Hostname: page-component-7c8c6479df-27gpq Total loading time: 0 Render date: 2024-03-28T04:03:29.944Z Has data issue: false hasContentIssue false

Risk factors associated with Campylobacter detected by PCR in humans and animals in rural Cambodia

Published online by Cambridge University Press:  23 June 2016

K. OSBJER
Affiliation:
Division of Reproduction, Department of Clinical Sciences, Swedish University of Agricultural Sciences (SLU), Uppsala, Sweden
S. BOQVIST
Affiliation:
Department of Biomedical Sciences and Veterinary Public Health, SLU, Uppsala, Sweden
S. SOKERYA
Affiliation:
Centre for Livestock and Agriculture Development, Phnom Penh, Cambodia
K. CHHENG
Affiliation:
National Institute of Public Health, Phnom Penh, Cambodia
S. SAN
Affiliation:
National Veterinary Research Institute, Phnom Penh, Cambodia
H. DAVUN
Affiliation:
National Veterinary Research Institute, Phnom Penh, Cambodia
H. RAUTELIN
Affiliation:
Department of Medical Sciences, Clinical Microbiology, Uppsala University, Uppsala, Sweden
U. MAGNUSSON*
Affiliation:
Division of Reproduction, Department of Clinical Sciences, Swedish University of Agricultural Sciences (SLU), Uppsala, Sweden
*
*Author for correspondence: Professor U. Magnusson, Box 7054, SE-750 07 Uppsala, Sweden. (Email: Ulf.Magnusson@slu.se)
Rights & Permissions [Opens in a new window]

Summary

Campylobacter are worldwide-occurring zoonotic bacteria, with the species Campylobacter jejuni and C. coli commonly associated with diarrhoea in children in low-income countries. In this cross-sectional study, the prevalence of C. jejuni and C. coli in human and livestock faecal samples was detected by PCR and zoonotic risk factors associated with human Campylobacter positivity were identified. In total 681 humans and 753 livestock (chickens, ducks, pigs, cattle) from 269 households were sampled. Children aged <16 years were more frequently Campylobacter positive (19%) than adults (8%) and multilevel logistic models revealed that human C. jejuni positivity was associated with the following household practices: home-slaughtering [odds ratio (OR) 2·4, P = 0·01], allowing animals access to sleeping and food preparation areas (OR 2·8, P = 0·02), and eating undercooked meat (OR 6·6, P = 0·05), while frequent consumption of beef was protective (OR 0·9, P = 0·05). Associations were stronger for home-slaughtering (OR 4·9, P = 0·004) with C. jejuni infection in children only. Campylobacter was highly prevalent in pigs (72%) and chickens (56%) and risk factors associated with human Campylobacter positivity were identified throughout the meat production chain. The findings underline the importance of studying source attributions throughout the production chain and the need for upgraded understanding of Campylobacter epidemiology in low-income countries.

Type
Original Papers
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
Copyright © Cambridge University Press 2016

INTRODUCTION

Gastroenteritis is a major public health concern, with over 800 000 fatalities in children annually, most occurring in Asia and Africa [Reference Liu1]. Despite a global decline, diarrhoeal mortality accounts for one in ten child deaths in resource-poor countries and gastroenteritis is known to be closely associated with malnutrition and underweight [Reference Liu1, Reference Checkley2]. Campylobacter, belonging to the most commonly detected pathogens in children with moderate-to-severe diarrhoea in Asia [Reference Kotloff3Reference Bodhidatta5], are the most common cause of human bacterial gastroenteritis worldwide [6, 7]. In campylobacteriosis symptoms range from acute abdominal pain, diarrhoea and fever to late sequelae such as reactive arthritis and, although rarely occurring neurological Guillain–Barré syndrome [Reference Janssen8]. Of all Campylobacter species, C. jejuni and C. coli are the most common causes of human infection [Reference Humphrey, O'Brien and Madsen9].

The epidemiology of human campylobacteriosis appears to differ between high- and low-income countries [Reference Coker10, Reference Havelaar11]. In high-income countries symptomatic infection occurs in all age groups [Reference Janssen8], whereas in low-income countries most symptomatic Campylobacter infections are diagnosed in young children and adults seem to acquire a level of protective immunity following repeated exposure [Reference Coker10, Reference Havelaar11]. The global distribution of Campylobacter is attributed to asymptomatic colonization of the intestinal tract in a wide range of livestock species [Reference Humphrey, O'Brien and Madsen9]. Zoonotic transmission to humans is significant and source-attribution studies in high-income countries have recognized direct contact with farm animals and consumption of chicken, unpasteurized dairy products and contaminated water as being important [Reference Domingues12, 13]. International travel, particularly to tropical regions, has however, been suggested as the most important risk factor in high-income countries, involving practices during travel such as eating vegetable salad and raw or undercooked pork [Reference Havelaar11, Reference Mughini-Gras14]. In low-income countries, sources have been less well examined. For rural households in Egypt, the presence of poultry manure, uncovered litter in house yards and lack of barriers to keep animals out of houses have been identified as risk factors for Campylobacter infection in children [Reference Hassan15, Reference Rao16], while a study carried out in Ethiopia identified exposure to domestic animals as a sufficient risk factor for infection [Reference Lengerh17].

In Cambodia, 80% of the population live in rural areas and smallholder farmers represent the majority of livestock producers [Reference Young18, Reference Nampanya19]. Livestock are predominantly reared in free-range systems, with close interaction between livestock and humans and thus enabling exposure to zoonotic pathogens [Reference Young18]. In such rural and often resource-scarce households, the burden of malnutrition and diarrhoeal disease is high, particularly in children aged <5 years [Reference Liu1, 20, Reference Darapheak21]. Nonetheless, data on enteropathogens and their source attribution is limited and the role of zoonotic transmission poorly understood. Few studies have focused on detection of enteropathogens in livestock, although Campylobacter have been detected by culture in 81% of the poultry carcases available on sale in Cambodian wet markets [Reference Lay22]. Additionally, in a recent study on livestock in neighbouring Vietnam, Campylobacter were detected by culture in 32% of poultry and 54% of pigs sampled on low-biosecurity farms [Reference Carrique-Mas23].

To the best of our knowledge, no previous study has examined factors associated with Campylobacter transmission between animals and humans in Cambodian households. The aim of this study was therefore to identify zoonotic risk factors associated with human Campylobacter positivity in rural Cambodian households for which the prevalence of C. jejuni and C. coli in human and livestock faecal samples had been detected by polymerase chain reaction (PCR).

MATERIAL AND METHODS

Study design and data collection

This cross-sectional study was based on our previous studies on household practices [Reference Osbjer24] and detection of Campylobacter by culture and PCR [Reference Osbjer25] conducted in three regions in Cambodia: Kampong Cham province (in May 2011), Battambang province (in July 2012) and Kampot province (in March 2013) (Fig. 1). In each region, 10 villages were included and in each village, 10 households were selected for interviews and collection of faecal samples. The purposive selection of regions, villages and households has been described previously [Reference Osbjer24]. The interviews, targeted towards the female head of the household, were carried out in Khmer using a household questionnaire consisting of questions on livestock management, meat consumption and household practices related to zoonosis transmission (Table 1). To enhance consistency between the three regions, the field team was trained in questioning and sampling prior to fieldwork [Reference Osbjer24]. Each village was visited for two consecutive days. On day 1, selected households were interviewed following consent to participate and provided with containers for human faecal samples. All members of the household were encouraged to provide a faecal sample, regardless of gender, age and history of gastrointestinal symptoms. On day 2, all human samples produced were collected and samples from 1–6 livestock, including chickens, ducks, pigs and cattle (Table 2). Livestock samples were selected depending on the species reared by the household with the aim of covering as many species and age groups as possible. In households farming more than one animal species, a minimum of one sample from each species, was obtained. For each person and livestock sampled, information on age was recorded. In addition, self-reported (or parental report for younger children) gastrointestinal symptoms within a 2-week period prior to sampling were recorded for each sampled person. Geographical position at the central point of the villages included in the study was recorded using a hand-held global positioning system (GPS; Garmin eTrex H).

Fig. 1. Map of Cambodia showing geographical distribution of the 30 villages included in the cross-sectional study in 2011–2013. Open Development Cambodia (www.opendevelopmentcambodia.net) and OpenStreetMap contributors (openstreetmap.org).

Table 1. Self-reported household practices in the 269 households included in the study (Cambodia, 2011–2013)

Table 2. Number of sampled humans and livestock per household (n = 269) (Cambodia, 2011–2013)

The target sample size was calculated based on sample size for expected Campylobacter prevalence at 95% confidence interval with 5% precision, using the formula presented by Thrusfield [Reference Thrusfield26]. The expected Campylobacter prevalence for human samples was set at 20% [Reference Lindblom27, Reference Mason28] and the overall prevalence for livestock samples at 35% [Reference Kassa, Gebre-Selassie and Asrat29]. An extra 15% was added to the human sample size to adjust for possible confounding and interaction in the statistical models [Reference Dohoo30]. To account for clustering of infection within households, the target sample sizes were adjusted for intra-cluster correlation, with a coefficient of 0·2 [Reference Otte and Gumm31]. The average number of humans and livestock sampled per household was set at 3. Thereby an expected Campylobacter prevalence of 20% in humans [Reference Lindblom27, Reference Mason28] gave a target sample size of 246, which was adjusted to 542 when taking household clustering into account. After adjusting for confounding and interactions, the final target sample size for human samples was set at 623. In livestock, an expected Campylobacter prevalence of 35% gave a target sample size of 350, which was adjusted to 840 when accounting for household clustering.

Self-collected human faecal samples were stored on ice-packs until faecal material was transferred by sterile cotton swabs into vials with bacterial freeze medium. Poultry samples were collected by insertion of a swab into the cloaca, while cattle and pig samples were collected by dipping a swab into faecal material collected manually from the rectum. All swabs were placed in vials containing bacterial freeze medium as previously reported [Reference Osbjer25] and stored in cooler boxes or refrigerated before transportation on ice to Phnom Penh within 1 day for storage at −70 °C pending shipment to Sweden for analysis. Extraction of DNA in livestock samples was carried out at the Swedish University of Agricultural Sciences, and DNA extraction in human samples and all PCR analyses were performed at Uppsala University. Identical multiplex PCR was performed on all human and livestock samples using two specific primers. For C. jejuni, the primer pair MDmapA1 upper and MDmapA2 lower targeting the mapA gene was used [Reference Denis32], and for C. coli we used the primer pair COL3 upper and MDCOL2 lower targeting the ceuE gene [Reference Denis32, Reference Gonzalez33]. A detailed description of the laboratory analyses can be found in Osbjer et al. [Reference Osbjer25].

Data management and statistical analysis

Data from questionnaires were independently translated by two translators from Khmer into English and compared for consistency before being transcribed into spreadsheets in Microsoft Office Excel 2010 (Microsoft Corp., USA). Statistical analysis was performed in SAS for Windows v. 9.3 (SAS Institute Inc., USA). Statistical tests including Pearson's χ 2, or Fisher's exact test when there were <5 observations per group, were used to analyse differences between age groups in the proportion of Campylobacter-positive samples and the proportion of people with gastrointestinal symptoms. An intra-cluster correlation coefficient (ICC) for human and livestock samples detected with C. jejuni and C. coli was calculated to estimate the correlation between two observations in the same household or village by building unconditional logistic models, extracting the village- and household-level variances and assuming that the person-level variance was 3·29 [Reference O'Connell34, Reference Vigre35].

To explore potential risk factors for C. jejuni and C. coli positivity, multilevel logistic models were run with human samples that tested positive for C. jejuni or C. coli as the outcome variable. Comparable models were also run using the subset: samples positive for C. jejuni or C. coli in children aged <16 years of age. Univariable models were run for the outcome variables human samples positive for C. jejuni or C. coli and any of the 11 self-reported household practices (presented in Table 1) as the explanatory variable. Multivariable models were run for the same outcome variables and one of the four groups of explanatory variables: the self-reported gastrointestinal symptoms in sampled humans (presented in Table 3); number of chickens, ducks, pigs and cattle reared in the household; C. jejuni- or C. coli-positive samples from chickens, ducks, pigs or cattle; and number of days per month that poultry, pork and beef was consumed by the household.

Table 3. Rate of self-reported (or parental report for younger children) gastrointestinal symptoms during the 2-week period prior to sampling (n = 681) (Cambodia 2011–2013)

Values given are n (%).

The statistical models had three levels of nested factors in the hierarchy, where each person sampled was clustered within households that were clustered within villages. Variables were considered candidates for multivariable analysis based on their biological plausibility and risk factors previously identified to be associated with human Campylobacter infection [Reference Domingues12, Reference Hassan15, Reference Rao16]. Random effects (variables) for households were assumed to be independent and the number of livestock reared and meat consumption were fitted as continuous variables in modelling, with smoothing loess plots applied to assess their functional form [Reference Dohoo30]. Due to considerable collinearity and interaction, only univariable analysis was performed on the 11 self-reported household practices. The statistical significance level was defined as a two-tailed P value ⩽0·05.

QGIS 2·0·1 software was used to map the distribution of villages in open-source base map layers obtained from Open Development Cambodia (www.opendevelopmentcambodia.net) and OpenStreetMap contributors (openstreetmap.org).

Ethical approval

Ethical approval (43 NECHR, 8 April 2011) was obtained prior to the survey from the National Ethics Committee for Health Research, Ministry of Health, Cambodia, and an advisory ethical statement (Dnr 2011/63) was obtained from the Regional Board for Research Ethics in Uppsala, Sweden. 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.

RESULTS

Description of included households

A household was defined as a group of people making common arrangements for food and shelter. Interviews and human samples were obtained from 269 households with a median household size of 5·0 people (range 1–17). Of these households, poultry were reared in 253 (94%), pigs in 148 (55%) and cattle in 177 (66%). As described in our previous household study [Reference Osbjer24], the majority of households reared poultry and cattle in a primarily free-range system, while pigs were reared in a primarily confined system. The mean number of days per month that meat was consumed in the household was 4 for poultry [standard deviation (s.d.) = 4·4], 9 for pork (s.d. = 7·4) and 2 for beef (s.d. = 4·1).

Self-reported gastrointestinal symptoms in sampled humans

Symptoms of abdominal pain, diarrhoea, fever and vomiting during the 2-week period preceding sampling, as defined by the respondent or for younger children by parental report, were recorded for each person sampled. Fever was the most commonly reported symptom (21%) and showed no statistically significant difference between age groups. However, diarrhoea, was more frequently reported in children aged <6 years than in adults and children aged 6–15 years (P = 0·0001) (Table 3).

Detection of Campylobacter in human samples

As reported earlier [Reference Osbjer25], of the 681 human samples, 82 (12%) tested positive by PCR; C. jejuni was detected in 66 samples (80%) and C. coli in 16 samples (20%) (Table 4). Children aged <16 years more often tested positive for C. jejuni or C. coli than adults (P<0·001), but no significant difference in the proportion of positive samples could be determined between the three age groups (<2, 2–5, 6–15 years). At least one positive sample was detected in 66 households (24%), with a quite strong clustering of positive samples within households (ICC = 0·14, variance estimate 0·47) and a weaker clustering within villages (ICC = 0·02, variance estimate 0·07).

Table 4. Detection of Campylobacter jejuni and C. coli by multiplex PCR in faecal samples from children and adults in rural Cambodia, 2011–2013

Detection of Campylobacter in livestock samples

Of the 763 livestock samples obtained from 229 different households, 324 (42%) tested positive for Campylobacter; C. jejuni was detected in 165 samples (51%), C. coli in 108 samples (33%), and both C. jejuni and C. coli in 51 samples (16%). C. jejuni, C. coli or both were detected in 56% of chickens, 22% of ducks, 72% of pigs and 5% of cattle as presented with stratification by age in Table 5. In the youngest age group of chickens, but not in that of ducks, pigs or cattle, C. jejuni/C. coli was more often detected than in the older age groups (P<0·001). The number of households with at least one livestock sample analysed by species and the percentage of sampled households with at least one positive sample was 197 (65%) for chicken, 69 for ducks (25%), 112 (78%) for pigs and 132 (6%) for cattle. Clustering of positive samples was weak within households (ICC = 0·05, variance estimate 0·17) and non-detectable within villages.

Table 5. Detection of Campylobacter jejuni and C. coli by multiplex PCR in faecal samples from different age groups of chickens, ducks, pigs and cattle in rural Cambodia, 2011–2013

* Fifty-one samples tested positive for both C. jejuni and C. coli.

Analysis of zoonotic risk factors associated with human Campylobacter positivity

In the multilevel models, no associations were found between the outcome variables C. jejuni or C. coli in human samples and self-reported gastrointestinal disease symptoms. Likewise, there were no associations between C. jejuni or C. coli in human samples and the number of chickens, ducks, pigs or cattle reared or detection of C. jejuni or C. coli in the household's chickens, ducks, pigs or cattle (Table 6). The household practices of slaughtering domestic animals at home, allowing animals into sleeping and food preparation areas and eating undercooked meat were associated with increased odds of human C. jejuni positivity, whereas frequent consumption of beef was associated with decreased odds. The probability of C. jejuni-positive samples was higher in the subset models of children aged <16 years for the household practice of home-slaughtering. None of the other household practices listed (Table 1) were associated with C. jejuni or C. coli in samples from children. Detection of C. coli was associated with frequent consumption of poultry, both when all the human samples were included in the model and when the child subset model was used. Frequent consumption of pork was associated with detection of both C. jejuni and C. coli in the child model (OR 1·1, P = 0·04). All models with significant associations between C. jejuni or C. coli detected in human samples and explanatory variables are presented in Table 6.

Table 6. Significant associations in generalized linear mixed models between the outcome variables detection of Campylobacter jejuni or C. coli by PCR in human samples (n = 681) and samples from children only (n = 272), and explanatory variables measured at the household level (Cambodia, 2011–2013)

OR, Odds ratio; CI, confidence interval.

* Quantitative explanatory variable.

DISCUSSION

Our findings suggest that household practices play a role in animal-to-human transmission of Campylobacter in rural Cambodian households. The practices of home-slaughtering, allowing animals access to sleeping and food preparation areas, consuming undercooked meat, and frequent consumption of poultry and pork were all associated with an increased probability of human C. jejuni or C. coli positivity. Children aged <16 years had more than twice the prevalence of C. jejuni and C. coli found for adults, whereas no difference was identified between older and younger children. Symptoms of diarrhoea were commonly reported, particularly in young children, but gastrointestinal symptoms were not associated with either C. jejuni or C. coli positivity. Finally, a high prevalence of C. jejuni and C. coli was detected in poultry and pigs.

In Cambodia, underweight and stunting, markers of acute and chronic malnutrition, are estimated to affect between 28% and 45% of children, with the highest burden seen in rural and resource-poor households [20, Reference Darapheak21]. Consumption of a diverse diet, in particular animal-based foods, is a protective factor in malnutrition [Reference Darapheak21], but poor control of zoonotic pathogens may jeopardize the health benefits. In this study gastrointestinal symptoms were frequently reported in adults and children. Symptoms were self-reported and based on personal perception rather than a set case definition as such method is suggested to reduce recall bias when a recall period of ⩾2 weeks is applied [Reference Baqui36, Reference Goldman, Vaughan and Pebley37]. The absence of associations between Campylobacter detection and gastrointestinal symptoms as seen here has been previously reported in low-income countries [Reference Randremanana38, Reference da Silva39], and is likely due to the development of protective immunity in endemic settings [Reference Havelaar11]. Frequent exposures to Campylobacter at a young age have been shown to boost the immune response with increasing age to protect against clinical disease, but not necessarily against transient positivity [Reference Havelaar11]. Regardless of symptoms, however, Campylobacter positivity is of importance in rural low-income areas, particularly in children, as some studies have also found asymptomatic Campylobacter infection to be associated bi-directionally with malnutrition and reduced growth [Reference da Silva39, Reference Lee40]. Some explanations for the absence of association between Campylobacter detection and symptoms may also be found in the methods used here. PCR is known to have a high sensitivity in detecting low numbers of live bacteria and also an ability to detect dead bacteria; however, neither of these may be indicative of clinical disease. Detected Campylobacter can also reflect convalescent phase as excretion of Campylobacter may last up to 10 weeks after infection [Reference Havelaar11].

In high-income countries the majority of human campylobacteriosis cases seem to be related to chickens [7, 13]. The effect of poultry rearing could, however, not be investigated in this study as nearly all households kept poultry. Livestock keeping per se was, in this study, not associated with an increased probability of human Campylobacter positivity, even when Campylobacter were detected in the livestock reared. Instead, the biosecurity measures and hygiene precautions applied within the household seemed more important. The ICC of 0·14 obtained also shows that human Campylobacter infections clustered quite strongly within households, but marginally within villages. The self-reporting used to quantify household practices and disease symptoms has possibly induced some over- and under-reporting resulting from perceived desired responses; however, this approach allowed inclusion of a larger number of households compared to structured observations [Reference Biran41].

Household involvement in slaughtering has not been previously reported as a risk factor for human Campylobacter positivity. In this study, the odds for children were higher than for adults, although the actual slaughter was carried out by adults. Possible explanations could be that children are in closer contact with slaughter waste during outdoor play and are less cautious with hand hygiene. Household risk factors associated with human Campylobacter positivity were detected throughout the meat production chain here, from free-ranging livestock and home-slaughtering, to unsafe meat preparation and consumption. Such results suggest future actions targeting the entire meat production chain for reduced burden of human Campylobacter infection. Moreover, as previously reported, livestock are mainly produced to generate an income and often sold by households [Reference Osbjer24]. Thus, efficient Campylobacter control ought to move beyond the households with improvements in hygiene practices targeting also external factors along the meat production chain, such as middlemen, abattoirs and consumers.

As described by others, associations identified between Campylobacter infection and meat consumption are most likely attributable to in-kitchen cross-contamination of food consumed raw, in addition to consumption of meat [Reference Evans, Ribeiro and Salmon42, Reference Boer and Hahné43]. Interestingly, in this study, consumption of poultry was associated with human C. coli, but not with C. jejuni positivity, which is remarkable since C. jejuni was detected in 45% of the poultry samples and C. coli only in 13%. Nevertheless, some care is needed before generalizing these results, as only 16 human samples tested positive for C. coli. Consumption of beef was found protective against human C. jejuni, although borderline significantly, but an explanation for this remains unclear. Our data did not support the theory that beef was more frequently consumed in affluent households, affording a higher hygiene standard, or that an increase in beef consumption corresponded to a decrease in poultry and pork consumption (data not shown). Seemingly low odds ratios were obtained for the meat consumption variables due to the unit of 1 day. Odds would increase considerably if meat was consumed 2–3 days extra or more per month. The high odds ratios presented for undercooked meat consumption should, however, be interpreted with caution as the association with C. jejuni positivity is borderline significant with a wide confidence interval. The estimated livestock prevalence should also be viewed with caution. Samples were collected at one occasion, thus any intermittent excretion of Campylobacter could have been missed. Additionally, the initial 853 livestock samples collected were reduced to 763 after excluding samples from the 31 households where no human samples were obtained, therefore the target sample size of 840 livestock samples was not met.

Surprisingly, and unlike in other studies [Reference Coker10, Reference Rao16], no differences were identified in Campylobacter detection between different age groups of children. One possible explanation could be the previously discussed high sensitivity of PCR detecting low numbers of Campylobacter in comparison with culture. Moreover, as previously reported [Reference Osbjer25], negligible differences were found in Campylobacter prevalence between the three regions, which were sampled in different seasons. Seasonal differences are therefore unlikely to have biased the results presented. However, the purposive sampling process in the study, including selection of households with many different livestock species, may have introduced some bias. However, given the high number of households and samples we assume that selection bias had only minor impact on the results and that our sample can serve as an approximation of a population-based design for species-diverse households.

CONCLUSIONS

Consumption of animal-based foods is important in reducing malnutrition in resource-poor households, but is hampered by the presence of zoonotic pathogens. In this study, C. jejuni and C. coli were frequently detected in humans, especially children, and in livestock, especially in pigs and chickens. Several self-reported household practices along the meat production chain from rearing of live animals to meat consumption were found to be associated with Campylobacter positivity in humans. These findings underline the importance of studying source attributions of zoonotic enteropathogens throughout the production chain. Finally, an upgraded understanding of the Campylobacter epidemiology in low-income countries may guide future interventions aimed at food and nutrition security.

ACKNOWLEDGEMENTS

We thank the Cambodian families who participated in the study. We also extend our thanks to the commune, district and provincial veterinarians in participating regions for assistance during the field work and to the invaluable field teams for collection of data. Finally, we thank Akofa Olivia Mac-Kwashie and Jamila Mohammad at Uppsala University and Annlouise Jansson at the Swedish University of Agricultural Sciences for assistance with sample analyses. The work was financially supported by The Swedish Civil Contingencies Agency (MSB) and the Swedish International Development Cooperation Agency (Sida), Sweden.

DECLARATION OF INTEREST

None.

References

REFERENCES

1. Liu, L, et al. Global regional and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000. Lancet 2012; 379: 21512161.CrossRefGoogle ScholarPubMed
2. Checkley, W, et al. Multi-country analysis of the effects of diarrhoea on childhood stunting. International Journal of Epidemiology 2008; 37: 816830.Google Scholar
3. Kotloff, KL, et al. Burden and aetiology of diarrhoeal disease in infants and young children in developing countries (the Global Enteric Multicenter Study, GEMS): a prospective, case-control study. Lancet 2013; 382: 209222.Google Scholar
4. Isenbarger, D, et al. Prospective study of the incidence of diarrhoea and prevalence of bacterial pathogens in a cohort of Vietnamese children along the Red River. Epidemiology and Infection 2001; 127: 229236.Google Scholar
5. Bodhidatta, L, et al. Case-control study of diarrheal disease etiology in a remote rural area in Western Thailand. American Journal of Tropical Medicine and Hygiene 2010; 83: 11061109.Google Scholar
6. World Health Organization. WHO estimates of the global burden of foodborne diseases. Geneva, Switzerland, 2015.Google Scholar
7. World Health Organization. The global view of campylobacteriosis. Geneva, Switzerland, 2013.Google Scholar
8. Janssen, R, et al. Host-pathogen interactions in Campylobacter infections: the host perspective. Clinical Microbiology Reviews 2008; 21: 505518.Google Scholar
9. Humphrey, T, O'Brien, S, Madsen, M. Campylobacters as zoonotic pathogens: a food production perspective. International Journal of Food Microbiology 2007; 117: 237257.Google Scholar
10. Coker, A, et al. Human campylobacteriosis in developing countries. Emerging Infectious Diseases 2002; 8: 237244.Google Scholar
11. Havelaar, A, et al. Immunity to Campylobacter: its role in risk assessment and epidemiology. Critical Reviews in Microbiology 2009; 35: 122.Google Scholar
12. Domingues, AR, et al. Source attribution of human campylobacteriosis using a meta-analysis of case-control studies of sporadic infections. Epidemiology and Infection 2012; 140: 970981.CrossRefGoogle ScholarPubMed
13. European Food Safety Authority (EFSA) and European Centre for Disease Prevention and Control (ECDC). The European Union summary report on trends and sources of zoonoses, zoonotic agents and food-borne outbreaks in 2014. EFSA Journal 2015; 13: 191 pp.Google Scholar
14. Mughini-Gras, L, et al. Campylobacteriosis in returning travellers and potential secondary transmission of exotic strains. Epidemiology and Infection 2013; 142: 12771288.Google Scholar
15. Hassan, KE, et al. The impact of household hygiene on the risk of bacterial diarrhea among Egyptian children in rural areas, 2004–2007. Journal of Infection in Developing Countries 2014; 8: 15411551.Google Scholar
16. Rao, M, et al. Pathogenicity and convalescent excretion of Campylobacter in rural Egyptian children. American Journal of Epidemiology 2001; 154: 166173.Google Scholar
17. Lengerh, A, et al. Prevalence, associated risk factors and antimicrobial susceptibility pattern of Campylobacter species among under five diarrheic children at Gondar University Hospital, Northwest Ethiopia. BMC Pediatrics 2013; 13: 19.Google Scholar
18. Young, JR, et al. Improving smallholder farmer biosecurity in the Mekong Region through change management. Transboundary and Emerging Diseases 2013; 62: 491504.Google Scholar
19. Nampanya, S, et al. Improvement in smallholder farmer knowledge of cattle production, health and biosecurity if southern Cambodia between 2008 and 2010. Transboundary and Emerging Diseases 2011; 59: 117127.CrossRefGoogle ScholarPubMed
20. World Health Organization. Neonatal and child health country profile (http://www.who.int/maternal_child_adolescent/epidemiology/profiles/neonatal_child/khm.pdf). Accessed 21 December 2015.Google Scholar
21. Darapheak, C, et al. Consumption of animal source foods and dietary diversity reduce stunting in children in Cambodia. International Archives of Medicine 2013; 6: 29.Google Scholar
22. Lay, KS, et al. Prevalence, numbers and antimicrobial susceptibilities of Salmonella serovars and Campylobacter spp. in retail poultry in Phnom Penh, Cambodia. Journal of Veterinary Medical Science 2011; 73: 325329.Google Scholar
23. Carrique-Mas, JJ, et al. An epidemiological investigation of Campylobacter in pig and poultry farms in the Mekong delta of Vietnam. Epidemiology and Infection 2014; 142: 14251436.Google Scholar
24. Osbjer, K, et al. Household practices related to disease transmission between animals and humans in rural Cambodia. BMC Public Health 2015; 15: 476.Google Scholar
25. Osbjer, K, et al. Detection of Campylobacter in human and animal field samples in Cambodia. APMIS 2016; 124: 508515.Google Scholar
26. Thrusfield, M. Veterinary Epidemiology, 3rd edn. Oxford: Blackwell Science, 2007 pp. 610.Google Scholar
27. Lindblom, G-B, et al. Campylobacter jejuni/coli and enterotoxigenic Eschericia coli (ETEC) in faeces from children and adults in Tanzania. Scandinavian Journal of Infectious Diseases 1995; 27: 589593.Google Scholar
28. Mason, J, et al. Campylobacter infection in children in Malawi is common and is frequently associated with enteric virus co-infections. PLoS ONE 2013; 8: e59663.Google Scholar
29. Kassa, T, Gebre-Selassie, S, Asrat, D. Antimicrobial susceptibility patterns of thermotolerant Campylobacter strains isolated from food animals in Ethiopia. Veterinary Microbiology 2007; 119: 8287.Google Scholar
30. Dohoo, IR, et al. Veterinary Epidemiologic Research, 2nd edn. Charlottetown: AVC Inc., 2003, pp. 865.Google Scholar
31. Otte, MJ, Gumm, ID. Intra-cluster correlation coefficients of 20 infections calculated from the results of cluster-sample surveys. Preventive Veterinary Medicine 1997; 31: 147150.Google Scholar
32. Denis, M, et al. Development of a m-PCR assay for simultaneous identification of Campylobacter jejuni and C. coli . Letters in Applied Microbiology 1999; 29; 406410.Google Scholar
33. Gonzalez, I, et al. Specific identification of the enteropathogens Campylobacter jejuni and Campylobacter coli by using a PCR test based on the ceuE gene encoding a putative virulence determinant. Journal of Clinical Microbiology 1997; 35: 759763.Google Scholar
34. O'Connell, AA, et al. Multilevel Modeling of Educational Data. Charlotte: Information Age Publishing Inc., 2008, pp. 199242.Google Scholar
35. Vigre, H, et al. Intra-unit correlations in seroconversion to Actinobacillus pleuropneumoniae and Mycoplasma hyopneumoniae at different levels in Danish multi-site pig production facilities. Preventive Veterinary Medicine 2004; 63: 928.CrossRefGoogle ScholarPubMed
36. Baqui, AH, et al. Methodological issues in diarrhoeal disease epidemiology: definition of diarrhoeal episodes. International Journal of Epidemiology 1991; 4: 10571063.Google Scholar
37. Goldman, N, Vaughan, B, Pebley, AR. The use of calendars to measure child illness in health interview surveys. International Journal of Epidemiology 1998; 27: 505512.CrossRefGoogle ScholarPubMed
38. Randremanana, R, et al. Case-control study of the etiology of infant diarrheal disease in 14 districts in Madagascar. PLoS ONE 2012; 7: e44533.Google Scholar
39. da Silva, Quetz J, et al. Campylobacter jejuni and Campylobacter coli in children from communities in Northeastern Brazil: molecular detection and relation to nutritional status. Diagnostic Microbiology and Infectious Disease 2010; 67: 220227.Google Scholar
40. Lee, G, et al. Symptomatic and asymptomatic Campylobacter infections associated with reduced growth in Peruvian children. PLoS Neglected Tropical Diseases 2013; 7: e2036.CrossRefGoogle ScholarPubMed
41. Biran, A, et al. Comparing the performance of indicators of hand-washing practices in rural Indian households. Tropical Medicine and International Health 2008; 13: 278285.Google Scholar
42. Evans, MR, Ribeiro, CD, Salmon, RL. Hazards of healthy living: bottled water and salad vegetables as risk factors for Campylobacter infection. Emerging Infectious Diseases 2003; 9: 12191225.Google Scholar
43. Boer, ED, Hahné, M. Cross-contamination with Campylobacter jejuni and Salmonella spp. from raw chicken products during food preparation. Journal of Food Protection 1990; 53: 10671068.Google Scholar
Figure 0

Fig. 1. Map of Cambodia showing geographical distribution of the 30 villages included in the cross-sectional study in 2011–2013. Open Development Cambodia (www.opendevelopmentcambodia.net) and OpenStreetMap contributors (openstreetmap.org).

Figure 1

Table 1. Self-reported household practices in the 269 households included in the study (Cambodia, 2011–2013)

Figure 2

Table 2. Number of sampled humans and livestock per household (n = 269) (Cambodia, 2011–2013)

Figure 3

Table 3. Rate of self-reported (or parental report for younger children) gastrointestinal symptoms during the 2-week period prior to sampling (n = 681) (Cambodia 2011–2013)

Figure 4

Table 4. Detection of Campylobacter jejuni and C. coli by multiplex PCR in faecal samples from children and adults in rural Cambodia, 2011–2013

Figure 5

Table 5. Detection of Campylobacter jejuni and C. coli by multiplex PCR in faecal samples from different age groups of chickens, ducks, pigs and cattle in rural Cambodia, 2011–2013

Figure 6

Table 6. Significant associations in generalized linear mixed models between the outcome variables detection of Campylobacter jejuni or C. coli by PCR in human samples (n = 681) and samples from children only (n = 272), and explanatory variables measured at the household level (Cambodia, 2011–2013)