I was in a meeting a couple of years ago at a global life-sciences company where I spent 20 minutes explaining the limitations of FHIR.
My role at the company was advising the technical architects as they put together their data flows and transformations, and incorporated FHIR into their systems.
The reason I was on the call was because the technical team was having difficulty getting the business to understand what FHIR could not deliver. Expectations were not being met and the business did not understand why.
The feeling was that an outside perspective – someone with no skin in the game – might have more success.
There were 40 people on the call representing different parts of the business. Product leads, tech leads, sales & marketing, and the project manager dialing in from Switzerland (I was told this was a big deal).
I started with a 5 minute intro to FHIR, explaining all the good stuff: interoperability, data sharing, a common data model and common terminology.
Then I gave them my “FHIR is not a database” line, and explained how common this misunderstanding was.
There were a few heads nodding at this point.
I then moved on to the detail and started to explain the true limitations of FHIR and what those limitations meant for their project.
I explained how the FHIR API could not be used to run complicated queries that databases excelled at. How it could not meet their requirements for analytical dashboards. How FHIR validation did not remove the need for data governance and business validation. How fine grained access control was not typically part of a FHIR server’s capabilities.
I explained how each of these was a separate project that needed to be scoped and delivered independently.
And I explained the problems with the managed FHIR server they were using. Missing features, slow queries, truncated result sets, broken search parameters.
By this stage no one was smiling and no one was nodding their head.
After I stopped speaking, the silence was broken by the project manager.
This is disturbing.
Direct and to the point.
The floodgates opened quickly after that.
FHIR project failures are almost never FHIR problems. They often come down to misunderstanding the role FHIR plays in a project. They’re business failures.
---