A cross-sectional study was carried out in Namwala, Choma and Monze districts of Southern province of Zambia (Figure 1). The province and study districts were purposively selected because they are the top cattle producers in Zambia (29). Furthermore, these areas are endemic for bovine brucellosis as documented by several serological surveys (24, 29, 30). Southern Province lies between latitudes 15°14′ S and 17°42′ S and longitudes 25°E and 28° S. It has a total land surface area of 85,283 Km² with an estimated human population of 1,907,784 and a cattle population of 2,105,891 (31). In these districts, a pastoral or nomadic cattle-grazing system is practiced, where animals are grazed in the Kafue flats/floodplains in dry seasons and moved to the upper areas during the wet season (30).

The study population comprised of persons occupationally at risk of exposure to Brucella infections in Namwala, Choma and Monze districts of Southern province of Zambia. The individuals were grouped into two categories (herdsmen and abattoir workers), depending on their level of daily activities that could lead to direct contact with either suspected Brucella infected animals or infected animal carcass.

A sample size of 153 was calculated using the Ausvet EpiTools software (http://epitools.ausvet.com.au/) based on the following assumptions: Assumed prevalence estimation of 5.03% (21), desired absolute precision of +/−2% and confidence level of 95%. Sampling was stratified according to the study districts and weighted using the cattle population as proxy weighting value for the persons to be sampled (Table 1).

A total of 153 individuals (125 herdsmen and 28 abattoir workers) were sampled from three purposively selected districts, namely Namwala, Monze, and Choma. Furthermore, these districts are subdivided into villages namely; Simaunbi, Mbabala, Mapanza, Katengwa, Maala, Batoka, Nakeempe, Kayuni, Muyoba, Siakasenke, Chitonga, Nakamboma, Nteme, Baambwe and Hakunkula. The herdsmen were selected based on their proximity to the abattoir, from locations adjacent to the abattoir to locations approximately 100 km from the abattoir. There were five active abattoirs in Namwala, five in Monze, and one in Choma. A list of all abattoir workers both in contact with the meat or slaughter of animals or both from an abattoir in each of the three districts, was obtained and used to randomly select individuals to screen. An informed consent was obtained from all study participants before blood collection. Four (4) ml of blood was collected via the median cubital vein by a clinical officer and stored in sterile plain tubes at +4°C. A semi-structured pre-tested questionnaire was then administered to collect information on the knowledge, attitudes, and practices of abattoir workers and herdsmen. The questionnaire was first developed in the English language and translated to the local language for better understanding of the questions by the participants.

Serum samples from 153 participants were tested in parallel to detect anti-Brucella antibodies using the i-ELISA (Ingezim Brucella Compac 2.0, Madrid, Spain) and c-ELISA (SVANOVIR®Brucella-Ab Boehringer Ingelheim Svanova, Uppsala, Sweden) test kits according to manufacturer's instructions. On i-ELISA and c-ELISA tests, the result of the sample was compared with the mean cut-off value which was >40% mean optical density and >30% mean optical density of the four conjugate control wells, respectively. Any test sample giving an optical density equal to or below this value was recorded as being positive. In this study, parallel interpretation of results was used. Therefore, any sera testing positive either on i-ELISA or c-ELISA was regarded as positive.

The data obtained were coded and entered in the Microsoft Excel 2016®, exported, cleaned and analyzed using STATA version 14® (Stata Corp., College Station, TX, USA). Categorical data were expressed in percentage, and seroprevalence was calculated by dividing the number of positive sera samples by the total samples analyzed. Using the cut-off of mean optical density ≥30% and mean optical density ≥40%, for c-Elisa and i-ELISA, respectively, the independent effects of categorical risk factors on anti-Brucella spp. seropositivity were assessed using Fisher's exact test. Variables with a p-value ≤ 0.25 from univariate analysis were used as candidate variables in the logistic model. Multivariable logistic regression model was used to calculate odds ratio at 95% confidence interval to see the degree of association between Brucella seropositivity and the risk factors. The validity of the model to the observed data was assessed by computing the Pearson chi-square goodness-of-fit test.

Ethical clearance was obtained from Excellence in Research Ethics and Science (ERES) before commencing the study (Ref No. 2018-Dec-004). Permission to conduct the study was obtained from the Ministry of Health and the National Health Research Authority. The aim and brief background of the study were explained to the study participants in the local language, thereafter written informed consents were obtained for blood sample collection and questionnaire interview. Participants were assured of confidentiality, anonymity and were free to withdraw from the study whenever they chose to do so without incurring any consequence.

The Pearson chi-square goodness-of- fit test (p = 0.1908) showed that the model fitted the data thus increasing its reliability in predicting. The Receiver-operating Characteristic curve ROC analysis demonstrated that the model was good in prediction (the area under the curve was 0.8377) (Figure 2). The model had rather high sensitivity and low specificity in classifying individuals as seropositive or seronegative (Figure 3).

The multivariate logistic regression model showed that occupation, age category, and district of residence were predictors of being seropositive to Brucella spp (Table 4). The odds of abattoir worker being seropositive to Brucella antibodies were 8.6 (95% CI: 2.6–28.2) higher than that of herdsmen being the reference group. The odds of age category 17–50 years being seropositive to Brucella antibodies were 7.0 (95% CI: 0.7–72.2) higher than being <16 years as the reference group. The odds of one having attained primary level of education being seropositive to Brucella were 1.3 (95% CI: 0.1–14.7) or secondary level of education were 6.2 (95% CI: 0.5–72.6) or tertiary level of education were 5.1 (95% CI: 0.2, 113.3) higher than that of no level of education as the reference group. Furthermore, the odds of a respondent being seropositive to Brucella antibodies were 4.5 (95% CI: 1.315.7) for Namwala and 4.9 (95% CI: 1.1–21.7) Monze higher than that of Choma as the reference group.

The estimated seroprevalence of Brucella antibodies was 20.3% (14.4% in herdsmen and 46.4% in abattoir workers). This is higher than that reported in a previous studies undertaken between August 2004 and July 2005 in the Southern province of Zambia, which found a 5.03% seroprevalence in humans (21). However, the cited study did not include abattoir workers, but rather focused only on livestock farmers. Abattoir workers were at an increased risk of exposure to Brucella because farmers in the study area tend to cull animals that are old or have reproductive challenges by slaughtering them. Furthermore, during the slaughtering process, abattoir workers are constantly exposed to aerosols and animal parts i.e., blood, tissues, fluids, with inadequate or poor use of personal protective equipment. Further, these workers were at a high risk of injury (knife-cuts) as compared to herdsmen, which increases the risk of exposure to the Brucella pathogen (20). A Brucella cattle herd seroprevalence of 28.5% has been previously been estimated in this study area (24). Comparable seropositive results among the study districts showed that Namwala had the highest (26.8%), followed by Monze (19.4%), while Choma had the lowest (11.4%). The high seropositivity in Namwala district can be attributed to its close proximity to the Kafue flood plains where most of the livestock come into close contact with wild animals. Migration of livestock from the sampled districts to the Kafue plains in search of pasture during the dry season is regular and usual phenomenon due to drastic change in the weather conditions thus increasing co-mingling of animals and transmission of infectious diseases such as brucellosis. Herdsmen tend to drink raw milk, hence the increased likelihood of exposure (5). The difference observed may also be due to the use of different diagnostic test and also the difference in the environmental condition of the region.

Our findings are higher than the 1.41% reported among butcher men, abattoir workers and herdsmen in Tanzania (20) and the 5.6 and 4.7% among abattoir worker in Cameroon (17) and Ethiopia (32), respectively. Handling of fetus and uterine contents was associated with increased risk of human brucellosis in Cameroon whereas in Ethiopia this was attributed to low levels of disease awareness and one working in the abattoir.

Similarly, our findings are higher than the reported seroprevalences among slaughterhouse workers (9.6%) in Ghana (33) and livestock farmers (4.4%) in Uganda (34) but lower than 33.3% observed among cattle herders in South Sudan (15). In contrast, our findings are comparable to 18.5% by (35), higher than the 33.5% reported among abattoir workers in Nigeria (36) and higher than the 7.9% among butcher and slaughterhouse workers in Iran (37).

One of the limitations of this study was that formal random sampling was not achieved as inclusion in the study was voluntary. Further, the participants in this study were not clinically examined for evidence of brucellosis. Therefore, the estimated seroprevalence does not imply persons were suffering from brucellosis, but that the persons had anti-Brucella antibodies, thus likely to be infected. Despite this, the results give a good indication of brucellosis situation in the sampled districts. In endemic areas, as prevailing in the study areas, cases of sub-clinical and self-limiting episodes of brucellosis are likely to show anti-Brucella spp. antibodies thus reducing assay specificity (38). However, the seroprevalence found in the current study parallels with bovine herd Brucella seroprevalence that has been reported to be 28.5% (24) and 20.7% by (29). It is most probable that the Brucella infected animals in our study area were likely to act as reservoirs for human brucellosis (39). Thus, high bovine Brucella seroprevalence in the region could explain the possible reason for the high seroprevalence in humans.

In this study, occupation, age category and district of residence were identified as risk factors for human brucellosis. Our current study revealed that seroprevalence was highest among abattoir workers 46.43% (n = 13). The odds of disease were 8.6 times higher in abattoir workers than in herdsmen, implying the former were more at risk.

The 17–50 years' age category (considered to be actively working with livestock) was the most commonly affected with Brucella in the study area although this was not statistically significant. Furthermore, a respondent who was in the 17–50 years' age category was 7 times more likely to be infected with brucellosis than those who were <16 years old (school going children). These did most of the physical work requiring direct contact with the livestock and livestock products. Similarly (17), reported an association between age and an increased seroprevalence. Similar findings have been reported in Pakistan where age was statistically associated to Brucella seropositivity (40).

Namwala district had 4-fold higher risk of brucellosis than Choma. This might be due to the fact that Namwala has the highest cattle density in the Country and animals are communally grazed on the Kafue plains (30). Furthermore, Mfune et al. (24) established that the odds of testing positive for Brucella in animals were high in Namwala district (OR = 8.55, CI: 2.66–27.44) compared to those from other districts.

A noteworthy association is the type of husbandry practices (Communal/individual) coupled with contact with the Kafue lechwe which have been documented to harbor brucellosis on the flood plains of the Kafue River in this ecosystem (30, 41). Assenga et al. (39) demonstrated the presence of anti-Brucella antibodies in humans, livestock, and wildlife in the Katavi- Rukwa ecosystem in Tanzania. Transmission of the infection between wildlife, livestock and humans is likely to continue due to increasing human activities in the human wildlife interface.

However, the spread out large odds ratios, with wide confidence intervals obtained in this study should be cautiously interpreted, given that the distribution of the individuals within the two occupational categories of the risk factors was not even.

Although this study implicated livestock breeding method (natural breeding) (p = 0.02) as a probable risk factor associated to Brucella seropositivity, Nguna et al. (34) reported that livestock breeding method was not an important risk factor among the herdsmen. Communal grazing coupled with natural breeding is widely practiced hence the risk of transmission of disease to susceptible animals. Similar studies were done by Kubuafor et al. (42) where a significant association between antibodies against Brucella and a history of abortions and retained placenta was observed.

History of blood splashes around the mouth was observed to be associated with seropositivity among abattoir workers (p = 0.01). Similar findings have been documented in Southwest Nigeria (43), Tanzania (44) and Egypt (45). Noteworthy difference is that breeding method and blood around the mouth were potential risk factors and statistically significant at univariate analysis but were dropped in the final multivariable logistic regression model. This was done to increase the predictability of the final model.

The odds of disease in participants older than 50 years was <1, implying old-age was a protective factor. In spite of the high seroprevalence of Brucella infection in humans, it was not considered for routine laboratory diagnosis in cases of acute febrile illness.