Identication of Canine Brucellosis in Three Province, Ethiopia

Background: In Ethiopia, brucellosis has been reported targeted on bovine, occasionally on shoat, and rarely on camels. An investigation of the disease Brucellosis in the neglected companion animals is scared in Ethiopia. The objective of this study was to identify canine brucellosis in Batu town, Alage and Naka village through cross sectional approaches. A total of 389 serum samples (207 from Batu, 107 from Alage, and 75 from Naka) were collected by restraining dogs with a portable and safe modied dog crush, invented by this author. Blood samples were collected from ear vein and sera were screened for Brucella antibodies using different serological tests. RBPT prepared from the smooth strain B. abortus antigen and CFT was used as a screening test and conrmatory test, respectively. Furthermore, all sera samples had also screened by RBPT canis antigen (rough strain); and those positive were considered the cause for B. canis infection. Results: Using RBPT smooth strains, 21(5.4%; CI: 3.35, 7.96) were positive and 19(4.88%; CI: 2.7, 7.0) were conrmed by CFT. Besides, 34 (8.74%; CI: 5.92, 11.56) were positive for RBPT canis rough strains. Relatively, higher proportion of anti B. canis antibodies had seen in Batu (11.59%) followed by Alage (5.61%), and Naka (5.33%). Sex, living condition, and history of obstetrical problem were signicantly associated with the occurrence of canine brucellosis (p< 0.05). Odd of canine brucellosis due to smooth and rough strains in outdoor dogs were 4.72 and 6.42 times higher compared with indoors, respectively. This is true the fact that outdoors had a chance of getting infected aborted wastes when roaming freely. Conclusion: This study suggests that canine brucellosis is prevalent in the province. The seropositivity could give an insight that, the awareness of the people toward the disease was also the gap in the study area. Hence, this warrants public education among the community is recommended.

Since the rst report of brucellosis in the 1970s in Ethiopia, the disease has been noted as one of the important livestock diseases and has been reported from different localities, commonly targeted on bovine, occasionally on sheep and goats, and rarely on camels (9,10) Observably, many Ethiopians keep dogs as a domestic pet both in rural and urban communities. However, the practice of providing due care to these human companions is very rare. Owners feed them family food leftover and taking pets to veterinary clinics for medical treatments is considered as something of a luxury. In urban areas, it is common to nd a large number of stray dogs roaming freely in the streets scavenging for their survival. If irresponsible owners continue to allow indiscriminate growth of the pet dog population, they will be added to the stray dog population. In this event, brucellosis infectivity rate in the stray dog population continues and infected dogs will increasingly contaminate the environment with aborted fetal tissue, vaginal discharges, faces, ejaculate and urine. The role of infected dogs in spreading of B. abortus, B. suis, and B. melitensis to neighboring herds, ocks and humans had reported by (11).
However, despite of these risk factors that serve as sources of infection to other domestic animals and human, to the best of knowledge, canine brucellosis had not studied in the country. Even in Africa, except in Nigeria (7), Zimbabwe (12), and South Africa (13), there is lack of information on canine brucellosis. This epidemiological investigation was therefore, designed to estimate the sero-prevalence and associated potential risk factors of brucellosis in dogs in selected areas of East Shoa Zone, Ethiopia

Study Areas
The study was conducted in Batu, Alage and Naka village as seen in (Fig. 1). The college is positioned at 217 km southwest of Addis Ababa and 32 km west of Bulbula town; near the Abijata and Shala lakes of the Ethiopian Rift Valley. Naka village is located at North east of Alage in Adami tulu jido kombolcha district, Oromia regional state. Both the college and Naka are geographically located at a longitude of 38°30 East and latitude of 7°30'North, with an altitude of 1600 m.a.s.l. The mean annual minimum and maximum temperature range from 11 to 32°C, respectively. The areas have the mean annual rain fall of 800 mm and three distinct seasons; a short rainy season (March to May), a long rainy season (June to September) and a dry season (October to February) (14,15) Whereas, Batu is a town and separate district, located on the road connecting Addis Ababa to Hawasa in the East Shewa Zone of the Oromia Region of Ethiopia. It has latitude of 7°56′North and longitude of 38°43′East, with an elevation of 1643 meters above sea level. According to CSA 2007, the Region had an estimated of 17,214,540 cattle, 6,905,370 sheep, 4,849,060 goats, 959,710 horses, 63,460 mules, 278,440 asses, 139,830 camels, 11,637,070 poultry, and 2,513,790 beehives. However, dog population is not de ned except those 300,000 dogs found in Addis Ababa (16).
A cross sectional study was carried out from November 2017 to May 2018 in Alage, Batu and Naka; using serological tests on dogs' sera; so as to determine the seroprevalence of canine brucellosis. The areas were conveniently selected based on the abundance of dogs. The study had also involved face to face interview.

Study Population
Owned dogs, which were older than 6 months, found in Alage, Batu and Naka; were the target population. None of them were also vaccinated against brucellosis. Study animals were performed through owner's oral consent and study was made with the after obtaining ethical clearance letter from the Board of Research and Ethical Review Committee. The oral-only consent from the owners was approved by the given legal ethics committee.

Sample Size and Sampling Method
The required sample size of dogs was determined based on expected prevalence of brucellosis and the desired absolute precision stated on (17).
Where: n = the required sample size; P = estimated prevalence = 0.5; z = level of con dence as 1.96 and d = desired precision level = 0.05.
Canine brucellosis hadn't studied previously in the study area and the country. Therefore, based on an estimate of 50% prevalence, 95% con dence interval, and 5% absolute precision, a sample size of 384 was calculated. There was not any recorded data indicating the total dog population, hence dog owners and their dogs found in each location had registered (sampling frame had prepared). A total of 826 owned dogs (443 in Batu, 224 in Alage, and 159 in Naka) were then identi ed. Thence, using proportional strati ed sampling method, the total population had divided in to three strata based on their locations; and the proportion of minimal number of dogs to be sampled in each location had been calculated, which resulted 206(53.63%) from Batu, 104(27.12%) from Alage, and 74(19.25%) from Naka.
As it was very di cult to take blood sample from those dogs by going to every owners' home, free dog's rabies vaccination campaign had prepared. After that, following establishing many and suitable vaccination centers in all areas, all dog owners had informed to freely vaccinate in nearby vaccination centers (detail of the procedure is described in the following page). However, considering some owners may not vaccinate due to inconvenience, random selection of individual dogs from the sampling frame was not performed. Instead, as described by Stevenson (2005) after calculating the sampling interval (k = 2 or from every second vaccinated dog), owners and their dogs had again registered during the vaccination program; and thenceforth a systematic sampling method had been used from the list; to sample individual dogs. Thus, a total of 107 (Alage), 75 (Naka), and 207 (Batu) owned dogs had sampled.

Data Collection Approaches
Data was collected using serological tests, and by use of interview.
Approach 1: Restraining approaches: For restraining dogs, a portable and safe modi ed dog crush was employed. The researcher invented the crash also evaluated and certi ed by Alage ATVET College.
Currently it had registered in Ethiopia Intellectual Property O ce.
Approach 2: Rabies vaccination campaign: In collaboration with Alage and National Veterinary Institute (NVI, Ethiopia), the researcher had prepared free and mobile rabies vaccination campaign in the study areas. To do so, consent had made with all administrators of the locations. As vaccinating all dogs in one place and at one time was very di cult, both for the researcher and for the owners who found at distant, suitable time schedule had prepared and different vaccination centers had established in residential areas near to the community. Thence, all residents of the locations were informed by loudspeakers, school mini-medias and notice; to freely vaccinate their dogs on their schedule. Parallel to the vaccination program, after describing the objective of the research, a verbal consent had made with owners of the dogs to take a blood sample. Additionally, a questioner was lled by face-to-face interview.
Approach 3: Blood sample collection: After proper restraining, about 6-8 ml of blood was aseptically collected through the saphenous vein of each sampled dog using syringe and plain vacationer tubes. Each sample was coded and had transported to Alage ATVET College Department of Animal Health, Microbiology laboratory. The blood samples were allowed to clot and centrifuged at 3000 rpm for ve minutes. Serum samples were then decanted, transferred in to labeled cryo-vials, and screened by RBPT in the college laboratory. Dogs' sera tested positive by RBPT were stored at -20 o C and lastly sent to National Veterinary Institute (NVI), Bishoftu, Ethiopia for further con rmation.
Approach 4: Serological tests: Rose Bengal's Brucella antigens (B. abortus and B. canis) and their control sera (sourced from China Institute of Veterinary Drugs Control), and CFT Brucella antigen, control sera, and complement (Bg vv, Germany and China Institute of Veterinary Drugs Control), were employed each for RBPT and CFT tests, respectively. Samples were then considered as positive for canine brucellosis (due to smooth strains), if they were positive using CFT. However, due to the lack of serological tests prepared from rough strains of Brucella species in Ethiopian laboratories, samples positive for RBPT with B. canis antigen were considered as positive for canine brucellosis due to rough strain (B. canis infection).
i. Rose Bengal Plate Test: All serum samples were screened by RBPT containing B. abortus antigen.
Furthermore, to know B. canis infection in dogs, a RBPT prepared from B. canis antigen was used as described by (18). Using a micropipette, 1 drop (30 µl) of the test serum was placed on one spot of the slide. Using another pipette, an equal volume of RBPT antigen was placed close to the test serum on the slide. Using an applicator stick, the antigen and the test serum were mixed thoroughly; the slide was then hand-rocked for about 4 minutes after which the slide was examined for agglutination under a good source of light. Formation of pink granules (agglutination) was recorded as positive while absence of pink granules (agglutination) was recorded as negative.
ii. Complement Fixation Test: Positive samples for RBPT (containing B. abortus antigen) were further con rmed by CFT. The con rmation of dog sera was undertaken at NVI, Department of Immunology. All the reagents required for CFT were evaluated by titration. A sheep Red Blood Cell (SRBC) suspension were prepared before being used in the test proper. The preparation of reagents and CFT procedures were performed according to the protocols of the Federal Institute for Consumer Protection and Veterinary Medicine Service Laboratory, Berlin, Germany (19). The CFT test was regarded as positive when the reading is as complete xation or partial hemolysis and as negative (0) when there is complete hemolysis (20).
Approach 5: Interview: The study had clearly explained to the respondents and informed consent was obtained. A questionnaire was prepared in English and using trained assistant researchers who had a mother tongue of the local language 'Affan Oromo', validation of the questionnaire had conducted by a pre-testing on total of 10 individuals; to analyze and validate the degree to which the questions were properly understood or misunderstood, the degree to which individuals within a group interpreted the questions differently, the effectiveness of the questions in soliciting the proper information, and any areas of information which were neglected by the proposed questionnaire. Once analysis has been completed, some questions were modi ed. Lastly, they were interviewed on their demographic factors, awareness toward the disease, and on history of the animals and possible factors associated with occurrence of brucellosis. Questions regarding religion, assisting bitches during whelping were not included due to being a sensitive topic culturally.

Variables collected
The information for hypothesized explanatory variables was gathered from the dog owners, swine farm managers/owners, farm employees and farm records. Factors like age, sex, history of obstetrical problems (Abortion, retained placenta, abnormal vaginal discharges, infertility…etc for females; and enlargement of testicle, scrotal edema, and scrotum dermatitis for males), and living status (for dogs) were recorded for each animal. Age of dogs were strati ed into three categories (≤ 2 years, 2-4 years, and > 4 years); and for pigs in to two (≤ 3 years and > 3 years). Furthermore, living status (condition) of dogs was categorized as indoors, outdoors, and semi-indoors (modi ed from (1) and (21).

De nition of terms
Based on the aspect of colonies on agar plates, which is in accordance with the cell surface and LPS structure, Brucella may occur either as smooth or rough species (22).
Smooth strains of Brucella species -Theses Brucellae that express full LPS molecule (S-LPS) that is anchored in the outer membrane. They carry complete S-LPS and have a smooth (S) phenotype, so termed after the smooth texture of the colonial surface. This includes the zoonotically more relevant Brucella species, B. melitensis, B. suis, and B. abortus (22)(23)(24). To detect Brucella infection serologically due to smooth strains, the antigen of the serological test should prepare from smooth strains (25)(26)(27).
Rough strains of Brucella species -They express R-LPS that lack the O-antigen, a trait linked to their reduced virulence and include B. ovis and B. canis. To serologically detect Brucella infection due to rough strains, the antigen of the serological test must prepared from either of the rough strains; as the surface antigens of smooth Brucella spp. do not cross-react with these (25)(26)(27). Generally Serologic tests that use suspensions of smooth phase Brucellae are useless in diagnosing B. canis infections (28).
Living condition/ living status/ maintenance condition -Dog's living condition varies from owner to owner, which has its own in uence for the epidemiology of canine brucellosis (29,30). Researchers had included it as a risk factor and studied about canine brucellosis in association with this factor. They had then further categorized it (eg. as outdoors, indoors, shelter, foster and stray). Therefore, in the present study, considering the existing dog's management system of Ethiopian dog owners, the factors had further categorized in to three sub categories; and are de ned below accordingly.
Indoors -In the present study, the term referred to those con ned dogs (i.e they didn't had any contact with outside animals and their food was provided by their owners).

Semi-indoors
-It referred to partially free dogs. They might chain (con ned) for half of the day or some hours, but also had freedom to go free, mostly at night. Hence, they had contact with other animals and though their food was principally provided by their owners, they sometimes scavenge outside.
Outdoors -The term indicated to those always free and mostly scavenging dogs. They differ from stray dogs because they had owners and can rarely feed at home.

Data analysis
The collected data had entered into Microsoft Excel Spread Sheet program and statistical analysis had computed using Fisher's Exact test and logistic regression using STATA-12 version. The total prevalence was then calculated by dividing the number of animals seropositive for CFT and RBPT (for B. canis infection); to the total number of animals sampled. The association between risk factors and seropositivity to anti Brucella antibodies was considered as signi cant at p < 0.05 and odds ratio (OR) had used to measure the magnitude of the association between each risk factor. Lastly, the demographics of respondents and their awareness and practices toward the disease were determined using descriptive statistics.

Results
As the result is summarized in ( Analysis for association between locations of the animals and Brucella infection was carried out using Fisher's exact test (Table 2). There was no signi cant association observed between the study areas and seroreactivity to both smooth and rough types of Brucella infection (p≥0.05). Even so, using CFT, a relatively higher proportion of anti B. abortus antibodies was observed in Alage (5.61%) followed by Batu town (4.83%) and Naka (4%); and using RBPT canis , higher proportion of anti B. canis antibodies had seen in Batu (11.59%) followed by Alage (5.61%), and Naka (5.33%).
Age groups had only signi cantly associated with seropositivity to anti B. abortus antibodies but not with B. canis infection. Other risk factors including sex, history of obstetrical problems and living condition of the dogs were however signi cantly associated with both rough and smooth Berucella specie's infection (P< 0.05). Using CFT for detecting anti B. abortus antibodies, 9.39% of female and 2.08% of male were tested positive. According to different age groups, 1.25% of ≤2 years old, 4.05% of 2-4 years old, and 13.58% of >4 years old were seropositive. Moreover, regarding to living condition of the dogs, 13.89% of indoors, 3.88% of semi indoors, and 10.59% of outdoors; and with respect to history of obstetrical problems, 18% of dogs with history of reproductive problems and 2.95% without such history, were positive (Table 3).
For the screening of B. canis infection using RBPT canis , 14.09% of female and 5.42% of male; 4.17% of indoors, 0.06% of semi indoors, and 18.82% of outdoors; 22% of with history of reproductive problem and 6.78% without that history; were detected positive ( Table 3). Furthermore, according to their living status, outdoor dogs were 8.4 times more likely to be seroposotive than indoor dogs. In contrast, infection of canine brucellosis in males was lower by 0.2 odds ratio than females (p <0.05).
Univariable logistic regression analysis of association of risk factors with anti B. canis antibodies also depicted that, age and location had not signi cant effect with seropositivity. However, sex, living condition, and history of obstetrical problems had signi cant association. As it is clearly shown in Table  4, outdoor dogs and dogs with history of obstetric problems had about 5.3 and 3.88 times higher odds of getting B. canis infection, respectively than those indoors and those hadn't history of obstetrical problems. However, B.canis infection of in males was lower by 0.35 odds ratio than females (p <0.05).

Multi-variable logistic regression analysis
Final selection of the best potential risk factors that would likely best explain the response of the predictor variable was done; based on a stepwise forward elimination procedure as presented in (Table 4 and 5). Accordingly, location of the animals, semi-indoor living condition, and age group between 2 and 4 years were removed from the simpli ed model of multi variable logistic regression (p≥ 0.05). Thus, male dogs had lower seropositivity than female dogs (OR=0.26, CI; 0.85, 0.79); and dogs older than 4 years had higher seropositivity than those ≤2 years (OR=7.78, CI; 2.59, 23.32    Generally, result of questionnaire survey revealed that the knowledge and understanding about brucellosis among the dog owners was very limited. The low awareness on canine brucellosis in the present study could be attributed to the dearth of health education (especially regarding zoonotic diseases). In Ethiopia, the veterinary medicine is widely recognized important only for livestock population. The attention given towards dogs by veterinarians is very poor and so much misunderstanding prevails with regard to preserving their healthy conditions. It is therefore important to establish an educational campaign in the study areas to enlighten the communities on the disease.

Discussion
In Ethiopia, brucellosis has been reported from different localities, commonly targeted on bovine, occasionally on sheep and goats, and rarely on camels (31). To the best of knowledge, canine brucellosis had not studied in the country.
This study investigates the seroepidemiology of canine brucellosis in selected areas of East Shoa Zone, Ethiopia. In the study, RBPT containing B. abortus antigen was employed for screening canine brucellosis. However, as RBPT prepared from smooth strains of Brucella cannot detect B. canis infection, a RBPT containing B. canis (rough strain) had used for screening of B. canis infection. Consequently, all dog serum samples were screened twice. Complement xation test containing B. abortus antigen had been used for con rmation of canine brucellosis due to smooth strains. However, B. canis infection had not serologically con rmed. The present nding revealed that, the overall seroprevalence of canine brucellosis due to smooth strains was 5.4% (CI: 3.35, 7.96) and 4.88% (CI: 2.7, 7.0) using RBPT and CFT, respectively. On the other hand, prevalence of canine brucellosis due to rough strain B. canis was 8.74% (CI: 5.92, 11.56) using RBPT canis as illustrated in (Table 2).
Detection of canine brucellosis due to smooth strains of Brucella species recorded in the study is almost equal with the report by Cadmus et al. (7)  The difference observed could be due to sampling method, sample size of the studies, and sensitivity and specify of serological tests used. It could be also due to the difference in the dog's rearing culture, awareness of the people, population of stray dogs, dogs keeping purpose, contact with other domestic animals, and health, housing, hygiene, breeding, and feeding management system of the dogs.
In addition to estimating the seroprevalence of canine brucellosis, the association of risk factors had also done. Consequently, no signi cant association between locations and seroreactivity to both smooth and rough types of Brucella infection had detected by Fisher's exact test and univariable logistic regression analysis. The nal simpli ed model of multivariable logistic regression analysis however indicated a signi cant association of dog's locations with seropositivity of the rough strain of Brucella. As observed in (Table 5), dogs in Batu were 3.86 times more infected with Brucella canis than those in Alage. In contrary, Momoh et al. (21) and Chinyoka et al.(12) Sowed no statistically signi cant association across study districts. However, it is agreed with the nding of Adedoyin et al. (38) who had also described signi cant association with the site of samplings.
Unlike Alage and Naka, Batu is urban and the higher prevalence of B. canis infection in Batu might be because of high population of stray dogs in the town. It is known that stray dogs are the museum of many diseases, including brucellois. In Batu, there is a municipal abattoir, many butcher shops and restaurants; thereby many dogs migrate from neighbor villages in search for food; which could increase the number of stray dogs in the town. Perhaps, the big lake ("Lake Zway") in surrounding of the town which produces thousands of tons of sh each year could also be the reason. This is because upon processing the sh, many sheries dispose the un-edible offal to the shore of the lake; resulting for assembling freely foraging dogs. Thereupon, many dogs come together and those infected could possibly transmit to others during mating and direct contact. This is supported by studies of Chikweto et al. (3) (33). This might be due to the fact that sexually mature dogs would have higher potential of contracting canine brucellosis through the venereal mode of transmission and the probability of contracting the disease from carrier and other infected animals and materials raises, as time goes.
With respect to maintenance or living condition of the dogs, a very strong statistically signi cant association with seropositivity of both types of strains of Brucella species was recorded. According to the result of the multi logistic regression, the odd of canine brucellosis due to smooth and rough strains in outdoor dogs were 4.72 and 6.42 times higher compared with indoors, respectively as listed in (Tables 5   and 6). This higher prevalence seen in outdoors might have resulted from the fact that unattended dogs are often in closer contact with infected materials. In the present study, obstetrical problems had strongly signi cantly associated with seropositivity of both anti B. abortus and anti B. canis antibodies. Higher prevalence was found in dogs with history of obstetrical problems (Abortion, infertility, retained fetal membrane, still birth, scrotum dermatitis, swelling of Scrotum, abnormal vaginal discharge) than those hadn't (Tables 5 and 6). This could be explained by the fact that such reproductive signs are typical outcomes of brucellosis (52-54). As it is described by The present study concluded that the overall prevalence of canine brucellosis was found high. The disease had signi cantly associated with living condition of the dogs and outdoor management system was a key risk factor for the occurrence of the disease. Furthermore, respondents had reported abortion cases and the disease had strongly associated with history of reproductive problems. Thence, they can be a source of infection to others. The increasing evidence that B. canis infections are endemic in the dog population also gives rise to the suspicion that, the zoonotic potential may be greater than suspected.

Figure 1
Map showing the study areas