"background": "Public health surveillance systems are critical for timely disease detection and response, yet their methodological rigour and operational reliability in resource-limited settings are often inadequately assessed. In Kenya, despite significant investment, systematic evaluations of surveillance data quality and system performance remain sparse. ", "purpose and objectives": "This case study aimed to conduct a methodological evaluation of the country's integrated disease surveillance and response system. Its primary objective was to assess the reliability of surveillance data across administrative levels and identify key predictors of reporting consistency. ", "methodology": "We employed a multilevel regression analysis on a national dataset of weekly notifiable disease reports. The core statistical model was a three-level negative binomial: \ () = \0 + \ X{ijk + uj + vk, where is the expected count for facility i in sub-county j and county k, X represents facility-level covariates, and u and v are random intercepts. Model inference used robust standard errors. ", "findings": "Analysis revealed substantial heterogeneity in reporting completeness, with a significant intra-class correlation of 0. 31 at the county level. Facilities with dedicated surveillance officers were 2. 4 times (95% CI: 1. 8 to 3. 2) more likely to submit complete reports. The strongest predictor of reliability was the frequency of supervisory visits. ", "conclusion": "The surveillance system's reliability is compromised by structural inconsistencies, particularly at sub-national levels. While the system architecture is sound, its operational performance is uneven, affecting data utility for public health action. ", "recommendations": "Implement regular, data-driven performance audits with feedback mechanisms. Strengthen sub-national capacity through targeted training and consistent supervision. Institutionalise the use of multilevel modelling for ongoing system monitoring. ", "key words": "surveillance evaluation, health information systems, multilevel modelling, data quality, health systems strengthening",
Mwangi et al. (Wed,) studied this question.