FEV₁/FVC ratio. When results were expressed in percentages of the predicted value, the reference values were obtained from Knudson and coworkers (11). The 2-yr decrements in spirometric measures were expressed as: the differences between the follow-up (1994) and baseline (1992) values for FEV₁ (ml), FVC (ml), and FEV₁/FVC (%).

Airway Responsiveness

Methacholine bronchial challenge tests were performed according to a standardized methodology using a Wright's nebulizer (Aerosol Medical Ltd, Colchester, Essex, UK) (output: 0.14 ml × min¹), and methacholine inhaled at tidal volume breathing for 2 min (12). An abbreviated protocol was applied in subjects with lung function results within normal range and with no history suggestive of asthma (13). The maximum methacholine concentration used was 32 mg × ml¹ if a 20% fall in FEV₁ had not been reached.

The three parameters of airway responsiveness considered for analysis in our previous study were kept: a PC₂₀ < 16 mg × ml¹ defining significant airway hyperresponsiveness (14), a PC₂₀ < 32 mg × ml¹ for a measurable PC₂₀, and the slope of the methacholine dose-response curve. A transformation was applied to the dose-response value calculated as previously described (8, 13): the slope of the dose-response curve, or dose-response slope (DRS), was expressed by the percentage decline in FEV₁/dose, where decline was expressed as a positive value and where dose in µmol was defined as the final cumulative methacholine dose administered (conversion from mg of bromide methacholine to µmol, using a corrective factor of 5.116); because the slope distribution was not normal, a logarithmic transformation of this ratio was then used; prior to logarithmic transformation, a constant of 0.1 was added to DRS values, because many subjects had a FEV₁ that remained stable or improved slightly during bronchial challenge and thus yielded a zero or negative DRS value. Higher positive values corresponded to more pronounced airway hyperresponsiveness. A subject was considered to have an increase in airway responsiveness at follow-up if the PC₂₀ became < 16 mg × ml¹ or if there was a 1.5-fold decrease in PC₂₀, when PC₂₀ had been < 16 mg × ml¹ at the first survey; and, similarly, if the PC₂₀ became < 32 mg × ml¹ or if there was a 1.5-fold decrease in PC₂₀, when PC₂₀ had initially been < 32 mg × ml¹.

Data Analysis

When comparing groups, chi-square test, Fisher's exact test, Student's t test, or one-way analysis of variance were used. McNemar's test was used to compare winthin-subject changes in exposure and airway responsiveness. The statistical analyses were carried out using the SPSS/ PC for Windows statistical software package (Chigaco, IL).

Multiple linear regression analysis was used to assess the effect of chlorine exposure and smoking on the 2-yr decrement in spirometric measures: FEV₁ and FEV₁/FVC considered as the outcome variables. The independent variables included in the regression models were baseline functional value, calculated as the mean of baseline (in 1992) and follow-up (1994) functional values [(F2 + F4)/2] (15, 16), accident report between 1992 and 1994 (0, 1); number of puffs with mild symptoms between 1992 and 1994 (0, 1-5, > 5) and smoking. The interactions between smoking and gassing incidents and between smoking and number of puffs were assessed.

Logistic regressions were performed and odds ratio (OR) and 95% confidence intervals (CI) were estimated for the effects of baseline dose-response slope, baseline FEV₁, accidental inhalations, estimated number of puffs with mild symptoms, and smoking, on the increase in airway responsiveness (defined above).

A p level of 0.05 or less for two-sided tests was considered to be statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Two hundred eleven workers participated in the second assessment (88.7%). Their mean age ± SD was 35.3 ± 6.1 yr, and two were women. Of the remaining workers, four refused to participate and 17 had left the company, most of them having been laid off because of a slowing down in the plant production; one worker was on leave of absence for sickness, and four were on temporary leave of absence for no particular health-related reason; one subject who underwent the spirometry tests but had the questionnaires administered several months later, was excluded from the analysis. Participants and nonparticipants were compared for a number of parameters obtained at the first visit: age, FEV₁, FEV₁/FVC, PC₂₀, dose-response slope, number of accidents reported to the first aid unit, number of puffs with mild symptoms, and smoking status; only slight differences were found for most variables. However, a greater proportion of participants (16.1%) had reported an accident to the medical unit, as compared with nonparticipants (3.6%). Smoking habits were different: 22.7% of participants were current smokers and 46.9% were nonsmokers, whereas 46.4% of nonparticipants were current smokers and 32.1% were nonsmokers.

Gassing Incidents

The number of accidental inhalations reported to the first-aid unit was lower in the 1992-1994 period than in the 1989-1992 period. Eight accidents were reported in the later period as compared with 34 in the first one. Seven of 179 subjects (3.9%) had an inhalation accident in the 1989-1992 period only, whereas 31 of 34 subjects (91.2%) who had suffered an inhalation accident in the first period did not suffer such an event in the second (McNemar's test, p < 0.001). An accidental inhalation was reported in the questionnaire by eight workers, one of whom did not report the event to the first-aid unit. One hundred twenty-four workers reported one or more puffs with mild immediate symptoms, and 74 reported one or more puffs with no symptoms. When considering the intensity of symptoms reported with the occurrence of puffs in 1992-1994 in comparison with the 1989-1992 period, 111 workers reported symptoms of same intensity, 22 reported more severe symptoms and 78 reported milder symptoms.

Chronic Respiratory Symptoms

The changes in the frequency of chronic respiratory symptoms reported in the questionnaires are shown in Table 1. Nearly 25% of workers reported at least one symptom in 1992 or in 1994. At least one symptom was present on both assessments in 8.1% of the workers; among those symptoms, wheezing was present on both assessments in 10 workers (4.7%); persistent wheezing was more frequently reported during the follow-up period among workers with mild or significant symptoms following puffs (p < 0.05, Fisher's exact test) (data not shown).

Changes in Spirometry Values

The overall changes over 2 yr in FEV₁% pred, FVC% pred, and FEV₁/FVC% pred, were not different according to a history of accidental inhalation reported to the first aid unit or of chlorine puffs with symptoms (either mild or significant) during the two-year follow-up (Table 2). The mean ± SD annual decline for FEV₁, FVC, and FEV₁/FVC for the whole group of workers was 54.4 ± 82.0 ml, 44.7 ± 86.8 ml, and 0.36 ± 1.1%, respectively. The mean decrement in these function measures was also unrelated to the number of reported puffs with mild symptoms. Mean FEV₁% pred, at follow-up was significantly lower in the eight workers who reported an accidental inhalation to the first-aid unit between 1992 and 1994 than in the others (90.4 ± 9.1% versus 97.8 ± 10.5%). In these workers, the mean baseline values for FEV₁% pred, and FEV₁/FVC% pred, were significantly lower than in the others (91.2 ± 7.1% versus 99.1 ± 10.6%, and 94.2 ± 3.6% versus 98.0 ± 5.5%). When all workers who reported a puff with either mild or significant symptoms (n = 132) were compared with those reporting no symptoms or no puff (n = 79), the mean FEV₁ at follow-up was significantly lower in the first group (96.5 ± 10.6% versus 99.3 ± 10.2%, p = 0.05); the baseline values for these two groups were not significantly different (100.1 ± 10.7% versus 98.0 ± 10.5%, p = 0.16). Finally, changes in functional values were not different according to the smoking status.

Linear Regression Models for Changes in Spirometry Values

In regression analyses, taking into account the average functional value, number of puffs with mild symptoms (1992-1994), but not accident report, were related to changes in FEV₁ in smokers of  20 packs-years only (r = 0.10, B = 63, p = 0.05) and changes in FEV₁/FVC (r = 0.26, B = 1.0, p < 0.01). Results from the multivariate linear regression analysis for change in FEV₁ (ml) and FEV₁/FVC (%) are presented in Table 3. There was a significant interaction between accident and smoking in pack-years with regard to changes in FEV₁. There was a diminution of 1.08% in the FEV₁/FVC ratio per category of number of puffs. Using the baseline functional values and not the average of the 1992 and 1994 values, we found similar results (8% change in FEV₁, p = 0.02 for the interaction accident and pack-years; 1.2% change in FEV₁/FVC ratio, p = 0.03 for the interaction number of puffs and pack-years).

Airway Responsiveness at Follow-up and Two-year Change

Eight subjects developed airway hyperresponsiveness at the follow-up, whereas three subjects no longer had airway hyperresponsiveness (McNemar's test, p = 0.2). The proportion of workers with a PC₂₀ of  16 mg × ml¹ or  32 mg × ml¹ and the mean values for the dose-response slope at baseline and at follow-up are presented by categories of reported exposure for the period 1992-1994 (Table 4). There was a significantly higher proportion of subjects with a measurable PC₂₀ among those with an accident report to the first aid unit than among the other workers. In addition, the proportion of those with an increase in airway responsiveness between the two assessments (d1) was greater (p < 0.05).

Subjects with airway hyperresponsiveness at the 1992 visit by comparison with those with normal airway responsiveness did not have significantly more accidental reports (5.9 versus 3.6%) and symptomatic puffs (64.7 versus 62.9%) at the follow-up visit.

Linear and Logistic Regression Analysis for Changes in Airway Responsiveness

To investigate the effects of gassing incidents and repeated exposures to chlorine puffs causing mild symptoms on the changes in airway responsiveness, a dichotomous outcome variable was first defined and referred to as change in measurable PC₂₀ (see METHODS). Univariate logistic regression analysis was performed to investigate the effects of functional values at first survey, accidental chlorine exposure, and smoking on change in measurable PC₂₀ (Table 5). The baseline dose-response slope (log), accident reports, and FEV₁ (% pred) (not shown in Table 5, unadjusted OR for a 10% decrease in FEV₁ = 2.0, 95% CI = 1.22 to 3.23) were strongly associated with the change in measurable PC₂₀. There was a tendency for association between increase in airway responsiveness and number of puffs with mild symptoms. The effects of accidental inhalations and of number of puffs remained the same after adjusting for the other variables in the model. The relation between slope of the DR curve at follow-up was also assessed in a linear regression model, but none of the independent factors examined above were significant predictors.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Two conclusions can be drawn from this 2-yr longitudinal study of workers at risk of accidental chlorine exposure. First, decrease in airway function parameters was related to accidental chlorine inhalation, mostly among heavy smokers (above 20 pack-years). Second, the change in airway responsiveness was related to accidental chlorine inhalation, taking into account baseline airway responsiveness and smoking.

Few longitudinal studies of airway function have been conducted to date in workers at risk of accidental chlorine exposure. Salisbury and coworkers (5) studied pulp mill workers twice at an 8-yr interval; chlorine exposure was assessed through reports to the first-aid unit of gassing incidents during this period. There was a greater decline in FEV₁/FVC ratio in the 37 workers who experienced at least one accident reported to the first-aid unit, as compared with their matched control subjects, irrespective of the smoking status. In the present study, we showed an effect of exposure on decrease in lung function, mostly in heavy smokers. This is consistent with the recent findings of Henneberger and coworkers (7) in a cross-sectional study of 300 workers at risk of chlorine exposure: incremental decline in FEV₁ was associated with gassing in current smokers only. Such an interaction of irritant exposure and smoking was also observed in their previous investigation of older men who had worked at the same pulp mill (6). These investigators speculated that smoking might provide yet another exposure that works in conjunction with accidental chlorine exposure to impair ventilation. Whether an improvement in lung function can occur when workers are no longer exposed to chlorine is a matter of interest. Data from our previous (8) and current studies suggest that both recent (1992- 1994) and past (1989-1992) accidental chlorine exposure had an effect on FEV₁ and FEV₁/FVC ratio. Therefore, it is reasonable to think that accidental chlorine inhalations can lead to long-lasting impairment of airway caliber.

We were unable, however, to demonstrate any effect of accidental chlorine exposure on longitudinal changes in FEV₁ in the whole group of workers. The difficulty of detecting an effect could be accounted for by within-subject variability. According to Becklake and coworkers (17) the coefficient of variation for within-subject reproducibility in FEV₁ measurements can be estimated at 4.5% (SD ~ 120 ml), thus the lower 90% CI would be at 1.64 × 0.12: 197 ml. In our population of mostly male workers, the mean difference in FEV₁ per year was 54.41 ml (SD, 82.0), and therefore the lower 90% CI: 54.4-(1.64 × 82.0): 189 ml coincides with the estimated lower 90% CI for within-subject variability. A longer period of follow-up may be needed to detect a change in FEV₁ that can be attributed to accidental chlorine exposure and also to chronic low-level exposure. However, our findings are comparable to those published by Humerfeld and coworkers (16); in their 20-yr follow-up study of a random sample of 1933 men 22 to 54 yr of age, the annual decline in FEV₁ ± SD was 51.9 ml ± 0.8 in subjects without any known exposure and 55.3 ml ± 2.5 in subjects with exposure to chlorine.

Another explanation for this absence of effect on longitudinal changes in FEV₁ in the whole group of workers may lie in the security measures set up in this modern plant, which started operations in 1989. Between 1992 and 1994, there was a decrease in the number of gassing incidents reported to the first-aid unit, as compared with the first 3 yr of plant operation (1989-1992), whereas the estimated number of puffs with mild symptoms or without symptoms reported by workers increased. This fall in the number of gassing incidents may be explained by a general improvement in exposure levels, with recent reduction in production, and increased security measures that were adopted by the company (information provided by their industrial hygienist). This diminution of the magnitude of effect of accidental events is confirmed by a recent report from our group in which only three of 14 subjects from the same population who presented to the first-aid unit had measurable and relatively short-lasting changes in airway responsiveness (18).

In the present study, as in others conducted in pulp and paper mills (4, 19), there is a lack of information on chlorine concentrations during an acute exposure, as current portable apparatus do not allow for quantitative assessment of peak chlorine levels. Therefore, exposure has been estimated through either self-reported accidents or reports to the first-aid unit, and these two sources of information were found to be consistent in the present study.

Whether or not chronic airway symptoms are related to chlorine exposure is a matter of controversy in previous studies (4, 8, 19). Enarson and coworkers (19) found a greater prevalence of wheezing in 392 pulp-mill workers as compared with rail-yard control workers. This was confirmed in further reports from the same group, in which an increased risk for workplace-associated symptoms of wheezing, chest tighness, and cough was related to acute chlorine inhalations reported to first-aid unit (4, 5). In our previous cross-sectional assessment, no relationship was found between persistent symptoms and the exposure variables studied (8). In the present study, workers with persistent symptoms of wheezing were more likely to report an accidental inhalation followed by mild or significant symptoms during the follow-up period; however, the effect of possible confounders, mainly smoking status, cannot be assessed in this small subgroup of 10 patients.

Whereas spirometry was examined prospectively in other studies, this is the first study, to the best of our knowledge, that repeatedly assessed airway responsiveness in workers exposed to chlorine in an epidemiologic setting. We found a significant relationship between the change in measurable PC₂₀ and recent (1992-1994) accidental chlorine exposure, taking into account baseline airway responsiveness and smoking. The magnitude of these differences might have been masked by some effect of recovery from acute inhalational events. Indeed, Malo and coworkers (23) found that airway hyperresponsiveness improved, whereas spirometry plateaued 2 to 3 yr after an inhalational accident. Therefore, the influence of relatively recent events can still be measurable, but they will not necessarily persist. The pattern of improvement of RADS, which is considered a form of occupational asthma (24), is quite similar to what happens in occupational asthma with a latency period (25).

It is our impression that work forces exposed to irritants should be assessed prospectively by both airway caliber and reponsiveness. Any case of significant acute change in airway caliber and/or responsiveness after a gassing event can be offered treatment with corticosteroids, which may have a beneficial effect on this condition according to preliminary data (26, 27). Our results also suggest that subjects with lower baseline airway caliber but not increased airway responsiveness, can be susceptible of reporting inhalational accidents, although these data warrant confirmation in a prospective design of new employees.