Referential Integrity
summary
FHIR does not require references to resolve and most real systems choose a middle ground: enforce referential integrity for clinically meaningful links (subject, medication) but allow broken or identifier-based links for secondary feeds, audit/provenance, or out‑of‑order histories. Practical options are validating at write time, using transaction bundles to atomically create dependencies, or rejecting deletes when other resources reference a target; implementations use things like a HAPI interceptor to customize rules. Important gotcha: a structure validator often cannot check existence because the validator can't resolve references against live APIs.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 0:33Topic Overview
- 0:57Referential Integrity
- 4:01Enterprise Information Systems
- 9:45Reference Types
- 16:28Versioning Reference Integrity
- 19:08Maintaining Referential Integrity
- 20:40Transactions
- 22:34Observations
- 23:24Maintaining Reference Integrity
- 26:58Circular References
- 28:30Happy
- 30:26I was totally slow
- 32:54Summary
- 33:56QA
- 39:02Conclusion
