Epidemiology of human and animal brucellosis in Kenya: A one health meta-regression and network analysis

Background: Brucellosis is a prioritised zoonotic disease in Kenya, but an in-depth quantitative review of its presence in humans and animals is currently lacking.

Methods: Systematic literature searches were performed in African Journals Online (AJOL), Embase, PubMed, Scopus, and Web of Science databases. Relevant data was extracted from prevalence studies and random-effects meta-analyses were applied to estimate pooled prevalence by sample and diagnostic types for five key species. Between-study heterogeneity was explored using meta-regression models. A seroprevalence-weighted bipartite network analysis was used to map brucellosis research in various host species to livelihood zones in Kenya.

Results: We found 65 brucellosis prevalence studies that increased with time, spanning 1969–2024 (55 years) and consisting of samples from 20,956 humans, 57,340 domestic animals, 727 wildlife from 16 species, and 235 Hippoboscid flies (<i>Hippobosca</i> spp.). Sampling methodologies comprised non-probability (n = 29, 44.6%), mixed (i.e., both probability and non-probability, n = 17, 26.2%), probability (n = 16, 24.6%), and census approaches (n = 3, 4.6%), and sample size calculations were reported in 38.5% (n = 25) of studies. The overall seroprevalence using indirect serological tests was: 4.8% (95% CI: 3.1–7.3) in sheep, 7.9% (95% CI: 4.5–13.6) in goats, 10.5% (95% CI: 7.1–15.4) in cattle, 10.8% (95% CI: 4.4–24.2) in humans from community settings, and 13.2% (95% CI: 7.6–21.9) in camels. There was high between-study heterogeneity in all hosts (I2 > 89.2). Direct tests had 3.29 times higher odds (95% CI: 1.61–6.71, p-value = 0.0012) of determining positive results in animals than indirect tests. There was no clear difference between seroprevalence in domestic and wild animals (p = 0.63). Network analysis revealed that pastoral (strength = 130.4) and agropastoral zones (strength = 69.6) had the highest brucellosis presence, with humans (strength = 48.4) and cattle (strength = 37.6) showing the highest cumulative seroprevalence among hosts. Goats were the most central hosts (betweenness = 40), and all wildlife were peripheral nodes (betweenness = 0) despite considerable seroprevalence. Over half (55.4%) of possible host-zone combinations were unstudied.

Conclusion: There is widespread brucellosis, especially in pastoral and agropastoral systems. Current published evidence has considerable geographic gaps and under-represents wildlife. The use of probability sampling approaches in future epidemiological studies, where possible, could ensure results are more representative of the population. A One Health approach integrating human, livestock, and wildlife populations is required in future studies for comprehensive risk assessment of brucellosis in Kenya.