Orrery Health

One patient record across forty-one clinical systems

Orrery Health runs diagnostics and outpatient care across twenty-three sites on forty-one clinical systems that had grown up separately. We put a single FHIR R4 gateway in front of all of them, so a clinician sees one patient history instead of six logins.

Client: Orrery Health
Sector: Healthcare
Run: 7 months · Jun 2024 – Jan 2025
Filed: 2025-03-11
Capability: Systems integration · Data engineering · Security and compliance
Stack: TypeScript · Node.js 22 · HL7 FHIR R4 · SNOMED CT · Azure · Azure API Management · PostgreSQL 16 · Redis · OpenTelemetry

Impact

1.9 s
Median full-record assembly, down from 22 seconds
268
Point-to-point HL7 v2 interfaces retired
£480k
Annual interface maintenance taken out of the IT budget

Challenge

Healthcare
7 months · Jun 2024 – Jan 2025

Assembling a full patient record meant opening six applications and reading across them by hand; the median was twenty-two seconds of waiting per system, and clinicians were printing results to avoid it. Underneath sat 268 bespoke HL7 v2 feeds built one at a time over eleven years, four of which had no owner left in the organisation. A 2023 data audit found roughly 31,000 duplicate patient identities, and nobody could say which record a given department was writing to.

Solution

We built a FHIR R4 gateway rather than another point-to-point feed, and wrote one adapter per source system so no clinical application had to change. A master patient index resolved identities with deterministic matching on NHS number first and probabilistic matching second, with every uncertain merge queued for a named data steward instead of resolved silently — that cleared 29,400 duplicates and referred 1,600 for human review. Terminology mapping normalised local codes to SNOMED CT so results from different labs became comparable. Access is role-scoped with break-glass emergency entry that pages an audit reviewer within sixty seconds, which is what carried the DSPT submission and the ISO 27001 surveillance audit in the same quarter.