Study design and sample size determination
A cross sectional study compared knowledge and practices related to brucellosis between participants from the predominantly nomadic pastoralists of Marsabit County and that of mixed farmers of Kiambu County (Fig. 1). Kiambu County neighbors Nairobi, the capital city of Kenya, and is located in a high potential agro-ecological zone with farmers practicing smallholder livestock production (keeping primarily cattle, sheep and goats) and crop farming. Marsabit County is located in the northern arid agro-ecological zone of the country and farmers practice nomadic pastoral livestock production mainly, keeping cattle, sheep, goats and camels. The estimated livestock population in Marsabit County is 2,731,407, of which 42% are goats, 35% are sheep, 16% are cattle and 7% are camels; whereas Kiambu County has a livestock population of 1,832,045 of which 39% are sheep, 38% are goats, 22% cattle and < 1% camels.
Kiambu County has good physical infrastructure with 35% of the roads tarmacked or on gravel, accessible medical and veterinary services, and is densely populated with over 630 persons per square kilometer, inhabited by a community with high literacy levels, more than 45% of them deriving livelihood from the Capital City of Nairobi [16]. In contrast, Marsabit County has only one major road with most areas inaccessible for medical or veterinary services and is sparsely populated with 4 persons per square kilometer, inhabited by a poor, nomadic pastoralist communities that derive their livelihood from rearing livestock, including cattle sheep, goats, and camels [17].
This study was part of a larger study on seroprevalence of and risk factors for brucellosis infection in humans and livestock in Kenya whose findings were published previously [10]. The sample size was calculated based on an estimated Brucella spp. seroprevalence of 5% in Kiambu County and 50% in Marsabit County, with an error margin of 2 and 5%, respectively, at 95% confidence level. A design effect of two and a factor of 10% were applied to account for clustering and non-response respectively, giving a minimum sample size of 730 individuals for Marsabit and 866 individuals for Kiambu counties.
Household selection and sampling
The study applied stratified random sampling to identify study households in each county [10]. In the first stage, sub-locations were stratified by predominant livestock production system and 10% of sub-locations were randomly selected from each stratum in each county. This resulted in 21 sub-locations in Kiambu County and 10 in Marsabit County. In the second stage, the number of households to be visited in each sub-location were determined proportionate to the total human population and assuming an enrolment of three persons per household. In order to identify households to recruit into the study, random geographical coordinates were generated using ArcGIS corresponding to number of households for each sublocation. The selected household coordinates were loaded into a global positioning system device used by each study team. When the coordinates did not correspond to a household, the nearest household was visited. In each household, up to three persons aged 5 years and above were randomly selected, consented/assented in line with the ethical approval, and a structured questionnaire, loaded on to a smartphone, administered to each participant and the household head. Nomadic pastoralists were defined as households whose livelihood was based primarily on domesticated livestock production and involved seasonal movement of dwelling. Mixed farming were households whose livelihoods depended on both livestock rearing and crop farming.
Data and sample collection and laboratory testing
We used an electronic interviewer administered structured questionnaire with standardized questions and scheme to collect data from household respondents on knowledge and practices that may be associated with increased risk of infection with Brucella spp. The questionnaire was pretested, and interviewers trained before data collection. The data collected included knowledge of human and animal brucellosis including transmission, symptoms and modes of prevention. The study also collected data on practices at individual level including drinking of unboiled milk, assisting in animal birthing, drinking raw blood, working with raw hides and skins. Weekly frequencies on selected variables were done to check on data quality. A blood sample was collected from all eligible persons and animals as previously reported [10]. After processing for sera, the specimens were tested for presence of anti-Brucella spp. IgG antibodies using IBL-America IgG enzyme-linked immunosorbent assay (ELISA) and Svanova Biotech AB ELISA kits for human and animal samples respectively as we previously reported [10].
Data analysis
Data were analysed using R statistical software, version 3.5.1 [18]. Categorical variables were presented as percentages and their associations assessed by Chi-square test while continuous variables were tested using the t-test. Knowledge on human and animal brucellosis by household heads was presented by production system practiced by the household (nomadic pastoralism vs mixed farming).
The prevalence of practices among participants from households practicing nomadic pastoralism or mixed farming was compared. We conducted a multivariable mixed effects logistic regression model with human Brucella spp. IgG seropositivity as the outcome variable and included the practices, sex, age, and education level as predictor variables. Household was included in the model as a random effect to account for possible clustering. P-values < 0.05 were considered significant. Missing values were excluded from the analysis and a goodness-of-fit test was conducted on the model using Hosmer-Lemeshow test (p > 0.05).
Ethical approval
The study received ethical approval by the Kenya Medical Research Institute Scientific Ethical Review Committee (No. 2193) and Centers for Disease Control and Prevention Institutional Review Board. Project approval was also obtained from the Kenya Ministry of Health, and the Ministry of Agriculture Livestock and Fisheries.