The first multi-sectoral, user-driven data pipeline to adaptively manage HIV prevention programs in Malawi


BACKGROUND: In Malawi, HIV prevention data are fragmented, managed by different custodians, and are difficult to analyze and interpret nationally and by frontline workers. District, city, and health facility managers have limited views and ownership of HIV data, both within the formal health system and in the community. Despite the importance of social determinants on HIV risk, available data are typically summary statistics without context. This results in underused data and missed opportunities for more targeted and effective prevention activities. We integrated HIV-related data sources into a 'data pipeline' with real-time, web-based, automated analyses and visualizations for decision makers in Blantyre, Malawi.
DESCRIPTION: We developed the Prevention Adaptive Learning and Management System (PALMS) for Blantyre. PALMS is a data pipeline which leverages the country's existing programmatic, surveillance, research, and donor data sources. As new data are available, PALMS ingests, transforms, analyzes, and visualizes the results. All PALMS development is driven by Blantyre leadership with processes to incorporate feedback, shifting priorities, and new data systems. To improve ease of use, PALMS data are automatically mined to highlight concerning statistics and notify users. Finally, PALMS is developed to 'plug and play' with the national enterprise architecture and is being absorbed by MOH technicians to improve sustainability and facilitate uptake to other interested districts.
LESSONS LEARNED: PALMS was rapidly adopted by decision-makers and used to monitor site performance, anomaly detection, and identify hotspots. Systems notifications are the most popular feature and have helped focus attention on facilities and programs that need the most support. The biggest barriers to PALMS development were not technological, but rather involved navigating the complex data ownership landscape and eliciting and integrating user feedback.
CONCLUSIONS: PALMS will expand to include additional data sources, multi-platform notifications, and interoperate with a new HIV incident management tracker. We believe PALMS will help usher in a new data-driven, event-based surveillance approach to HIV prevention in Blantyre. The PALMS approach can be generalized to other settings and can hone attention and support resource allocation efforts. Such user centered designed, sustainable, and scalable data systems that satisfy user expectations are feasible and desperately needed for HIV programmes throughout Africa.

Download the e-Poster (PDF)