Ethiopia is estimated to have a chicken population of 42 million  of which over 95% are kept under a traditional backyard scavenging system [Reference Tadelle, Kijora and Peters2]. Backyard poultry represents an important part of the national economy and provides about 98·5% and 99·2% of national egg and poultry meat production, respectively [Reference Tadelle, Kijora and Peters2]. Lack of knowledge about poultry production, limitations of feed resources and prevalence of diseases (Newcastle, coccidiosis, etc.) are among the constraints of backyard poultry production in Ethiopia [Reference Ashenafi3].
Toxoplasma gondii is an obligate intracellular protozoan that infects humans and a wide range of mammals and birds [Reference Dubey4]. Felines are the only definitive hosts that shed oocysts in their faeces, although intermediate hosts can harbour infective tissue cysts. Most feline infections occur post-natally through ingestion of infected tissue cysts or rarely oocysts, although congenital infections can occur [Reference Dubey and Jones5].
Humans acquire T. gondi infections from ingestion of oocyst-contaminated soil and water, from tissue cysts in undercooked meat, by transplantation, blood transfusion, laboratory accidents, or congenitally [Reference Dubey4, Reference El-Awady, El-Hosseiny and Ismail6, Reference Dubey7]. The prevalence of Toxoplasma infection is estimated to range from 30% to 50% worldwide, but the majority of infected humans remain asymptomatic [Reference El-Awady, El-Hosseiny and Ismail6]. However, it should be noted that T. gondii infection may cause severe illness in immunodeficient people and pregnant women [Reference Dubey4].
The prevalence of T. gondii infection in free-range chickens is a good indicator of the prevalence of T. gondii oocysts in the environment, due to the habits of chickens that feed by scratching the earth, which facilitates greater access to hidden faeces of cats [Reference Dubey7]. Thus, free-range chickens play an important role in the epidemiology of T. gondii in the rural environment, perhaps more so than rodents, because chickens are clinically resistant to T. gondii and live longer than rodents. Moreover, T. gondii-infected chickens are an efficient source of infection for cats, because cats fed naturally infected chicken tissues may shed millions of oocysts [Reference Dubey7, Reference Dubey8].
A thorough knowledge of the extent of contamination of the environment or areas surrounding human habitation by T. gondii oocysts is important for control purposes. In the developing world T. gondii seroprevalence in free-range chickens may be up to 65% and tissue cysts have been found in 81% of seropositive birds [Reference Kijlstra and Jongert9]. In Africa, little research has been conducted on T. gondii infection of free-range chickens. Lindstrom et al. [Reference Lindstrom10] from Uganda, and Dubey et al. [Reference Dubey11] from Ghana, report prevalences of 47% and 64% in chickens, respectively. The epidemiology of T. gondii infection in humans and animals is poorly understood in Ethiopia. Chickens are a good source of protein for humans and very much associated with food, economy, culture and religion [Reference Reta12], yet they pose a strong disease risk for humans via T. gondii. However, there seems to be only one previous study available regarding T. gondii status in Ethiopian chickens. That study was conducted in one area only (Addis Ababa) on free-range chickens and reported a prevalence of 38·2% from 125 chicken sera examined by modified agglutination test (MAT) and viable T. gondii was isolated from only 1/43 (2·3%) bioassayed seropositive chickens [Reference Tilahun13]. Therefore, the present study was conducted with the objective of estimating animal- and flock-level seroprevalence and assessing the potential risk factors associated with T. gondii infection in free-range chickens (Gallus domesticus) of East and West Shewa zones, Central Ethiopia.
MATERIALS AND METHODS
Description of the study area
The study was conducted in three selected districts (Ambo, Adea, Fentale) in East and West Shewa zones of Oromia Regional State, Central Ethiopia (Fig. 1). The study areas are separated from each other by 150–289 km and represent the highland [⩾2300 m above sea level (masl)], midland (1500–2300 masl) and lowland (<1500 masl) agro-ecologies of the region.
Ambo district (longitude 37° 32′ to 38° 3′ E; latitude 8° 47′ to 9° 20′ N) is located in Western Shewa Zone of Oromia Regional State. The altitude within the district ranges from 1400 to 3045 masl. The majority of the locations in Ambo district are highland. The range of annual rainfall and temperature is 800–1000 mm and 15–29 oC, respectively. The mean temperature is 18·6°C.
Adea district (longitude 38° 58′ E to 39° 22′ E; latitude 08° 22′ N to 8° 56′ N) and Fentale district (longitude 39·93° E to 39° 56′ 0″ E; latitude 8·975° N to 8·58′ 30″ N) are located in East Shewa Zone at a distance of 45 km and 190 km from Addis Ababa, respectively. The altitude of Adea district ranges from 1500 to >2000 masl, although the majority of the territory is midland. Its average rainfall is ~839 mm with temperature ranging from 7·9°C to 28°C.
Fentale district ranges in altitude from 955 to 2007 masl, but comprises mainly lowland. Its average rainfall and temperature are ~553 mm and 29–38°C, respectively. Fentale district has an arid to semi-arid climate and the production system is predominantly pastoral and agro-pastoral. Sedentary farming, dominated by an extensive type of management system, is a feature of the highlands and midlands of Ambo and Adea districts. However, semi-intensive farming is practised in some urban and peri-urban areas.
Study animal population and production system
Previous reports on T. gondii infection in sheep [Reference Gebremedhin14], goats [Reference Zewdu15] and women of childbearing age [Reference Gebremedhin16] from the same study districts initiated this investigation into the level of T. gondii oocyst contamination of the environment by using free-range chickens as indicators.
The biological characteristics and management properties of village chickens in Ethiopia have previously been well described [Reference Reta12, Reference Fisseha, Azage and Tadelle17]. The chicken production systems in the selected districts are dominated by extensive/free-range system, where main feed sources are insects, worms, cereals and plants. In this system chickens are set free during the day and spend the night on perches in or around the main house.
Study design and sample size
A cross-sectional study design was conducted from October 2012 to May 2013 to estimate the seroprevalence and to identify the potential risk factors in the epidemiology of T. gondii infection in free-range chickens of the three study districts.
In the absence of any well-designed previous studies on chicken toxoplasmosis in Ethiopia, a 50% expected prevalence (p) and a 95% confidence interval (Z = 1·96) with a 5% desired absolute precision (d) was considered to arrive at the required sample size, using the formula: n = p(1−p)Z 2/d 2) [Reference Thrusfield18]. However, the calculated sample size (n = 384) was increased to 601 for better accuracy. The total chicken population of the study area was 196 644. The required sample size was distributed on the basis of the chicken population of each respective district: Ambo (n = 286), Adea (n = 209) and Fentale (n = 109). The study ‘kebeles’ (kebele refers to the smallest administrative unit) from Ambo (n = 7), Adea (n = 5) and Fentale (n = 4) were chosen purposely based on accessibility. The average number of chickens per household in Ethiopia is 2–6 hens and 1–3 cocks [Reference Reta12, Reference Fisseha, Azage and Tadelle17]. The lower ratio (i.e. 1 male : 2 females) was considered in order to sample three chickens per household using a simple random sampling technique. Accordingly, 88, 65 and 37 households were sampled from Ambo, Adea and Fentale districts, respectively. The households of selected kebeles were sampled by considering the relative distance from one cluster to another, since rural communities live in clusters. Finally, 2–3 households per cluster were sampled.
A questionnaire survey was used to collect data regarding the potential risk factors including breed (local, cross, exotic), age (⩽6 months, 7–12 months, ⩾13 months), sex (male/female), flock size (small, ⩽4; medium, 5–10; large, ⩾11 chickens), management systems (extensive, semi-intensive, intensive), altitude (lowland, ⩽1400 masl; midland, 1500–2200 masl; highland ⩾2300 masl), presence of cat (yes/no), presence of dog (yes/no), presence of feral cat (yes/no), type of residence (rural/peri-urban), owner education (literate/illiterate) and water source (tap, river, pond, well). The questionnaire also included origin and housing of chickens, type of feed and housing for cat(s); and awareness and knowledge of the respondents about toxoplasmosis.
Sample collection and transportation
About 2–3 ml blood samples were collected from the wing vein of each chicken, using a disposable plain vacutainer tube and needle. Samples were labelled and placed in a slant position for a few minutes until clotting. Centrifugation at 2250 g for 5 min was performed for some sera which failed to separate overnight. The sera were collected in 1·5 ml Eppendorf tubes, labelled and transported in an ice box with ice packs to the Ethio-Belgium laboratory of the College of Veterinary Medicine and Agriculture, Addis Ababa University, Debre-Zeit and stored at −20°C until tested.
All sera samples collected were examined for the presence of antibodies (IgG) against T. gondii by MAT (Toxo screen DA, bioMérieux, France) according to the manufacturer's instructions. Sera were assayed at a screening dilution of 1 : 40 and 1 : 4000. A titre of 1 : 40 or 1 : 4000 or both was considered indicative of T. gondii exposure. Sedimentation of antigen at the bottom of the well, and clear agglutination above half of the well at either dilution were recorded as negative and positive results, respectively. MAT-positive samples were titrated to obtain the endpoint titre. Positive and negative controls were included in each test.
Data management and analyses
The data generated from the field and laboratory investigations were entered and coded using Microsoft Excel 2007 and analysed using Stata v. 11.0 for Windows (StataCorp., USA). Descriptive statistics were used to summarize the data. Flocks containing at least one seropositive animal were considered positive. Seroprevalence was calculated by dividing the total number of seropositive chickens by the total number of chickens tested. Similarly, flock level seroprevalence was calculated as the number of flocks with at least one positive animal, divided by the total number of flocks tested. The variables considered as potential risk factors for T. gondii seropositivity were selected based on existing literature. During analyses, categories of variables, e.g. cross-breed and exotic breed, urban and peri-urban locations, intensive and semi-intensive management were merged. Association of risk factors (independent variables) with dependent variable (T. gondii seropositivity) was assessed initially using cross-tabulation. Strength of association was assessed using univariable logistic regression to compute the odds ratio associated with potential risk factors. Non-collinear variables, that presented a P value ⩽0·25 in univariable analyses, were included in the multivariable logistic regression model. In line with this, the sex, flock size, presence of feral cats and source of water were not entered in the animal-level multivariable model, due to a P value >0·25 in the univariable analyses, while feed was omitted for management due to collinearity. Similarly, during flock-level analysis, district was dropped from joining the model due to collinearity with altitude. Model fitness was assessed by the Hosmer–Lemeshow goodness-of-fit test [Reference Dohoo, Martin and Stryhn19]. The receiver-operating characteristic curve (ROC) was used to evaluate the reliability of both individual- and flock-level models. The 95% confidence level was used and results were considered significant at P ⩽ 0·05.
Overall seroprevalence and antibody end titre in free-range chickens
The overall seroprevalence of T. gondii infection in the study districts were 30·5% [95% confidence interval (CI) 26·76–34·14] and 54·2% (95% CI 47·06–61·36) at animal and flock level, respectively. The highest (animal level 41·4%, flock level 72·3%) and lowest (animal level 9·2%, flock level 21·6%) seroprevalences were recorded in chickens from Adea and Fentale districts, respectively. There was a statistically significant difference in seroprevalence of T. gondii infection between districts, both at animal (χ 2 2 = 34·9632, P ⩽ 0·001) and flock (χ 2 2 = 24·4104, P ⩽ 0·001) levels.
Titration was conducted for each seropositive serum to evaluate the endpoint titre at which the MAT antibody is no longer detectable. Subsequently, 183/601 chickens were seropositive with an end titre of 1 : 60 in 46, 1 : 180 in 28, 1 : 540 in 29, ⩾1 : 1620 in 48, 1 : 6000 in 22, 1 : 18 000 in five, 1 : 54 000 in one, and ⩾1 : 162 000 in four.
Risk factors of T. gondii infection in free-range chickens
The variables altitude, age, management, and presence of cats were all significant factors associated with T. gondii seropositivity in both univariable and multivariable logistic regression analyses (P ⩽ 0·05). Presence of dogs and breed were significantly associated with T. gondii seropositivity only in univariable and multivariable logistic regression model, respectively (P ⩽ 0·05, Table 1).
N, Number tested; prev., prevalence; OR, odds ratio; CI, confidence interval; WBCS, wheat bran, cereals and supplements.
Univariable logistic regression analysis revealed that the variables altitude, location, and presence of cats and dogs were all significantly associated (P ⩽ 0·05) with T. gondii seropositivity. However, only altitude and presence of cats were found to be independent predictors of seropositivity by the multivariable logistic regression model (Table 2). From the flock-level multivariable logistic regression model it was evident that not all factors were equally responsible as a risk factor for T. gondii seropositivity, implying that only some factors are sufficient risk factors. Table 3 shows the best-fitting model that contains all the appropriate risk factors for T. gondii seropositivity at both animal and flock levels.
N, Number tested; prev., prevalence; OR, odds ratio; CI, confidence interval; WBCS, wheat bran, cereals and supplements.
OR, Odds ratio; CI, confidence interval.
Animal level: Hosmer–Lemshow χ 2 = 0.34, P = 0.95, receiver-operating characteristic curve = 0.68.
Flock level: Hosmer–Lemshow χ 2 = 0·08, P = 0·96, receiver-operating characteristic curve = 0·71.
Respondents' awareness of T. gondii infection and its risk factors
From a total of 190 respondents interviewed, 95 (50·0%) owned domestic cats. Table 4 shows that 90·5% of respondents reported that their cats lived completely outdoors. All (100%) respondents indicated that the cats had no separate accommodation while 85·3% reported that the cats ate raw animal products as well as household leftovers. Most respondents do not have adequate knowledge about the health risk posed by cats to animals and humans. This health risk to animals showed no significant difference between districts (χ 2 2 = 1·6161, P = 0·446). However, there was a significant difference with respect to awareness about health risks of cats to humans across districts, in that awareness was considerably higher in Fentale (37·8%) than Ambo (10·2%) and Adea (15·4%) districts (χ 2 2 = 14·1072, P = 0·001). The sanitation of chicken houses was judged as poor in 93·7% of cases, while 97·9% of respondents reported that they did not have a separate house for chickens (free ranging during the day and spending the night on perches in or around the owners' home). Dead chickens were simply thrown into open fields by 72·1% of the respondents (Table 4).
T. gondii occurs widely in humans and animals, including domestic poultry throughout the world. The distribution of T. gondii is believed to be best detected by using its levels of distribution in free-range chickens [Reference Dubey7]. To the best of the authors' knowledge, this is the first detailed report on seroprevalence and risk factors of T. gondii infection in free-range chickens in different agro-ecologies of Ethiopia. The choice of MAT as a diagnostic tool for the current study was made because of its high sensitivity (96·22%) and specificity (98·8%) in chickens compared to other serological tests [Reference Dubey7, Reference Dubey20].
An overall animal- and flock-level seroprevalence of 30·5% and 54·2%, respectively, was documented. The highest animal-level (41·4%, 95% CI 34·62–48·06) and flock-level (72·3%, 95% CI 61·27–83·34) seroprevalences were recorded from Adea district while the lowest animal-level (9·2%, 95% CI 3·72–14·63) and flock-level (21·6%, 95% CI 8·09–35·16) seroprevalences were recorded from Fentale district. The variation in seroprevalence between districts can be defined by the difference in temperature, moisture and management practices across the studied districts.
The high overall animal-level seroprevalence of the present study is in agreement with the report of Tilahun et al. [Reference Tilahun13] from Addis Ababa, Ethiopia (38·4%). The high seroprevalence of T. gondii in free-range chickens in the present study also corroborates well with the high prevalence of oocyst shedding (19%, 7/36) previously reported in feral cats which are commonly fed on dead avian carcases in Addis Ababa [Reference Dubey21]. Surveys conducted in different countries around the world report prevalences of T. gondii infection in free-range chickens ranging between 5·5% and 100% [Reference Lindstrom10, Reference Dubey22–Reference Dubey, Weiss and Kim24]. The differences in the seroprevalence observed between the current and above-mentioned studies may relate to differences in relative cat densities, management, study design, the number of chickens examined, sanitation, housing of chickens, type of serological tests used and the cut-off value reported [Reference Dubey4, Reference Dubey7, Reference Beltrame25].
Epidemiological data on risk factors of T. gondii infections in chickens are generally limited across the globe. In the present study, the potential risk factors of T. gondii infection in free-range chickens were assessed. Accordingly, altitude, presence of cats, age, management and breed were to a large degree associated with T. gondii seropositivity (P < 0·05), whereas risk factors like sex, location, presence of feral cats, education level of owner, type of feed, source of water and flock size did not show statistical association with T. gondi infection (P > 0·05).
The likelihood of acquiring T. gondii infection in chickens reared in the midland areas is twice that of chickens reared in lowland areas (P = 0·026). This may be due to the fact that the moist and warm macro-climate of midland areas supports survival of oocysts and its higher infection rate, compared to the dry and hot climate of lowland areas which decreases the survival and infection rate of T. gondii [Reference Dubey4, Reference Tenter, Heckeroth and Weiss26]. Relatively lower seroprevalence has previously been reported in lowland sheep and goats compared to highland and midland areas [Reference Gebremedhin14, Reference Zewdu15]. The influence of the environment on the epidemiology of toxoplasmosis has been well documented [Reference Dubey20, Reference Dubey22, Reference Tenter, Heckeroth and Weiss26, Reference Yang27].
The seroprevalence (45·8%) in older chickens was significantly higher than the seroprevalence in young chickens (21·6%, P < 0·001). This difference is partly due to an increase in the cumulative effect of exposure to the parasite by adult chickens, i.e. as the age of chickens increases the likelihood of encountering oocysts of T. gondii from the environment also increases [Reference Dubey4]. There have been numerous reports indicating age-related increase in seroprevalence in chickens worldwide [Reference Tenter, Heckeroth and Weiss26, Reference Alvarado-Esquivel28].
The odds of acquiring T. gondii infection by chickens kept under an extensive management system was nearly seven times higher compared to chickens managed under semi-intensive and intensive management systems (P = 0·021). This variation may be explained by the fact that chickens kept under an extensive system are exposed to T. gondii oocysts shed by cats in the environment since chickens scratch the soil to collect feed, whereas chickens under intensive and semi-intensive management systems receive better hygiene and have limited or no access to the contaminated environment, since they are kept in confinement [Reference Dubey4, Reference Zhu29]. The influence of management on seroprevalence of chicken toxoplasmosis has previously been well documented [Reference Xu30].
A significantly high seroprevalence of T. gondii infection in free-range chickens kept in areas where cats are present (P = 0·009) is obviously an indication of the dissemination of cat-originated oocysts in the soil close to the households. The frequent cohabitation of chickens with cats in the same environment was observed during sample collection. An increased risk of infection in the presence of cats has been reported by different studies [Reference Dubey7, Reference Tenter, Heckeroth and Weiss26, Reference Zhu29].
Local breeds of chickens have been claimed to have better disease resistance than exotic breeds [Reference Tadelle, Kijora and Peters2]. In this study, the likelihood of getting T. gondii infection by cross-bred and exotic-bred chickens (50·0%) was significantly higher than for locally bred chickens (29·3%, P = 0·006). A possible reason for increased risk of infection in the exotic and cross-breeds may be due to higher susceptibility to T. gondii infection or the preference of owners to keep these chicken breeds for a longer period as potential breeders for high production purposes, hence there is an increased chance of exposure to the possible source of infection [Reference Tadelle, Kijora and Peters2, Reference Dubey8]. However, the role of breed in predisposing chickens to T. gondii infection needs further investigation.
With regard to the owners' awareness about the role cats play in the transmission of disease to humans and animals, 17·4% (33/190) of respondents reported association of cats in the transmission of diseases like asthma and parasites to humans, but none of the interviewed owners were aware of the health risk of cats to other animals. None of the respondents were aware of toxoplasmosis. Similarly, none of the respondents kept their cats in a separate house. Cats were fed with raw animal products like viscera of ruminants and chickens slaughtered for human consumption, and meat of dead animals, which might have T. gondii cysts, hence boosting the spread of toxoplasmosis. Moreover, 97·9% of respondents keep their stock free range (outdoors) during day-time hours and at night used a perch housing system, where the sanitation is poor (93%). Furthermore, about 92% of respondents used unsafe methods when disposing of chicken carcases (throw away into fields or fed to pets). Poor housing, feeding and management of free-range chickens and cats together with poor sanitation and the lack of awareness of toxoplasmosis create favourable conditions for the high prevalence of the disease and successful continuation of the life-cycle of the parasite in the study areas.
In view of home slaughtering of free-range chickens, limited refrigeration facilities and poor sanitation in Ethiopia in general, and the study districts in particular, the high seroprevalence in free-range chickens indicates a high likelihood of transmission to humans, unless chicken meat is well cooked and good kitchen sanitation is practised [Reference Dubey7].
In conclusion, T. gondii antibodies are widespread in free-range chickens in Central Ethiopia suggesting high contamination of the living environment with oocysts of T. gondii and that the meat from free-range chickens may be an important source of infection for humans. Management, age, presence of cats, altitude and breed are epidemiologically important predictors for T. gondii seropositivity. The general absence of awareness of toxoplasmosis may play an important role in the epidemiology of the disease. Education of farmers about toxoplasmosis and further studies to elucidate the burden of toxoplasmosis in animals and humans warrants consideration.
The financial support of Addis Ababa University (Combating HIV/AIDS and promoting maternal and child health project) as well as the partial financial support to Maria Vitale (grant no. RF 2007/RC 2009 from the Italian Ministry of Health) are greatly appreciated. The study could not have been conducted without this support. We are grateful to Margareta Langbacka Walker, advisor and trainer in the English Language Improvement Centre at Ambo University, for her valuable comments on the language aspect. The authors also extend their gratitude to the owners of chickens for their collaboration during the study.
DECLARATION OF INTEREST