Stream: implementers
Topic: Seperate Profiles vs. Hierarchical ValueSet
Noemi Deppenwiese (May 28 2019 at 10:14):
Hello, we're currently working on FHIR profiles for a biobanking data model. We plan to use the Specimen resource for our samples and want to distinguish between different "superclasses" of samples (e.g. liquid, tissue), each of which has several sub-types (e.g. blood, tears). The only difference between the classes profile-wise is that is that for liquids, there is no bodySite attribute. Is it "better" FHIR to define different profiles for each superclass OR use a hierarchical ValueSet and one common profile with a FHIRpath constraint (using the "subsumes" operator) on bodySite ? Thanks!
Lloyd McKenzie (May 28 2019 at 13:07):
The hierarchy would be in the code system rather than the value set and that's how it would typically be done. Profiles never convey "data" - the fact it was a liquid must be conveyed by the codes already in the instance. That said, typically you need to differentiate arterial vs. venous blood, might care which abscess fluid was drawn from, etc. - so I'm not sure if it's wise to say that fluid will never have a body site (though accept that it might not be mandatory).
Grahame Grieve (May 28 2019 at 13:08):
I'd use one common profile with FHIRPath
Noemi Deppenwiese (May 28 2019 at 13:52):
Thank you! We will use a single profile & CodeSystem hierarchy then.
Last updated: Apr 12 2022 at 19:14 UTC
