INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although the increase in diagnosed asthma in recent decades is real enough (1), most of this increase remains unexplained. Recently, it was stated that there is an urgent need to understand environmental influences bearing on the development of asthma (2). The author stressed that occupation is not only important as a direct cause of asthma but also as a valuable model for investigating potential interactions in its causation. When studying chronic disease in the population at large, knowledge of the relations between exposure and disease in specific occupational groups is important. This is especially the case for types of exposure that can be found in many situations. In the case of chronic respiratory disease, attention is focusing more and more on the role of bacterial endotoxin, both in a large number of work situations (3) and in the home environment (8). A number of cross-sectional studies reported associations between exposure to endotoxin as measured in dust and baseline lung function (3, 9) and/or changes in lung function across a work-shift (12). In several experimental studies short-term effects of controlled exposure to endotoxin in purified form (LPS) were shown, demonstrating that it has a strong potency to induce inflammation in the airways (18). An association between endotoxin exposure and lung function decline was reported by Schwartz and colleagues (23) in a 2-yr follow-up study in farmers. It remains unclear from this study at which levels of exposure effects were observed, nor was an exposure-response curve published. Bias from various sources makes it difficult to interpret cross-sectional studies reporting associations between endotoxin exposure and lung function. For a correct assessment of the effects of such exposure, it is imperative that data from longitudinal studies become available. The present study evaluates the association between endotoxin exposure, as measured in dust, and longitudinal decline of lung function in a 3-yr follow-up in a cohort of 171 pig farmers.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

In 1990, a questionnaire survey was held among 1,504 pig farmers (24). To be able to study exposure-response relations, a study population was formed with sufficient contrast in respiratory morbidity, and therefore presumably also in exposure. In 1991, 200 randomly selected subjects from among the respondents with one or more chronic respiratory symptoms (chronic cough, chronic phlegm, shortness of breath, ever wheezing, frequent wheezing, chest tightness/asthma), and 199 randomly selected subjects without such symptoms were invited for a medical examination that included an extensive interview. Of the 200 farmers who reported one or more chronic symptoms in the postal questionnaire, 115 qualified for the category of chronic obstructive pulmonary disease (COPD) and/or asthma, on the basis of further information obtained in the interview at the medical examination. They were considered to be consistently symptomatic. Of the 199 farmers without chronic symptoms in the postal questionnaire, 145 revealed no chronic symptoms either in the more extensive interview. They were considered to be consistently asymptomatic. Ninety-eight of the consistently symptomatic and 100 of the consistently asymptomatic were randomly selected for a follow-up program which included measurements of environmental and personal exposure in their own confinement units. All worked at least 5 h/d in pig farming. This cohort was invited in June 1992 (25) and June 1995 for medical examinations which included a measurement of lung function. In 1995, participants again completed a questionnaire on symptoms. The number of farmers that participated on both occasions was 171 (82 of the consistently symptomatic and 89 of the consistently asymptomatic). Reasons for nonparticipation in 1995 were mostly "no time/too busy" or "no interest any more." One farmer had died from a nonpulmonary disease.

Lung Function

Lung function was measured in 1992 with a Vicatest dry rolling-seal spirometer and in 1995 with a Vicatest (n = 94) or a Sensormedics Pulmonet3 water-sealed spirometer (n = 77). All measurements were taken according to European Respiratory Society (ERS) guidelines (26) , and corrected to body temperature and pressure, saturated with water vapor (BTPS). The study was approved by the Committee for Ethical Research of the University of Nijmegen.

Exposure

Measurements of exposure were carried out on the farms of all participants during full work-shifts of on average 8.3 h on 2 d in summer 1991 and winter 1992. Personal exposure to inhalable dust was determined using a dust sampler with a 6-mm diameter inlet opening and an airflow of 2 L/min. Endotoxin in the inhalable dust samples was analyzed with a modified kinetic Limulus Amoebocyte Lysate test (27). Methods for measurement of dust and endotoxin are described in detail elsewhere (28, 29). Because the day-to-day variations of exposure to dust between days in individual participants were considerable compared with the variations of exposure in the entire group, the long-term average exposure was predicted by a mathematical modeling technique: long-term average exposure to dust and endotoxin of each individual farmer was estimated using data on farm characteristics and time spent on activities in pig farming of all cohort members combined. This is described in detail elsewhere (30). From the total number of 171 cohort members, complete data on exposure to dust and endotoxin became available for 146 participants.

Analysis

As exposure was lognormally distributed, logtransformed values with base 2 were used in the analyses of associations with health effects. Declines in lung function were normally distributed. Associations between exposure and lung function decline were tested with linear regression analysis, adjusting for age, baseline FEV₁ (baseline FVC for analyses with FVC), and smoking behavior, defined as pack years of cigarettes. For current and past cigarette smokers the numbers of pack years were calculated by the number of cigarettes smoked per day multiplied by the number of years smoked, divided by 25.

One participant contracted severe pulmonary disease in the follow-up period, resulting in a 2.5 liter loss of lung function. To avoid distortion of the results, he was excluded from the analyses of associations with exposure.

Computations were completed with Statistix for personal computers.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Personal Characteristics

Mean age of the 171 farmers was 39.6 yr at the start of the observation period (Table 1). Mean number of years worked in pig farming was 16.7. Five participants were no longer active as pig farmers in 1995, three of them stopped partly because of respiratory problems. Percentage of cigarette smokers was 25.

Lung Function

Mean FEV₁ and FVC of the 171 participants was 3.97 L and 5.06 L, respectively, in 1992 (Table 1). Farmers with chronic symptoms in 1991 had a significantly lower lung function than those without. Mean decline in lung function was 73 ml/yr for FEV₁ and 55 ml/yr for FVC. This was not different for farmers consistently symptomatic or asymptomatic in 1991.

Decline in FEV₁ was smaller for those tested with the water-sealed spirometer in 1995 than for those tested with the dry rolling-seal spirometer on both occasions. After leaving out one participant who had a difference of 2.5 L in his lung function, the mean decline was 57 ml/yr (SD 68) versus 79 ml/yr (SD 82), p = 0.06.

Exposure and Associations with Respiratory Effects

Estimated long-term average exposure to inhalable dust was 2.63 mg/m³ (geometric mean) and to endotoxin this was 105 ng/m³ (Table 1).

After adjusting for age, baseline FEV₁ or FVC and smoking, decline in FEV₁ during the 3-yr period between 1992 and 1995 was significantly associated with endotoxin exposure alone, whereas decline in FVC was associated with both endotoxin exposure and inhalable dust (Table 2). The association between endotoxin exposure and annual decline in FEV₁, based on these data, is represented in Figure 1. Over the whole range of exposure to endotoxin in this group, annual decline in FEV₁ varied from 40 ml for the lowest exposed farmers to 100 ml for the highest exposed ones.

View larger version (8K): [in this window] [in a new window]   Figure 1.   Predicted association between logtransformed endotoxin exposure and annual rate of decline of FEV1 (with standard error), corrected for age, baseline FEV1, and pack-years of smoking. Based on 171 pig farmers.

The associations between exposure and lung function decline were stronger for the group tested with the dry rolling-seal spirometer on both occasions than for the group tested with the water-sealed spirometer in 1995 (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this 3-yr follow-up of 171 pig farmers, we showed a large decline in FEV₁ and FVC. Long-term average exposure to endotoxin was found to be associated with decline of FEV₁ and FVC, whereas exposure to dust was associated with decline of FVC alone.

The mean annual decline of FEV₁ of 73 ml is large compared to the expected age-related decline of 29 ml/yr (26), and equal to that of 73 ml in pig farmers reported by Iversen and colleagues (31). The selection procedure did not affect this figure, as decline was similar for farmers with or without symptoms. This is probably due to the fact that at the start of follow-up symptomatic farmers had a mean lung function only slightly different from normal. The group tested with a different spirometer at the end of the follow-up, showed a smaller (though still large) mean decline than the group tested with the same spirometer on both occasions. As these groups were not different in any relevant feature at the start of the study, we assume that the true decline in FEV₁ is higher than the 73 ml/yr on average found.

Long-term average exposure to endotoxin was an important factor associated with this decline. Logtransformed exposure with base 2 was used in analysis. This means that an increase in exposure with a factor 2 was associated with an extra decline of FEV₁ of 19 ml/yr. In multivariate regression, pig farmers with the lowest levels of exposure showed an average predicted decline similar to the expected age-related decline, whereas average predicted decline of the highest exposed pig farmers was as much as 100 ml per year. Part of the group was tested with a different spirometer in 1995. This probably introduced some underestimation of the exposure-response relationship: associations were stronger in the subgroup tested with the same spirometer both in 1992 and 1995. Modeled long-term average exposure was used rather than the mean of the two measurement days per farmer. Exposure assessment taking repeated samples at each farm would have required at least 30 measurements per farm to obtain a reliable estimate of the long-term average exposure. This is because of the large variation in dust and endotoxin exposure from day to day, when compared with the differences in exposure between farms. We applied a recently described method (30) that predicted exposure, using information on farm characteristics (e.g., type of floor and feeding method) and tasks (e.g., controlling, cleaning, teeth cutting). A small validation study was performed among six farmers for which six to nine valid monthly measurements were available. This showed that the estimated long-term average exposure gave a better reflection of the long-term exposure (mean of monthly measurements) than did the average of the two measurement days available for the entire group (30).

In cross-sectional studies, adverse effects of endotoxin exposure on baseline lung function have been shown before in pig farmers (3, 5, 11) and other occupational groups (4, 6, 7, 9, 10, 13). One study group reported an association between endotoxin exposure and increased loss of lung function in a longitudinal observation of 2 yr on average (23). However, in that study no clear associations between levels of exposure and rates of decline in lung function were shown. This is the first study that shows a clear exposure-response relation of endotoxin with longitudinal decline of lung function. Occupational threshold exposure levels of 100 and 30 ng/m³ (32, 33) have been proposed, to prevent respiratory disease. The health effects seen in our study support the lower of these proposals.

The balance of studies on pig farmers suggests development of obstructive rather than restrictive pulmonary disease (31, 34). In studies in various occupational circumstances endotoxin exposure correlated better with (changes in) lung function when compared with dust exposure (3, 5, 7, 10, 13). In our study exposure to inhalable dust correlated with lung function decline measured as FVC, but not with FEV₁, whereas endotoxin correlated with both. Although inhalable dust is a more generally available and cheaper marker for exposure, we conclude that dust exposure alone is insufficient as predictor for lung function decline in pig farmers. We established a clear link to exist in a longitudinal study between lung function and occupational endotoxin exposure in a basically healthy group of workers. As strong effects were demonstrated in healthy workers at levels previously not proved harmful, it becomes increasingly important also to know the role played by endotoxin exposure in the development of respiratory disease in a domestic environment. There is evidence that susceptible groups, especially asthmatics, can be affected by chronic exposure at lower levels. Michel and colleagues confirmed that natural exposure to endotoxin in house dust is related to the clinical status of asthmatics (8). Sparse information available on endotoxin exposure in domestic environments indicated airborne levels of up to 1.2 ng/m³ (35), much lower than in the working situation. Research is needed on relevant levels of exposure to endotoxin (airborne or in dust from beds or floors?) and the mechanisms by which exposure leads to respiratory disease.