A Bayesian Hierarchical Model for Assessing the Reliability of Public Health Surveillance Systems in Uganda, 2000–2026

{ "background": "Public health surveillance systems are critical for disease control, yet their reliability in low-resource settings is often uncertain. Existing evaluation frameworks lack robust quantitative methods to integrate heterogeneous data sources and account for spatial and temporal dependencies in system performance.", "purpose and objectives": "This study aimed to develop and apply a novel Bayesian hierarchical model to quantitatively assess the reliability of national public health surveillance, defined as the probability of correctly detecting and reporting a notifiable disease event.", "methodology": "We developed a Bayesian latent variable model integrating case notification data, laboratory confirmation reports, and health facility survey data. The core reliability metric is modelled as (\ρit) = \α + \β Xit + ui + \, where \ρit is the reliability for district i in period t, Xit are covariates, $ui$ are spatial random effects, and \ are temporal effects. Inference used Markov chain Monte Carlo sampling.", "findings": "Posterior estimates revealed substantial spatial heterogeneity in system reliability, with a median national reliability of 0.72 (95% credible interval: 0.68–0.76). Reliability was strongly associated with health facility density (posterior probability > 0.95) and exhibited a declining temporal trend in northern regions.", "conclusion": "The proposed model provides a rigorous, data-driven tool for quantifying surveillance reliability, revealing significant and spatially structured gaps in performance.", "recommendations": "Resource allocation for surveillance strengthening should be prioritised based on quantitative reliability estimates. The model should be integrated into routine performance reviews to enable targeted interventions.", "key words": "Bayesian inference, disease surveillance, health systems, hierarchical model, latent variable, reliability, sub-Saharan Africa", "contribution statement": "This paper introduces a novel Bayesian latent variable framework for quantifying public health surveillance reliability, providing the first spatially explicit, probabilistic assessment for a national system in