INTRODUCTION
Globally, diarrhoea is the second greatest cause of mortality in children aged <5 years after pneumonia [1]. Children in low- to middle-income countries are at risk of frequent diarrhoea episodes [Reference Fischer Walker2]; the 2011 national survey in Ethiopia estimated that 13·4% of children in the <5 years age group had suffered an episode of diarrhoea in the preceding 2 weeks, although prevalence was >20% in some regions [3]. Numerous infectious agents cause diarrhoea; many of which are zoonotic, with Campylobacter being one of the main bacterial causes, especially in children aged <2 years [Reference Coker4–Reference Mitike6]. While the organism is also frequently isolated from healthy children and adults, and morbidity appears to decline with age, adults with HIV infection are also at increased risk of Campylobacter-associated diarrhoea and bacteraemia [Reference Coker4]. Different routes by which humans are exposed to Campylobacter include ingestion of contaminated food and water, direct contact with infected animals or carcasses, from environments contaminated with animal faeces or sewage, or direct person-to-person transmission [Reference Coker4, Reference Whiley7]. In countries with a highly developed intensive poultry industry, there is considerable epidemiological evidence that chickens are the main source of human Campylobacter infection [Reference Wagenaar, Mevius and Havelaar8–Reference Adak10]. In developing countries, far less is known about the epidemiology of Campylobacter, but it has been found at high levels in retail poultry meat [Reference Osano and Arimi11], and strains isolated from chickens have been observed to correlate with human isolates in terms of biotype and serotype [Reference Adegbola12].
Around 83% of Ethiopia's population, currently estimated at 96 million, live in rural households [13] where sharing space with domestic livestock is common practice. Chicken production is an integral part of most rural families’ livelihoods (an estimated two-thirds of Ethiopian villagers keep poultry) and birds are commonly kept at night on perches within the family dwelling, frequently in the kitchen [Reference Dessie and Ogle14, Reference Halima15]. In Ethiopia, it has been reported that chickens and poultry meat have a higher Campylobacter prevalence compared to other farm animal and meat products [Reference Kassa, Gebre-selassie and Asrat16, Reference Dadi and Asrat17], and exposure to pets and livestock, including chickens, has been associated with increased odds of Campylobacter isolation in diarrhoeal children aged <5 years [Reference Lengerh18]. Furthermore, in a study from Egypt, Campylobacter diarrhoea was more common in households where animals were present in food preparation areas [Reference Rao19]. It might, therefore, be anticipated that an important route by which rural families are exposed to Campylobacter is from the faeces of infected chickens, which may contaminate the shared living environment.
The national poultry population in Ethiopia is estimated at 49·3 million [20], and is constituted of about 97% indigenous chicken ecotypes [21]. These birds are predominantly maintained under a traditional free-range, scavenging or semi-scavenging system with minimal inputs for housing, feeding or healthcare. Despite their low production performances in terms of egg and meat output, indigenous chickens are well adapted to the local environmental conditions and fulfil a variety of social and cultural functions, in addition to being both nutritional and economic assets [Reference Dessie and Ogle14].
In the last 20 years, poultry production in Ethiopia has started to be seen as a profitable venture, and more families in urban and peri-urban areas have started to keep small- to medium-sized flocks (about 50–1000 birds) under semi-intensive management [21]. Entrepreneurs have also started to invest in the poultry industry by setting up larger flocks of exotic breeds kept under semi-intensive/intensive management, particularly in the cities of Dukem, Debre Zeit and Nazeret, close to the capital, Addis Ababa [Reference Mammo22]. Efforts have also been made by the Government of Ethiopia since the early 1990s to boost the productivity of indigenous birds kept by poor rural families through its genetic improvement programme, based on the introduction and distribution of exotic breeds by poultry multiplication centres throughout the country [23]. Several non-governmental organizations are also involved in the distribution of intensively reared chickens to smallholder farmers. Although still accounting for a small proportion of Ethiopia's poultry, this on-going shift to more intensive production using commercial breeds may alter the dynamics of zoonotic infections such as campylobacteriosis. While housing chickens may limit their exposure to environmental pathogens and reduce direct and indirect contacts between people and chickens, Campylobacter is known to spread very rapidly in housed flocks and interventions to limit exposure and transmission are likely to be ineffective, impractical or unaffordable in village poultry production settings. Stresses, such as heat stress, or food and water restriction, is known to be detrimental to the chicken's immune function, altering tissue invasion of intestinal bacteria [Reference Leitner and Heller24, Reference Mashaly25]. However, different types or lines of chickens show variation in both the effect of heat stress on their immune response [Reference Tirawattanawanich26] and their ability to regulate the immune response to Campylobacter [Reference Humphrey27]. The links between management system, chicken welfare, stress and the biology of Campylobacter infections remain poorly understood, but have raised concerns about how farming systems may impact on public health risk [Reference Humphrey28–Reference Humphrey30].
As yet in Ethiopia there has been little research examining how the changing dynamics of chicken production may have consequences for public health. The aim of this study was to investigate the impact of production settings and systems in Ethiopia on the epidemiology of Campylobacter, and to consider the potential implications of this evolving production system for human health in the country. Therefore, the main objectives were (i) to determine the environmental prevalence of Campylobacter in different geographical areas, including rural and peri-urban areas; and (ii) to investigate whether Campylobacter prevalence varied in chickens from the same area reared under different production systems.
MATERIAL AND METHODS
Study design
To investigate differences in Campylobacter prevalence in village chickens between geographical areas, a cross-sectional study was conducted between October 2012 and April 2013 in three woredas (administrative regions) in the Oromia region of Ethiopia. Woredas are further subdivided into kebeles, which are the lowest administrative unit in Ethiopia, and may contain several villages or sub-areas. Woredas were purposively selected based on known characteristics of their poultry production systems and bird ecotypes and, for the two rural sites, knowledge that there had been no recent poultry development programmes in the areas.
Study area
Horro woreda is located about 310 km west of the capital, Addis Ababa, and Jarso is about 560 km east. Over 95% of the population live in rural households in both woredas [31]. Horro is considered an area of high agricultural potential and produces an annual food surplus [32], whereas, in Jarso, agricultural land is limited and the area tends to have a food deficit every year, with the poor reliant on food aid to cover the food gap [33]. Debre Zeit (also called Bishoftu) is an urban centre located 47 km east of Addis Ababa within the Ada'a woreda in the East Shewa zone. This woreda contains a number of towns and is food secure [34].
Sampling was conducted in four kebeles in each of the two rural districts, where over 70% of households were estimated to own chickens, but where there were no known semi-intensive or intensive poultry farms. Twenty households were recruited to participate from each kebele; names of potential participants were randomly selected from a list of farmers in each kebele, provided by local development agents. Development agents are employees of the local agricultural extension service, whose role is to provide farmers with access to training, research and technologies (http://www.moa.gov.et/policies-and-strategies). Households were excluded from the study if they did not own at least two chickens: 30% extra additional names of potential participants were included in the random selection to allow for exclusions or non-participation. In the peri-urban area of Debre Zeit, five kebeles were selected for inclusion in the study, all of which contained at least one intensive chicken farm. As chicken ownership was believed to be markedly lower in this woreda, random selection of households was not considered feasible; instead, convenience sampling of households was undertaken based on transect walks through the kebeles and identification of households which kept free-ranging scavenging chickens. A total of 60 households across the five kebeles were recruited in this manner. In addition, in order to investigate differences in Campylobacter prevalence between different production systems, 20 intensive/semi-intensive farms within the same Debre Zeit kebeles were purposively selected for sampling during the same time period as the cross-sectional study. Sample sizes were determined based on available logistic and laboratory resources (i.e. DNA extraction kits) between the three study sites.
Sample collection
The unit of sampling was defined as either a backyard chicken-producing household (i.e. a household that kept at least two chickens under free-ranging scavenging conditions, with or without feed supplementation) or an intensive poultry farm. For the purposes of this study, an intensive farm was defined as one which kept at least 50 birds under confined conditions and provided all feed. On the day of sampling, flock information such as flock size and type was collected by interviewing the householder/farmer. Flock type was broadly grouped into four categories: indigenous ecotypes; Rhode Island Red (RIR), an established multi-purpose breed, or RIR hybrids (normally RIR crossed with local ecotypes, which are difficult to distinguish morphologically from pure RIR birds); Cobb 500, a commercial fast-growing broiler breed; and mixed flocks, which contained both local ecotypes and pure or hybrid RIR birds.
For each unit, a single sample was collected from the ground by walking through the household environment where chicken faeces were concentrated in the case of the extensive system or inside the chicken shed in the more intensive system, using one disposable fabric overshoe (boot sock) worn over the footwear. Boot socks were pre-moistened with sterile physiological saline before use, in order to allow maximum uptake of Campylobacter. To avoid cross-contamination between different flocks, for each sampling a clean disposable plastic overshoe was worn to separate the footwear from the fabric boot sock. After collection, each boot-sock sample was carefully placed into a clean ‘ziplock’ sterile plastic bag and kept under refrigerated conditions until processing. Samples were processed within 6 h to 30 days depending on the district of collection.
Campylobacter detection
Boot socks were processed individually in the Microbiology Laboratory of Addis Ababa University College of Veterinary Medicine and Agriculture. Each boot sock was placed in a sterile plastic stomacher bag with 200 ml sterile saline water, and the mixture was treated in a stomacher at medium speed for 1 min to release detached matter. The boot sock was then left for about 10 min at room temperature to allow the solid material to settle. Without disturbing the sediment on the bottom, 1 ml of the suspension was then transferred to a 1·5 ml sterile Eppendorf tube. The suspension was centrifuged for 7 min at ~12 000 g . The supernatant was discarded and the bacterial DNA was extracted from the pellet using a commercial DNA extraction kit (Promega Wizard Genomic DNA purification kit, USA). After DNA extraction, 50 μl of the supernatant, containing the suspended DNA, was added to 450 μl sterile distilled water, and stored at 4 °C. Samples were then exported under license to the University of Liverpool, where they were analysed with a PCR assay specific for the genus Campylobacter on the basis of 16S rRNA gene sequences (16S rDNA; Table 1) [Reference Linton, Owen and Stanley35, Reference Katzav36].
Primers used in the 16S rRNA PCR assay for identification of the genus Campylobacter (product size 857-bp DNA)
Isolates were confirmed and identified to species level (C. jejuni or C. coli) using a multiplex PCR as described by Klena et al. based on differences in the lpxA gene (Table 2) [Reference Klena37]. All primers used in this work were obtained from Eurofins MWG Operon (Germany) and all PCR constituents were supplied by Thermo Scientific (UK).
Primers used in the lpx gene PCR assay for Campylobacter isolates’ speciation (product size: C. coli 391 bp and C. jejuni 331 bp)
Data analysis
Basic descriptive statistics, graphs and multilevel logistic regression modelling were performed using R software, using the core functions, the plotrix library, the mgcv library and the lme4 library (R Foundation for Statistical Computing, Austria). A multilevel multivariable model was constructed to investigate whether area, production system or breed kept had any effect on the odds of a flock being detected as Campylobacter positive. As Cobb 500 chickens were only kept under intensive systems, and all other types were only kept under backyard scavenging systems, the variables ‘breed’ and ‘production system’ were combined to create a variable called ‘type of flock’, with four categories: ‘intensive-Cobb 500’, ‘scavenging-RIR’, ‘scavenging-mixed’, and ‘scavenging-indigenous’. The functional form of the relationship between flock size (the only continuous explanatory variable) and the outcome was assessed using generalized additive models and suggested inclusion of a linear relationship was appropriate. Backward stepwise model selection was used to identify fixed effects to be retained in the model. The full model included the presence or absence of Campylobacter (at the genus level, as detected by PCR) as the outcome variable and the type of area (urban or rural), the type of flock and the flock size variables as fixed effects. Two-way interaction terms were also considered. Kebele was included as a random variable to account for clustering within geographical locations. Variables with a likelihood ratio test statistic P value ⩽0·05 were retained in the model.
Ethical standards
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
A total of 239 boot-sock samples were collected from the three districts: 79 from Horro, 80 from Jarso and 80 from Debre Zeit. The overall Campylobacter prevalence detected across the three geographical areas using PCR diagnostic techniques was 18·4%. PCR assay for assignment of Campylobacter species was performed on all 44 Campylobacter-positive samples. In total, only 16 (36·4%) samples gave positive results using this multiplex PCR assay, and for all 16 samples a positive result was obtained for C. jejuni.
In the rural Horro region, all flocks sampled consisted purely of indigenous bird ecotypes, while in Jarso, despite a reported absence of current distribution programmes for exotic birds in this rural region, 12 (15%) of the 80 flocks sampled were made up either partially (n = 9), or entirely (n = 3) of RIR or RIR hybrids. In the peri-urban kebeles around Debre Zeit, 28% of the scavenging flocks consisted partly (n = 10), or exclusively (n = 7) of RIR or RIR hybrids. All 20 of the intensive farms sampled reared Cobb 500 broilers.
Higher Campylobacter prevalence was found in the peri-urban (46·3%) compared to the rural (4·4%) environments (Fig. 1; χ 2 = 105, P < 0·001). The prevalence was still significantly higher in the peri-urban flocks (41·7%), even when the intensive flocks were excluded (χ 2 = 45·5, P < 0·001). Of the two rural areas, Horro had the higher prevalence (8·9%) of positive flocks, while no flocks were detected as positive by PCR in the Jarso region.
Proportion of flocks tested positive by PCR for Campylobacter, showing location, bird and farm type. Points are scaled relative to the number of flocks of each type tested.
The final multi-level multivariable model (Table 3) suggested that flocks in the peri-urban area were at considerably greater odds of being detected Campylobacter positive [odds ratio (OR) 21·9, 95% confidence interval (CI) 2·0–235·0], even after controlling for the type of flock. After controlling for type of area, there was insufficient evidence to reject the null hypothesis of no difference in risk between mixed exotic/indigenous and indigenous-only backyard flocks. By contrast, backyard flocks which consisted only of RIR or RIR hybrids and housed flocks of Cobb 500 chickens were at increased risk compared to the backyard mixed flocks (OR 18·1, 95% CI 1·8–183·9 and OR 6·5, 95% CI 1·1–38·7, respectively). Inclusion of flock size or any two-way interaction terms did not significantly improve the model (likelihood ratio test statistic P values >0·05); hence these were not included in the final model.
Final multi-level multivariable logistic regression model of factors associated with detection of Campylobacter by PCR from environmental samples collected from chicken production sites in three regions of Ethiopia
s.e., Standard error; OR, odds ratio; CI, confidence interval; RIR, Rhode Island Red.
* Likelihood ratio statistic P value for overall variable.
DISCUSSION
The greatest risk factor identified in this study for detection of Campylobacter in the environment of the sampled chicken flocks was location in the peri-urban area, where many farms are starting to intensify their production systems. Campylobacter is highly prevalent in intensive production systems in developed countries, with birds often having high levels of infection, posing a public health risk. While it is therefore perhaps unsurprising that the more intensively farmed flocks in the peri-urban areas were at increased risk, it is of particular concern that the greater risk in the peri-urban area persisted despite controlling for the type of flock, suggesting increased risk in all flock types in the peri-urban area, irrespective of breed and production type. Although the high levels of Campylobacter in intensive and semi-intensive flocks are clearly of concern for farm and abattoir workers and for public exposure through contaminated meat (as is the case in more economically developed countries), this study suggests that attention should also be paid to other potential routes of transmission in situations where scavenging poultry remain commonplace. The backyard flocks, which live closely with people, can potentially contaminate the human living and food preparation environment, and infections may be shared with livestock and other animals (including peri-domestic pests) in the household.
Further work is needed to identify the reasons for the increased detection of Campylobacter in flocks in the peri-urban area, compared to the rural areas. Key questions to be addressed should include the impact of the close proximity of backyard production to large intensive farms, the potential effect of the distribution of exotic or improved birds from poultry multiplication centres to smallholder producers, the effect of movement of birds through trade, and other factors which may be associated with differences in the ecology of Campylobacter between peri-urban and rural areas. Furthermore, where there is close inter-species contact, such as was evident in many households in this study, the potential role of transmission from humans or other peri-domestic animals and livestock to chickens should also be considered.
An alternative explanation for the higher prevalence of detection in the peri-urban area, independent of the type of flock, is that there is some other local factor in this region which increases the risk of detection of Campylobacter in this area, such as a more favourable environment, climate or increased presence of wild bird, insect or rodent vectors. One of the features of Debre Zeit is the presence of numerous permanent and seasonal lakes, which provide habitat for birds and insects. The positive flocks in Horro were also located in kebeles where bodies of water are a feature, whereas Jarso, where no positive flocks were identified, has a much more arid climate and no substantial water bodies were present in the kebeles where sampling was carried out. Only a single peri-urban region was sampled in this study, therefore these results may not be generalizable to other areas. However, it is of note that other major urban centres of poultry production and distribution in Ethiopia, such as Awassa and Bahir Dar, are also located close to large permanent lakes.
Of the backyard flocks, the odds of Campylobacter detection in the environmental samples was greatest for those flocks consisting of only RIR birds, or their hybrids. The cause of this is unknown, but factors underpinning this finding may include a genetic susceptibility of this breed, that this breed type is subjected to particular stress when kept under backyard management, or differences in the Campylobacter-exposure risk between these and indigenous flocks, such as bird sources, contact patterns or behaviours. It has previously been reported that survival of RIR-type birds in backyard conditions is poor, compared to indigenous breeds, and they are poorly adapted to scavenging and highly susceptible to disease [Reference Dana38]. It should be considered that such stress might have implications beyond the health and welfare of the bird, if it alters carriage of zoonotic pathogens such as Campylobacter.
A potential limitation of this study was that samples collected from the two rural areas had, of necessity, a longer gap between collection and processing than samples collected in the peri-urban area, which was in close proximity to the laboratory facility. It has not been possible, in this study, to analyse whether this time lag had any impact on the detection rate of Campylobacter from the study samples. While viable Campylobacter decline rapidly in stored samples, especially at temperatures >4 °C, it has been shown that DNA-based PCR methods continue to detect Campylobacter in manure for at least 30 days even in samples stored at temperatures up to 52 °C [Reference Bui39]. However, this also highlights that the PCR-based method used in this study detected Campylobacter DNA, not viable bacteria, so further work is required to better evaluate the public health risk of these findings.
Only just over one-third of Campylobacter-positive samples could be identified to the species level. This is not surprising, as the multiplex PCR [Reference Klena37] was not designed for speciation on DNA extracted from boot-sock samples and, given the amount of DNA present in such samples, sensitivity may be an issue. That all 16 samples identified to the species level were found to be C. jejuni may suggest that this is the most prevalent species. However, the possibility that other species are present should not be discounted, particularly as these may include some species less pathogenic to people. Further investigation with more sensitive methods is needed in order to clarify the relative importance of different Campylobacter spp. and molecular epidemiological approaches are required to determine the extent to which the Campylobacter identified in each area and flock-type are similar.
There is currently much interest in Ethiopia in developing the poultry industry and enhancing the productivity of backyard chickens, and current programmes often advocate the adoption of high-producing breeds and changing from scavenging to semi-intensive management. This can be readily observed in the peri-urban areas, such as Debre Zeit, where exotic chickens such as the RIR have been incorporated into the scavenging flocks, with the aim of enhancing egg production. There is also ongoing research to retain the desirable characteristics of the indigenous poultry types while raising the output of eggs and meat from the scavenging system by selective and cross-breeding of the local ecotypes with exotic breeds. Since this study was performed, ‘improved’ indigenous birds selected over seven generations for increased productivity traits (body weight at age 16 weeks and cumulative egg number at week 45 of lay) in Debre Zeit have been placed on farms in the Horro and Debre Zeit regions to evaluate their performance under semi-intensive conditions. While higher productivity is desirable to increase food availability, it is important that measures to enhance food security are also evaluated in terms of their ability to produce safe food.
Backyard chickens are an important component of families’ livelihoods, and likely to remain so for many years to come [Reference Alders and Pym40], even in the face of increasing urbanization. While chicken production can contribute to improved livelihoods and nutrition, careful consideration is required of all impacts of programmes to develop the chicken production sector if negative human health effects may also occur. Here, we identify a greater risk of environmental detection of Campylobacter in a major peri-urban production area in Ethiopia, compared to rural areas. Further work is needed to determine whether other zoonotic pathogens may be similarly affected. If chicken production in urban/peri-urban areas increases the risk of environmental contamination with Campylobacter (and, potentially, other zoonotic pathogens), then managing these risks will be a vital component of veterinary public health services in these areas. Furthermore, agricultural development programmes aimed at improving chicken production in rural areas, especially where this involves distribution of intensively reared birds or major changes to the management system, need to carefully consider the potential for perturbation to the ecology of Campylobacter in these areas.
ACKNOWLEDGEMENTS
We thank Biotechnology and Biological Sciences Research Council (BBSRC), the UK Department for International Development (DFID) and the Scottish Government for providing funding for the ‘Reducing the impact of infectious disease on poultry production in Ethiopia’ project under the Combating Infectious Diseases of Livestock for International Development (CIDLID) programme (BB/H009396/1). Further funding was provided by the Institute of Infection and Global Health at the University of Liverpool. We thank the Chicken Health for Development project team members and the study participants for their time and assistance. Funders had no involvement in study design; the collection, analysis and interpretation of data; the writing of the manuscript; or in the decision to submit the article for publication.
DECLARATION OF INTEREST
None.
Open access
