Modeling hepatitis incidences in Jos, Nigeria using Bayesian truncated Poisson regression

Hepatitis remains a significant public health burden in Nigeria, with Jos, Plateau State, representing a critical urban center for epidemiological study. Traditional Poisson regression often fails to account for under-reporting, a common issue in disease surveillance systems, leading to biased estimates. This study aimed to model the incidence rates of Hepatitis in Jos, Nigeria, utilizing a Bayesian Truncated Poisson Regression framework to correct for potential under-reporting and provide robust estimates of disease determinants. We analyzed secondary surveillance data of Hepatitis cases (A, B, and C) reported in Jos from 2018 to 2023. A Bayesian Truncated Poisson model was developed, incorporating covariates such as age, gender, socioeconomic status, and seasonal variation. The truncation point was set at zero to account for the possibility of unobserved (unreported) cases. Models were fitted using Markov Chain Monte Carlo (MCMC) methods via Stan/RStan. The Bayesian Truncated Poisson model demonstrated a superior fit to the data compared to the standard Poisson model (as indicated by Widely Applicable Information Criterion (WAIC) values). Key findings identified age (25-44 years, Posterior Mean Odds Ratio [PM-OR]: 2.3, 95% Credible Interval [CrI]: 1.8–3.0) and low socioeconomic status (PM-OR: 1.9, 95% CrI: 1.4–2.5) as significant risk factors for Hepatitis incidence. The model estimated a mean under-reporting rate of approximately 30%.The application of Bayesian Truncated Poisson Regression provides a more realistic model for Hepatitis incidence in a resource-constrained surveillance setting. The findings highlight high-risk demographics and quantify under-reporting, offering actionable insights for targeted public health interventions and improved data collection strategies in Jos and similar contexts.