This mixed-methods study evaluated Zimbabwe’s national Impilo electronic health record (EHR) to determine its capacity to support locally driven learning for integrated, patient-centred care in the context of rising multimorbidity. The assessment used HIV-hypertension as a tracer condition pair to examine how routine care and EHR use might enable adaptive changes at the primary care level.
The study combined documentary review, ethnographic observation, patient journey mapping, and interviews with frontline health workers and key stakeholders. Analysis was guided by Friedman's socio-technical infrastructure model and interpreted through Learning Health Systems (LHS) theory. These frameworks oriented the team to both the technical features of the EHR and the organisational, social and practice-level arrangements needed for learning to stabilise and inform system adaptation.
Frontline learning for person-centred multimorbidity care was actively generated in routine practice. Health workers used a combination of the EHR and paper-based artefacts—such as facility registers and patient-held booklets—to interpret patient trajectories and adjust care based on experience. Coordination across HIV and hypertension services emerged through these practices, with clinicians making incremental adjustments within encounters.
This day-to-day interpretive work produced valuable, practice-relevant insights about individual patient pathways and service interactions. However, the study found that this learning largely remained local to encounters and individual practitioners rather than becoming a durable, shared resource for the health system.
Although frontline actors generated learning from patient trajectories, Impilo did not routinely deliver several features necessary to support integrated multimorbidity care at scale. Specifically, the system lacked:
Because these elements were absent or weak, the EHR’s data and the learning produced at the bedside did not translate into broader changes in care pathways, routines, or system design. Instead, adaptations were predominantly incremental and confined within existing workflows.
The assessment highlights a central insight: the main barrier to developing a Learning Health System in this context is not an absence of data or frontline learning capacity. Rather, it is the lack of socio-technical arrangements that enable learning to stabilise, aggregate, and inform system-level adaptation. In other words, digitalisation on its own—implementing an EHR—does not automatically produce the organisational processes, analytics, or collective practices required for an LHS.
This distinction underscores the importance of attending to the non-technical dimensions of digital health interventions: governance for shared reflection, design of practice-facing analytics, workflows that surface longitudinal patient information, and mechanisms that convert individual clinician insight into institutional change.
The study concludes that to support adaptive, integrated multimorbidity care, EHR deployment should be accompanied by deliberate socio-technical design and organisational mechanisms that stabilise frontline learning. One proposed pathway is co-production with frontline health workers to better align digital system design with routine care realities, ensuring tools provide usable longitudinal views and practice-facing analytics and that institutionalised feedback and reflection mechanisms are established.
Overall, the findings caution that digitalisation alone is insufficient to create adaptive multimorbidity care systems; meaningful system change requires both technical features and organisational arrangements that make frontline learning visible, shareable and actionable.
The Medical Research Council of Zimbabwe (MRCZ/A/3315) and the London School of Hygiene & Tropical Medicine Research Ethics Committee (Ref: 32628) provided ethical approval. Written informed consent was obtained from participants. The work was funded by the Wellcome Trust (224309/Z/21/Z). The article is a preprint posted July 19, 2026, on medRxiv and has not been peer reviewed. The authors declared no competing interests.