INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Worker exposure to aerosols of fluids used to lubricate and cool metal machining operations (metalworking fluids) has been shown to be associated with specific occupational asthma (1), increased respiratory symptoms (4), hypersensitivity pneumonitis (8, 9), and an acute reduction in air flow rates over one work shift (7, 10, 11). Metalworking fluids (MWF) consist of complex mixtures of oils or hydrocarbons, additives, and water. Three general classes of fluids are in common use: those which contain no water (referred to as straight oils), those which consist of a water and oil emulsion (soluble oils), and those which contain no petroleum-based oils (synthetic fluids). Combinations of these types are also found. Specific formulations differ from manufacturer to manufacturer and according to the specific purpose for which the fluid is intended. In use, fluids may change as a result of other additives being applied by the operator, by contamination from the metals being worked, from machine and hydraulic oils, and through thermal degradation. In addition, as fluids are generally recirculated and may be changed only infrequently, microbiological contamination of water-containing MWF is almost universal (12, 13).

Specific occupational asthma from MWF exposure has been attributed variously to fluid components such as colophony, metals, and ethanoloamines. Specific agents responsible for the other adverse respiratory outcomes are not well characterized, although endotoxin has been implicated in some studies (11).

In a previously published report (10), a cross-shift reduction in FEV₁ of 5% or greater was seen in 23% of machinists working in automobile parts manufacture compared with 10% of assembly workers in the same plants (p < 0.05). Although an exposure response trend was seen (with increasing prevalence of an acute response associated with increasing exposure level), the effect appeared to be of similar magnitude regardless of the type of fluid being used. Similar results have been reported recently (7, 11) from two other studies of automobile industry employees. It is unknown if this acute response to exposure is related to specific occupational asthma or is a nonspecific response to irritants present in the aerosol. In one study of French automobile workers, no increase in bronchial responsiveness was found when gearbox machining workers were compared with assembly workers (5), however recently Massin and colleagues (14) studied 114 workers exposed to soluble MWF at a French ball bearing plant and found increased airway responsiveness associated with cumulative MWF exposure.

In order to investigate further the potential relationship between MWF exposure and the early development of bronchial hyperresponsiveness (BHR) and asthma, we initiated a prospective study of newly apprenticed machinists in the Canadian province of British Columbia. This study reports the results for factors associated with bronchial responsiveness at baseline and after 2 yr of follow-up.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

All newly apprenticed machinists attending the provincial technology school during a 2-yr period were invited to be tested. A total of 116 machinists agreed (participation rate: 91%), of whom 99 met the eligibility criterion of being employed (in any job) for less than 5 yr. The comparison group included 234 apprentices, similarly recruited, from construction painting, insulation, and electrician apprenticeship classes (participation rate 83%), of which 207 were eligible. As we were interested in the new development of asthma and BHR, a further four machinists and five control subjects with physician-diagnosed asthma at the time of the first visit were also excluded from the analyses. The study protocol was approved in advance by the University of British Columbia Clinical Screening Committee for Research and Other Studies Involving Human Subjects, and informed written consent was obtained from each participant.

Baseline testing was carried out while participants were attending their Year 1 apprenticeship training class at the school. Follow-up testing (approximately 2 yr after baseline) was conducted when each participant returned to the school for his or her Year 3 training class. Participants who had quit the apprenticeship program after the baseline test were located and asked to come for testing at our laboratory. Several participants who lived at some distance from Vancouver were tested at their homes with the assistance of our mobile laboratory. All participants had been away from work for at least 1 wk prior to any testing.

Identical testing procedures and equipment were used for baseline and follow-up studies. An expanded version of the American Thoracic Society questionnaire for use in epidemiologic studies (15) was administered by a trained interviewer. The expanded questionnaire supplemented each of the respiratory symptom questions with the additional query: "Is there any thing or situation which makes your (. . . `symptom') worse (and if so, describe)?" Asthmalike symptoms were considered present if a participant responded positively to one or more of: usual cough (aggravated by dust or fumes), wheeze apart from colds (aggravated by dust or fumes, or worse at night), or chest tightness (aggravated by dust or fumes, or if associated with difficulty in breathing). The questionnaire also included a detailed history of current and past work practices, hours spent at selected machining tasks, and information about types of MWF used for each task.

Allergy skin prick tests were conducted using three common environmental antigens (Dermatophagoides pteryonyssinus, mixed Pacific grasses, cat epidermal antigen) and positive and negative controls (histamine and saline, respectively). The wheal diameter was read at 15 min and a positive test was recorded if the wheal diameter was 3 mm or more larger than the saline control. Subjects were defined as being atopic if one or more skin test was positive by this criterion.

Spirometry was performed, using a dry rolling seal spirometer (S&M Instruments Ltd., Doylestown, PA), with subjects seated and wearing noseclips, following the American Thoracic Society protocol (16). The same two trained technicians conducted all tests. A minimum of 3 acceptable forced expiratory maneuvers were obtained from each subject on each test occasion and the maximal FEV₁ and FVC were used for analysis.

Bronchial responsiveness was measured by methacholine challenge performed on a separate day, after all other spirometry tests had been performed. This test was rescheduled if the participant reported a recent upper respiratory tract infection. The methacholine challenge test protocol followed the tidal breathing method (17) with normal saline and methacholine concentrations from 0.01 to 64 mg/ml being nebulized into a face mask for 2 min using a Bennett-Twin nebulizer with an output of 0.15 ml/min. A forced expiratory maneuver was performed at 30 s and 3 min after each concentration and the lowest FEV₁ recorded, provided the maneuver was technically satisfactory. The test was terminated when FEV₁ fell to 20% of the lowest postsaline level or the maximum concentration was reached. The linear slope of the least squares regression line from the relationship between FEV₁ (in ml) and methacholine concentration (in mg/ml) was calculated (methacholine slope) and the concentration associated with a 20% drop in FEV₁ (PC₂₀) determined by linear interpolation or extrapolation.

Exposure Evaluation

As most of the apprentices worked in very small shops and the worksites were distributed widely throughout the province, it was not possible to conduct exposure monitoring at all worksites. Therefore, representative full shift personal monitoring for "total aerosol mass" was conducted for specific machine-coolant combinations at 13 different shops (n = 68 samples). For this testing, each monitored machinist wore a personal air sampling device which consisted of a 0.8 µm pore size, 37-mm-diameter cellulose ester membrane filter (Nuclepore, Corning, NY) in a plastic cassette, positioned at the worker's lapel, attached to a constant flow personal sampling pump operating at an airflow rate of 2 L/min, calibrated before and after sampling with a Gilibrator soap film flow meter. Filters were equilibrated for 24 h before and after sampling in an environmentally controlled chamber and pre- and postweighed (triplicate weighings) on a semi-micro balance (0.01 mg sensitivity). A value of one-half the detection limit was substituted for values below detection limits for calculation of mean values.

The shops monitored represented a good cross-section of machine shop types (large and small production shops, automotive machine shops, remanufacturing shops, and jobbers). The shops chosen were selected systematically from lists of potential shops that hire apprentices and from the actual workplaces of the apprentices enrolled in the study.

Duration of exposure to each of several MWF fluid type/task combinations (prior to baseline testing, and between tests) was estimated by each machinist in response to questioning about the average amount of time per day (in hours) and days per month spent at each of these combinations, for each job held.

Statistical Analysis

Bronchial responsiveness was evaluated using several measures: methacholine dose-response slope, log-transformed [ln (20-slope)], at baseline and at follow-up; difference between the log-transformed slope values at follow-up and baseline; the proportion of subjects with PC₂₀  8 mg/ml (at baseline and follow-up); and the clinical criterion: proportion of subjects with a "doubling dose increase in PC₂₀"; i.e., an increase in bronchial responsiveness that resulted in a shift in PC₂₀ (toward the hyperresponsive end) of more than one dose category (i.e., PC₂₀ < 2, 2-8, 8-16, 16-32, 32-64, > 64).

Statistical testing was carried out using SAS-PC version 6.12 (SAS Institute, Cary, NC). Simple comparisons of means and rates between groups were tested using Student's t test, chi-square analysis, and Fisher exact test. To evaluate host and environment characteristics associated with bronchial responsiveness at baseline, we carried out multiple linear regression analyses with log-transformed methacholine slope as the dependent variable and demographic characteristics (smoking status and amount, sex, race, age), factors associated with asthma (family history, parental smoking, childhood symptoms), atopy (positive allergy skin test at the time of the visit, history of hay fever), pulmonary function (FEV₁/FVC%), and exposure variables (duration of work in machining jobs and duration of exposure to MWF prior to baseline, usual respirator use) as potential predictor variables. For analysis of predictors of methacholine slope (log-transformed) at follow-up, the same variables were offered as for the baseline tests (with the follow-up value substituted for the baseline value where appropriate), plus the following additional variables: time between tests, methacholine slope at baseline, indicator variables for having quit or started smoking in the past 2 yr, indicator variables to identify persons who developed a new positive (or negative) allergy skin test during the follow-up period, an indicator variable to identify persons who quit their trade between tests, an indicator variable for exposure to any MWF between tests, and separate exposure variables for duration of exposure to straight oils, soluble oils, synthetic fluids, and tool and cutter grinding. Possible interaction between positive skin test response and exposure to MWF was also examined (both with an interaction term in the model and by stratification). Linear regression models were also constructed for change in methacholine slope as the dependent variable, using similar predictor variables; and similar logistic regression models were constructed to investigate factors associated with BHR (PC₂₀  8 mg/ml), at baseline and at follow-up.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Of the 95 eligible machinists and 202 eligible control subjects, complete baseline data were obtained for all machinists and 189 control subjects (methacholine challenge testing was not completed for 13 control subjects because of unavoidable scheduling difficulties). After the 2-yr follow-up testing, a complete baseline and follow-up data set was obtained for 82 machinists and 157 control subjects. Table 1 displays the baseline demographic and health characteristics of the group with a complete data set (i.e., those included in this analysis) compared with those excluded from this analysis for various reasons.

Participants with incomplete data (either at baseline or follow-up) did not differ from subjects with complete data with respect to demographics or baseline test results. Apprentices who were lost to follow-up between the two test occasions did not differ significantly from those included in the analysis, except for the presence of slightly more nonwhites in the "lost" group (p < 0.05). Subjects lost to follow-up were more likely to be current smokers and had somewhat greater (but not statistically significant) bronchial responsiveness, on average, at baseline.

Characteristics of the participants at baseline and at follow-up are shown in Table 2. Machinists were more likely to be nonsmokers than the comparison population (p < 0.05), and to have slightly lower forced vital capacity (FVC) (p = 0.06), although FEV₁/FVC ratio did not differ between groups. There were no differences in prevalence of respiratory symptoms (cough, phlegm, wheezing, chest tightness, dyspnea) or in rates of parental asthma or parental smoking (not shown in the table).

As shown in Table 2, the duration of follow-up was similar for the two study groups; however, a significantly higher proportion of machinists had left their trade between tests. At the time of follow-up testing, the lower average FVC value seen at baseline was still present, although, as at baseline, there was no difference in the FEV₁/FVC ratio, nor in the change in FEV₁ between the two tests. The prevalence of a positive allergy skin test did not differ between groups, although in both groups, a greater proportion developed a new positive test than the proportion who were positive at the first visit and negative at the second.

Work and Exposure Characteristics of Machinists and Machine Shops

Most machinist apprentices had worked for at least a few months in a machine shop prior to testing, as entry into the apprentice program requires being employed by a machining trades employer. Months of prior employment ranged from 0 to 36, with a mean of 10.2 (SD: 8.9). During the follow-up period, 81 of the 82 machinist apprentices were employed for at least 1 d in a machining trade job with exposure to MWF, with almost every apprentice employed by a different employer and over half employed by more than one employer during the interval. The median days of employment in a job with exposure to MWF was 189 d (range 0 to 658). A total of 70% of the machinists reported exposure to straight oils (median duration of exposure: 92.5 h; range 0 to 1,979), 71% were exposed to soluble oils (median 380 h; range 0 to 3,952), 66% were exposed to synthetic fluids (median 202 h; range 0 to 3,960), and 22% reported working with computer numeric controlled (CNC) machines, primarily using synthetic fluids. None of the apprentices in the comparison group reported exposure to MWF or machining work before the baseline testing or during the follow-up period.

During the workplace exposure monitoring from representative machine shops, straight oil was never used exclusively in any of the shops tested, although it was used in combination with synthetic and soluble fluids. Exposure concentrations ranged from < 0.7 mg/m³ to 3.65 mg/m³, with an arithmetic mean value of 0.46 mg/m³ and a geometric mean of 0.31 mg/m³ (GSD = 2.39). These data were not used in the analyses presented in this study, but are included here to provide information about the general level of exposure these machinists would likely have encountered during the follow-up period.

Change in Bronchial Responsiveness from Baseline to Follow-up

Bronchial responsiveness, whether measured as methacholine slope or as proportion with a PC₂₀ value  8 mg/ml, did not differ between the groups at baseline (Table 3). In contrast, at follow-up, the mean level of bronchial responsiveness had increased (almost 2-fold) among the machinists but had not changed among the control population, and twice as many machinists as control subjects had a PC₂₀ value  8 mg/ml. A total of 13% of machinists and 7% of control subjects had a "doubling dose" change in PC₂₀. Combining a doubling dose change in PC₂₀ with the presence of asthmalike symptoms (as defined in the METHODS section above), we found six machinists (7%) and three control subjects (2%) with this combination (p < 0.05). None of the apprentices had been diagnosed by their physicians as having occupational asthma. Of the 10 apprentices (six machinists and four others) who had a PC₂₀ > 8 at baseline and a PC₂₀  8 at follow-up, only three (all machinists) had a new diagnosis of asthma from their personal physician.

Characteristics Associated with Increased Bronchial Responsiveness

Results from multiple linear regression analysis, including both host and environment characteristics potentially associated with bronchial responsiveness at baseline and at follow-up, are shown in Table 4. The table shows results from the best fitting models. Variables described in METHODS, but not listed in Table 4, were tested but were not significantly associated with the outcomes, nor did their inclusion (or exclusion) substantially change the coefficients for variables left in the model.

To summarize the results of the models displayed in Table 4, at baseline, increased methacholine responsiveness was associated with a positive allergy skin test, with younger age, and with a lower FEV₁/FVC ratio. No other variables were significant in the model. After 2 yr of follow-up, positive allergy skin test at the time of the test (but not change in skin test response) and lower FEV₁/FVC ratio remained associated with increased responsiveness, but age was no longer a predictor. Increased methacholine responsiveness at baseline, having quit smoking between tests, and working in a job with exposure to MWF were all associated with increased bronchial responsiveness. There was no evidence of interaction between atopy (positive skin test) and exposure to MWF, indicating that the increase in bronchial responsiveness associated with MWF exposure was independent of atopic status. Although a significantly greater proportion of machinists had quit the trade between tests, this was not associated with increased bronchial responsiveness, either at baseline or at follow-up.

Table 5 shows results from a similar model for methacholine slope at the time of follow-up, in which duration of exposure to various tasks and types of MWF was considered in the model. The results indicate that increased bronchial responsiveness at follow-up was associated with increased duration of exposure to both soluble and synthetic MWF, but not with straight oils or with tool or cutter grinding.

Various other models were investigated to examine the stability of the findings reported here. These included similar models, but without baseline responsiveness included; models with the difference between methacholine slope at baseline and follow-up as the outcome variable; and models with duration of exposure which included an additional indicator variable for machinists (as well as the exposure duration terms). In every model, machinists had significantly greater methacholine responsiveness (or a greater change in responsiveness) at follow-up compared with other apprentices. The coefficient for duration of exposure to synthetic fluid remained about the same (or even larger) and statistically significant in all models. The coefficient for soluble fluid was slightly less stable, but, remained at about the same value.

Results from logistic regression models for predictors of BHR at follow-up (defined as having a PC₂₀  8 mg/ml) are shown in Table 6. These results are similar to those seen in the linear regression models; however, as might be expected, childhood asthma appeared to be a more important (although not statistically significant) predictor of hyperresponsiveness than seen in the linear models. In this model, machinists demonstrated almost 5 times the odds of being hyperresponsive at follow-up and this effect appeared to be limited to exposure to synthetic MWF (when duration of exposure was included in the models in place of the indicator variable for machinists).

All the analyses were repeated after excluding all subjects with baseline PC₂₀  8 (n = 12), and results were essentially unchanged.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In summary, we followed a small group of newly apprenticed machinists (with no, or very little, prior exposure to MWF) for a period of 2 yr and found a significant increase in BHR during the follow-up period, compared with nonmachinist apprentices. Further, the incidence rate for a clinically relevant increase in bronchial responsiveness accompanied by asthmalike symptoms was over twice as high among the machinists compared with other apprentices. The increase in bronchial responsiveness was associated with duration of exposure to water-based MWF.

To our knowledge, this is the first longitudinal study of workers exposed to MWF, and also the first to investigate the respiratory effect of this exposure early in the career of these workers. Our results suggest that a clinically relevant adverse effect of this exposure on the airways can be detected even in the first few years of employment in the machining trade. The only other prospective study of machinists of which we are aware, a study of hand dermatitis among metalworker apprentices, found a 10-mo incidence rate of 10% for mild skin irritation (18).

We excluded from analysis the small number of workers with a current diagnosis of asthma at baseline as we were interested in investigating the possibility of an association between MWF exposure and the development of new asthma or asthmalike conditions. The fact that the results were unchanged even when we excluded all workers with a PC₂₀ of 8 or less at baseline, and when we included (or excluded) baseline bronchial responsiveness in the regression models, is further indication that the exposure may be associated with the induction of BHR.

There is no gold standard for the diagnosis of asthma in epidemiologic studies so we investigated the incidence of BHR using several different approaches (i.e., PC₂₀  8 mg/ml with or without asthmalike symptoms, and doubling dose change in PC₂₀) and consistently found machinists to have an increased incidence rate. It is unlikely that the increased bronchial responsiveness seen was a transient response to exposure as these workers were tested at the technical school during a 1-mo training course.

Other possible explanations for finding a differential increase in BHR among the machinists include differences in baseline characteristics or differences during the follow-up period in smoking behavior, frequency of upper respiratory viral infections, or exposure to other environmental irritants or allergens. We found no differences between the groups for the baseline characteristics we measured, nor in the prevalence of recent (reported) upper respiratory tract infection. It is difficult to hypothesize other baseline characteristics that would be differentially distributed according to choice of apprenticeship. We also found no significant difference in change in smoking behavior between groups during the follow-up period, and we did not detect any effect of smoking on BHR, except to note a relationship between increasing bronchial responsiveness and the decision to quit smoking. We have no data on exposures outside the workplace, but there is no reason to suspect any differential distribution according to apprenticeship choice.

One of the limitations of this study is the small sample size. This is a particular problem given the relatively low prevalence (and relative instability of the measurement) of BHR in general. However, the associations seen were stable with various approaches to analysis, lending further support to our conclusions.

Although we have no clinical confirmation that these workers have developed occupational asthma, our results are consistent with case reports and case series of occupational asthma among workers exposed to MWF (1, 2, 19) and with Eisen and colleagues' report of an increased prevalence of asthma among U.S. automobile workers exposed to MWF (20). In Eisen's study, when the investigators compared asthma prevalence in exposed and unexposed groups, they found no increase in the exposed group. However, when they looked at the job held at the time of onset of the asthma, there was a strong association with MWF exposure, particularly synthetic fluids. This suggested an asthma-related shift away from exposed jobs. We did not see any evidence in our study that leaving the trade was associated with increased bronchial responsiveness (although machinists were significantly more likely to leave the trade than other apprentices). This might be due to the short follow-up time in this study.

Our findings are similar to those recently reported by Massin and colleagues (14), who found increased bronchial responsiveness to methacholine among exposed compared with control workers in a cross-sectional study of French bearing manufacturing workers. They also found a significant association between bronchial responsiveness and cumulative exposure to soluble oil mist. In that study, all workers were employed at one plant and were exposed only to soluble oils. The measurement of exposure was based on area samples, using the dichloromethane soluble fraction of the aerosol collected on the filters (measured gravimetrically). Geometric mean values were 2.2 mg/m³ prior to 1989 and 0.65 mg/m³ after 1989, in machining areas.

We were unable to estimate exposure concentration for each worker in our study, owing to the large number of worksites represented. However, our representative sampling from 13 workplaces indicated personal exposure concentrations were likely to be considerably lower, on average, than those seen by Massin and colleagues, despite the fact that our study measured total aerosol and theirs only measured the dichloromethane soluble fraction. It should be pointed out that our measurements were restricted to workshops in which the employer was willing to admit our research team, therefore, it is possible that higher average exposures might have been seen had we been able to conduct truly random sampling among machine shops.

Our finding of a significant association with duration of exposure is similar to Massin and colleagues' finding of a significant association with cumulative exposure, given the strong correlation usually seen between these two measures of exposure over time. An important difference is that in the French study, average exposure duration was 15 yr, whereas in ours, the maximal exposure duration was 2 yr.

It is not possible to identify specific components of the MWF which are responsible for the increased bronchial responsiveness seen in this study, except that the effect appeared to be associated only with exposure to water-based fluids (soluble and synthetic fluids). In the analysis of BHR, synthetic fluid exposure was the most strongly implicated. This is consistent with other studies of occupational asthma in this trade (19, 20), although studies which have measured increased prevalence of irritant symptoms and cross-shift change in FEV₁ have implicated straight oils as well as water-based fluids. This suggests that straight oils, while irritating, may be less likely to be associated with BHR and asthma than the water-based fluids. Possible explanations for this include the presence of allergenic substances in biocides or other additives or the presence of microbiological allergens themselves. This would imply that, from the perspective of monitoring for risks for occupational asthma, exposure monitoring efforts need to focus not only on the oil component of fluids, but on the total aerosol and on specific potential aeroallergens. Continued research with larger study populations will be needed to help clarify the irritant and the allergenic responses to these complex fluid mixtures.