12 hard truths you won’t learn in the FHIR documentation.
- You will almost certainly have to build a proxy in front of your FHIR server.
- There are big differences between the “brand name” server providers: Azure, AWS & Google.
- Many FHIR servers truncate search results. You will have to deal with this.
- Smart-on-FHIR is never as easy to implement as the sample projects suggest.
- Your FHIR project team is almost certainly missing key people.
- “Off the shelf” FHIR servers cannot be used to perform analytical queries.
- R5 may be the current version, but that doesn’t mean you should use it.
- Managed FHIR servers introduce new bugs all the time. You need early access!
- You should not store AuditEvent resources in the same FHIR server you’re auditing.
- Not all FHIR servers support Subscriptions — a key FHIR feature for consumers.
- You may have to build RBAC and ABAC into your FHIR server yourself.
- FHIR expertise and terminology expertise rarely go hand-in-hand.
Every production solution built on top of FHIR has had to address these issues. Collectively, they can turn your 1 year project into a 3 year project.
Numbers 1, 2, 6, 10, and 11 are major development projects in themselves. They don’t come for free with commercial FHIR servers.
Numbers 5 and 12 are the easiest to miss and the most costly to fix.
You won’t read about any of this in the FHIR documentation.
---