Hybrid surveillance and Bayesian modelling to estimate brucellosis incidence in a livestock-keeping population in northern Tanzania
- 1University of Glasgow
Abstract
Brucellosis is a globally distributed and widely under-appreciated zoonotic bacterial disease that causes infertility and abortion in livestock and febrile illness in humans, sometimes with debilitating sequelae. Brucellosis incidence estimates typically focus on acute presentations and range from 50 to 100 per 100,000 people per year for endemic settings, showing marked variation within and between countries. Estimating brucellosis incidence in low-resource settings is complicated by lack of access to and utilisation of health care and high-quality diagnostic testing for many people in rural communities. To estimate brucellosis incidence in a pastoralist population in northern Tanzania, we developed a Bayesian model that integrates multiple data sources including diagnostic test results from a hospital-based prevalence study performed in 2016-2017, test sensitivity and specificity, and data from a community-based healthcare utilisation survey in 2019-2020. We show that this population experienced a high incidence of brucellosis with wide uncertainty, with estimates of 387.6 (95% HPDI: 103.3 - 903.5) illnesses per 100,000 people per year for the total population, 199.7 (95% HPDI: 16.2 - 584.6) for individuals aged >15 years, and 632.4 (95% HPDI: 180.1 - 1379.4) for children aged [≤]15 years. These incidence estimates were among the highest in the world to date. The wide density intervals for these incidence estimates reflect uncertainty in the underlying parameter values used. Sensitivity analyses show that healthcare utilisation influenced incidence estimates the most. This result indicates that accounting for low levels of healthcare utilisation is essential to avoid under-estimating disease incidence. Our findings illustrate that brucellosis is a frequent cause of illness in this population. Increased access to treatment, prevention, and accurate diagnostic testing would greatly improve understanding of brucellosis epidemiology and the health of many people living in similar at-risk communities.
Related
- An integrated One Health methodology to study Rift Valley fever, Crimean-Congo haemorrhagic fever and brucellosis at the human-livestock-wildlife interface in Uganda☆
- Occupational brucellosis among livestock farmers in Northeastern Thailand: A cross-sectional study of knowledge, attitudes, and practices from a One Health perspective
- Facilitators and barriers to community-based livestock abortion reporting for Rift Valley Fever surveillance in Uganda: A COM-B analysis
- Population-based incidence rate of in-hospital delirium in older adults in Norway—a HUNT study
- Sex differences in immune variables and respiratory infection incidence in an athletic population
BibTeX
@misc{holmes2026hybrid,
title = {{Hybrid surveillance and Bayesian modelling to estimate brucellosis incidence in a livestock-keeping population in northern Tanzania}},
author = {A. L. Holmes and W. A. de Glanville and W. A. Katiti and A. J. F. Mendes and R. F. Bodenham and A. Lukambagire and B. T. Mmbaga and M. P. Rubach and J. A. Crump and S. Cleaveland and M. Viana and J. E. B. Halliday},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.24.26363878},
url = {https://www.medrxiv.org/content/10.64898/2026.09.24.26363878v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Hybrid surveillance and Bayesian modelling to estimate brucellosis incidence in a livestock-keeping population in northern Tanzania
AU - A. L. Holmes
AU - W. A. de Glanville
AU - W. A. Katiti
AU - A. J. F. Mendes
AU - R. F. Bodenham
AU - A. Lukambagire
AU - B. T. Mmbaga
AU - M. P. Rubach
AU - J. A. Crump
AU - S. Cleaveland
AU - M. Viana
AU - J. E. B. Halliday
PY - 2026
DA - 2026/09/24
PB - medRxiv
DO - 10.64898/2026.09.24.26363878
UR - https://www.medrxiv.org/content/10.64898/2026.09.24.26363878v1
AB - Brucellosis is a globally distributed and widely under-appreciated zoonotic bacterial disease that causes infertility and abortion in livestock and febrile illness in humans, sometimes with debilitating sequelae. Brucellosis incidence estimates typically focus on acute presentations and range from 50 to 100 per 100,000 people per year for endemic settings, showing marked variation within and between countries. Estimating brucellosis incidence in low-resource settings is complicated by lack of access to and utilisation of health care and high-quality diagnostic testing for many people in rural communities. To estimate brucellosis incidence in a pastoralist population in northern Tanzania, we developed a Bayesian model that integrates multiple data sources including diagnostic test results from a hospital-based prevalence study performed in 2016-2017, test sensitivity and specificity, and data from a community-based healthcare utilisation survey in 2019-2020. We show that this population experienced a high incidence of brucellosis with wide uncertainty, with estimates of 387.6 (95% HPDI: 103.3 - 903.5) illnesses per 100,000 people per year for the total population, 199.7 (95% HPDI: 16.2 - 584.6) for individuals aged >15 years, and 632.4 (95% HPDI: 180.1 - 1379.4) for children aged [≤]15 years. These incidence estimates were among the highest in the world to date. The wide density intervals for these incidence estimates reflect uncertainty in the underlying parameter values used. Sensitivity analyses show that healthcare utilisation influenced incidence estimates the most. This result indicates that accounting for low levels of healthcare utilisation is essential to avoid under-estimating disease incidence. Our findings illustrate that brucellosis is a frequent cause of illness in this population. Increased access to treatment, prevention, and accurate diagnostic testing would greatly improve understanding of brucellosis epidemiology and the health of many people living in similar at-risk communities.
ER -