DHIS2 in production: lessons from Kenya

Adaobi: Wanjiru, set the scene for listeners who do not know DHIS2. Dr Kamau: DHIS2 is open-source health management software developed by the University of Oslo and used by health ministries in more than seventy countries. It is not a complete solution. It is a platform. The work of running it well is the work of customization, deployment, and operations — which is invisible from the outside. Kwame: How do you scale to nine thousand facilities? Dr Kamau: Slowly and then suddenly. We scaled the data model first. We rationalized our indicator dictionary down from about four thousand to twelve hundred indicators. That alone made the system faster and the analyses more useful. Then we scaled the infrastructure: we moved from a single Postgres to a primary plus three read replicas, and we sharded the analytics workload off to a separate data warehouse. The trickiest part was scaling the human network. We trained two thousand health information officers over three years. Adaobi: Mobile-first? Dr Kamau: Two years ago we shipped the Android capture app to every facility with a county-issued tablet. Compliance went from sixty-three percent of monthly reports submitted on time to ninety-one percent in six months. The lesson is that the entry barrier matters more than any back-end optimization. Kwame: Reporting cadence and political pressure? Dr Kamau: When the minister wants a number by Friday, the system has to produce that number by Thursday. We learned to anticipate the questions and to prepare the data products before the political demand arrived. Now there is a weekly dashboard that anticipates the most common questions and the system survives the surges. Adaobi: How to be a good DHIS2 customer? Dr Kamau: Contribute back. Every customization we make that is general goes upstream. Pay your community membership fees. Hire engineers who participate in the global community calls. Be the customer the project deserves and the project will be there for you in a crisis. Kwame: Wanjiru, thank you. Dr Kamau: Thank you.