Respiratory bronchial reactivity among pig and dairy farmers.

ratory symptoms and bronchial reactivity among pig and dairy farmers. Scand J Work Environ Health 1994;20:48-54. OBJECTIVES - This study assessed the prevalence of respiratory manifestations among French pig and dairy farmers and determined the relationship between bronchial reactivity and respiratory manifestations. METHODS- The pig farmers included 102 men working more than half time inside swine confinement buildings. There were 51 male dairy farmers and 81 male referents. The demographic characteristics of the three groups were similar except for more smokers among the referents. Each subject completed a standardized questionnaire. Pulmonary function tests were per formed before and after a methacholine challenge (cumulative doses 80, 240, and 560 ug), Airborne dust, ammonia, and carbon dioxide were measured inside 28 swine confinement buildings. RESULTS - The pig farmers were exposed to a total dust level of 2.4I mg . nr-'. The respirable particle con centration was low. The pig and dairy farmers had a significantly higher prevalence of cough and morning phlegm than the referents. Before the methacholine challenge, the dairy farmers had nonsig nificantly lower mean lung function values than the other groups. Among the subjects with no history of asthma, nonspecific bronchial hyperreactivity was significantly higher among the pig and dairy farm ers than among the referents. There was a fall in the forced expiratory volume in I s (FEY, ,) that was greater than 10% in 6.7% of the referents, 17.9% of the swine workers, and 35.6% of the ~airy farm ers. This result was unchanged after adjustment for the initial FEV] o: CONCLUSIONS - The preva lence of respiratory symptoms was significantly higher among the pig'farmers without base-line lung function impairment. However, both the pig and the dairy farmers had increased bronchial reactivity. demonstrated an in creased risk of respiratory dysfunction among and, especially, among workers in swine tified the following characteristics of pig farmers in different countries: (i) a higher prevalence of respi ratory symptoms, including chronic bronchitis and asthma than in a reference group, (ii) nonrespirato ry symptoms consistent with the organic dust toxic syndrome; (iii) similar or only slightly lower lung function values as compared with reference groups, (iv) no change or only small dec reases in lung fun

tified the foll o win g cha ra cter is tics of pig farme rs in different co untries : (i) a higher preval en ce of re spirato ry sy mpto ms, including chro nic bronchitis and asth ma th an in a re fer en ce gro up, (ii) nonrespiratory sympto ms cons istent with the or ganic du st toxic syndro me; (i ii) similar or only slightly lower lun g function values as compared wi th reference groups, (iv) no cha ng e or only sm all dec reases in lun g fun c- 48 tion va lues du ring th e wo rkshi ft . Th e foll owing airborne ag en ts in sw ine confinement bu ildings have been suggested as being re sponsible for these re spiratory manifestations: organic dust, mi croorganism s, endotox ins, and gases including ammonia and hyd rogen s ulfi de (8 , 10, 15, 20, 2 1, 23, 25-30). Th e pathoge nesis of these sympto ms has not ye t been establ ish ed (24,31).
How ever, pr eviou s studies ha ve repo rte d an increase in nonspec ific airway responsiveness am ong pig farmers. Acute exposure in swine confinement buildings caused bronchi al hyperresp onsiveness am on g six healthy su bjec ts (30). S win e co nfi ne ment worker s have slig htly high er non specific airw ay responsiveness than outdoor blu e-collar wo rkers (22) and referent wo rke rs (32) . Moreover, nonspeci fic bronch ial reactivity correl ates wi th duration of exposure among pi g farmers ( 17) . However, the hyperreactivity found am ong pig farmers is not significantly different from that found am ong dairy far mers (17,32). These studies co mpa ring pig farmers with ei ther dairy farme rs or other re fer ents suggest an inc reased nonsp eci fic airway responsi ven ess am on g pig fa rmers and pro bably also am ong dairy farme rs. Onl y one study did not fin d b ronchial hyperresp on si veness in pig farmers (33).
Th e we stern re gion of Fr ance ha s a particularly high den sit y of pig farms . Th ere are abo ut 150 000 people working on p ig farm s, incl udi ng 30 000 wh o work in swine confinement buildings. No data is currently available about the respiratory health of French pig farmers. Therefore, we initiated a cross-sectional epidem iologic study to assess the prevalence of respiratory manifestation s among French pig and dairy farmers and to determine the possible role of bronchial reactivity in the respiratory manifestations by comparing farmers with people not exposed to pigs or the farm environment.

Population
All of the subjects were selected from the Rural National Health Service register for two districts of western France. The group of pig farmer s was chosen first. All swine confinement producers with more than 50 sows or more than 800 pigs were selected. In farms of this size at least one worker spends more than half of his or her worktime inside the swine confinement buildings. Each pig farmer was visited at his or her residence and asked to participate in the study. One hundred and sixty-four pig farmers were visited, and 138 (84.1%) were included in the study. Women were excluded because they had different patterns of possible confounding factors. They were older than the men (43.2 versus 39.7 years, P = 0.08), the prevalence of smokers was lower (5.6 versus 28.4%, P< 10-4), and the type of work and duration of exposure inside the swine confinement buildings were differ ent. The study group was thus 102 men (group P).
Two reference groups were then selected from the same register according to a similar recru iting program. The men were not confin ement swine producers. They were matched for age (±5 years) and county of residenc e.
One hundred and twenty-one farmers other than pig produ cers were chosen, of which 101 persons (83.5%) responded . All 36 women and 14 subjects who were ex-swine producers or poultry breeders were excluded. The group of dairy farmers thus contained 51 men (group D).
Scand J Work Environ Health 1994, vol 20, no 1 A second reference group was drawn from the list of workers in the dairy industry who were not exposed to any air contaminants. A sample of 125 subjects was selected, and 115 persons (92.0%) were studied. Twenty-five women and nine men were exswine producers, poultry breeders , or dairy farmers and were therefore excluded. The group of referen ce worker s was thus 81 men (group R).
The study was conducted over three weeks in May 1989.
The demographic characteristics of the three groups are presented in table 1. The smoking habits of groups P and D were similar. The percenta ge of smokers in group R was higher than in group s P (P< 10-4) and D (P < 10-4).

Questionna ire
Each subject completed a questionnaire about personal characteristics, respiratory symptoms, smoking habits, and occupational history . Nonsmokers were defined as those with a lifelong abstin ence from smoking and ex-smokers as those who had ceased smoking completely at least six months before the study. The questions on respirato ry symptoms were translated from the questionnaires of the British Medical Research Council and of the International Union Against Tuberculosis and Lung Disease. They addres sed, among other things, the follow ing item s: usual and morning cough , phlegm , wheezing durin g the last 12 months, shortness of breath, and a description of breathin g. Additional questions on work-related symptoms such as rhinitis, irritation of the eyes, and respiratory symptoms were included .

Pulmonary fun ction tests
Pulmon ary function tests were performed on each subject with the use of a compute rized pneumotachograph Fleish no 3 (Spiromatic, ets MSR). The spirometer was calibrated each day with a 1.5-1 syringe. At least three satisfactory forced maximal expiration s were performed by each subject while he was seated and wearing a noseclip.
The highest forced vital capacity (FVC), highest forced expirator y volume during 1 s (FEVl.o)' and highest peak flow rate (PF) were recorded, not necessarily from the same test. The other forced expiratory flow rates were taken from the curve with the highest sum of the FVC and FEV Lo : forced expiratory flow between 25% and 50% of the FVC (MEF), and maximal expiratory flow rates at 75%, 50%, and 25% of the FVC still to be expired (FEF 75 , FEF 5o ' and FEF 25 , respectively). All of the values were expressed as percentages of predicted values from the European Committee for Coal and Steel and Bouhouys. To avoid bias from age, height, and smoking habits in the lung function measurements, standardized deviations were used for the statistical analyses (ie, the observed value minus the predicted value divided by the residual standard deviation).

Methacholine challenge
Bronchoconstrictor challenge was performed with an automatic device (FDC 88, Et Mediprom) that vaporizes a dose of methacholine at the beginning of inhalation. The solution of methacholine (I mg· ml") was vaporized by a De Vilbiss 646 nebulizer powered by an air compressor. The inhalations were done while the subject was seated and wearing a noseclip. Each dose contained 40 ug of methacholine, and the spirometric variations were recorded 2 or 3 min after cumulative doses of 80, 240, and 560 ug of methacholine. The challenge was not performed if the subject had an observed FEVlo:predieted FEV ratio of less than 70%, and it was 10 stopped after a decrease of 20% of the FEV 10' For the analysis of the methacholine challenge, we compared the prevalence of the subjects with a fall in FEV LO that was greater than 10, 15, and 20%. They were considered responders. (See table 5 in the Results section.) The decrease in FEV 10 was less than 10% for many of the subjects. The dose-response slopes were calculated for each subject as the ratio between the percentage of the variation from the initial value of the FEV LO and the final cumulative methacholine dose administered as follows (34)  50 Environmental assessments Airborne dust, ammonia, and carbon dioxide were measured inside 28 swine confinement buildings on six pig farms (table 2). Samples were taken from sites considered to represent the typical work environment in the building. Total dust, inspirable particles, and the respirable fraction were determined for each site (35). Inspirable particles in the breathing zone of the workers were also determined with personal sampling pumps. The dust was collected on 25-mm polyvinyl chloride filters with or without cyclone separators. Some of the dust samplers were equipped with Nuclepore filters for electronic microscopy. The particles were identified by scanning electron microscopy with xray energy spectrometry.
Ammonia in the sites from which the dust samples were taken was measured using Drager colorimetric tubes, and the carbon dioxide concentration was determined by spectrophotometry (Binos I). Ammonia was also assayed by personal sampling. Filters were impregnated with sulfuric acid, and the ammonia was measured by ionic chromatography (Dionex).
The measurement of endotoxins is not possible in France. The environment was not assessed on the small dairy farms, which are scattered over the region. The dairy farmers had varied activities. They do not work in confinement buildings but often work in the open air.

Statistical analysis
Standard descriptive statistics were used to represent the responses. Chi-squared tests were used with 2x2 contingency tables to determine whether the relationships between the qualitative variables were statistically significant, and the Mantel-Haenszel analysis was used to determine the relative risk. Discrete variables were corrected for smoking. An analysis of variance was used to examine the relationships between pulmonary function and respiratory symptoms, work exposure, and smoking. The continuous variables were corrected for age, height, and smoking.
The treatment of the data was performed with an SAS (Statistical Analysis System) package.

Results
The workers on the pig farms were exposed, on the average, to a total dust level of 2.41 mg· rrr", However, the amount of respirable particulate matter was low (table 2). Electron microscopy indicated large organic particles (figure I). The amount of ammonia varied from 0.60 to 5.9 mg . m' between buildings. The concentration of carbon dioxide was low (1000 to 5000 ppm). The peak concentrations of carbon dioxide occurred when the pigs were excited by the farmer entering the buildings (figure 2).
Before the methacholine challenge, the mean lung function values we re normal accordin g to the European refere nce val ues. The dairy farmers had lower values, but there was no significant difference between the groups (table 4). No significant corre lation was observed betw een duration of employment and the lung function values, adjusted for age, height, and smoking habits.
Two subjects in group P, two in group D, and one in group R could not undergo bron choconstrictor challenge becaus e their initial pulm onary function values were too low . One subject in group P, one in group D, and two in group R stopped the cha llenge at cumulative doses belo w 560 ug.
Th e meth acholine challenge indu ced a significant decrease in lung function in all three groups.
Bronchial hyperreactivity to methacholine cha llenge was significa ntly higher among the dairy and pig farme rs than among the referents, with highe r preval ences of respond ers and lower negati ve mean slopes (table 5). Onl y fou r subjects (1.8%) had a decrease in FEV LO that was greater than 20%. The findings were unaffected by adjustment for the initi al Th e distribut ion of positive res ponses to the question "Have you ever had asthma ?" was simil ar in the three groups, four subjects in group P, three in group D, and thre e in group R. Excluding asthmatics did not change the demo graph ic characte ristics of the groups.
The prevalence of respiratory symptoms did not differ bet ween group P and D exce pt for work-related fits of coughing (P < 0.02) (table 3). These two groups had a slightly higher prevalence of cough and morn ing phlegm than group R did. These differences were sign ificant even after adjustment for smoking habit s.
Respirat ory symptoms were significantly more frequent during work in groups P and D than in group R (table 3). Very few farmers rep orted dela yed , work-related symptoms.    Table 4.  . ..._----Scand J Work Environ Health 1994, vol 20, no I Table 5. Prevalence (%) of subjects with a decrease in forced expiratory volume in 1 s equal to or greater than 10% (responders 10) and 15% (responders 15) and the mean of the slope of the bronchial reactivity by group. The reactivity was higher in groups P and D than in group R. (group P = pig tarrners , group D = dairy farmers , group R = referents)

Discussion
Several airborne contaminants are present in pig confinement buildings. In this study, the mean levels of ammonia, carbon dioxide, and dust were lower than the threshold limit values. However, there may be synergistic effects between toxic agents. The level s of the se agents were similar or lower than the concentrations observed in Canada (10 ), Sweden (8,15), the United States (25 , 26), Finl and (29 ), The Netherlands (27), and Yugoslavia (23). The dust was mainly composed of organic particles. The size of these particles, evaluated with the aid of both samplers and electronic microscopy, were about 5-10 urn. This finding is consistent with those of Haglind & Rylander (8), Crook et al (20), and Donham et al (26,36). The inhalation of endotoxins may playa role in the occurrence of febrile reaction and respiratory manifestations among swine confinement workers (8,18,20,21,27). The endotoxin level correlates with base-line PVC (2 1) and also with change in FEVl.o over a work period (15 ). End otoxins may thu s increase bronchial reactivity. Howe ver, it is not known whether endotoxins alone are responsible for the change in bronchial responsiveness caused by inhaled swine dust (28,30,32). For technical reasons, expo sure to endotoxins was not asse ssed in our study. We compared pig farmers, dairy farm ers, and referents. Man y previous studies ha ve compared pig farmers to only one reference group, either dairy farmers or other referents. Th e use of nonp ig farmers as referents is debatable, as they may be expo sed to various airborne contaminants. They have a higher prevalence of respiratory symptoms and function abnormalities than nonfarm workers (2--4, 10, 19). On the other hand, the pre valence of smokers was lower among the farmers than among the nonfarmers in our study population, and in Ital y (2) and Canada (19 , 22).
This cro ss-sectional study may have been biased by the "healthy worker effect." However, we identified a higher prevalence of respiratory symptoms among the pig farme rs and among farmers not involved in intensive breeding than among nonfarm- Group P Group D GroupR ing references workers. There was no significant differenc e between the prevalence of respiratory symptoms in the two farming groups. However, respiratory symptoms were more prevalent in both groups of farmers than in the group of nonfarmers, despite lower percentages of smokers. The pig farmers especially reported symptoms of cough, fits of coughing , and phlegm. The prevalenc e of wheezing, che st tightness, shortness of breath , and asthma were not significantly different between the pig farmers and referents. Various respiratory symptoms have been identi fied among pig farmers in previous studies. For example, a higher prevalence of airway irritation and bronchitis was found in several studies (8,9,12,14,15,17,19,22,23 ), and sometimes a higher pre valence of asthma (1 I, 16) and symptoms of bron chial hyperreactivity (13) were revealed.
In the three groups, the mean base-line FVC and FEVl.o were similar to the European reference values. The mean MEF, FEF 5o ' and FEF 25 were slightly lower than the reference values but without significant differences between the three studied groups even when smoking habits were taken into account.
Lung function values were also slightly lower for the reference farmers than for the pig farmers in the study of Holness et al (10). Lower lung function values were also found by Dalphin et al (3) among French dairy farmers.
Bongers et al (9) found me an pulmonary fun ction values, exc ept for PVC, that were lower than the reference value s among 132 pig farm owners. However, con sistent with our study, there was no significant correlation between duration of exposure and pulmonary function . They did not study other farmers or reference groups. In contrast, Iver sen et al (16) demonstrated tha t a low FEV l.n correlated with the number of years in pig farm ing and also with bronchial hyperreacti vity in a subgroup of symptoma tic farmers without asthma. Therefore, bronchial hyperreactivity has been suggested to be inv ol ved in the higher prevalence of respiratory disease among farmers, esp ecially among pig producers (16 , 36).
Our doses of methacholine may have been too low to identify the hyperreactive subgroup unambiguou sly. The doses were howe ver sufficient to reveal higher hyperreacti vity in the farming groups than in the nonfarming group. Th is find ing confirms the higher hyperreacti vity observed by Zhou et al (22) among 20 pig farmers as compared with outdoor urban workers and by Rylander et al (32) among 36 pig farmers as compared with 16 referents, but not between the pig farmers and 23 dairy farmers. Iversen et al (16) studied bronchial reacti vity by determining the pro vocative concentration of histamine causing a 20 % decrease in peak exp iratory flow . They found that most of the hyperreactivity among farmers wa s explained by a low FEVl.o' Moreover, they found that the hyp erresponsiveness wa s slightly higher among pig farmers than among dairy farmers, but the difference was not significant ( 17) .
The lower initia l FEVI. Odid not completely explai n our re sults. The in crease in the nonspecific hyperreactiv ity among th e pi g and dair y farm er s was indep endent of the initial FEV I.o' Further stu d ies ar e required to c ha rac te rize th e c ausative agents of re sp iratory sym pto m s and hyperreactivity am ong farmers.