HL7 v2 vs FHIR

HL7 v2 and FHIR are both HL7 standards, built a generation apart for different jobs. HL7 v2 (1989 onward) is event-driven messaging: pipe-delimited segments pushed between systems when something happens (an admit, an order, a result). FHIR R4 (2019) is a resource-oriented HTTP API: JSON or XML resources (Patient, Observation) fetched and searched over REST. Inside hospitals, v2 still carries the bulk of live interface traffic; FHIR is the standard for app-facing APIs and for federal API mandates such as CMS-0057-F. Most integration work is not choosing between them but running both.

Example

The same patient, both standards, one engine, one seed.

HL7 v2 (one PID segment of an ADT^A01 message):
PID|1||EA1VOECPYL^9^ISO^^MS|EA1VOECPYL^9^ISO^^MS|Byrd^Rodolfo^^^^^^^|Jones^Jayne^^^^^^^|19360121||Byrd^Rodolfo^^^^^^^||258 Walnut Street^^Boston^AZ^10001^USA^H^^^|...

FHIR R4 (a Patient resource, trimmed to its first elements):
{
  "resourceType": "Patient",
  "id": "patient-d2b41bdcba282d7083d379d378a143c4",
  "identifier": [{
    "use": "usual",
    "type": { "coding": [{ "system": "http://terminology.hl7.org/CodeSystem/v2-0203",
               "code": "MR", "display": "Medical Record Number" }] },
    "system": "http://hospital.example.org/mrn",
    "value": "8X6X22R9P3"
  }],
  "name": [{ "use": "official", "family": "Robinson", "given": ["Lise"] }]
}

Both blocks are engine output at seed 42 (`pidgeon generate ADT^A01 --seed 42` and `pidgeon generate Patient --seed 42`), each trimmed as labeled. Note the FHIR identifier.type coding system: FHIR reuses HL7 v2 table 0203 for identifier types, one of many places the two standards deliberately interlock.

Validation context

Strict and compatibility validation notes land here with the phase-1 data export.