Emily Liddell

Digital health technologies have untapped potential to transform community health systems, and they are underutilized for chronic HIV care and electronic data management in low- and middle-income countries. We describe the design, development, and use of a mobile health application in community-based HIV prevention, care, and treatment programs implemented in Ethiopia since 2017. Initially, these programs utilized aggregate data collection methods, which were fragmented, non-uniform, and largely paper-based, posing challenges to data quality and data use. Local software developers applied user-centered methods to build a Unified Data System (UDS), an innovative digital health solution aimed at standardizing case management and routine data collection procedures. Developers involved end-users, including frontline community health workers (CHWs) and program managers, in the design and development process. They also conducted field visits to understand and assess user activities and needs. Developers and program staff conducted field testing, and they made design changes iteratively to incorporate user feedback. Local implementing partners deployed 950 CHWs to use the UDS for HIV service delivery and client-level data collection offline. Data were stored centrally in the CommCare HQ web service backed by a local analytics server. The UDS was linked to Power BI for advanced data analytics and visualization. Data validation tests were performed prior to indicator computation and reporting to the next level. 1,766 mobile accounts were opened for CHWs, of which 1,628 (92.2%; 95% confidence interval [CI]=90.8%, 93.4%) have submitted data. An additional 368 of 392 web accounts (93.9%; 95% CI=91.0%, 96.0%) have been used by program managers and monitoring and evaluation experts to access the dashboard. The UDS has eliminated the tedious, manual, paper-based work associated with data (dis)aggregation and reporting by frontline CHWs and local implementing partners. Hence, the number of days to access client-level data by the local partners was reduced from 1 month to just 1 day, and the time to report on activity performance was shortened from 5 days to a half-day. Data completeness was nearly 100% and data validation tests demonstrated no invalid data elements or errors. Generally, there was a high degree of digital health technology utilization by frontline CHWs and local implementing partners for standardized HIV care, high-quality data collection, and making data-informed decisions. Lessons learned from the UDS implementation could be adapted not only to support community HIV programming but also to strengthen national electronic community health information systems.