Introduction

Brucellosis is a zoonotic disease that affects humans, domestic animals, wildlife and marine mammals [1, 2]. Wildlife and livestock are the main hosts; humans are usually accidental hosts [3, 4]. Transmission to humans is through direct contact with infected materials like afterbirth and blood, through ingestion of infected animal products such as raw milk and meat, and by inhalation of airborne agents [5, 6]. Human-to-human infection does not normally occur although there have been reports of transmissions through breast-feeding, blood transfusion and tissue transplantation [7, 8]. Brucella melitensis and B. abortus are the most important zoonotic Brucella species [5, 9, 10]. Brucella suis and B. canis are also zoonotic. The disease is endemic in many developing countries including countries in sub-Saharan Africa. Brucellosis is of economic importance to farmers due to losses associated with abortion, death of young ones, stillbirth, retained placenta or birth of weak calves, delayed calving, and reduction in milk yield. Brucellosis is also associated with male infertility and localized chronic conditions such as epididymitis, hygroma, arthritis and bursitis [11]. All these lead to food insecurity, poverty and loss of ability to sell animal products to international markets. In humans, brucellosis causes acute febrile illness (AFI) with undulant fever [12, 13], long-term signs and symptoms may include fatigue, recurrent fevers, arthritis, endocarditis and spondylitis and sometimes neurological symptoms [14]. Human brucellosis is often debilitating if not diagnosed early and treated correctly. It has a long duration and convalescence period, leading to economic losses in terms of treatment cost and lost man-hours. Treatment is normally expensive and extensive, lasting six weeks or longer [13].

Even though the disease has been eliminated in a number of countries, its occurrence is still on the increase in developing countries such as in sub-Saharan Africa causing a serious zoonotic risk [15, 16]. In Kenya, the disease is widely spread and endemic, especially among livestock kept by the pastoralist communities [17]; many cases have been reported in the annual reports of the Ministry of Agriculture, Livestock and Fisheries and elsewhere [18–20]. The prevalence has been shown to range from 1% -25% for bovine brucellosis, 7% -22% for caprine and ovine brucellosis and 17% -32% for human brucellosis [9, 21, 22]. Despite the growing global recognition of the economic importance of brucellosis in livestock and humans, the disease is still among the neglected, poorly understood and uncontrolled diseases [23] particularly among pastoral and agro-pastoral area in Kenya, such as Narok [18].

Opportune and correct diagnosis of brucellosis is important for proper management of the disease in humans. Isolation of the microorganism is taken as the gold standard for brucellosis diagnosis [24] in humans because it gives definitive diagnosis. However, it has serious drawbacks; it is time consuming, complicated, laborious, needs highly skilled personnel and high containment laboratory facilities (Biosafety Level 3) and has a long turnaround time [25, 26]. Furthermore, Brucella isolation requires a large number of viable bacteria in clinical samples, proper storage and quick delivery to the diagnostic laboratory [27]. Few studies have reported isolations of Brucella species from bovine milk and vaginal discharges [28]. Alternative methods used in diagnosis of human brucellosis are serological assays, including Rose Bengal Plate test, serum agglutination tests, complement fixation test, Enzyme-Linked Immunosorbent Assay and gel precipitation tests [20, 29, 30], and molecular biology techniques such as polymerase chain reaction (PCR) [31–33].

Polymerase chain reaction (PCR), although relatively rapid and accurate, is expensive and requires specialized equipment and trained staff [34]. It is currently preferred where possible and has been documented in many national and international publications [31–33]. Multiplex PCR was first developed in 1994 [25] and is still used for diagnosis and identification of Brucella [35]. The Multiplex PCR was named AMOS PCR using the first letters of Brucella species it detects, that is B. abortus, B. melitensis, B. ovis and B. suis. Improved AMOS PCR can detect and identify Brucella S19 and RB51 vaccine strains, therefore the tests differentiate B. abortus field strains from vaccine strains [36, 37]. The PCR method is independent of the disease phase [38] and can be used even when antibiotics have been administered prior to clinical specimen collection [39]. Since PCR method does not differentiate between DNA from live and dead organism [40], there is potential for detection of residual nucleic acid in patients after brucellosis has resolved; leading to false positives. PCR is fairly expensive to run and many clinical laboratories lack capacity to run it.

Brucellosis infections are mainly diagnosed using serological tests, even though it has been shown that serological results could be affected by cross-reactions with other bacteria, such as Escherichia coli sero-group O:157, Yersinia enterocolitica serovar O:9, and Salmonella serotypes, Francisella tularensis, Pseudomonas maltophilia, and Vibrio cholerae [20, 32, 41, 42]. This often leads to false-positive results and subsequently, to over-diagnosis of brucellosis, resulting in patients who are not suffering from brucellosis taking the long Brucella treatment unnecessarily [43–45]. Since no single serological test has been identified as the test of choice for human brucellosis diagnosis [31, 46], some studies recommended that three serological tests, be carried out to increase chances of picking seropositive samples [30, 47].

Most of the healthcare facilities in Kenya use FBAT for quick diagnosis and prompt treatment. These tests give results within a few minutes and do not require complicated infrastructure or sophisticated training. They are simple, easy to carry-out and relatively inexpensive when compared to PCR and Brucella isolation. Rose Bengal Plate test (RBPT) has been used as screening test for animals brucellosis but its results should be confirmed using culture or other serological test and has been recommended by some international organization for diagnosis of human brucellosis [48]. Rose Bengal Plate test has been described as the ideal test for brucellosis diagnosis in small and understaffed hospitals and laboratories, shown to be effective in the diagnosis of human brucellosis [49].

Healthcare facilities in Kenya and East African region use FBAT that are sold under different commercial names. The tests are simple and give rapid results and are affordable [6, 50, 51]. Studies have shown that commercial FBAT give high positivity rates for brucellosis cases compared to other serological tests [6, 50–52]; some of the positive cases could be due to cross-reactivity, hence presenting a misdiagnosis of the disease. Some FBAT kits have two antigens one for Brucella abortus and Brucella melitensis. However studies have demonstrated that there are no respective differences in antigenicity for the two species [35, 53, 54] because all smooth Brucella species share common epitopes in the O-polysaccharide (OPS) and almost all serological tests use Brucella antigen as whole cell [55]. It is only in detection of B. ovis and B. canis infections, rough lipopolysaccharide (RLPS) antigen is used; because the two species lack OPS component [56].

This study demonstrated Brucella positivity status of humans attending healthcare facilities in Baringo County, Kenya, using Febrile Brucella agglutination antigen test (FBAT), and Rose Bengal Plate agglutination test (RBPT), and compared the two tests with respect to sensitivity and specificity. The study also identified the Brucella species present in some of the samples using PCR. The study was carried out in the year 2014.

Materials and methods

Study design

This was a cross-sectional laboratory based study. Sample size calculation was done using the method by Martin et al [57]. The confidence interval used was 95% and the assumed brucellosis prevalence was 13.7%, based on study done in Narok, Kenya a pastoralists community [58], which like Baringo is mainly inhabited by pastoralist and agro-pastoralist communities. Participants were recruited from adult patients who visited any of the study hospital’s laboratories with a clinician’s request for Brucella test. Inclusion criteria was patients 18 years and older and consented to participate in the study. Exclusion criteria was patients less than 18 years old and prisoners. Patients who consented were interviewed using questionnaire to collect data based on knowledge and attitudes associated with transmission of human brucellosis.

Identification of Brucella species in the blood specimens

DNA was extracted from the blood clots of all the blood samples. To determine the presence and quality of extracted DNA, the DNA was resolved using electrophoresis through 0.8% agarose gel in 1 x TAE buffer stained with ethidium bromide solution (0.05mg/ml) and visualized under UV transilluminator and photographed. Due to a limitation on availability of PCR reagents, the first 71samples perceived to have higher quality DNA as determined by the gel image were amplified using AMOS PCR carried out according to Bricker and Halling [25] a positive control was extracted from B. abortus S19 vaccine strain and a negative control was extracted from a serum sample known to be negative from Brucella.

The PCR reaction was carried out as follows; a 20 μL reaction was prepared having 10 μL 2 × PCR buffer (50 mM Tris, 1.5 mM MgCl₂, 10 mM KCl, 50 mM (NH₄)₂SO₄, pH 8.3) (TopTaq Master Mix, QIAGEN), 6 μL double distilled sterile water, 1μL cocktail forward primers of B.abortus, B. melitensis, B. ovis and B. suis (Table 1) at a final concentration of 0.25 μM each, 1μL IS711 reverse primer at a final concentration of 1μM, and 2 μL DNA template for each sample. Two control reaction mixtures were included, positive control containing 2 μL Brucella abortus DNA and a negative control containing 2 μL double distilled water instead of specimen DNA.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Polymerase chain reactions primer sequences.

https://doi.org/10.1371/journal.pone.0269831.t001

The amplification was carried out as follows, initial denaturation at 95°C for 5 min, 40 cycles of 95°C for 15 seconds, 52°C for 30 seconds, 72°C for 90 seconds and a final extension at 72°C for 5 minutes and the PCR products were stored at 4°C [59]. The amplified products were analyzed by electrophoresis through 1% agarose gel stained with ethidium bromide solution (0.05mg/ml) and visualized under UV transilluminator and photographed.

Results

Participants

A total of 182 patients met the inclusion criteria during the period from June to August 2014, but 16 samples did not have questionnaire, therefore were excluded from the study. Of the 166 patients who consented, 116 (69.9%) were females and 50 (30.1%) were males. The patients enrolled were 84 participant from Marigat, 52 from Kabarnet, and 30 from Eldama Ravine District Hospitals (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Results of three serological tests used on the serum samples collected from Baringo Hospitals.

https://doi.org/10.1371/journal.pone.0269831.t002

Overall, 88.6% (147/166) of the participants were aware that there is a disease although most of them referred to it using the Kiswahili term, ugonjwa ya maziwa. Some of the participants knew some causes of the disease, 50.6% (84/166) drinking unpasteurized (unboiled) milk and 40.4% (67/166) consumption of raw meat and/or raw blood. Some of the participants 9.0% (15/166) did not know the causes of the disease. Some of the participants 78.9% (131/166) had taken medication for the illness before enrollment to the study. This included 32.5% (54/166) who had taken traditional herbs, 21.7% (36/166) paracetamol, 18.1% (30/166) antibiotics, 6.6% (11/166) anti-malaria drugs, and 21.1% (35/166) unidentified medication. The medicine was either self-prescribed or obtained from a healthcare facility.

Sensitivity and specificity and predictive values of Brucella diagnostic tests

The sensitivity of FBAT was 76.5%, 95% CI 52.74–90.45 and specificity was 79.2%,95% CI 71.98–84.94 the positive predictive value was 29.5%, 95% CI 28.16–44.22 and negative predictive value was 96.7% when compared to RBPT (Table 3). The agreement of FBAT and RBPT was 32.7% indicating fair agreement [61].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Sensitivity, specificity and predictive values of FBAT compared to RBPT.

https://doi.org/10.1371/journal.pone.0269831.t003

Identification of infecting Brucella species

Seven out of 71 (9.9%) specimens analyzed were positive for B. abortus (498 bp). No amplicons were observed that corresponded with the expected amplicon sizes for B. melitensis (731 bp), B. ovis (976 bp) or B. suis (285bp). Fig 1, shows negative samples, positive sample, positive control (S19 B. abortus) and negative control (double distilled water) and Fig 2, showing the seven positive samples (B. abortus (498 bp).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Agarose gel electrophoresis showing 100bp molecular marker (M), negative samples 1, 2, 3, 4, positive samples 5, positive control 6 (S19 B. abortus) and negative control 7 (double distilled water).

https://doi.org/10.1371/journal.pone.0269831.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. AMOS PCR, gel electrophoresis showing 100bp molecular marker (M), positive human samples 1,2,3,4,5,6,7 (B. abortus (498bp) negative human samples 8, blank well 9 and negative control 10.

https://doi.org/10.1371/journal.pone.0269831.g002

Discussion

The study revealed that brucellosis is present in Baringo as demonstrated by presence of positive samples in the three study hospitals. This was supported by study done on livestock by Kosgei and Lokamar et al [62, 63]. The two tests used had varied positivity: FBAT detected the higher overall positivity of 26.5%; RBPT positivity was 10.2%. The same was the case when addressed per study hospital: FBAT consistently detected higher positivity of 35.7%, 17.3%, and 16.7% for Marigat, Kabarnet and Eldama-Ravine hospitals, respectively, than RBPT. The high percentage recorded by FBAT compared to RBPT indicated a possibility of overestimation of positivity. Since FBAT were the ones used for Brucella diagnosis at the study hospitals, the possible over-diagnosis of brucellosisin patients, could explain the high brucellosis cases recorded in the study hospitals (respective hospital records); some of the patients being incorrectly treated for the disease. The finding in this study agrees with a study carried out in two health facilities in Busia County in Kenya [50]; the recorded positivity rate was 19.6% using FBAT, 1% using RBT, 0.62% using serum agglutination test (SAT), 1.9% using Coombs test; also, in this study, out of the positive FBAT samples, only 3.4% were positive using the rapid IgM lateral flow immunochromatography assay (LFA). A study carried out in Ijara district hospital in Garissa in Kenya showed positivity rate of 32% using FBAT and 15.4% using real-time PCR(RT-PCR) [6]. Another study on twelve hospitals in Western Kenya showed positivity rate of 13.3% using FBAT and 1.7% using RBT [51]. In Tanzania a study on performance of four commercial FBAT showed the positivity rates ranging between 25.2% and 45.0% while that for RBT was 6.0% and 8.7% and for cELISA was 16.3% [64]. These studies demonstrated that FBATs have a tendency of overestimating brucellosis cases which mean many patients go through brucellosis treatment while actually they are not infected.

In this study, using RBPT, Kabarnet District Hospital and Eldama Ravine District Hospital both in agro-pastoralist areas, had positivity rate of 5.8% and 3.3%, respectively; these were lower compared to 7.0% to 15.3% rates reported in agro-pastoralist areas. However Marigat District Hospital positivity rate was 15.5%, which was within the range of 14.4% and 46.5% documented in a systematic review of human brucellosis in local pastoralist communities in Kenya [9]; RBPT was the test used in most of the studies. Marigat District Hospital, which has its main catchment among pastoralist communities, recorded the highest positivity of the three study hospitals (35.7%) using FBAT; this compared well with study at Ijara, Garissa [6] which is local pastoralist area.

The FBAT used in this study were supplied as two antigens, one for B. abortus and the other for B. melitensis, however, the two antigens gave identical results when tested. The sample that was positive for B. abortus was also positive for B. melitensis and the sample negative for B. abortus was also negative for B. melitensis. This confirming that, as expected that the two antigen B. abortus and B. melitensis could not be differentiated using these test kits [35, 53, 54]. The authors observed that validation of the commercial FBAT prior to use for diagnosis at the hospitals was not routinely carried out. Further, there were no specific guidelines provided by the health authorities on laboratory diagnosis for Brucella at the time of this study. This meant that each healthcare facility procured and used the kits that they perceived to be most cost-effective, resulting in different kits being used in different hospitals and in some cases in the same hospital.

Studies have demonstrated that FBAT produced by various manufacturers, are used widely in healthcare centres in Kenya [6, 50, 51] despite their poor performance. It has also shown that the sensitivity of these FBAT is low in the region where this study was carried. Since there are possibilities that FBAT would pick some false positives, there need to be cautions on interpretations of respective results in healthcare facilities in Kenya. For the OIE/WHO-recommended rapid test, RBPT, Diaz-Aparicio et al [49] have shown that cross-reactivity with Yersinia or any other antigens disappears with increased serum dilution; agglutination at 1:8 dilution and above was shown to indicate definite positive reaction for brucellosis. This way, the specificity of the RBPT can be greatly improved, to give tangible results; however, caution needs to be taken when dealing with low Brucella titres, which may have end-points at lower dilutions; in this case, one will need to also do one or two other tests to rule out the cross-reactivity with Yersinia or other antigens. This dilution method has now gained appraisal by a number of researchers [3, 49]; who have shown that the RBPT dilution method improves its specificity and considerably reduces the need for additional serological tests. Another point in support for RBPT is that the test is affordable and easy to run; hence it can easily be made available to healthcare facilities. Compared to other serological tests, RBPT has been shown to be exceedingly cheap; Moriyon [65] and others, compared the costing for RBPT and ELISA (which is currently the mostly used) in Spain and showed that while competitive ELISA and indirect ELISA cost 6 and 5 euros per test, respectively, RBPT costs a mere 5 cents (0.05) of an euro per test. Correct and timely diagnosis of brucellosis will save patients who are not suffering from brucellosis the long and expensive treatment.

Brucella abortus was the species identified from the tested samples in this study; this was demonstrated by positive PCR amplification of B. abortus (498bp) in seven patient specimens. Similar results were obtained from a study which was concurrently run on livestock in the area; B. abortus was detected in samples collected from cattle and sheep [62]. The fact that B. abortus, was isolated from cattle and patients in the study area confirm the source for the human disease. This transmission of brucellosis from animals to humans is documented [20, 22, 62]. There is, therefore, need to educate the Baringo residents on how to safeguard themselves from getting infected; from the questionnaire data, some people in Baringo County did not know what causes the disease thus they could be easily get infected.

The study, however, was not without limitations; first, being a cross-sectional hospital-based; the results cannot be generalized to the general population. Second, the method of recruitment of participants was based on a clinician’s recommendation and differences in clinicians’ understanding/view on brucellosis may have led to differences in patients’ recruitment. Third, due to the exclusion of children (<18 years) the study cannot be generalized to the general population. It has been shown that there is an increased probability of brucellosis in individuals of lower age who herd cattle, sheep and goats [66], activities that are associated with pastoralism. Fourth, most of the patients sampled were females. This is possibly due to differences in health-seeking behavior between males and females; a study in Kibera, Kenya, showed that males preferred self-treatment rather than visiting healthcare facilities [67]; this study reported more females seeking treatment than males as more women were recruited during the study period. A randomized community based study needs to be carried out in future to address this limitation. The fifth RBPT was used to compare FBAT, though recommended by OIE as a better screening test the test also has challenges. The sixth PCR was not done on all the DNA samples extracted therefore we cannot rule out the presence of other Brucella species.

In conclusion, this study shows that human brucellosis is present in Baringo County, the healthcare facilities use FBAT for brucellosis diagnosis; the FBAT are not validated by the hospitals before use, and the sensitivity of these FBAT is low compared to RBPT. There is need to validate FBAT used and the kits should have strong positive, weak positive and negative control samples for use in validation of the kits, if they are still to be used in healthcare facilities or their use should be discontinued. The study recommends the use of Rose Bengal Plate test (RBPT) as it is the one recommended by OIE and WHO; it is validated and is relatively inexpensive and easy to use. Further, the study has shown that B. abortus causes human brucellosis in Baringo County. Individual in these areas should be sensitized on control and preventive measures to stop animal to human transmission.

Supporting information

Acknowledgments

Thanks to University of Nairobi, Biochemistry Department and Baringo County administration, medical superintendents and the entire laboratory staff of the three hospitals and all the study participants. We also thank Dr. Peninah Munyua of CDC-Kenya and Dr. Joseph Kamau of the University of Nairobi for technical assistance.