Background

In low- and middle-income countries (LMICs), effective cancer prevention and early detection depend on timely, high-quality surveillance data, yet subnational cancer intelligence systems remain weak. In Kenya’s devolved health system, counties are responsible for implementing cancer control strategies, but fragmented reporting pathways, incomplete pathology data, and limited digital integration constrain local decision-making. Kiambu County, a rapidly urbanizing region with substantial public and private diagnostic activity, provides a critical setting to evaluate and strengthen county-level cancer surveillance in alignment with the WHO Global Initiative for Cancer Registries (GICR).

Objectives

To assess the performance of the cancer data ecosystem in Kiambu County and to evaluate the feasibility and implementation impact of digital and AI-enabled interventions for improving cancer case ascertainment, data quality, and use in prevention and early detection programs.

Methods

We conducted a mixed-methods evaluation embedded within routine county health operations through collaboration between Kiambu County health authorities, cancer registry personnel, diagnostic facilities, and technical partners. Cancer case reporting workflows were mapped across the diagnostic continuum. Quantitative audits of pathology reports and registry abstraction records assessed completeness (case capture, morphology, topography, stage), timeliness (diagnosis-to-registration intervals), and data consistency. Semi-structured interviews with clinicians, pathologists, health records officers, and registrars identified operational and workforce constraints.
Digital interventions—including structured histopathology reporting templates, automated case-finding tools, and AI-assisted coding and abstraction algorithms—were implemented and assessed using implementation outcomes (feasibility, adoption, workforce impact, and data quality improvement). Capacity-building was integrated through hands-on training, co-design of workflows, and mentorship of county and facility staff. Findings were triangulated to identify priority system improvements.

Results

Baseline assessment revealed substantial under-ascertainment of cancer cases, driven by incomplete pathology submissions, non-standardized diagnostic reporting, manual abstraction processes, and weak linkage between facility-based systems and the population-based cancer registry. These gaps resulted in delayed and incomplete surveillance outputs. Implementation of structured pathology reporting improved completeness of key diagnostic variables by an estimated 15–25%. Integration of digital case-finding and AI-assisted abstraction tools increased overall case capture by 20–35% and reduced registrar workload, accelerating case consolidation. Capacity-building activities enhanced local ownership and routine use of cancer data for monitoring cervical, breast, and colorectal cancer early detection initiatives.

Conclusions / Implications

Strengthening cancer intelligence systems in Kiambu County is both feasible and impactful when digital innovations are paired with embedded capacity-building and county ownership. This hybrid evaluation–implementation approach provides a scalable model for other Kenyan counties and LMIC settings seeking to translate national cancer control commitments into effective, data-driven local action.