Evaluation of respiratory effects related to high-pressure cleaning in a piggery with and without robot pre-cleaning

Evaluation of respiratory effects related to high-pressure cleaning in a piggery with and without robot pre-cleaning. Scand J Work Environ Health. 2009;35(5);376–383. Objective Exposure in connection with cleaning piggeries induces airway inflammation. The aim of this study was to compare the health effects related to two different pre-cleaning processes in a piggery. Methods In a cross-over study design, 12� su��ects were randomly exposed for three hours during the cleaning of su��ects were randomly exposed for three hours during the cleaning of a piggery with a high pressure water �et, with and without pre-cleaning using a ro�ot. We assessed lung function, �ronchial responsiveness, symptoms, �ody temperature, and exhaled nitric oxide, and performed �lood sampling and nasal lavage �efore and after �oth exposures. Results Compared with ordinary cleaning without the use of a ro�ot, pre-cleaning with a ro�ot significantly reduced the increase in �ronchial responsiveness (P=0.049), total cell num�er (P=0.002�9), and pro-inflamma-tory cytokines (IL-8 level [P=0.016]) in nasal lavage, and diminished the increase in neutrophils (P=0.002�9) in the �lood. Conclusion Pre-cleaning of a piggery with a ro�ot reduced exposure to dust and endotoxin, and resulted in an attenuation of the increase in �ronchial responsiveness and the airway inflammatory response compared precleaning without a ro�ot.

People working in the pig farming industry have an increased prevalence of respiratory symptoms and diseases such as chronic �ronchitis and chronic o�structive pulmonary disease (1)(2)(3)(4). Exposure of healthy, previously unexposed volunteers in the farming environment has resulted in intense airway inflammation and increased �ronchial responsiveness. The exposureinduced airway inflammation in volunteers is characterized �y a multifold increase of inflammatory cells (mainly neutrophilic granulocytes), pro-inflammatory cytokines (IL-6 and IL-8) in �lood, nasal, and �ronchoalveolar lavage fluid, and increased levels of exhaled nitric oxide (NO) (5)(6)(7). The reaction in �oth farmers and volunteers is most likely caused �y the high levels of air�orne organic dust in the piggery (8).
Larsson et al has shown that the cleaning of the interior of the sta�le is a working situation that causes a high dust exposure with a greater proportion of respi-ra�le dust compared with exposure when weighing pigs (9). After each completed �reeding period, the sta�les are normally soaked with sprinklers during the night and su�sequently manually cleaned using a high-pressure water �et. The high proportion of respira�le dust measured during this cleaning procedure could originate from the fragmentation of dust particles �y the highpressure cleaner's water �et that, together with dampness in the sta�le, intensifies the exposure as previously shown. Even though the exposure to dust was approximately 2�0 times lower during the cleaning of the sta�le compared to when weighing pigs, the post-exposure increase in �ronchial responsiveness is similar on �oth occasions. It has also �een shown that wearing a mask attenuated, �ut did not a�olish, the effects of exposure in healthy volunteers (9)(10)(11).
Recently, an automatic cleaning ro�ot (Ramsta Ro�otics, Uppsala, Sweden) has come into use in piggeries. The ro�ot pre-cleans the sta�les during the night and the farmer finishes the cleaning procedure with a water �et on the su�sequent day. This procedure lowers the exposure time for the farmer up to 75% (12�-13). The ro�ot Hiel et al is, however, expensive and has to �e specially ad�usted for each sta�le �y an expert, which is a time-consuming procedure. It is unknown whether the use of the ro�ot influences the health impact of the cleaning procedure on the farmer. Even though the dust exposure and exposure time is reduced when using the ro�ot, the higher proportion of the respira�le dust fraction might indicate a deeper penetration of small particles into the lungs, which in turn may cause more severe health effects.
Using a cross-over design, the aim of this study was the aim of this study was to evaluate the acute effect of the cleaning processes, with and without pre-cleaning using a ro�ot, on previously unexposed volunteers.

Subjects
Twelve healthy, non-smoking volunteers participated in the study [nine women, mean age 2�7 (range 2�0-37) [nine women, mean age 2�7 (range 2�0-37) years]. The participants had no respiratory symptoms or . The participants had no respiratory symptoms or diseases, and no allergies as confirmed �y a questionnaire and skin prick tests using 17 common aeroallergens. Other inclusion criteria included normal results of a physical examination, spirometry, and �ronchial responsiveness to methacholine. None of the volunteers had previously �een exposed to farm dust. All volunteers gave their informed consent. The study was approved �y the ethics committee at the Karolinska Institutet (2�005/1163-31/4).

Study design
The su��ects were twice exposed for three hours in a piggery for �oth exposure settings. During the exposure, the farmer was cleaning the sta�le with a high-pressure cleaner. On one occasion, the cleaning ro�ot had carried out an automatic cleaning on the previous day. On the other occasion, cleaning was preceded �y soaking the sta�les with a sprinkler installation (ie, no ro�ot was used). The study was performed in a randomized, crossover design, with a wash-out period of at least one month �etween the two exposures. Six persons were exposed on each occasion, resulting in two exposures of pre-cleaning with a ro�ot and two exposures of pre-cleaning without a ro�ot. During each exposure, 2�-3 of the volunteers wore personal samplers for exposure monitoring. Two weeks prior to the exposure, all su��ects underwent spirometry, �ronchial methacholine provocation, and nasal lavage and �lood sampling; exhaled nitric oxide (NO) was also measured. The same tests were repeated seven hours after the start of each exposure. Immediately prior to and after exposure, we measured peak expiratory flow (PEF), symptoms, and oral temperature.
After weighing, we extracted the filters and analyzed them for endotoxins using a kinetic technique version of limulus ame�ocyte lysate (LAL) assay (Endosafe Endochrome-K, Coatech AB, Kungs�acka, Sweden) with E. coli 0.111:B4 as standard (14). To ensure that b-glucans, which are most likely present in the dust, did not interfere with the assay, we used an endotoxinspecific �uffer (Charles River Endosafe, Charleston, USA) to dilute the LAL assay according to manufacturer to dilute the LAL assay according to manufacturer according to manufacturer recommendations.
For all duplicates, an intra-assay variation of <15% was accepted. The detection limit for endotoxins was 0.02� EU/ml corresponding to 40 ng/m 3 .

Symptoms
We assessed general and airway specific symptoms using a 100 mm visual analogue scale. The volunteers were requested to put a cross on the scale where 0 indicated "no symptoms" and 100 indicated "un�eara�le symptoms". We measured oral temperature directly �efore exposure and then every hour up to eight hours after the start of exposure.

Lung function and bronchial responsiveness
We measured vital capacity and forced expiratory volume in one second (FEV 1 ) using a wedge spirometer (Vitalograph®, Buckingham, United Kingdom) according to the criteria of the American Thoracic Society (15). Local reference values were used (16)(17). We measured PEF immediately prior to and after exposure and every hour up to five hours following exposure with a mini-Wright peak flow meter (Clement Clarke Ltd, London, United Kingdom). We registered the �est out of three measurements.
Bronchial responsiveness to methacholine was measured as has �een previously descri�ed (18). The results were expressed as the cumulative dose causing a 2�0% decrease in FEV 1 (PD 2�0 FEV 1 ). The dose/response slope of change in FEV 1 was calculated �y linear regression including all data points, as the percent change of FEV 1 as a function of the cumulated methacholine dose (19).

Exhaled nitric oxide
We measured NO in exhaled air using a single-�reath exhalation with 50 ml/s flow rate, according to the American flow rate, according to the American , according to the American American Thoracic Society recommendations (2�0). The analysis was (2�0). The analysis was conducted after reaction with ozone (NIOX®, Aerocrine, Stockholm, Sweden) as descri�ed earlier (11). To decrease descri�ed earlier (11). To decrease (11). To decrease contamination from the oral cavity, preceding the meas-preceding the measurement, the mouth was rinsed with water and a sodium the mouth was rinsed with water and a sodium �icar�onate solution (10%) for 30 seconds each (2�1).

Nasal lavage
Nasal lavage was performed using a procedure descri�ed �y Bascom et al, with minor modifications (2�2�). The lavage samples were centrifuged, and the num�ers of cells were counted in a Bürker cham�er. The supernatant was frozen �efore analyses. IL-6 and IL-8 were determined in the samples using an enzyme-linked immunosor�ent assay developed at our la�oratory using commercially availa�le anti�ody pairs (R&D Systems Ltd, Europe, A�ingdon, United Kingdom) (2�3). The lower detection limit of IL-6 and IL-8 were 3 pg/ml and 50 pg/ml, respectively. For duplicated samples, we accepted an intra-assay coefficient of variation of <10% and an inter-assay coefficient of variation of <2�0%.

Blood leukocyte analysis
Whole peripheral �lood was collected in ethylene diaminetetra-acetic acid vacutainer tu�es (BD Bioscience, New Jersey, USA) and tested within two hours of the sampling.
The samples were analyzed in Attractors TM (BD Bioscience, San Jose, CA, USA) to perform a five-part white �lood cell differential, as previously descri�ed (11).

Statistics
Lung function, symptoms, and cell distri�ution in �lood are presented as means and 95% confidence intervals (95% CI). Statistical comparisons �etween the exposures were assessed �y analysis of variance (repeated measures) followed �y the Student's t-test for paired o�servations as a post hoc test when appropriate. For not normally distri�uted data, the results are presented as medians and 2�5 th -75 th percentile (exhaled nitric oxide, �ronchial responsiveness, and nasal lavage fluid results). Statistical comparisons �etween exposures were then assessed �y the Friedman test followed �y the Wilcoxon signed rank test when appropriate or �y the Spearman rank correlation test. We considered a P-value of <0.05 significant.

Exposure measurements
Significantly lower levels of inhala�le dust (P=0.02�3) were o�served when the sta�le was pre-cleaned with the ro�ot, whereas respira�le dust levels were similar for �oth exposures (figure 1). Endotoxin levels in inhala�le (P=0.0062�) and respira�le dust (P=0.014) were also lower when the sta�le was pre-cleaned with the ro�ot.

Lung function and bronchial responsiveness
Pre-exposure lung function of vital capacity and FEV 1 were normal for all participants. Pre-exposure percentage of predicted value for vital capacity was mean 97% (95% CI 90-103), and for FEV 1 mean 98% (95% CI 91-105), respectively. Both exposures caused a significant decrease in PEF (P<0.01), �ut not in other lung function parameters (ta�le 1). No significant difference �etween the two exposures was found. The PD 2�0 FEV 1 decreased significantly only after the exposure without ro�ot pre-cleaning (P=0.034), with no significant difference �etween the two exposure days (ta�le 1). The dose response slope increased significantly only without ro�ot pre-cleaning (P=0.0076), and significantly more than when the ro�ot was used (P=0.049).

Exhaled nitric oxide
The level of exhaled NO increased without the use of a ro�ot although not significantly, �ut significantly more than when pre-cleaning with a ro�ot (P=0.0049).

Nasal lavage
After pre-cleaning the sta�le with the ro�ot, the num�er of inflammatory cells in nasal lavage fluid dou�led (P=0.02�2�) (figure 2�); without the use of a ro�ot, a more than five-fold increase was o�served (P=0.002�2�; P=0.002�9 �etween exposures). No significant differences were found in the concentrations of IL-6 �efore and after the two exposures (P=0.30). The concentration of IL-8 increased following exposure without ro�ot precleaning (P=0.0033) and significantly more than after pre-cleaning with a ro�ot (P=0.016). There was a significant correlation �etween IL-8 concentrations and the cell num�er in nasal lavage fluid (Rho=0.58; P=0.0006).

Blood leukocyte analysis
The total num�er of leukocytes in peripheral �lood increased significantly after �oth exposures (P≤0.02�0), �ut significantly less using the ro�ot (P<0.001). The change was mainly due to a greater increase in neutrophilic granulocytes which was significantly higher when not using the ro�ot compared with ro�ot pre-cleaning (P=0.003) (figure 3). The num�er of monocytes, and Thelper cells increased significantly after �oth exposures. The increase in monocytes were higher after exposure without using the ro�ot (P<0.0002�), �ut no significant difference in �lood T-helper cells was found �etween the two exposures. Table 1. Results of the measurements of lung function and exhaled nitric oxide (NO). Pre-exposure values were >80% of predicted in all participants (within 80-120% of predicted value). Peak Expiratory Flow (PEF) fell significantly following exposure, with no significant difference between the two exposures. Bronchial responsiveness to methacholine are presented as PD 20 FEV 1 (ie, the cumulative dose) causing a 20% fall in FEV 1 (forced expiratory volume in 1 second) or the dose/response slope (DRS) of change in FEV 1

Discussion
Manual cleaning of a piggery inevita�ly results in exposure to air�orn material. In this study, it has �een demonstrated that pre-cleaning with a ro�ot reduces the inflammatory reaction compared to soaking the sta�le over night with a sprinkler (ie, pre-cleaning without the use of a ro�ot). Pre-cleaning the sta�le with a ro�ot yielded reduced inhala�le dust levels and lowered endotoxin levels in inhala�le and respira�le dust. The use of a ro�ot also significantly reduced the increase in �ronchial responsiveness, the inflammatory response in the nose, and the systemic reaction that was assessed as increase of circulating neutrophils and monocytes Cleaning of the interior of the sta�le after a completed �reeding period induced an inflammatory reaction in the respiratory tract as shown in this and a previous study (9). The magnitude of increase in �ronchial responsiveness was similar to that which has �een found in su��ects exposed during the weighing of pigs (7, 8, 10, 11, 2�3). This was a somewhat surprising finding considering that the exposure to inhala�le dust was reduced up to 2�0 times during the cleaning procedure. However, the respira�le fraction including endotoxin content, which consequently can penetrate into the lower airways, was on the other hand of the same magnitude as, or even higher than, during the weighing of pigs (2�4-2�5). These findings suggest that the water �et can �reak the particles into smaller pieces.
Apart from a reduction of PEF, no changes in lung function (vital capacity and FEV 1 ) could �e detected after exposure compared to pre-exposure values, in agreement with a previous study (9). The apparent discrepancy �etween the findings of a significant decrease in PEF, �ut not FEV 1 could �e due to the time of the measurement. Since PEF was measured hourly throughout the day and FEV 1 was only measured seven hours after exposure, it is possi�le that lung function was already normalized at the time of FEV 1 measurements.
The level of NO in exhaled air did not increase significantly after either exposure, although a significant difference �etween the exposures was found. The lack of a significant increase after cleaning without the use

Hiel et al
of ro�ot is pro�a�ly due to the variation in pre-exposure values. The ro�ot apparently attenuated the effect of cleaning on exhaled NO. As NO is considered to �e a �iomarker for lower airway inflammation, the use of a ro�ot thus seems to attenuate the pro-inflammatory effect of exposure. When the sta�le was not pre-cleaned using the ro�ot, we found a significant decrease in PD 2�0 FEV 1 and an increase in the slope compared to pre-exposure values. However, only the slope increased significantly more when not using a ro�ot compared to pre-cleaning with a ro�ot. It thus seems as if pre-cleaning with the ro�ot leads to a smaller increase in �ronchial responsiveness although this could only �e demonstrated for one parameter due to the small sample.
A previous study has shown that lipopolysaccharide (LPS) most likely is not the ma�or cause of the increase in �ronchial responsiveness, following exposure in a pig �arn (14). That study found that �ronchial responsiveness in farmers, smokers, and control su��ects increased more after exposure in a piggery while weighing pigs, compared to a pure LPS-provocation. The concentration of LPS was 15 times higher during the LPS-provocation than the concentration of endotoxin in the respira�le dust while weighing pigs. It has also �een shown that mice with defective TLR4 (a toll-like receptor for LPS) have an attenuated inflammatory response in the lung, �ut there was no effect on �ronchial responsiveness after exposure in a piggery compared to wild-type (WT) mice (2�6). The low impact of LPS as a causative agent of increased �ronchial responsiveness is also supported �y Sund�lad et al who showed that, despite wearing a respirator with filters protective for �oth particles and gas which reduced endotoxin exposure �y 99%, an increase in �ronchial responsiveness could still �e o�served (11). Ultra fine particles, uptake through the skin, and exposure to organic dust prior to entering the piggery may contri�ute to the increased �ronchial responsiveness. These results indicate that LPS is not the sole cause of the increased �ronchial responsiveness following exposure in a pig house (11, 14, 2�6).
The results of our study showed that the use of  a cleaning ro�ot reduced the increase in �ronchial responsiveness apparently �y decreasing the exposure to agents that cause such an increase. The study's crossover design could have made this possi�le in contrast to previous comparative studies that included two parallel groups. This study design was chosen as a consequence of the study �y Strand�erg et al where it was proven that the final post-exposure �ronchial responsiveness is independent of pre-exposure values (2�5). This study has shown that the use of a ro�ot reduced the inflammatory reaction after three hours of exposure during cleaning procedures in healthy, unexposed volunteers. Differences �etween the two exposures was o�served for systemic parameters in peripheral �lood as well as inflammatory parameters in �oth lower (exhaled NO, �ronchial responsiveness) and upper (nasal lavage) airways. It is, however, still unknown if chronically exposed farmers will react in the same way as unexposed volunteers.
In addition to the reduced inflammatory response o�served following exposure after pre-cleaning with the ro�ot, it should �e considered that the use of the ro�ot also lowers the duration of exposure for the farmer �y approximately 75%, a fact that might �e even more important in health aspects (12�-13). The sta�le must �e cleaned after each �reeding period and is normally a working task which results in high exposure to organic dust. The dampness of the sta�les, which makes it difficult for the farmers to protect themselves with respirators, the continuous duration of exposure (up to two days), and the high proportion of respira�le dust due to the use of a high-pressure water �et, indicate that piggery cleaning is a ma�or work environment pro�lem that needs to �e improved. The possi�ility of using a ro�ot during the cleaning process, which �oth reduces the duration and �iological effects of the exposure, might, in the long run, improve the work environment for pig farmers.