Infection occurs via contaminated feed or water, by inhalation, through the conjunctiva, or by contact with infected aborted materials; while calves become infected in utero or via infected colostrum or milk [8]. Venereal transmission has also been reported [9]. In fully susceptible herds, abortion rates vary from 30 to 70% [5]. Infection may be lifelong, and during subsequent pregnancies there is invasion of the gravid uterus and allantochorion; abortion rarely recurs, but uterine and mammary infection recurs [10]. Since the reproductive performance of these carrier animals is unaffected, they are retained in herds in Nigeria despite the presence of pathognomonic clinical signs in some cases, making effective control programmes extremely difficult.

Prevalence of bovine brucellosis varies widely across Nigeria, and between herds in the same area [11,12], with reported seroprevalences of 0.2% to 80.0% [13,14]. In institutional farms, abattoir surveys and other ranches or dairy farms in southern Nigeria, prevalence in cattle ranged between 3.7% and 48.8% [15-18], while in the traditional nomadic Fulani cattle herds, the prevalence was between 0.4% and 26% [3,11,12,19]. Recently, a within-herd prevalence of 32.2% on a prison cattle farm [20]; and 19.5% seropositive and 25.3% positive milk samples [21] were reported in northern Nigeria. Prevalence studies in other surrounding countries indicated 8.4% in Cameroon [22], 7% in Chad [23], 41% in Togo [24] and 6.6% in Ghana [25]. Brucellosis has also been reported in many other parts of Africa [26-32], although detailed information on its prevalence is still lacking for most countries [2].

Although prevalence is high and variable in many countries, surveillance for the disease is generally poor [2,3,14,33]. Factors assumed to be responsible for variation in prevalence include purchase of infected cattle from the market for replacement or upgrading, nature of animal production, demographic factors, regulatory issues, climate, deforestation and wildlife interaction [19,26,33-35]. Furthermore, one major factor contributing to the spread of the disease is the free movement of animals practiced by the nomadic Fulani herdsmen, who own about 95% of all food animal populations in Nigeria [3,19,36]. Other factors that may influence the prevalence of brucellosis in Nigeria include management system [11,37], the herding of different species together [3,7,20], use of common pastures and water sources [14], age [12,18,21], breed [17,18], sex [21,37], lactation status [21] and season [11,14]. However, other variables such as pregnancy status and state have not been assessed. All these risk factors need to be taken into consideration in designing and execution of effective control programmes in Nigeria.

In the development of a livestock industry, disease eradication and control are paramount. The continuous movement of cattle as a result of trade and for grazing is a common practice in Nigeria, putting many herds at risk of brucellosis infection. So too do the sharing of bulls and use of open-range grazing. Reports have indicated that trade cattle in and from the northern states, and also those from across the northern borders of Chad and Niger showed evidence of infection [14,17,18]. More than 20% of the trade cattle came from outside Nigeria [17]. The nature of the communal grazing systems used by the nomadic Fulanis, as well as porosity of the borders, predisposes livestock in the study area to infection, with a serious risk to human health [11,17,36]. Even in developed countries, despite the preventive and control measures that exist; there is still a high potential for transmission and spread of Brucella organisms via animals and their products [38]. Recent estimates of losses in meat and milk production as a result of brucellosis are $800 million annually in the USA [39], in excess of $224 million in Nigeria [16] and $37.5 million in South Africa [40].

Many of the studies conducted on brucellosis in Nigeria have been from the humid southern part [15-18] and were mainly abattoir surveys which were not representative of the general population. Abortions in Adamawa Province in the late 1940’s caused so much concern that B. abortus strain 19 vaccination was implemented in 1949, 1953 and 1956 [41]. Since then no surveys to our knowledge have been done on brucellosis in Adamawa Province, except that of Atsanda and Agbede [37] who used the Rose-Bengal plate agglutination test (RBPT) and serum agglutination test (SAT), and very recently Bertu et al.[14] who analysed some samples collected from sick animals using RBPT and a rapid field test. Most studies in northern Nigeria were based on small sample sizes or suspect samples submitted to laboratories, or were done in restricted locations or in abattoirs [3,4,12-14,19-21]. In addition, most studies in Nigeria have relied on the relatively non-specific RBPT and few have used a more specific confirmatory test such as enzyme-linked immunosorbent assay (ELISA) [12]. Reported sensitivity and specificity of RBPT in Zambian cattle were 90.0% and 75.0% [42] and for c-ELISA they were 98.0% and 99.0%, respectively [43].

A total of 4,745 samples was tested with RBPT, of which 1,735 (36.6%) were positive for brucellosis. Of these, 1,137 (65.5%) were confirmed to be seropositive for brucellosis upon further testing by c-ELISA, giving 34.5% overall false positives (42.8%, 24.7% and 27.1% in Adamawa, Kaduna and Kano states, respectively) (Table ​(Table2).2). Based on c-ELISA, the estimated overall survey adjusted true animal-level seroprevalence was 26.3% (95% CI, 22.1%-31.0%) (Table ​(Table3).3). The prevalence for Adamawa state was 29.2% (95% CI, 22.5%-36.9%), for Kaduna state 23.3% (95% CI, 18.9%-28.3%) and for Kano state 26.7% (95% CI, 18.8%-36.7%) (Table ​(Table33).

Overall mean within-herd seroprevalence of brucellosis was 24.6%. Within positive herds, the mean prevalence of seropositive animals, adjusted for survey design, was 30.2% (95% CI, 25.3%-35.1%) and ranged from 3.1% to 85.7%. The distribution of within-herd seroprevalence in infected herds for the various management systems is shown in Figure ​Figure2.2. Mean within herd seroprevalences in the different management systems, from lowest to highest, were: commercial 18.5% (95% CI, 10.2%-26.8%), agro-pastoral 25.2% (95% CI, 19.0%-31.4%), pastoral 43.0% (95% CI, 38.6%-47.5%) and zero-grazing 51.1% (95% CI, 46.9%-55.3%).

The seroprevalence differed significantly between males (38.2%; 95% CI, 31.7%-45.2%) and females (24.7%; 95% CI, 20.4%-29.5%) (P<0.001), between cattle <4 years (13.5%; 95% CI, 8.9%-19.9%) and >7 years of age (35.0%; 95% CI, 28.5%-42.3%) (P<0.001), between commercial farms (15.9%; 95% CI, 9.5%-25.5%) and pastoral farms (45.1%; 95% CI, 38.6%-51.9%) (P<0.001), and between non-pregnant (27.8%; 95% CI, 22.9%-33.5%) and pregnant females (17.2%; 95% CI, 13.6%-21.5%) (P<0.001). There was no significant difference in seroprevalence between lactating (25.3%; 95% CI, 20.7%-30.5%) and non-lactating (23.2%; 95% CI, 19.3%-29.9%) animals (P=0.635) or between breeds (P=0.392), although Bos taurus had the lowest prevalence of 15.1% (95% CI, 6.6%-31.0%) and Bunaji had the highest prevalence of 27.5% (95% CI, 22.5%-33.2%) (Table ​(Table33).

The associations between animal-level factors and brucellosis seropositivity, adjusted in the multivariable model for confounding both by the other animal-level factors and by state and management system, are shown in Table ​Table4.4. The greater odds of seropositivity in males (OR=1.98; 95% CI, 1.54-2.54; P<0.001) remained, as did the monotonic increase with increasing age (e.g., for >7 y vs. < 4 y: OR=3.82; 95% CI, 2.72-5.36; P<0.001) and the increased odds in non-pregnant compared to pregnant cows (OR=1.84; 95% CI, 1.49-2.27; P<0.001). Neither state nor management system acted as confounders of the effects of the above variables, as their exclusion resulted in <10% change to coefficients. Although there was ultimately no significant effect of breed, there was some confounding both by state and by management system, with their exclusion from the model resulting in up to 66% and 180% change in coefficients, respectively. Similarly, there was some confounding of the effect of lactation status on seropositivity by both state and management system. The odds of seropositivity were significantly higher in the pastoral management system than in the other management systems (e.g., vs. agro-pastoral: OR=3.52; 95% CI, 2.50-4.95; P<0.001). The random effects in the hierarchical model showed that there was significant variation in brucellosis seropositivity between herds within ward and between wards within LGA, but not between LGAs within state.

Livestock production in Nigeria was dominated by nomadic pastoralism long before the advent of colonial era. In the 1930s, the government established stock farms for dairy herds by selective breeding. In the same period, mixed farming policy hence agro-pastoral production system as well as range management were introduced for livestock improvement in Nigeria [56]. In the 1940s to 1950s, government investigation and breeding centres in settled herds all over the country and artificial insemination were established. It was also within this period that exotic breeds of cattle were introduced to upgrade the local stock [56]. Brucellosis infection rate of over 30% was reported during the 1940s at various livestock centres in Nigeria characterized by abortion storms [15,17]. Attempts were made to vaccinate cattle against brucellosis but it was limited and irregular [17,18]. Between 1970s and 1990s, about 96% of the cattle were zebu-type cattle, most of which were tended by traditional Fulani pastoralists [57]. In addition, in 1970s, 30 to 40% of the beef consumed in Nigeria was imported from Niger, Chad, and other neighboring countries [57]. These factors may have influenced the increase in prevalence of bovine brucellosis in Nigeria and elsewhere [3,38].

An overall adjusted animal-level true prevalence of 26.3% (95% CI, 22.1-31.0%) was obtained in this study (Table ​(Table3).3). Of the three states sampled, Adamawa state had the highest apparent animal-level seroprevalence of 29.2%, although this was not significantly higher than in Kano (26.7%) or Kaduna (23.3%). However, after adjustment for confounding, the difference between Adamawa and Kaduna approached significance (P=0.07). In addition, Adamawa state showed the highest number of cattle exhibiting hygroma, seen in 54 of 84 positive herds (Table ​(Table1).1). Adamawa state borders on Cameroon, and constant trans-border movement of cattle has been reported to result in transmission of contagious bovine pleuropneumonia [58]. Cross-border movement has been implicated in the transmission of brucellosis by previous investigators in Nigeria [17,19] and elsewhere [25,59], and although not directly implicated by this study, it is possible that it may be a risk factor for brucellosis in Adamawa state.

The variations in the results of the two tests showed that many of the RBPT results were falsely positive because of its relatively low specificity, with Adamawa state showing the greatest discrepancy between the two tests (Table ​(Table2).2). The c-ELISA of all the samples was done in the same laboratory at the same time, while screening using RBPT was done in two different laboratories at different times but by the same investigator, thereby reducing the possibility of laboratory error or subjective interpretation. Since none of the herds sampled had been vaccinated against brucellosis, the antibodies responsible for the false positives were likely from other sources. Some bacterial pathogens such as Yersinia enterolitica serovar IX, Vibrio cholera, Escherichia coli O:157, Salmonella spp. and Sternotrophomonas maltophilia have been reported to produce cross reacting antibodies to brucellosis [60,61], with Y. enterolitica being the most significant cause of false positives. It is possible that the prevalence of one or more sources of cross-reacting antibodies was higher in Adamawa than in the other two states. In a Zambian study, for every three positive RBPT animals, only one tested positive on c-ELISA, except among animals which had aborted, where the ratio was close to one [30]. The c-ELISA test, with a higher specificity than RBPT, complement fixation test and florescent polarization assay, and therefore an ideal confirmatory test [42], has however rarely been used in Nigeria in naturally infected cattle, and never in a large study including different production systems.

The animal-level prevalence reported in this study (26.3%) is higher than recent reports from northern Nigeria [21]. Furthermore, the prevalence is much higher than the 9.8% and 18.6% using RBPT and MRT respectively in indigenous cattle in abattoirs in western Nigeria [18], 20% in government farms in the north using SAT [62], and 6.6% in cattle herds in northern Nigeria using ELISA [12], but lower than a recent report of 45% from samples of sick animals in Adamawa state, Nigeria, using RBPT and rapid field test [14]. However, the result is consistent with the 32% within-herd prevalence reported in one prison farm in northern Nigeria using SAT and MRT [20]; and the 38.0% using RBPT and SAT in cattle in government Livestock Investigation and Breeding Centers in Kano [19]. Studies elsewhere showed prevalences between 49% and 60% among breeding cows and heifers, dairy farms and abattoir surveys in the southern and western states of Nigeria on the basis of SAT, with CFT on doubtful results [15,17]. The animal-level prevalence obtained in this study was also much higher than those reported in South Africa [32], North Africa [63] and East Africa [28,64]. However, a higher prevalence of 41% was reported in Togo, West Africa [24]. McDermott and Arimi [2] also reported a higher prevalence in sub-Saharan Africa. Although some of the variation in results between studies may be due to the use of different diagnostic techniques, considering only those studies that used the same serological test as the present study, the prevalence of bovine brucellosis appears to be increasing in northern Nigeria. Several reports have previously indicated that brucellosis is on the increase in Nigeria [3,19,21] and other developing countries [2,33]. Lack of proper surveillance and control measures in most parts of Africa may be contributing to this increase, as may the importation of animals and their products from more developed countries despite the preventive and control measures in such countries [2,33,38]. Nevertheless, despite reports showing the extent of brucellosis in Nigeria, there is no record of a proper brucellosis control programme in the country [3,14].

Traditional Fulani herds practicing nomadism or pastoralism showed the highest prevalence followed by the zero-grazing and agro-pastoral systems, with the lowest prevalence being recorded in commercial herds. The findings by Nuru and Dennis [11], Bale and Kumi-Diaka [19], and Ocholi et al. [12] who reported prevalences of between 0.4% and 26% in traditional nomadic Fulani herds; and Atsanda and Agbede [37] who reported a slightly higher infection rate of 5.1% among nomadic cattle herds than among settled cattle herds (4.4%) in Adamawa state, are consistent with the findings of our study. Furthermore, Ocholi et al. [3] and Rikin [36] reported the prevalence of brucellosis to be rising among pastoral and semi-pastoral herds which comprise about 95% of the cattle population in Nigeria [11]; Bale and Kumi-Diaka [19] indicated free movement of the pastoral Fulani herdsmen and interaction of cattle with those of other Fulani herdsmen as major factors in spreading brucellosis. These observations agree with our findings, as do reports by McDermott and Arimi [2] and Matope et al. [31] of highest occurrence in pastoral production systems in arid and semi-arid areas in Zimbabwe and other parts of Africa, and Bernard et al. [27] in Uganda and Berhe et al.[28] in Ethiopia who also reported a higher seroprevalence in the transhumance system than in sedentary cattle.

The odds of brucellosis seropositivity were 3.5 times greater amongst pastoral herds than agro-pastoral herds in this study. The high prevalence of brucellosis in a pastoral management system may partly be attributed to long distance movement of cattle in search of pasture and water and co-mingling in communal grazing areas and at watering points, particularly during the dry season. Musa et al. [26] observed in Sudan that clinical manifestation of brucellosis often began during adverse weather conditions and famine. During such times animals become concentrated on scarce pastures and around watering points, which may become contaminated with aborted foetal materials or fluids from infected normal calvings. Many pastoralists do not isolate cows during parturition or dispose of the placenta following calving, resulting in contamination of the environment and transmission of brucellosis within and between herds. Other possible risk factors for brucellosis associated with the pastoral management system in Nigeria include bull sharing which may result in venereal transmission of brucellosis [9], purchasing of livestock from markets without quarantine [19] and interaction of cattle with wildlife [34,35].

The prevalence of brucellosis in zero-grazing systems in this study was also high. This is contrary to the report by McDermott and Arimi [2] of low prevalence due to very low level of between-herd contacts. However, Bayemi et al.[22] and Karimuribo et al.[29] observed a high prevalence in intensively managed herds. Cattle in zero-grazing systems in Nigeria are generally bought from the open market for a fattening programme, which may explain the high prevalence in such systems.

Reports indicate that about 20% of infected pregnant animals do not abort, while 80% of animals that abort as a result of B. abortus infection, do so only once [65] and thereafter will usually carry the pregnancy to full-term and appear healthy. In herds that have chronic brucellosis and do not introduce new animals, very few or no abortions occur and the disease is almost impossible to recognise clinically [66]. The emphasis in livestock production in Nigeria is on the ability of the females to produce calves; as long as cows produce, farmers tend to keep them, even if they have a history of abortion. In bulls, brucellosis causes no impairment of libido or breeding capacity [6] and the disease is subclinical in most animals [5]. For these reasons, farmers seldom cull infected animals from their herds, contributing to the high prevalence observed in this study. These apparently healthy cattle that are reproducing normally serve as permanent carriers of brucellosis. Some cattle may get rid of the infection within a few months, while others may remain infected for life, thereby transmitting the disease at subsequent parturitions [8,65]. This scenario could make control of the disease in Nigeria an extremely difficult task, requiring a well-designed and coordinated eradication policy and good cooperation of all sectors of the industry. Strategies such as immunization and the identification of and selection for genetic resistance factors may be required to make significant progress in control of the disease.

Cows with visible hygromas, but reproducing normally are also left in the herds. All forms of hygromas were encountered in this study including fluid accumulation in some infected animals on the cervical region, between the nuchal ligament, shoulder, flank, primary thoracic spines and most commonly the carpal and stifle joints. An additional picture file shows this in more detail (see Additional file 1). Over 23% of all herds sampled and 30% of the infected herds had hygroma (Table ​(Table1).1). This is the first report of this manifestation of the disease in Nigeria. Similar clinical signs have been reported elsewhere [26,33]; these authors also used the term hygroma for fluid accumulation in locations other than the joints. The hygromas are localized in carpal and other bursae and contain large numbers of the organisms [67,68]. The traditional Fulani cattle rearers practice ‘firing’ of the hygroma lesions, by using a hot knife to incise the swelling through the capsules, when large numbers of the localized brucellae are discharged from the hygroma and contaminate the environment, further encouraging the spread of the disease (see Additional file 1). The herd that had hygroma and was serologically negative is consistent with a previous report that 13% of brucellosis positive animals were serologically negative [69]. It is also possible that some of the hygromas observed in this herd may have been due to another aetiology, such as intermittent mild trauma to the precarpal area caused by lack of bedding or a poorly designed feed bunk [70]. This may partly explain the very high prevalence of hygromas seen amongst the seropositive animals in Adamawa state. Nevertheless, the presence of hygroma in one or more animals in a herd appeared to be a fairly specific predictor of herd seropositivity, with estimated specificity of 98.4%. It could therefore be used in participatory disease surveillance, although its estimated sensitivity of only 29.5% would mean that other signs of herd infection would also have to be considered.

Although not statistically significant, the prevalence of brucellosis was somewhat lower in Bos taurus breeds than amongst indigenous breeds. This finding is consistent with reports by Kubuafor et al.[25] in Ghana. Karimuribo et al.[29] in Tanzania stated that the proportion of seropositive animals was significantly higher in indigenous than in crossbred cattle. However, Muma et al. [42] reported no association between Brucella seropositivity and cattle breed. The better management in the exotic herds, stall or intensive feeding that minimizes contact between animals and herds may be responsible for the low prevalence. The distribution of breeds between management systems in this study varied, with highest number of Bunaji found in agro-pastoral followed by pastoral systems; Gudali in agro-pastoral and commercial systems; and Bos taurus were mainly in commercial farms. However, adjustment for management system did not change the result. Esuruoso [17] reported that the Ndama breed was the most affected breed in western Nigeria, while Cadmus et al.[18] found the Red Bororo breed to have the most positive reactors followed by Bunaji. Junaidu et al. [21] reported the Sokoto Gudali breed to have the highest prevalence, followed by Azuwarq, with Bunaji having the least prevalence. Genetic variation is an important factor in conferring resistance or tolerance of cattle breeds to a wide range of diseases, and the antibody response of animals classified as resistant to infection by B. abortus differed significantly from that of susceptible animals [71]. Significant genetic variability in resistance/susceptibility to brucellosis has been detected in cattle and associated with a 3’ untranslated polymorphism in the Slc 1a1 gene [72]. This aspect needs further studies in Nigeria.

The prevalence of brucellosis was significantly higher in males than females, and this did not change after adjusting for age, management system, state or breed. This difference between sexes is consistent with reports by Chimana et al.[73] who recorded more seropositive cases in bulls (12.5%) compared to females (8.1%). However, our findings are contrary to other reports that showed significantly higher prevalence in females than males [20,21] or no difference between sexes [12,22,25]. Fifteen percent of the infected herds had animals with clinical evidence of orchitis and/or epididymitis (Table ​(Table1).1). Reports in Nigeria and elsewhere indicated that testes, epididymis and other accessory sex organs may be affected [5,6].

In the present study, the prevalence of brucellosis increased with age, with the odds of having brucellosis 3.8 times greater amongst cattle >7 years than those <4 years old. This is consistent with previous reports [21,25,74]. The higher prevalence of brucellosis in older cattle can be attributed to constant exposure of the cattle over time to the infectious agent. However, Cadmus et al. [18] observed no difference between cattle >3 years and 1–3 years old, whereas Matope et al. [31] reported decreased frequency of brucellosis with increasing age, with 2–4 years old having higher odds of being seropositive compared to those >7 years. They concluded that some older cows may not exhibit detectable antibody titres possibly due to latency, which is common in chronic brucellosis.

The significantly higher prevalence in non-pregnant compared to pregnant animals in this study did not change after adjusting for age, breed, state and management system. This finding is consistent with the observation in Ethiopia by Ibrahim et al. [64] but contrary to reports by Mekonnen et al.[75]. Pregnant cattle above five months of gestation are more susceptible to Brucella infection due to the preferential localization of Brucella in the uterus in which allantoic fluid factors such as erythritol stimulate the growth of Brucella[5]. However, the greater probability of abortion in infected animals could explain the higher seroprevalence in non-pregnant animals.