domain-healthcare:ehr-integration

EHR Integration

When to use

• Integrating with Epic, Cerner, or Allscripts via FHIR R4 or proprietary APIs
• Processing HL7v2 ADT, ORM, ORU, or other message types from an EHR feed
• Selecting or configuring an interface engine (Mirth, Rhapsody)
• Implementing CCD or C-CDA document generation, parsing, or reconciliation
• Setting up or debugging SMART on FHIR application registration
• Testing integrations against vendor sandboxes

Core principles

1. FHIR R4 first, HL7v2 as legacy — all new integrations use FHIR; HL7v2 is maintenance mode for existing feeds
2. Interface engine as single point of control — route all EHR traffic through one engine; direct point-to-point integrations become unmanageable fast
3. ACK or it didn't happen — every HL7v2 message needs acknowledgment; assume delivery failure without it
4. Dead letter queue from day one — messages that fail processing will happen; design the failure path before the happy path
5. Test against real sandboxes, not mocks — vendor-specific quirks (Epic's ADT extensions, Cerner's FHIR pagination) only show up against actual endpoints

Reference Files

• references/epic-cerner-allscripts.md — Epic FHIR R4, App Orchard, MyChart, Interconnect; Cerner Millennium, Code Console, CDS Hooks; Allscripts Open API and Unity API
• references/hl7v2-messaging.md — ADT/ORM/ORU/MDM/SIU/DFT/RDE message types, segment structure, ACK pattern, field mapping best practices, dead letter queue design
• references/interface-engines-document-exchange.md — Mirth Connect channel architecture, Rhapsody routing, interface engine best practices, CCD/C-CDA exchange, Direct messaging, integration testing tools