Study design and setting
This was a facility-based cross-sectional study conducted between May and June 2018 in Ngorongoro District, Arusha region of Northern Tanzania. The district plays host to parts of the wildebeest migration at the same time cattle, goat and sheep rearing is a common practice. The population of the Ngorongoro District is around 130,000 and the major ethnic groups are the Masai and Sonjo.
The Ngorongoro District has 20 public health facilities including 14 dispensaries, four health centres, and two hospitals. Each of the two hospital records between 25- 40 new antenatal clinic attendances per week. The study involved six health facilities that provide antenatal services including: Wasso designated district hospital, Sakala and Loliondo health centers as well as Muholo, Sale, and Samunge dispensaries.
Study population, sample size, and sampling procedure
Pregnant women attending the antenatal clinic at selected health facilities were invited to participate in the study. The study enrolled pregnant women who lived in the study area for more than three months and accepting to participate by signing written informed consent were enrolled. The sample size was estimated using Kish Leslie formula , at 95 % confidence interval (CI) considering 7.7% seroprevalence of Brucella infection in Arusha Tanzania  and a 3% margin of error. Eligible clients were consecutively enrolled in the study until reaching a representative sample size.
A structured questionnaire (additional file 1) was used to collect the required information from each participant. Data for socio-demographic and obstetric characteristics included: age, marital status, education level, occupation, location, gestation age, gravidity, parity and history of spontaneous abortion. Factors with potential risk for Brucella transmission related to animal care, animal product consumption and presence or absence of exposure at the individual level were also collected. The questionnaire included contact with animals and animal products, involvement in milking, sharing water sources with animals, assisting animals to give birth or drink animal fresh milk.
The dependent variable was Brucella serostatus as defined using the Rose Bengal Plate Test result and independent variables were behavior and practices with potential risk for Brucella infection. Regular contact with animal manure was defined as unprotected exposure to manure at least once in every week in the past three months. Participants were counted to contact the placenta if assisted animals giving birth at least once in the past three months. Washing animals at home was counted when performed at least once every week for three months. Preference of foodstuffs like animal fresh milk, raw animal blood, and raw meat was defined as consumption of the same at least once every week in the past three months.
Experienced health personnel working at the facilities collected 4 ml of venous blood aseptically using a plain vacutainer system. The collected specimens were labelled with the specific participant's identification number. Serum samples were separated from whole blood by centrifugation at 3,000 rpm for five minutes. The specimens were kept at room temperature for 30 min then at 2 - 8 OC up to 24h before processing
Rose Bengal Plate Test
The Brucella serology was first determined by Rose Bengal Plate Test (RBPT) a rapid agglutination test as previously described . The test does not differentiate antibodies against different Brucella species like Brucella abortus and Brucella melitensis. Briefly, a drop of serum (50μl) was taken using a clean micro-pipette onto the test plate beside an equal (50μl) drop of RBPT antigen. The drops of serum and antigen were mixed using applicator stick then rocked manually for 4 minutes before examination. The presence of any visible reaction was considered to be positive .
Enzyme-Linked Immunosorbent Assay
Positive samples were kept at minus 20 OC before transportation to the reference laboratory in Dar es Salaam for the detection of Immunoglobulin M and G antibodies. The commercially available test kits of enzyme-linked immunosorbent assay (ELISA), SERION ELISA classic Brucella IgG/IgM/IgA (Institut Virion/Serion GmbH) was used to detect IgM and IgG antibodies. The technique was performed according to the instructions from the manufacturer. In brief, 100 ml of diluted serum samples and ready to use control were added to the micro test wells containing antigen. The assays were then incubated at 37 OC for 60 minutes, after which the first wash was performed. Later, anti-human IgM or IgG conjugated with an enzyme was added and incubated for 30 minutes at 37 OC. All wells were washed to remove excess conjugate, followed by a new incubation for 30 min at 37 OC with the enzyme-substrate. Finally, the reaction was stopped by adding 100 ml of stopping solution. The enzyme reaction with the Substrate yields a coloured product. The colour intensity is proportional to the amount of specific antibody and can be measured by the photometric method.
Categorical variables were summarized as frequencies and proportions while continuous variables were summarized as median and inter-quartile range (IQR). Group differences were examined using Pearson’s Chi-square test. Bivariate analysis was conducted to determine the variables associated with Brucella seropositivity and crude odds ratio (cOR) with 95% CI. Multivariable logistic regression was performed to examine the associations between the outcome variable and independent variables after adjustment for other variables as fixed effect and the cluster variable facility as random effect. Likelihood ratio tests (LRT) was used to simplify the final multivariable model so that only variables that are significantly associated with the outcome are retained in the final model. Associations in the multivariable logistic models were presented as adjusted odds ratios (AOR) with 95% CI. Interactions between independent variables were examined, and the Wald test was used to test the associations of the variables and interactions. The Hosmer-Lemeshow test was used to examine the overall fitness of the model. Statistical Package for Social Sciences version 23 was used for all data analyses. The level of significance was specified at 0.05.