INTRODUCTION
In humans, infections with extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae are associated with increased burden of disease and costs [Reference Pitout and Laupland1]. Livestock can carry ESBL-producing Enterobacteriaceae with bla CTX-M-1 as the most prevalent ESBL gene found in Europe [Reference Ewers2]. ESBL-producing Enterobacteriaceae can be transferred from animals to humans through food or direct contact [Reference Dohmen3, Reference Kola4]. Direct contact with livestock mainly occurs in an occupational setting. In farmers, carriage of ESBL-producing Enterobacteriaceae is associated with the presence of ESBL-producing Enterobacteriaceae in animals [Reference Dohmen3, Reference Dierikx5, Reference Hammerum6]. Slaughterhouse workers might also be occupationally exposed to ESBL-producing Enterobacteriaceae. Depending on the job task, slaughterhouse workers have frequent contact with live animals, animal carcasses or animal products. ESBL-producing Enterobacteriaceae have frequently been found in intestinal content of pigs at the slaughter level [Reference Geser7–Reference Randall9]. Besides that, ESBL-producing Enterobacteriaceae can also be detected on pig carcasses at the slaughterhouse [Reference Bardon10]. A higher prevalence of antimicrobial resistant Escherichia coli in pig slaughterhouse workers has been reported [Reference Nijsten11–Reference Alali13]. However, carriage of ESBL-producing Enterobacteriaceae in slaughterhouse workers has not been established yet. In addition, E. coli contamination on carcasses seems to be reduced during the slaughter process [Reference Wu14, Reference Bouvet15]. Therefore exposure, and as a consequence prevalence of carriage, of ESBL-producing Enterobacteriaceae might be dependent on the working area and job task within a slaughterhouse.
We investigated the prevalence of ESBL carriage in pig slaughterhouse workers and the association with occupational exposure.
MATERIALS AND METHODS
Study design
A large pig slaughterhouse in the southern part of the Netherlands (Vion, Boxtel) was visited by the researchers during 1 week in June 2015. All slaughterhouse workers with a slaughter process related job were asked to participate in the study by the use of flyers and information on screens. In total, 1781 eligible slaughterhouse workers were employed by the slaughterhouse at the time of the study. Participants were asked to provide a faecal specimen and to fill out a consent form and a questionnaire containing items on personal and occupational information, including job function in the slaughterhouse. Due to the heterogeneity in nationality of the slaughterhouse workers (14 different nationalities), the flyer, questionnaire and consent form were provided in 11 different languages. Participants were motivated for participation with a voucher worth of 25 euro. Faeces samples and documents were handed in to the researchers at the slaughterhouse within the same week. At the same day, for the purpose of detecting ESBL-producing Enterobacteriaceae, <1 g faeces was collected from the tube by the use of a swab and stored in the refrigerator immediately. All swabs were sent refrigerated in one batch to the laboratory. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. The Medical Ethical Committee of the University Medical Centre Utrecht confirmed that the Medical Research Involving Human Subjects Act did not apply for this study and that therefore an official approval by the Medical Ethical Committee was not required (Protocol no. 14–346/C). All participants gave written informed consent.
Laboratory analysis
At the day of arrival in the laboratory, all faecal specimens were analysed for the presence of ESBL-producing Enterobacteriaceae by selective plating. Samples were suspended in 10 ml LB-medium with cefotaxime (1 µg/ml) and incubated overnight at 37°C. Approximately 10 µl of each suspension was streaked on MacConkey agar plates with cefotaxime (1 µg/ml) and incubated overnight at 37°C. Individual colonies with different morphology were selected for bacterial species identification performed by MALDI-TOF MS. All isolates suspected of producing ESBLs were selected for further molecular analysis to confirm the presence of ESBL genes. DNA was isolated using DNeasy 96 Blood & Tissue Kit (Qiagen, Hilden, Germany). A β-lactamase microarray assay (Check-MDR CT101, Checkpoints, Wageningen, the Netherlands) was used to detect genes encoding carbapenemases (KPC and NDM), ESBLs (CTX-M groups 1, 2, 8/25 and 9, TEM, SHV) and AmpCs (CMY-1 /MOX, ACC, DHA, ACT/MIR, CMY-2, FOX). DNA from ESBL/AmpC microarray positive isolates was amplified and sequenced with group-specific primers to determine the exact gene type [Reference Dierikx16]. DNA sequences were interpreted with Basic Local Alignment Search Tool (National Center for Biotechnology Information). Plasmids encoding the ESBL (or AmpC) genes were determined using a transformation-based approach. Purified plasmid DNA was electro-transformed into E. coli DH10B cells (Invitrogen, Van Allen Way, CA, USA) under the following conditions: 1·25 kV/cm, 200 Ω, 25 µFar [Reference Liakopoulos17]. When strains recovered from workers were not giving transformants (workers 12 and 15), conjugation was performed using the plasmid-free rifampin-resistant E. coli E3110 as a recipient strain for liquid-mating assays in 1:1 ratio as previously described [Reference Liakopoulos17]. Transformants were selected on lysogeny broth (LB) agar supplemented with cefotaxime (1 µg/ml), whereas transconjugants on LB agar with rifampin (100 µg/ml) and cefotaxime (1 µg/ml). PBRT (PCR-based replicon typing) on the transformants and multilocus sequence typing (MLST) on the parental strains were performed to analyse genetic similarities of isolates and plasmid content [Reference Liakopoulos17].
Statistical analysis
Slaughterhouse workers were classified as ESBL positive, if any ESBL gene was detected in their faeces sample. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). The association between carriage of ESBL-producing Enterobacteriaceae in humans and job function was calculated with logistic regression analysis (proc GENMOD). For this analysis, participants were divided in two groups based on potential exposure within their specific job task. One group with an assumed higher exposure to ESBL-producing Enterobacteriaceae consisted of job tasks in the early stages of the slaughter process (e.g. stable work, stabbing, dehairing, removal of organs, i.e. direct contact with pigs and pig carcasses). The other group, assumed to have lower exposure to ESBL-producing Enterobacteriaceae, consisted of job tasks in the latter stages of the slaughter process (e.g. refrigeration, packaging, expedition, i.e. contact with meat or meat products) (see Table 1). When participants reported to have more than one job function, they were assigned to the first job function at the slaughter line. Other potential risk factors such as nationality, consumption of raw meat, hospitalization, use of antimicrobials, living on a farm, travelling, and potential confounder's age, gender and smoking were analysed univariately as well and selected for multivariate analysis when P-value was below 0·2. Model selection was performed by a backward procedure. Model fit was checked with the QIC-statistic (Quasi-likelihood under the Independence model Criterion). The final model retained variables significant at P ⩽ 0·05.
RESULTS
Human stool samples from 334 employees were obtained. Most of the participants were male and did not originate from the Netherlands (Table 2). The overall prevalence of ESBL carriage was 4·9% (16/334), and all ESBL-producing isolates were E. coli. The ESBL genes detected were bla CTX-M-1 (n = 8), bla CTX-M-15 (n = 3), bla CTX-M-27 (n = 2), bla CTX-M-24 (n = 1), bla CTX-M-55 (n = 1) and bla SHV-12 (n = 1). In addition, three other slaughterhouse workers carried an AmpC gene [bla CMY-2 (n = 2) and bla DHA-1 (n = 1)]. Plasmids encoding ESBL genes were assigned to multi-F (n = 6), I1α (n = 4), N (n = 3) and K (n = 1) replicon types. For two isolates, transformation and conjugation experiments failed to give transformants and transconjugants suggesting the chromosomal location of these genes. MLST revealed that ESBL-producing E. coli belonged to 11 different sequence types (STs), with ST131 (n = 4), ST88 (n = 3) and ST93 (n = 3) as the most frequent ones. All molecular aspects are listed in Table 3.
Ref., Reference category; NE, Not estimated due to empty cells.
* Per 1 year increase.
† Not from the Netherlands: Cape Verdean (n = 1), Ghanese (n = 1), Hungarian (n = 46), Latvian (n = 3), Polish (n = 119), Portugese (n = 12), Romanian (n = 41), Slovakian (n = 20), Turkish (n = 2).
* No successful transformation or conjugation.
ESBL carriage was not equally distributed across different jobs in the slaughterhouse. A higher prevalence of ESBL seemed to appear in workers involved in ‘removal of lungs, heart, liver, tongue’ (33%) and ‘removal of head and spinal cord’ (25%) (see Table 1). In the ‘higher exposure’ group, ten out of 95 (10·5%) were carrying ESBL vs. six out of 233 (2·6%) in the ‘lower exposure’ group. Human ESBL carriage was significantly associated with job exposure in the slaughterhouse (odds ratio (OR) 4·45, CI 1·57–12·62). Smoking was univariately and negatively borderline significantly associated with ESBL carriage (OR 0·31, CI 0·09–1·12) as well as travelling (OR 0·36, CI 0·13–1·04). None of the other potential risk factors (nationality, consumption of raw meat, hospitalization, use of antimicrobials, living on a farm) or confounders (age, gender) were significantly associated with ESBL carriage. In multivariate analysis, the effects of smoking and travelling appeared independent of job exposure (only a slight change in OR) and did not remain in the final model. Therefore the final model only contained job exposure as a risk factor for ESBL carriage.
When looking at ESBL carriage by nationality, Romanian workers had a higher prevalence (five out of 41 (12·2%)) compared with the other three well-represented nationalities: Hungarian (one out of 46 (2·2%)), Dutch (three out of 60 (5·0%)) and Polish (five out of 119 (4·2%)). However, 23 out of 41 (56·1%) of the Romanians worked in the ‘higher exposure’ group, which is a considerably higher percentage than within Hungarian, Dutch and Polish slaughterhouse workers (seven out of 46 (15·2%), nine out of 60 (15·0%) and 30 out of 119 (25·2%) respectively). When in the final model the effect of job exposure was adjusted for Romanian nationality, the association between ESBL carriage and job exposure hardly changed (OR 4·85, CI 1·54–15·24). No significant interaction between Romanian nationality and job exposure was observed for ESBL carriage.
Focusing on bla CTX-M-1 gene solely, the prevalence was 7·4% (7/95) in the ‘higher exposure’ group and 0·4% (1/233) in the other group. When analysis was restricted for bla CTX-M-1 alone, the association between ESBL carriage and exposure group was confirmed (OR 18·46, CI 2·24–152·16).
From the 74 participants who reported to have more than one job title, 11 participants mentioned job titles not within the same exposure group. This misclassification of exposure might have changed the association between exposure and ESBL carriage. When a sensitivity analysis was performed by switching these 11 participants from the ‘higher exposure’ to the ‘lower exposure’ group, the observed association between ESBL carriage and occupational exposure did not change significantly and had a similar point estimate (OR 5·36, CI 1·89–15·25).
DISCUSSION
The study revealed that slaughterhouse workers were more likely to carry ESBL when working in the early slaughtering steps (before chilling of the pig carcasses) than slaughterhouse workers working from this slaughter step forward, i.e. working in the cooling, cutting and deboning area. Overall prevalence of ESBL carriage was 4·8%, which is more or less comparable to numbers found in two recent Dutch studies in residents of Amsterdam (8·6%) and residents living in the vicinity of livestock farms (4·5%) [Reference Reuland18, Reference Wielders19]. However, the ESBL prevalence differed between the assumed higher and lower exposed groups (10·5 vs. 2·6%) pointing at higher occupational exposures during specific tasks performed by individual slaughterhouse workers.
A previous study in the same slaughterhouse showed a decline in presence and counts of Salmonella on skin surface and the exterior of carcasses from bleeding till chilling [Reference van Hoek20]. In addition, a decline in presence of E. coli during the slaughter process, especially after chilling of the pig carcasses, has been described in literature [Reference Wu14, Reference Bouvet15, Reference Lenahan21, Reference Tamplin22]. It is likely that ESBL-producing E. coli shows the same decline as E. coli in general, although we can only assume since exposure to ESBL was not measured in this study. This could explain the higher carriage of ESBL among slaughterhouse workers when involved in the slaughter steps before chilling. For some slaughter steps only low numbers of participants could be included. As a consequence, it was not possible to estimate differences in ESBL carriage in more detail. However, when focusing on the descriptive figures, ESBL carriage is considerably higher when working in the slaughter steps ‘removal of lungs, heart, liver, tongue’ and ‘removal of head and spinal cord’. Exposure to ESBL might be higher for workers in these slaughter steps, because both steps involve handling of the throat area, including the pharyngeal tonsils which are colonized with high counts of several bacteria (potentially including ESBL-producing E. coli) [Reference Lăpuşan23, Reference Lowe24].
In this study, bla CTX-M-1 was the most predominantly detected ESBL gene in pig slaughterhouse workers. Out of the ESBL-positive slaughterhouse workers, eight were carrying bla CTX-M-1 (50%), which is significantly higher than the proportions of bla CTX-M-1 reported in two Dutch studies. In ESBL-positive residents of Amsterdam and ESBL-positive residents living in the vicinity of livestock farms 26 out of 145 (18%) and 13 out of 99 (13%) carried a bla CTX-M-1 gene, respectively [Reference Reuland18, Reference Wielders19]. In addition, bla CTX-M-1 is also the most frequently found ESBL gene in slaughter pigs in the Netherlands [25]. Moreover, pig farmers were found to carry the same ESBL gene type as exclusively or predominantly detected in their pigs, of which bla CTX-M-1 was most frequently found in both of them [Reference Dohmen3]. In contrast, bla CTX-M-15 has been found mostly in humans in a clinical setting and in the Dutch human population, but not that often in livestock [Reference Ewers2, Reference Reuland18, Reference Voets26]. A great diversity was seen in STs; therefore clonal transmission between slaughterhouse workers is probably not the dominant route of transmission. Besides, half of the STs (ST10, ST38, ST88, ST93, ST354) have been previously associated with E. coli from pig origin [Reference Ramos8, Reference Garcia-Cobos27, Reference Herrero-Fresno, Larsen and Olsen28]. In addition, IncN and IncI1 plasmids are known to play a role in the dissemination of bla CTX-M-1 in E. coli in livestock, including pigs [Reference Dolejska29–Reference Zurfluh31]. Although no samples from carcasses or slaughter products were obtained, these results regarding ESBL gene types, plasmids and STs, suggesting transmission from animals to humans.
Besides occupational exposure, none of the analysed determinants (nationality, consumption of raw meat, hospitalization, use of antimicrobials, living on a farm, travelling, age, gender and smoking) were found to be risk factors for ESBL carriage in slaughterhouse workers. ESBL carriage was higher in slaughterhouse workers with a Romanian nationality compared with other nationalities, although the association between ESBL carriage and occupational exposure remained when adjusting for Romanian nationality. Besides that, the effect of being in the ‘higher exposure’ group is similar for Romanian slaughterhouse workers as for the other nationalities, considering the non-significant interaction. Also, in three out of five Romanian ESBL carriers the ESBL gene type detected was bla CTX-M-1. Smoking and travelling were protective factors for ESBL carriage univariately, although they did not retain in the final model. From the 173 non-Dutch slaughterhouse workers who reported their travelling destiny, 88% travelled to their country of origin. Although travelling frequencies were not assessed, travelling might be interpreted as long distance commuting. To our knowledge, there is no explanation described in literature for any effect of smoking on ESBL carriage.
Although ESBL carriage was detected in slaughterhouse workers, the risk of acquiring ESBL-producing E. coli when handling meat intensively seems to be limited. When this is extrapolated to consumers, exposure to ESBL in pork through handling might not be of a high public health concern. In combination with environmental exposure data, this information can be used in formal risk assessments. Although not addressed in this study design, duration and persistence of ESBL carriage can vary greatly [Reference Tham32, Reference Birgand33]. Duration of ESBL carriage is not only important for assessing the risk of infection, but for the risk of transmission of ESBLs from slaughterhouse workers into the general population as well.
The results suggest that the overall carriage in slaughterhouse workers is comparable with the Dutch population, but the proportion of the bla CTX-M-1 gene (commonly found in livestock) is higher. Differences exist between slaughterhouse workers depending on their job tasks. A higher prevalence of ESBL carriage was found in slaughterhouse workers working before the chilling process of the carcasses compared to workers in the cooling area, cutting and deboning departments.
ACKNOWLEDGEMENTS
The authors would like to thank all participated slaughterhouse workers and other involved employees from Vion in Boxtel for participation and all field workers and laboratory workers for assistance. The authors would also like to thank Alejandro Dorado-Garcia for assistance on data analysis. This work was supported by the 1Health4Food (1H4F) project under the ESBLAT consortium (TKI-AF-12067); and by the European Commission, 7th Framework Programme for Research, technological Development and Demonstration (FP7/2007-2013–613754).
Preliminary results from this study were presented in an oral presentation at the 26th Conference on Epidemiology in Occupational Health in Barcelona, Spain, 2016 (abstract 121).
DECLARATION OF INTEREST
None.