This community innovation has been accepted at the 2026 DHIS2 Annual Conference and will be in abstract track/lightning talk.
From DHIS2 Silos to Cross-Program Patient Journeys
In MSF-France facilities, routine data are recorded at the service-level (e.g. ER, ICU…). Each service captures its data in a separate DHIS2 program, documenting only one part of the patient’s care. As a result, information for the same patient is spread across multiple sources, making it difficult to follow continuous hospital stays and detect readmissions, complications, or breaks in care. The goal of this work is therefore to reconstruct complete patient journeys by linking records from all services, so that each patient can be tracked from admission to discharge, and reliable indicators can be produced, while also highlighting data-quality issues. We designed a longitudinal data model where each row represents one patient event in one service, ordered over time. In event programs (ER/OT), admission and discharge are recorded as a single row, while in tracker programs (ICU/IPD) stays are recorded as two rows. Data were extracted from Tracker API sources, then standardized into a common schema. A deterministic linkage strategy built a robust patient_key across programs using normalized case number, cleaned initials, and sex, with age-compatibility checks. The outcome of the reconstructed patient pathways, allowed for the visualization of patient journeys from first admission to final discharge from MSF care. Furthermore, analysis was conducted on the entire patient journey of trauma and burn patients through applying data-quality controls and running negative binomial and Cox models.. From the initial facility assessed, the two key findings were that severe burns were associated with ~3× longer stays, while higher bed occupancy was associated with ~10% shorter stays, consistent with discharge pressure under capacity constraints.
Primary Author: Lilit ISAKHANYAN
Keywords:
DHIS2, , Tracker, , API, Extraction, Pipeline, Python, Pandas, Patient Key, Identifier, Standardization, Harmonization, Bed Occupancy, Bed Pressure, Survival Analysis,Length of Stay, Mortality, Regression, Negative binomial, Overdispersion, Survival, Cox, Hazard ratio, Estimation, Determinants, Data cleaning, Ethics, Governance, Decision-making, Economics, Health, Humanitarianism
4 Likes