Regional healthcare group

The challenge

The group had grown by acquisition and inherited a different patient administration system at almost every site. Patients who moved between branches were registered again, appointments could not be transferred, and clinicians could not see a record created two kilometres away. Group reporting was assembled by hand each month from exports that never quite matched.

There was an internal engineering team, and it was capable. It was also fully occupied keeping the existing systems alive, which meant the consolidation programme had been on the roadmap for two years without starting. The group did not want a supplier to own its clinical platform; it wanted enough senior capacity to finally do the work itself.

Any change had to happen around live clinical operations. Reception opens early, clinics run at capacity, and a booking system that is unavailable at the start of a session costs the site its whole morning.

What we did

We placed senior engineers directly into the group's own squads, working to their technical lead and their change process. Four to begin with, covering backend, integration, mobile and quality, scaling up and down over the engagement as the workstreams moved. The group's architect stayed accountable for the design; our engineers brought the delivery capacity and the experience of doing this in a regulated setting.

The first workstream was a master patient index with deduplication and a human review queue for probable matches, because nothing else could be consolidated until patient identity was settled. Group-wide scheduling followed, then encounters, then the integration layer to laboratory, imaging and pharmacy systems, then submission to the relevant health information exchange as a queued and retryable integration with a visible failure log.

Production patient data never entered development or test environments. We built synthetic datasets that reproduced the awkward cases, and put automated regression and load testing into the release pipeline early, since the release window for a clinical system is narrow and unforgiving.

The outcome

Sites moved onto the shared platform one at a time, each with dual running and a rollback path. Patient identity is now resolved across the group, appointments move between branches, and clinicians see a record that follows the patient rather than the building. Group reporting comes from one dataset instead of a monthly assembly exercise.

Active users of the platform rose 45% as clinical and administrative staff moved off the systems they had been working around, and as functions that previously happened on paper came into the platform. Availability has held at 99.95% across clinic hours since the second site went live, which is the number that matters to a reception desk at eight in the morning.

Two of our engineers remain embedded with the group's team on the remaining site migrations and the analytics workstream. The platform, the code and the infrastructure definitions belong to the group.

Results

Stack we use