Estimating the seroprevalence of t. Parva infection and determination of associated risk factors in the cattle population of Mbeere District, Kenya
The most important tick-borne disease (TBD) of cattle in Kenya is East Coast Fever (ECF) caused by Theileria parva, and transmitted by the tick Rhipicephalus appendiculatus. A study was carried out in Mbeere District, Kenya to estimate the seroprevalence of T parva infection and determine the associated risk factors. Previously, no well-structured observational tick-borne diseases' study have been conducted in a mixed crop-livestock system such as that found in marginal semiarid rangeland in Mbeere District. Tick-borne diseases are among the important livestock production constraints in Mbeere District. The objectives of this study were therefore to identify potential livestock production constraints, estimate T parva seroprevalence, and to identify and assess risk factors associated with T parva seropositivity in Mbeere District in a cross-sectional study. As the study data exhibited a nested hierarchy, further objectives included evaluating and quantifying variance components corresponding to three levels of cattle population organization (herd, sub-location and division) and using the variance components to identify levels in which there could be substantial opportunities for targeted interventions. The farms were selected by a multistage random sampling method. All the four administrative divisions (Siakago, Gachoka, Evurore and Mwea) in the district were included in the study. The 39 sub-locations in the district were listed according to divisions and two sub-locations from each division were randomly selected giving a total of 8 sub-locations. In collaboration with the Assistant-chief and village elders from each selected sub-location, a list of all households owning cattle was compiled and ten of them from each sub-location were randomly sampled using random number tables to end up with eighty farms. Sampling of individual animals in the farm used proportional allocation approach. A constant 50% proportion of animals in each herd were selected using systematic random sampling approach. All animals in a herd were, however, sampled if the herd size was 6 or less. A total of 440 animals were sampled for this study. Risk factor information on the farming system data and on animal and farm level management practices was gathered using a standardized questionnaire. Approximately 5 ml of blood were collected from the selected animals for serum preparation. Serum samples were assayed for antibodies to T. parva by Enzymelinked immunosorbent assay (ELISA). Potential livestock production constraints were analyzed descriptively by proportions and rankings. The relationship between T. parva seropositivity and i risk factors was assessed by multivariable logistic regression models. To investigate the contribution of the levels of organization to the total variance of T parva seropositivity, this study applied multilevel models using Schall's algorithm. The 8 sub-locations selected represented 20% of all sub-locations in the district (n=39) whereas the 80 farms selected represented approximately 1.7% of all farms in the sampling frame (n=4783). The 440 animals selected comprised of 66% of all animals in the selected farms. Ninety-four percent (94%) of the households depended on both crops and livestock as the main farm activity while the rest depended on livestock only. Sixty-three percent (63%) of the households depended on both crops and livestock for household cash income. Overall, 90% of the households kept only indigenous cattle breeds and their crosses reared under open grazing management systems. Reasons advanced for keeping cattle in order of importance were milk, homestead "security" banks, traction and manure. Feed and water availability and diseases were ranked as the most important farm constraints in all divisions. East Coast fever was ranked first only in Mwea division. The overall ECF seroprevalence was 19.3% (range: 3.9% to 48% across divisions) in the district [95% CI: 13.7%, 24.9%]. Regression analysis found four major factors (fixed effects) associated with seropositivity: presence of the vector tick on the farm (odds ratio (OR) = 3.8), frequency (number of times) of calf tick control before 6 months of age (for frequency of >5 times, OR = 3.9 relative to frequency of :::;5times), herd size (OR for herd size category '6-10 cattle' = 2.7; OR for over herd size category 'over 10 cattle' = 0.95 relative to herd size category' 1-5 cattle') and division (ORs for Siakago, Gachoka and Mwea Divisions = 0.3, 0.21 and 5.1 respectively relative to Evurore Division). There were distinct differential herd tick control practices across the district. Siakago and Gachoka Divisions had the least T. parva seroprevalence. Farmers in these two divisions had characteristically large herd sizes. On the other hand, Evurore and Mwea Divisions had the highest T. parva seroprevalence. In contrast, farmers in these two divisions had characteristically relatively smaller herd sizes. 'Division', as an area-level variable, had the most important and large effects associated with T. parva seroprevalence. As divisional boundaries are administrative in nature, a multivariable model without variable 'Division' was built to investigate significant effects which could have been masked by effects of 'Division'. Three additional effects became significant: age at which calf tick control was started, Normalized Differential Vegetation Index (NDVI) and agroecological zone (AEZ) indicating the importance of area-level effects. Upon evaluation of sources of variation in T. parva seropositivity in the cattle population of Mbeere District, the absolute values of variance components for the hierarchical levels were as follows for random effects models (no fixed effects): herd (2.314), sub-location (0.393), division (2.151) and variance due to error (0.532). Presence of the significant fixed effects changed the herd-level variance component value only slightly to 2.2185 while the values for sub-location (0.71) and division (l.6878) changed substantially. The variance due to error for the mixed model was 0.519. The study findings demonstrated that traditional mixed crop-livestock farming system was an important enterprise in Mbeere District. Although prevalence alone is not an adequate measure to make concrete conclusions on T parva infection endemicity status, the low T parva seroprevalence in Mbeere District indicated that ECF most likely occurred in the district in an endemic instability state. Thus, this requires more proof in a longitudinal study. The differential herd tick control strategies across the district probably arose out of differences in perceptions of ECF occurrence and importance. Substantial variation of T parva seroprevalence in Mbeere District rested at herd and division levels though the variance due to herd was larger than that due to geographical areas in multilevel models. The large variation at the herd level appeared to be related to the differential tick control management practices across the district. It also indicated that outcomes within a herd had a common cause and/or strong dependency in terms of parasite exposure. The large geographical variation could be related to possible differential ecological and climatic variability in vector suitability habitats in the district. These observations suggested that T parva seroprevalence in Mbeere District was mainly influenced by herd and environmental effects. These results implied that both among-farm and -division T parva transmission factors were important in designing ECF control strategies in the district. Thus, practical ECF intervention strategies should be directed at all farms within the two high risk divisions of Mwea and Evurore for the control efforts to have the greatest impact.