https://orcid.org/0009-0004-0257-7732
Figures
Abstract
Campylobacter and non-typhoidal Salmonella (NTS) are among the most common foodborne pathogens found in chickens at any production stage and cause gastroenteritis in humans. This study aimed to estimate the prevalence of Campylobacter spp. (C. coli and C. jejuni) and NTS in broiler production and distribution networks (PDNs) using a Bayesian approach. It also investigated the NTS serotypes circulating in these PDNs. A cross-sectional study was conducted in four provinces in northern Vietnam between March 2021 and March 2022. A total of 102 sites, including live bird markets, slaughter facilities (slaughterhouses and slaughter points), and their supplying farms, were randomly selected for sampling. Cecal and environmental samples were cultured for isolation of Campylobacter and NTS, with serotypes of NTS determined by targeted analysis of whole genome sequences. Bayesian models were developed to estimate the prevalence of Campylobacter at two levels (bird-level and site-level) and NTS at site-level. The selected best-fitted models indicated that C. jejuni prevalence was primarily influenced by site type, while C. coli was affected by both province and site types. For NTS, only site type was included. The highest overall prevalence of infected broilers was estimated on farms for C. coli (26.2% [95% High Density Interval (HDI): 19.0-36.0%]) and C. jejuni (19.9% [95%HDI 13.0-27.0%]). Slaughter points (97.6% [95%HDI 63.3-99.9%]) and wholesale markets (91.7% [95%HDI 28.2-99.9%]) had the highest probability of C. coli and C. jejuni contamination, respectively, but retail markets had the highest proportion of infected broilers at contaminated sites. NTS contamination was more frequent in markets and slaughter facilities (42.8% [95%HDI 30.8-57.1%]) than on farms (18.6% [95%HDI 9.5-30.1%]). Among 16 detected NTS serotypes, S. Infantis and S. Kentucky were the most common. These findings highlight the widespread contamination of broiler PDNs with Campylobacter and NTS in northern Vietnam, emphasizing the need for enhanced surveillance and control measures in PDNs to mitigate the risk of foodborne transmission.
Author summary
Campylobacter spp. and non-typhoidal Salmonella (NTS) are enteric pathogens that are key global pathogens causing diarrheal diseases. The Campylobacter species that are most frequently responsible for human diseases (campylobacteriosis) are Campylobacter jejuni (C. jejuni) and Campylobacter coli (C. coli). The severity of the salmonellosis depends on host factors and the serotype of NTS. Poultry and poultry products are recognized as major sources of NTS and Campylobacter infections in humans. In Vietnam, these pathogens have been detected in the broiler production and distribution networks (PDN). Factors that may increase the risk of transmission of those pathogens include low biosecurity practices, inadequate waste management, and high farm density. In this work, we estimated the prevalence of Campylobacter spp. (C. coli and C. jejuni) and NTS in PDN markets and slaughterhouses and farms supplying these facilities in four provinces in northern Vietnam. We estimated a high prevalence of Campylobacter contamination in slaughter points and wholesale markets and higher NTS contamination in slaughter facilities and markets than in farms. S. Infantis was one of the most frequently identified serovars. However, it was not reported as being frequently detected in humans in Vietnam. Therefore, serovar may vary depending on the source and region of contamination. Our results help us look at levels of infection/contamination with Campylobacter and NTS in the broiler PDNs, thus informing stakeholders to prevent and control campylobacteriosis and salmonellosis.
Citation: Xin C, Hoang TT, Truong DQ, Tran NT, Pham HTT, Dang STT, et al. (2025) Prevalence of Campylobacter and non-typhoidal Salmonella along broiler chicken production and distribution networks, Northern Vietnam. PLoS Negl Trop Dis 19(11): e0013615. https://doi.org/10.1371/journal.pntd.0013615
Editor: Francesca Schiaffino, University of Virginia, UNITED STATES OF AMERICA
Received: March 11, 2025; Accepted: October 1, 2025; Published: November 5, 2025
Copyright: © 2025 Xin et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data supporting the findings are included in the manuscript and S1 Data file.
Funding: This study was funded by the UKRI GCRF One Health Poultry Hub (Grant No. BB/S011269/1), one of twelve interdisciplinary research hubs funded under the UK government’s Grand Challenge Research Fund Interdisciplinary Research Hub initiative, awarded to a consortium led by FT and including GF, DB, DP, NPT, RS and PLC. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Campylobacter spp. and non-typhoidal Salmonella (NTS) are major bacterial pathogens causing intestinal infections and gastroenteritis in humans, accounting for an estimated 80,000 deaths globally in 2010 [1]. Food-producing animals, particularly poultry, serve as asymptomatic reservoirs for these pathogens [2]. Campylobacter spp. can colonize and persist in the intestines of apparently healthy birds [3]. Campylobacter jejuni (C. jejuni) is the most frequently isolated species in human bacterial foodborne illness, accounting for around 90% of reported infections, while Campylobacter coli (C. coli) is responsible for most of the remainder [4,5]. Similarly, NTS, which refers to Salmonella enterica serovars other than S. Typhi and S. Paratyphi [6], is frequently detected in poultry gastrointestinal tracts [2,7]. Poultry and poultry products are recognized as major sources of NTS and Campylobacter infections in humans [8]: broiler meat is considered the main food-borne source of human campylobacteriosis [9], while S. Enteritidis, the NTS serovar most frequently associated with salmonellosis, is primarily linked to consumption of contaminated eggs and poultry meat [9,10]. Additionally, human NTS infections may also be acquired through contaminated environments and water [11].
In Vietnam, poultry production has undergone rapid intensification over the last two decades [12], leading to the emergence of heterogenous production and distribution networks (PDNs) [13,14], in which poultry are bred, raised, traded, and consumed. Most of Vietnam’s broiler chicken meat production is sourced from slow-growing broilers, also known as colored broilers, which account for an estimated 72% of the total broiler population [15]. While fast-growing industrial broiler breeds are typically raised for 30–45 days, slow-growing broilers are hybrids of industrial and local breeds that are raised for 70 days to several months within diverse farming systems. Their distribution to consumers involves multiple intermediaries operating at a variety of sites, from farms to PDN endpoints, with the latter referring to sites where chickens are slaughtered or last traded alive (e.g., slaughter facilities, and live bird markets) [15]. These factors, compounded by low biosecurity practices, inadequate waste management, and high farm density, may promote the transmission of pathogens, including Campylobacter and NTS [16–19]. Understanding how the prevalence of Campylobacter and NTS varies along the PDN is essential to inform the development of targeted surveillance programs and effective risk mitigation interventions.
In Vietnam, a previous study conducted in the Mekong Delta region (southern Vietnam) estimated a prevalence of C. jejuni (28.4%) in farmed chickens [20]. In Ha Noi, a high proportion (74%) of retail chicken meat samples were found to be contaminated with Campylobacter, predominantly C. coli [21]. Similarly, NTS was also highly prevalent (45.9%) in chicken carcasses from retail markets in Vietnam [22]. Although Campylobacter and NTS have been detected at multiple stages of Vietnamese poultry PDNs [23–25], there is a lack of studies that assess pathogen prevalences in the diverse types of endpoints and their supplying farms.
Therefore, our objective was to estimate the prevalence of Campylobacter spp. (C. coli and C. jejuni) and NTS in farms and at the PDN endpoints they supply in four provinces in northern Vietnam. NTS serotypes circulating in these PDNs were also identified.
Materials and methods
Ethic statement
This study was approved by the National Institute of Veterinary Research (Vietnam) (020–433/DD-YTCC) and the Royal Veterinary College (UK) Ethics and Welfare Committee (URN: 2020 1983–3). All participants informedly agreed to participate in the study. Oral consent was obtained from the farmer to sample their chickens.
Study area and study design
A cross-sectional study was conducted in four provinces in northern Vietnam between March 2021 and March 2022. Northern Vietnam accounted for 36.5% of the country’s human population in 2020 [26] and around 67.3% of the poultry population [12,27]. Ha Noi, Bac Giang, and Hai Duong are among the most densely populated provinces both in terms of humans and poultry in northern Vietnam [26,27], while Quang Ninh is a bordering province with China (Fig 1).
The project aimed to assess the transmission of pathogens between farms and PDN endpoints; the sample size needed to be defined based on the expected size of targeted bacterial populations, which were unknown. Number of sites to select was set to sample chickens as widely as possible. Fast- and slow-growing broilers were sampled on 50 farms and 52 supplied PDN endpoints (S1 Table), which included markets and slaughter facilities. Markets were defined as authorized open spaces where at least two vendors operate at least once a week. They were classified as ‘wholesale’ if some vendors sold chickens to other traders who resold them elsewhere, and ‘retail’ if all chickens were sold directly to consumers and/or restaurants and caterers. Slaughter facilities were classified as slaughterhouses and slaughter points. Slaughterhouses have automated or partially automated processing chains capable of handling large numbers of chickens. Slaughter points involve manually preparing dozens to hundreds of chickens daily with or without plucking machines.
Site selection
For endpoints’ selection, a stratified cluster sampling design was adopted, with a stratification by provinces. First inclusion criterion was set as endpoints supplying urban areas. Retail markets and slaughter points fitting this criterion were located in urban areas or around urban centers (live bird markets are prohibited in Ha Noi’s urban center), while wholesale markets and slaughterhouses were located in peri-urban or rural areas. Meanwhile, slaughter points needed to process at least 30 chickens a day to ensure access to five chickens to sample. In the four provinces, lists of eligible endpoints (markets and slaughter facilities located in urban and peri-urban areas) were initially compiled through consultation with province and district-level Department of Animal Health officers, and stakeholder interviews in visited markets and slaughter facilities. Finally, due to their small numbers, all eligible wholesale markets (N = 11) and slaughterhouses (N = 6) that processed slow-growing broiler chickens in the four study provinces were recruited. Retail markets (N = 16) and slaughter points (N = 19) were selected through stratified cluster sampling: four to five wards (i.e., urban sub-districts) were randomly selected from each province. In each ward, one retail and one slaughter point were chosen based on the presence of slow-growing broilers. Fast-growing broilers were sampled when available on the site (S1 Table).
Endpoint supply areas, i.e., the provinces where supplying farms were located, were identified through structured interviews of vendors and slaughter facility managers. All farms located within the supply areas and raising at least 100 slow-growing broilers to a late production cycle stage (i.e., ≥ 70 days old) were eligible for inclusion and listed with the assistance of district-level Department of Animal Health officers. To select a paired supplying farm for each endpoint, one commune (i.e., peri-urban or rural sub-district) within the supply areas was randomly selected, and then, a single eligible farm was randomly chosen in that commune.
Chicken and environment sampling
All selected sites were visited, and oral consent to participate was obtained from the owner or manager of each site. GPS coordinates of sites were recorded, and standardized structured questionnaires were utilized. At each farm and slaughter facility, five slow-growing broilers were randomly selected and purchased for terminal sampling (sampling after euthanasia). In markets, five stalls were randomly selected, and one slow-growing broiler was purchased from each for terminal sampling. When available, five fast-growing broiler chickens were sampled as well. Cecal contents were collected from chicken carcasses and stored at 4°C in a cool box. Environmental samples were collected from each farm using boot socks or overshoes in the broilers’ living areas. At each endpoint, samples were obtained by collecting fecal material from the bottom of cages or enclosures where terminally sampled chickens were kept, using an inverted plastic food bag to grab a handful of fecal material from each area. After sampling, each sock, overshoe, or bag was placed into a plastic bag or sampling bottle containing 140 ml of brain heart infusion broth and stored upright in a cool box for further processing. All samples were transported and then processed immediately upon arrival at the laboratory of the National Institute of Veterinary Research (NIVR). The time from collection to processing was within 24 hours.
Isolation and identification of Campylobacter spp. and NTS
Cecal content samples were processed following standard protocols for Campylobacter isolation and identification (ISO 10272–1: 2017). Briefly, one loop (10 μl volume) of cecal content was directly placed into Preston agar (Oxoid, UK) and modified charcoal-cefoperazone-deoxycholate agar (mCCDA, Oxoid, UK), then incubated for 44 ± 4 h at 41.5 ± 1 °C in a gas jar with microaerophilic conditions using CampyGen (Oxoid, UK). Five colonies from these plates were screened by Gram-stain and wet mount for typical Campylobacter morphology and motility. Typical colonies were subjected to physiological and polymerase chain reaction (PCR) confirmation. For physiological confirmation, after sub-culturing isolates on Columbia sheep blood agar (Oxoid, UK), oxidase, catalase, and hippurate tests were applied [29]. DNA extracts prepared from fresh plate cultures using the Gene JET Genomic DNA Purification Kit (Thermo, USA). Genus and species identification was verified by PCR using primers listed in Table 1, as previously described [30].
For NTS, environmental samples were pre-enriched into brain-heart infusion broth (BHI, Merck, Germany) by incubation at 37°C for 16–18 h. Then, samples were selectively enriched in modified semisolid Rappaport Vassiliadis (MSRV, Merck, Germany) agar with incubation at 41.5°C for 24 to 48h. Positive cultures were streaked onto plates of xylose lysine deoxycholate (XLD, Merck, Germany) agar and incubated at 37°C for 24h. The isolates that showed growth with black or center-black coloring were presumptively identified as Salmonella spp. Genomic DNA was extracted using QIAamp DNA Mini Kit (Qiagen, Germany), following the manufacturer’s protocol. DNA concentration was quantified using the Qubit dsDNA High Sensitivity (HS) Assay Kit (Thermo Fisher Scientific) prior to library preparation. Confirmation of NTS identity was performed using PCR with primers ST11 and ST15 (Table 1).
Serotyping of the isolated NTS
Library preparation was performed according to the Illumina TruSeq Nano DNA sample preparation protocol as previously described [31–33]. In brief, DNA samples were sheared to ~450 bp using a Covaris E220 ultrasonicator, following the manufacturer’s guidelines. Libraries were uniquely tagged with Illumina’s IDT for Illumina-TruSeq DNA Unique Dual Indexes to enable multiplexing during sequencing. Finished libraries were quantified using Promega’s QuantiFluor dsDNA assay, and the average library size was determined on an Agilent Tapestation 4200 (Agilent, USA). Library concentrations were normalized to 4 nM and validated by qPCR on a QuantStudio-3 real-time PCR system (Applied Biosystems) using the Kapa library quantification kit for Illumina platforms as recommended by the manufacturer (Kapa Biosystems). Equimolar pools of the prepared libraries were sequenced on the Illumina HiSeq X platform at a read length of 150 bp paired-end.
Raw read quality filtering was performed using Cutadapt (version 2.10) to remove adapter sequences and low-quality bases. The filtering parameters were set at an error rate of 0.1 (-e 0.1), a quality cutoff of 20 (-q 20), and a minimum read length of 30 (--minimum-length 30). Trimmed reads were then de novo assembled using SPAdes (version 3.15.3) to generate draft genomes. The assembled genomes were analyzed using tools from the Centre for Genomic Epidemiology (CGE) to characterize various genomic features. Serotyping of Salmonella isolates was then determined using SeqSero2. The analytical approach used in this study follows established WGS-based methodologies for foodborne pathogen characterization [31–33].
Bayesian analysis
Model description.
Bayesian logistic regression models were developed to estimate the prevalence of Campylobacter at two levels (bird-level and site-level) and NTS at site-level. Separate models were specified for C. coli, C. jejuni, and NTS. Specifically, we developed two-level Bayesian hierarchical logistic regression models for both C. coli and C. jejuni, treating the site as a random effect. In each model, bird- and site-level factors were integrated to assess the risk of infection in broilers. The contamination status of a site follows a Bernoulli distribution, with the probability of a site being contaminated depending on its type. Then, the probability of a broiler testing positive in a contaminated site was formulated using logistic regression with a site-specific random intercept to account for unobserved heterogeneity at each site and explanatory fixed variables: chicken type, provincial location, and site type. As one environmental sample was collected at each site to test for the presence of NTS, the probability of a site being contaminated with NTS was modeled using logistic regression, with provincial location and site type as explanatory variables. Chicken type was not included as a variable because some broiler compositions were represented by only a small number of sites, with two sites exclusively housing fast-growing broilers. Site types were differentiated into (i) farm and endpoint, (ii) farm, market, and slaughter facility, or (iii) farm, retail market, wholesale market, slaughter point, and slaughterhouse. Chicken type was distinguished between slow- and fast-growing broilers. For NTS models, the provincial location referred to the location of the sampling site, as the contamination levels at these sites might vary across different provinces. In contrast, for Campylobacter models, the provincial location could refer to either the location of the site or the province of the supply area. This distinction is important because the province of the supplying farms could influence the prevalence of Campylobacter spp. in broilers, which were transported along different levels of the PDNs. The province of the supply area was not included in the NTS model due to the limited number of observations. Site locations were Ha Noi, Bac Giang, and Hai Duong/Quang Ninh, as most sites in Hai Duong and Quang Ninh were located close to their shared border (Fig 1), and supply areas were Ha Noi, Bac Giang, Hai Duong/Quang Ninh, and other provinces. Note that each site supply area was a dichotomous explanatory variable as a given site could be supplied by several of these areas. Weakly informative priors were specified: Beta(1,1) for the probability of a site being contaminated, Normal(0, 4) for the fixed regression coefficients, and an InverseGamma(1, 1) for the variance of the random intercept.
Model development
Models with all combinations of explanatory variables were fitted (N = 12 each for C. coli and C. jejuni models, and N = 5 for the NTS model). Provincial location was only considered in models for which all site types (farm, retail market, wholesale market, slaughterhouse, and slaughter point) were accounted for, as these types of endpoints and farms were heterogeneously distributed across the four provinces. Models were run using Markov Chain Monte Carlo simulation (MCMC) in JAGS (Just Another Gibbs Sampler) [34] and R.4.4.2 [35]. For each model, we ran four separate chains with an initial burn-in period of 10,000. The Deviance Information Criterion (DIC) was computed for each model. A model with a lower DIC value was considered a better fit and better supported than others if the difference in DIC scores was at least five units (∆DIC ≥ 5). Convergence and mixing within the Markov chain were assessed by examining trace plots. All model effective sample sizes were also higher than 1000, given that a random intercept was considered, which is sufficient for stable estimates [36], and the Gelman-Rubin convergency statistics were rounded to or less than 1.01, with successful convergence based on Rhat values (all < 1.1) [37]. Pair plots of the posterior distributions were generated, enabling us to assess parameter correlations and understand the distribution of each parameter’s posterior values. Moreover, posterior predictive checks (PPCs) were created to assess the capability of the models to replicate observed data patterns. Finally, we conducted Bayesian retrospective power analyses for Campylobacter and NTS models to evaluate the models’ ability to detect differences of 10% in prevalence and contamination probability between the different explanatory variables. The HDI of the estimated power was also computed [38].
Results
Descriptive results for C. coli and C. jejuni
A total of 500 slow-growing broilers and 65 fast-growing broilers were sampled in 102 sites from four provinces. Among these, 146 (25.8%) tested positive for C. coli and 101 (17.9%) for C. jejuni (Table 2). Campylobacter coli was most prevalent in slaughterhouses (n = 18, 45.0%), while C. jejuni was most prevalent on farms (n = 56, 22.4%). Conversely, wholesale markets had the lowest proportion of broilers positive for C. coli (n = 8, 11.4%), and no broilers from slaughterhouses tested positive for C. jejuni (n = 0, 0%).
Notably, the site location associated with the slightly highest proportion of positive broilers was Ha Noi province for C. coli (n = 67, 29.1%) and Hai Duong/Quang Ninh provinces for C. jejuni (n = 32, 19.4%), as shown in Table 3.
Descriptive results and serotypes of NTS
A total of 102 environmental samples were collected at 102 sites (Table 4), with 32 (31.4%) testing positive for NTS, representing 16 distinct serotypes (Table 5). The highest occurrence of NTS was in environmental samples collected from wholesale markets (n = 6, 54.5%), and the lowest in farm samples (n = 9, 18%). The most frequently detected serotypes were S. Infantis and S. Kentucky, with S. Infantis identified in two markets and two slaughter facilities and S. Kentucky in two farms and two slaughter facilities. Salmonella Agona, S. Albany/Duesseldorf, and S. London were each detected in three sites, and other serovars were detected at either two sites or just one site.
Bayesian model results of C. coli and C. jejuni
Best models and post-predictive checks.
Twelve models were specified for C. coli and C. jejuni (Table 6). Convergence was successfully achieved for all models, with Gelman-Rubin convergency statistics rounded to or below 1.01. The trace (n chains = 4, n iterations = 50000) of these two selected models (one for C. coli and one for C. jejuni) displayed tight, consistent horizontal bands, indicating no visual signs of non-convergence within the chains (S1 and S2 Figs). Pairs plots for parameters of two selected C. coli and C. jejuni models sampled via the MCMC algorithm from four chains are shown in S3 and S4 Figs. The selected C. coli model differentiated for all site types (farm, retail, wholesale, slaughterhouse, and slaughter point) and accounted for supply areas (Ha Noi, Bac Giang, Hai Duong/Quang Ninh, and other provinces). This model was 4.2 DIC units lower than the model that used site provincial locations instead of supply areas while maintaining similar site-type coefficient estimates. A model that included chicken type had a slightly lower DIC (∆DIC = 0.7), but this additional parameter value was not significantly different from zero, and other parameter estimates remained consistent with the selected model. The posterior predictive checks of the selected model (S5 Fig) revealed a notable discrepancy for slaughterhouses, where the observed proportion of chickens positive for C. coli (42.6%) far exceeded the model’s predicted median (9.8% [95%High Density Interval (HDI): 2.5-31.6%]). For C. jejuni models, the models differentiating all site types achieved DIC scores at least 12.7 DIC units lower than any other models with a different site-type specification. The selected model (Table 6) also excluded chicken type and provincial location, as the inclusion of these variables did not substantially improve the model fit, and did not alter site-level parameter estimates. The PPCs of the selected C. jejuni model show that posterior predictions for wholesale markets underestimated the observed proportion of C. jejuni-positive broilers (predicted: 4.1% [95% HDI 1.0–13.3%]; observed: 13.2%) (S6 Fig).
Posterior prevalence estimate
The estimated prevalences are reported in Tables 7 and 8. C. coli had the highest overall prevalence of positive broilers on farms (Mode 26.2% [95% HDI 19.0-36.0%]), while C. jejuni showed the highest prevalence on farms (Mode 19.9% [95% HDI 13.0-27.0%]) and in retail markets (Mode 19.8% [95% HDI 9.0-34.0%]). For C. jejuni, slaughter points and wholesale markets had the highest levels of site contamination, but the proportion of positive broilers at these contaminated sites was much lower than on farms and in retail markets, where nearly one in three broilers tested positive when the site was contaminated. For C. coli, nearly all farms were expected to be contaminated, and the proportion of positive broilers at contaminated sites showed less variability than for C. jejuni, with retail markets having the highest prevalence, comparable to that of C. jejuni. Both C. coli and C. jejuni prevalences were substantially lower in slaughterhouses compared to other endpoint types. For C. coli, having broilers supplied from Ha Noi (OR 1.3 [95% HDI 0.4-2.8]) and Hai Duong/Quang Ninh (OR 1.4 [95% HDI 0.4-3.1]) increased the likelihood of broilers testing positive more than supplies from Bac Giang (OR 0.5, [95% HDI 0.2-1.1]) and other provinces (OR 0.7, [95% HDI 0.2-1.6]).
The ability of the model to detect meaningful differences in Campylobacter prevalence was estimated to have a power of 68% (95% HDI 59–77%) for farm vs endpoints at the site level, 35% (95% HDI 26–45%) at the bird level, 28% (95% HDI 20–75%) for slow-growing vs. fast-growing broiler and 61% (95% HDI 51–70%) for province.
Bayesian model results of NTS
A total of five models were examined for NTS (Table 9). Convergence was successfully achieved for all models, with Gelman-Rubin convergency statistics below 1.01. The trace (n chains = 4, n iterations = 50000) showing consistent horizontal bands, and pairs plots for parameters of the selected NTS model are shown in S7 and S8 Figs. The selected NTS model (Table 9, highlighted with a star symbol) differentiated between farms and endpoints and had 7.6 DIC units lower than a model that included the provincial location. A model that categorized sites into three types (farm, market, and slaughter facility) had a slightly higher DIC score (∆DIC = 1.3) and did not improve model fit, as the posterior predicted prevalence in markets and slaughter facilities remained similar and was equivalent to the predicted prevalence for endpoints in the two-category model. In this selected NTS model, no discrepancy was observed in the post-predictive check of the Bayesian model for NTS infection (S9 Fig).
The environment of markets and slaughterhouses (Mode 42.8% [95% HDI 30.8-57.1%]) was more likely to be contaminated by NTS than farms (Mode 18.6% [95% HDI 9.5-30.1%]) (Table 10). The estimated power of the NTS model to detect meaningful differences of contamination between farms and endpoints was 79% (95% HDI 70–86%).
Discussion
Our study shows that prevalence of Campylobacter spp. in broilers, and NTS environmental contamination of broiler facilities, are both high in northern Vietnam. While the prevalence and contamination vary between sites, it is noted that both pathogens were detected in all types of sites, namely farms, markets, and slaughter facilities. These results are consistent with previous studies in southern Vietnam [20,39] or Ha Noi’s markets [21,40]. For example, in Ha Noi a high prevalence of Campylobacter spp. (41.1%) was reported on fresh chicken carcasses at retail markets, with the predominant species being C. coli, followed by C. jejuni [40]. Another study from 2012 in the Mekong Delta region (southern Vietnam) reported a high prevalence of NTS colonization (45.6%) in 204 backyard chicken farms [39]. In our study, the most commonly identified NTS serotype was S. Infantis, one of the Salmonella enterica serovars associated with infections in both humans and broilers [41]. It was one of the top five serotypes in NTS human infections in Europe in 2017 [42]. In contrast, S. Infantis has not been reported as being frequently detected in humans in Vietnam; the serovar most frequently isolated in non-invasive NTS human disease was S. Typhimurium [43], which was not the case in our findings. Similarly, S. Weltevreden, the second most detected serovar in humans [43], was not as prevalent in our study either. This may be explained by variations in contamination sources across different regions. For example, S. Typhimurium was more common in pork production in Vietnam [44]; this could explain its high detection in humans (high consumption of pork in Vietnam). To investigate these discrepancies, surveillance of the NTS with serotyping of the strain in humans and poultry would be beneficial in identifying the sources of the contamination. Moreover, our overall results raise concerns about the spread of foodborne bacterial pathogens and their impact on food safety.
Notably, we have found the overall prevalence of Campylobacter infection in broilers to be generally high in farms, and retail markets, with the highest prevalence estimated to be on farms for C. jejuni (19.9%) and C. coli (26.2%). The higher prevalence on farms may be attributed to potential contamination from suppliers, such as a high prevalence of Campylobacter in the supplied day-old chicks. The typical timing of C. jejuni colonization in chickens occurs when they are approximately seven days old [45]. Another possible cause of Campylobacter spp. could be low biosecurity in these farms and the proximity of other livestock production. An earlier study in the Mekong Delta of Vietnam reported that it was common for farms to have a combination of animal species, with 35% of pig farms also raising chickens and 18% of poultry farms keeping pigs. In these settings, C. jejuni was the most common species detected in chickens (20%) in poultry farms with low biosecurity and mixed-species farming, contributing to its spread [20].
The highest prevalence of positive broilers of C. jejuni (37.7%) and C. coli (35.8%) in contaminated sites were estimated at retail markets. Similarly, in a study conducted in 2005 in Ha Noi, Vietnam, high contamination rates were observed in retail chicken products, with the most frequently isolated Campylobacter being C. jejuni (45.2%), followed by C. coli (25.8%) in retail fresh chicken carcasses [46]. Broilers at retail markets may have spent longer periods being transported through the PDN compared to other endpoints. For instance, slow-growing broiler chickens in Vietnam are first collected from different farms in multiple locations by mobile traders with the assistance of middlemen, who later sell these broilers to other traders in local wholesale live bird markets, where broilers are further sold to retailers [15]. This highly connected transport and longer broiler transportation process may promote the transmission of pathogens (Campylobacter spp.) within and between different network sites through inadequate sanitized transport vehicles and cages, as well as through traders and intermediaries. Extended transit periods and longer transport distances can also increase transport-induced stress in broilers, enhancing the growth and shedding of Campylobacter spp. [47]. Meanwhile, broilers from multiple origins are generally mixed in retail markets. Infected broilers can introduce pathogens into the retail market population and contaminate environments, increasing the risk of cross-contamination. Therefore, it’s important to regularly and properly clean and disinfect retail markets, as well as the transport material used for broiler distribution.
Another key finding was the high proportion of C. coli (97.6%) and C. jejuni (87.7%) contaminated slaughter points. An earlier study from Senegal also found a high prevalence (64.4%) of C. jejuni among four Campylobacter species in chicken neck‐skin samples at points of slaughter, where 95% of the slaughtering was directly conducted by the seller [48]. At slaughter points, live chickens are kept and slaughtered in the same or adjacent areas, making contaminated surfaces a potential source of infection for other live chickens [49]. In contrast, in slaughterhouses, the lairage area is physically separate from the slaughter line, decreasing chances for surface contamination during evisceration, which could prevent cross-contamination and further reduce Campylobacter infection in broilers. Moreover, the waiting time at the slaughter point may be longer with a mix of chicken sources (as for retail markets), increasing the contact between contaminated environments, infected birds, and susceptible birds. Conversely, broilers at slaughterhouses are produced on contract farms with stricter biosecurity measures and are sent directly to the companies’ slaughterhouses for processing, making their transport time in PDNs shorter than for slaughter points. Therefore, broilers are less likely to get Campylobacter infection in slaughterhouses. However, the extremely low prevalence of C. jejuni infection may also stem from other factors, such as the small number of slaughterhouses sampled (N = 6). Nevertheless, good slaughtering practices should be emphasized in all slaughter facilities, and the time broilers are held should be minimized.
The proportion of NTS contamination in markets and slaughter facilities (42.8%) was higher than in farms (18.6%). One study conducted between 2014 and 2015 in wet markets in Penang and Perlis, Malaysia, also found that Salmonella was highly prevalent (100%) in poultry processing environments in wet markets and small-scale processing plants [50]. One potential explanation for the higher prevalence of NTS contamination in endpoints may be associated with poor hygiene practices and insufficient sanitation at markets and slaughter facilities. These conditions could provide favorable environments that serve as reservoirs for NTS and facilitate their growth and spread. Furthermore, the stress associated with transporting poultry before slaughter has been shown to also increase NTS populations in both the intestinal tract and on the exterior of the carcass [51], ultimately contributing to the environmental contamination of markets and slaughter facilities. Thus, measures must be taken to reduce stress of broilers during transportation.
Our study has some limitations. Firstly, the modeling approaches differed between Campylobacter spp. and NTS, primarily because detecting NTS at the individual level is challenging [52]. Therefore, it is not possible to accurately assess the individual prevalence of NTS in broilers, nor to compare Campylobacter prevalence and NTS contamination probability. Secondly, our model did not account for seasonal variations (autumn, winter, spring, and summer) as a risk factor. Elevated temperatures in summer create a favorable environment that significantly boosts bacterial growth. Our sampling period lasted 12 months, covering all four seasons from March 2021 to March 2022. However, this seasonal effect may not act as a confounder as all types of sites were sampled during all seasons. Additionally, other farm-level and endpoint-level risk factors were not assessed, given their heterogeneous nature across farms and endpoints. Thirdly, our site selection process may have introduced selection bias because we lacked an exhaustive list of endpoints, particularly retail markets and slaughter points. We addressed this by conducting snowball sampling interviews to identify additional eligible endpoints; however, some bias remain possible. Moreover, we collected samples from only six slaughterhouses and eleven wholesale markets. This limitation stems from the small number of such facilities in the study area. Given the small sample size, our estimation, based on the Bayesian framework with weakly informative priors, might introduce limited bias in posterior estimates [38] and led to a relatively low power for some variables. The absence of a difference between type of birds may then be due to this low statistical power. However, despite these limitations, the strength of this study lies in the Bayesian approach, which integrates prior knowledge with observed data, as well as using probability distributions instead of single-point estimates to maintain uncertainties and variations in the model parameters and allows for continuous improvement of the model’s accuracy through the integration of new data [38]. Also, the observed prevalence and contamination probabilities may be underestimated. False negatives can occur due to failure in amplification during PCR or qPCR assays, even under optimal conditions. We minimized this risk by ensuring short sample transport times and validating PCR protocols rigorously. Finally, this study did not include human health surveillance data to directly link the observed pathogen contamination in chickens with human disease burden in Vietnam. However, given that contaminated chicken products are a known major source of human Campylobacter and NTS infections [2], our findings provide valuable information on the prevalence and contamination levels of foodborne pathogens in the chicken food system.
In summary, our study reflects the epidemiological situation of foodborne pathogens in different network sites in PDNs in northern Vietnam, indicating a high prevalence of Campylobacter contamination in slaughter points and wholesale markets. We estimate that markets and slaughter facilities were more likely to be contaminated by NTS than farm environments, with a high diversity of NTS serovars across different sites. These findings highlight the need for integrated intervention measures to enhance food safety, particularly in farms, retail markets, and slaughter points. Interventions to control risk factors associated with broiler transportation along the PDNs are needed. Such interventions could include developing and enforcing strict hygiene and sanitation protocols with regulations for markets where broilers from different geographical origins are mixed and could, therefore, be hotspots of disease transmission. Transport vehicles, cages, traders, and intermediaries, which all pose potential fomite risk, should be regularly sanitized and monitored to help prevent the spread of infections. Continuous monitoring of foodborne pathogens and investigation of associated risk factors is crucial to improving poultry production in Vietnam.
Supporting information
S1 Table. Number of selected sites and the number of chickens sampled by different site types.
https://doi.org/10.1371/journal.pntd.0013615.s001
(DOCX)
S1 Fig. Trace plots with four chains for the selected ideal C. coli model.
This figure shows the trace plots (n chains = 4, n iteration = 50000) of the selected C. coli model, which displayed tight, consistent horizontal bands, indicating no visual signs of non-convergence within the chains. The parameters p_site_1, p_site_2, p_site_3, p_site_4, and p_site_5 represent the C. coli contamination probabilities for farms, retail markets, wholesale markets, slaughterhouses, and slaughter points, respectively. The parameter intercept_bird represents the baseline probability of a broiler testing positive for C. coli at a contaminated site, assuming no effect from site type and province of supply areas. The fixed-effect coefficients beta_ret, beta_slh, beta_slp, and beta_who quantify the effects of retail markets, slaughterhouses, slaughter points, and wholesale markets on the probability of a broiler testing positive for C. coli at a contaminated site, respectively. The fixed-effect coefficients beta_cbac, beta_chdqn, beta_chon, and beta_cot quantify the effects of supply areas—Bac Giang, Hai Duong/Quang Ninh, Ha Noi, and other provinces—on the probability of a broiler testing positive for C. coli at a contaminated site, respectively. The parameter sigma_1 means the variance of the random intercept.
https://doi.org/10.1371/journal.pntd.0013615.s002
(TIFF)
S2 Fig. Trace plots with four chains for the selected ideal C. jejuni model.
This figure shows the trace plots (n chains = 4, n iteration = 50000) of the selected C. jejuni model, which displayed tight, consistent horizontal bands, indicating no visual signs of non-convergence within the chains. The parameters p_site_1, p_site_2, p_site_3, p_site_4, and p_site_5 represent the C. jejuni contamination probabilities for farms, retail markets, wholesale markets, slaughterhouses, and slaughter points, respectively. The parameter intercept_bird represents the baseline probability of a broiler testing positive for C. jejuni at a contaminated site, assuming no effect from site type. The fixed-effect coefficients beta_ret, beta_who, beta_slh, and beta_slp quantify the effects of retail markets, wholesale markets, slaughterhouses, and slaughter points on the probability of a broiler testing positive for C. jejuni at a contaminated site, respectively. The parameter sigma_1 means the variance of the random intercept.
https://doi.org/10.1371/journal.pntd.0013615.s003
(TIFF)
S3 Fig. Pairs plot with four chains for the selected ideal C. coli model.
This figure shows the pairs plot of the selected C. coli model, which showed the distribution of each parameter’s posterior values and its interactions with other parameters. The parameters p_site_1, p_site_2, p_site_3, p_site_4, and p_site_5 represent the C. coli contamination probabilities for farms, retail markets, wholesale markets, slaughterhouses, and slaughter points, respectively. The parameter intercept_bird represents the baseline probability of a broiler testing positive for C. coli at a contaminated site, assuming no effect from site type and province of supply areas. The fixed-effect coefficients beta_ret, beta_slh, beta_slp, and beta_who quantify the effects of retail markets, slaughterhouses, slaughter points, and wholesale markets on the probability of a broiler testing positive for C. coli at a contaminated site, respectively. The fixed-effect coefficients beta_cbac, beta_chdqn, beta_chon, and beta_cot quantify the effects of supply areas—Bac Giang, Hai Duong/Quang Ninh, Ha Noi, and other provinces—on the probability of a broiler testing positive for C. coli at a contaminated site, respectively. The parameter sigma_1 means the variance of the random intercept.
https://doi.org/10.1371/journal.pntd.0013615.s004
(TIFF)
S4 Fig. Pairs plot with four chains for the selected ideal C. jejuni model.
This figure shows the pairs plot of the selected C. jejuni model, which showed the distribution of each parameter’s posterior values and its interactions with other parameters. The parameters p_site_1, p_site_2, p_site_3, p_site_4, and p_site_5 represent the C. jejuni contamination probabilities for farms, retail markets, wholesale markets, slaughterhouses, and slaughter points, respectively. The parameter intercept_bird represents the baseline probability of a broiler testing positive for C. jejuni at a contaminated site, assuming no effect from site type. The fixed-effect coefficients beta_ret, beta_who, beta_slh, and beta_slp quantify the effects of retail markets, wholesale markets, slaughterhouses, and slaughter points on the probability of a broiler testing positive for C. jejuni at a contaminated site, respectively. The parameter sigma_1 means the variance of the random intercept.
https://doi.org/10.1371/journal.pntd.0013615.s005
(TIFF)
S5 Fig. Post-predictive check for the selected C. coli model: The observed (solid circle shape) and predicted (triangle shape) proportion of broiler samples testing positive by different sites and from different catchment provinces for C. coli.
https://doi.org/10.1371/journal.pntd.0013615.s006
(TIFF)
S6 Fig. Post-predictive check for the selected C. jejuni model: The observed (solid circle shape) and predicted (triangle shape) proportion of broiler samples testing positive by different sites for C. jejuni.
https://doi.org/10.1371/journal.pntd.0013615.s007
(TIFF)
S7 Fig. Trace plots with four chains for the selected ideal NTS model.
This figure shows the trace plots (n chains = 4, n iteration = 50000) of the selected NTS model, which displayed tight, consistent horizontal bands, indicating no visual signs of non-convergence within the chains. The fixed-effect coefficient beta_end quantifies the effect of endpoints on the probability of a site being contaminated with NTS. The parameter intercept_site represents the baseline NTS contamination probability of a site when site type has no effect.
https://doi.org/10.1371/journal.pntd.0013615.s008
(TIFF)
S8 Fig. Pairs plot with four chains for the selected ideal NTS model.
This figure shows the pairs plot of the selected NTS model, which showed the distribution of each parameter’s posterior values and its interactions with other parameters. The fixed-effect coefficient beta_end quantifies the effect of endpoints on the probability of a site being contaminated with NTS. The parameter intercept_site represents the baseline NTS contamination probability of a site when site type has no effect.
https://doi.org/10.1371/journal.pntd.0013615.s009
(TIFF)
S9 Fig. The observed (solid circle shape) and predicted (triangle shape) proportions of environmental samples testing positive by farms, endpoints and all sites for NTS.
https://doi.org/10.1371/journal.pntd.0013615.s010
(TIFF)
S1 Data. Three datasets with their variable dictionaries containing the data used in this manuscript.
A document ‘README.text’ describes the structure of the datasets.
https://doi.org/10.1371/journal.pntd.0013615.s011
(ZIP)
Acknowledgments
We would like to thank all the people who helped with field collection, sample archiving, and analysis. We thank the research team from the National Institute of Veterinary Research (NIVR), Vietnam, who assisted with the fieldwork, in particular with sample collection and laboratory analysis. We thank the farmers, sellers, and all the many other actors of the productions distribution networks who took part in the study.
References
- 1. Havelaar AH, Kirk MD, Torgerson PR, Gibb HJ, Hald T, Lake RJ, et al. World Health Organization Global Estimates and Regional Comparisons of the Burden of Foodborne Disease in 2010. PLoS Med. 2015;12(12):e1001923. pmid:26633896
- 2. Thomas KM, de Glanville WA, Barker GC, Benschop J, Buza JJ, Cleaveland S, et al. Prevalence of Campylobacter and Salmonella in African food animals and meat: A systematic review and meta-analysis. Int J Food Microbiol. 2020;315:108382. pmid:31710971
- 3. Sadek SAS, Shaapan RM, Barakat AMA. Campylobacteriosis in Poultry: A Review. jwpr. 2023.
- 4. Lastovica AJ, Allos BM. Clinical Significance ofCampylobacterand Related Species Other ThanCampylobacter jejuniandCampylobacter coli. Campylobacter. Washington, DC, USA: ASM Press. 2014. p. 123–49.
- 5. Mirelis B, Llovet T, Muñoz C, Navarro F, Prats G. Resistance of Salmonella and Campylobacter species to antimicrobial agents. Eur J Clin Microbiol Infect Dis. 1999;18(4):312. pmid:10385026
- 6. Fleckenstein JM, Matthew Kuhlmann F, Sheikh A. Acute Bacterial Gastroenteritis. Gastroenterol Clin North Am. 2021;50(2):283–304. pmid:34024442
- 7. Barrow PA, Simpson JM, Lovell MA. Intestinal colonisation in the chicken by food-poisoning Salmonella serotypes; microbial characteristics associated with faecal excretion. Avian Pathol. 1988;17(3):571–88. pmid:18766717
- 8. Kagambèga A, Thibodeau A, Trinetta V, Soro DK, Sama FN, Bako É, et al. Salmonella spp. and Campylobacter spp. in poultry feces and carcasses in Ouagadougou, Burkina Faso. Food Sci Nutr. 2018;6(6):1601–6. pmid:30258603
- 9. European Food Safety Authority, European Centre for Disease Prevention and Control. The European Union Summary Report on Trends and Sources of Zoonoses, Zoonotic Agents and Food‐borne Outbreaks in 2012. EFS2. 2014;12(2).
- 10. Much P, Pichler J, Kasper S, Lassnig H, Kornschober C, Buchner A, et al. A foodborne outbreak of Salmonella Enteritidis phage type 6 in Austria, 2008. Wien Klin Wochenschr. 2009;121(3–4):132–6. pmid:19280139
- 11. Nhung NT, Phu DH, Carrique-Mas JJ, Padungtod P. A review and meta-analysis of non-typhoidal Salmonella in Vietnam: Challenges to the control and antimicrobial resistance traits of a neglected zoonotic pathogen. One Health. 2024;18:100698. pmid:38468610
- 12.
Birhanu MY, Geremew K, Esatu W, Worku S, Getachew F, Nguyen VD. Poultry production, marketing, and consumption in Vietnam: A review of literature. 2021. https://doi.org/10.13140/RG.2.2.10982.14407
- 13. Burgos S, T. Hong Ha P, Roland-Hol D, A. Burgos S. Characterization of Poultry Production Systems in Vietnam. Int J of Poultry Science. 2007;6(10):709–12.
- 14. Hennessey M, Fournié G, Hoque MA, Biswas PK, Alarcon P, Ebata A, et al. Intensification of fragility: Poultry production and distribution in Bangladesh and its implications for disease risk. Prev Vet Med. 2021;191:105367. pmid:33965744
- 15. Thi Dien N, Thi Minh Khue N, Ebata A, Fournié G, Huyen LTT, Van Dai N, et al. Mapping chicken production and distribution networks in Vietnam: An analysis of socio-economic factors and their epidemiological significances. Preventive Veterinary Medicine. 2023;214:105906.
- 16. Neelawala RN, Edison LK, Kariyawasam S. Pre-Harvest Non-Typhoidal Salmonella Control Strategies in Commercial Layer Chickens. Animals (Basel). 2024;14(24):3578. pmid:39765482
- 17. Rukambile E, Sintchenko V, Muscatello G, Kock R, Alders R. Infection, colonization and shedding of Campylobacter and Salmonella in animals and their contribution to human disease: A review. Zoonoses Public Health. 2019;66(6):562–78. pmid:31179637
- 18. Taha-Abdelaziz K, Singh M, Sharif S, Sharma S, Kulkarni RR, Alizadeh M, et al. Intervention Strategies to Control Campylobacter at Different Stages of the Food Chain. Microorganisms. 2023;11(1):113. pmid:36677405
- 19.
World Health Organization, Food and Agriculture Organization of the United Nations, World Organisation for Animal Health. The global view of campylobacteriosis: report of an expert consultation. Geneva:World Health Organization. 2013. https://iris.who.int/handle/10665/80751
- 20. Carrique-Mas JJ, Bryant JE, Cuong NV, Hoang NVM, Campbell J, Hoang NV, et al. An epidemiological investigation of Campylobacter in pig and poultry farms in the Mekong delta of Vietnam. Epidemiol Infect. 2014;142(7):1425–36. pmid:24067502
- 21. Huong LQ, Chisnall T, Rodgers JD, Cawthraw SA, Card RM. Prevalence, antibiotic resistance, and genomic characterisation of Campylobacter spp. in retail chicken in Hanoi, Vietnam. Microb Genom. 2024;10(1):001190. pmid:38294872
- 22. Ta YT, Nguyen TT, To PB, Pham DX, Le HTH, Alali WQ, et al. Prevalence of Salmonella on chicken carcasses from retail markets in Vietnam. J Food Prot. 2012;75(10):1851–4. pmid:23043836
- 23. Ha TAD, Pham TY. Study of Salmonella, Campylobacter, and Escherichia coli contamination in raw food available in factories, schools, and hospital canteens in Hanoi, Vietnam. Ann N Y Acad Sci. 2006;1081:262–5. pmid:17135522
- 24. Thai TH, Hirai T, Lan NT, Yamaguchi R. Antibiotic resistance profiles of Salmonella serovars isolated from retail pork and chicken meat in North Vietnam. Int J Food Microbiol. 2012;156(2):147–51. pmid:22497836
- 25. Tu LTP, Hoang NVM, Cuong NV, Campbell J, Bryant JE, Hoa NT, et al. High levels of contamination and antimicrobial-resistant non-typhoidal Salmonella serovars on pig and poultry farms in the Mekong Delta of Vietnam. Epidemiol Infect. 2015;143(14):3074–86. pmid:25778282
- 26.
General Statistics Office GSO. Average population by province, sex and residence in 2020. 2024. https://gis.gso.gov.vn/?lang=en#5.08/15.78/105.88
- 27.
GSO GSO. Statistics: Agriculture, forestry and fishing. Hanoi, Vietnam: GSO. 2020. https://www.gso.gov.vn/en/homepage/
- 28.
GADM (Database of Global Administrative Areas) maps and data. https://gadm.org/index.html. 2018.
- 29. Steinbrueckner B, Haerter G, Pelz K, Kist M. Routine identification of Campylobacter jejuni and Campylobacter coli from human stool samples. FEMS Microbiol Lett. 1999;179(2):227–32. pmid:10518720
- 30. Wang G, Clark CG, Taylor TM, Pucknell C, Barton C, Price L, et al. Colony multiplex PCR assay for identification and differentiation of Campylobacter jejuni, C. coli, C. lari, C. upsaliensis, and C. fetus subsp. fetus. J Clin Microbiol. 2002;40(12):4744–7. pmid:12454184
- 31. Tay MYF, Pathirage S, Chandrasekaran L, Wickramasuriya U, Sadeepanie N, Waidyarathna KDK, et al. Whole-Genome Sequencing Analysis of Nontyphoidal Salmonella enterica of Chicken Meat and Human Origin Under Surveillance in Sri Lanka. Foodborne Pathog Dis. 2019;16(7):531–7. pmid:31099590
- 32. Guo S, Aung KT, Leekitcharoenphon P, Tay MYF, Seow KLG, Zhong Y, et al. Prevalence and genomic analysis of ESBL-producing Escherichia coli in retail raw meats in Singapore. J Antimicrob Chemother. 2021;76(3):601–5. pmid:33331642
- 33. Guo S, Tay MYF, Aung KT, Seow KLG, Ng LC, Purbojati RW, et al. Phenotypic and genotypic characterization of antimicrobial resistant Escherichia coli isolated from ready-to-eat food in Singapore using disk diffusion, broth microdilution and whole genome sequencing methods. Food Control. 2019;99:89–97.
- 34.
Plummer M. Jags: A program for analysis of Bayesian graphical models using gibbs sampling. https://mcmc-jags.sourceforge.io/. Accessed 2023 July 25.
- 35.
Urbanek SR. R for macOS. https://cran.r-project.org/bin/macosx/. 2023. Accessed 2023 August 23.
- 36.
Makowski D, Lüdecke D, Ben-Shachar MS, Patil I, Wilson MK, Wiernik BM. Effective sample size (ESS). bayestestR: Understand and describe Bayesian models and posterior distributions. 2024.
- 37.
Plummer M. JAGS Version 4.3. 0 user manual. 2017.
- 38.
Kruschke JK. Doing bayesian data analysis: A tutorial with R and BUGS. New York, NY, USA: Academic Press. 2010.
- 39. Trung NV, Carrique-Mas JJ, Nghia NH, Tu LTP, Mai HH, Tuyen HT, et al. Non-Typhoidal Salmonella Colonization in Chickens and Humans in the Mekong Delta of Vietnam. Zoonoses Public Health. 2017;64(2):94–9. pmid:27152998
- 40. Minh VD, Meeyam T, Unger F, Gölz G, Ngoc PT, Giang TTH, et al. Prevalence of Campylobacter spp. on retail fresh chicken carcasses in Hanoi, Vietnam. VIS. 2022;21(1):221–7.
- 41. Montoro-Dasi L, Lorenzo-Rebenaque L, Marco-Fuertes A, Vega S, Marin C. Holistic Strategies to Control Salmonella Infantis: An Emerging Challenge in the European Broiler Sector. Microorganisms. 2023;11(7):1765. pmid:37512937
- 42. European Food Safety Authority and European Centre for Disease Prevention and Control (EFSA and ECDC). The European Union summary report on trends and sources of zoonoses, zoonotic agents and food‐borne outbreaks in 2017. EFS2. 2018;16(12).
- 43. Raju R, O’Neil L, Kerr C, Lehri B, Sarkar S, Soni T, et al. Non-typhoidal Salmonella in humans in India, Vietnam, Bangladesh and Sri Lanka: a systematic review. JAC Antimicrob Resist. 2024;6(6):dlae190. pmid:39600875
- 44. Holohan N, Wallat M, Hai Yen Luu T, Clark E, Truong DTQ, Xuan SD, et al. Analysis of Antimicrobial Resistance in Non-typhoidal Salmonella Collected From Pork Retail Outlets and Slaughterhouses in Vietnam Using Whole Genome Sequencing. Front Vet Sci. 2022;9:816279. pmid:35425826
- 45. Gibbens JC, Pascoe SJ, Evans SJ, Davies RH, Sayers AR. A trial of biosecurity as a means to control Campylobacter infection of broiler chickens. Prev Vet Med. 2001;48(2):85–99. pmid:11154782
- 46. Luu QH, Tran TH, Phung DC, Nguyen TB. Study on the prevalence of Campylobacter spp. from chicken meat in Hanoi, Vietnam. Ann N Y Acad Sci. 2006;1081:273–5. pmid:17135525
- 47. Whyte P, Collins JD, McGill K, Monahan C, O’Mahony H. The effect of transportation stress on excretion rates of campylobacters in market-age broilers. Poult Sci. 2001;80(6):817–20. pmid:11441852
- 48. Garin B, Gouali M, Wouafo M, Perchec A-M, Pham MT, Ravaonindrina N, et al. Prevalence, quantification and antimicrobial resistance of Campylobacter spp. on chicken neck-skins at points of slaughter in 5 major cities located on 4 continents. Int J Food Microbiol. 2012;157(1):102–7. pmid:22607809
- 49. Corry JE, Atabay HI. Poultry as a source of Campylobacter and related organisms. Symp Ser Soc Appl Microbiol. 2001;(30):96S-114S. pmid:11422565
- 50. Nidaullah H, Abirami N, Shamila-Syuhada AK, Chuah L-O, Nurul H, Tan TP, et al. Prevalence of Salmonella in poultry processing environments in wet markets in Penang and Perlis, Malaysia. Vet World. 2017;10(3):286–92. pmid:28435190
- 51. Verbrugghe E, Boyen F, Gaastra W, Bekhuis L, Leyman B, Van Parys A, et al. The complex interplay between stress and bacterial infections in animals. Vet Microbiol. 2012;155(2–4):115–27. pmid:21963418
- 52. Carrique-Mas JJ, Davies RH. Sampling and bacteriological detection of Salmonella in poultry and poultry premises: a review. Rev Sci Tech. 2008;27(3):665–77. pmid:19284036