What FHIR Validation Does NOT Check
A FHIR bundle passes validation. That means the structure is correct. It does not mean the meaning has arrived.
A FHIR bundle passes validation. That means the structure is correct. It does not mean the meaning has arrived.
What validation checks
A validator answers one precise question: does this resource comply with the profile’s rules? It checks structure and data types, cardinalities, bound terminologies, slices, constraints and invariants. If everything fits: green. That is valuable and necessary — and it is exactly one level.
What it does not check
Whether the information survived the transfer. The validator sees the resource that arrives — not the one that should have arrived. Loss is invisible to it as long as what remains is well-formed. In our work on SILD, these silent losses reduce to four canonical patterns — and no FHIR validator reports any of them:
- Type Narrowing — a precise concept is mapped onto a coarser one (the specific SNOMED code ends up as an unspecific ICD entry). The target element is valid; the precision is gone.
- Temporal Collapse — a time interval is reduced to a single point in time (a course from 08:12 to 11:47 becomes an all-day entry). The date is formally correct; the temporal reference is lost.
- Attribute Dropping — a qualifying attribute is missing: unit, status or negation. “No known allergy” and an empty bundle look the same to the validator.
- Reference Severing — a link between resources breaks, and the context that carries the statement is lost. The finding remains, its order reference does not.
Each of these losses happens inside a technically successful, formally valid transfer.
Why “validated” misleads
Validation is a guardian of structure, not of meaning. It confirms that the form is correct — not that the content survived. Anyone measuring data quality by the green check mistakes the one for the other.
What it takes instead
A second layer of inspection with a different question: did the meaning survive? That question is not a validation task — it is a governance task. This is exactly where SILD comes in: on the basis of the mathematical model FM-4 it detects the four loss patterns formally and verifiably, as an integrated part of CAIRN. “Green” thus becomes a checkable finding.
We measure the meaning loss in productive FHIR exports — quantified, reproducible, against the FM-4 model. If you want to know what your validation overlooks: request a SILD audit.
The formal foundation is documented at aion-clinical.eu; the detector is open source at github.com/fmatten/SILD.