Loading…
A 14-year-old claims system replaced by an event-driven platform that pays providers in days, not months, for a national insurance scheme covering 31 million people.
“We replaced a fourteen-year-old Oracle system that nobody on staff still understood with a platform our clinicians actually want to log in to. That is the win.”
The Ministry's existing NHIS claims platform was a fourteen-year-old Oracle Forms system that took clinics an average of 96 days to get paid. We replaced it in fourteen months with an event-sourced .NET and PostgreSQL platform, migrated 220 million historical claims with full audit fidelity, and rolled it out to 4,800 accredited facilities across the country. Average claim-to-payment time fell to 11 days, and fraudulent claim detection went from a sampling-based audit to a real-time scoring service.
The National Health Insurance Scheme covered 31 million citizens but ran on an Oracle Forms system commissioned in 2011. Claims were submitted on paper at the facility, keyed by data clerks at regional offices, and routed through a batch process that touched fourteen separate stored procedures before a provider could be paid. The average claim took 96 days to clear. Clinics in rural counties had simply stopped submitting low-value claims because the cost of pursuing them exceeded the payout.
The Ministry's leadership had two non-negotiables. First: no claim data could be lost, not even a corrupted row from 2013. Second: the migration could not interrupt payments to providers for more than two weeks at any point. The previous procurement attempt had collapsed because the vendor's plan required a four-month payments freeze.
We proposed a strangler-fig migration: stand up the new event-sourced platform alongside the legacy system, route new submissions through it from day one, and migrate historical claims in the background. Every legacy stored procedure was rewritten as a domain event in the new system, and a reconciliation harness ran continuously against the Oracle source until both systems agreed on every claim for thirty consecutive days.
We worked in joint pods with the Ministry's HIS team and three of the largest regional referral hospitals. Clinicians and claims officers were embedded in the design weekly — not as stakeholders to be informed, but as people whose forms we were redesigning. Accessibility, low-bandwidth modes, and offline-first capture for rural facilities were built in from the first sprint.
The new platform is an event-sourced claims system in .NET 8 with PostgreSQL as the write store and a denormalised read model for analytics. Every state change to a claim is captured as an immutable event, which gave us the audit fidelity the Ministry's auditors required and made fraud scoring tractable. Providers now submit through a web portal, a mobile-first PWA, or a HL7/FHIR API for hospital systems that wanted system-to-system integration.
“The first time a clinic in Turkana submitted a claim from a tablet and saw it paid in eight days, we knew the project had stopped being IT modernization and started being healthcare policy.”
All 4,800 accredited facilities were onboarded over a phased fourteen-month rollout that began with the three pilot hospitals and ended with the smallest rural clinics. The legacy Oracle system was decommissioned on schedule with every one of the 220 million historical claims migrated and reconciled. Average claim-to-payment time settled at 11 days. The fraud-scoring engine flagged $14M in suspicious claims in its first six months — more than the previous sampling-based audit had identified in the prior three years combined.
Provider satisfaction, measured by an independent survey commissioned by the Ministry, rose from 22% to 81%. The Ministry is now using the platform as the reference architecture for two adjacent national systems: civil registration and the pharmaceutical supply chain.
Strangler-fig migrations work when you can prove reconciliation. The boring infrastructure — a reconciliation harness that ran every fifteen minutes against the legacy source for the full duration of the cutover — was the single piece of work that built leadership confidence to retire the Oracle system. Without it, the political risk of any one mismatched claim would have stalled the decommission indefinitely.
Tech Stack
Services Engaged
Get the long-form case study including architecture diagrams, operating cadence, and the unabridged interviews.
Tell us a little about yourself and we'll send you the PDF.
Let's talk. We'll walk you through what an engagement looks like end to end.
Talk to our team