Stream: Orders and Observation WG
Topic: Identifier as a valid type of Observation.component.value[x]
Diego Bosca (Oct 20 2019 at 16:20):
I'm modelling a bunch of genetic related information and stumbled with some identifiers that won't really fit into any of the existing Reference attributes that the Observation profile has. After having a look at http://build.fhir.org/ig/HL7/genomics-reporting/ they seem to have the same issue (see dbSNP-id). They have decided to make it a CodeableConcept. If Identifier was a valid type in Observation.component.value[x] these problems could be solved. Are there any current plans to support Identifier type in Observation.component.value[x]?
Lloyd McKenzie (Oct 20 2019 at 17:04):
Identifier is used when you're referring to a specific instance, not a kind. I.e. if you have 3 identical SNPs but they were taken from different cells in the same individual, they'd all have the same "code", but they'd have different "identifiers". The latter isn't terribly useful. So CodeableConcept is the correct choice of data type for what's being represented.
Diego Bosca (Oct 20 2019 at 17:52):
Nice to know :)
I have still plenty of examples, ranging from identifiers for the phenopacket interpretation to family/paternal/maternal identifiers of a given person. I have also identifiers to things like hgvs allele, htsFile, spdi allele, and vcf allele, but maybe some of these can be seen as CodeableConcept (although not really sure it really fits in some of these cases)
Lloyd McKenzie (Oct 20 2019 at 19:18):
A specific file might have an identifier. Alleles would generally have codes. Though you can always have an identifier on the Observation when you make a particular genetic assertion about a specimen.
Diego Bosca (Oct 20 2019 at 19:41):
yeah, specimen, patient, etc. ones are the ones I got clear from the other thread. Doesn't help that everything I just mentioned is (or has) an identifier in the original data. Seems like it's Extension time
Lloyd McKenzie (Oct 20 2019 at 20:12):
Just because something is called an identifier, that doesn't necessarily make it an identifier. An identifier means it's a globally unique identifier such that if something with a given identifier exists at one place in the world, there's no other thing anywhere else in the world at the same time with the same identifier. If you can have multiple people with the same allele and the alleles can have the same identifier, then that's not an identifier, it's a code.
Diego Bosca (Oct 20 2019 at 23:15):
Yes, but the "entry" that groups the allele/alleles has an identifier by itself, even if the allele is coded
Lloyd McKenzie (Oct 21 2019 at 01:57):
That would generally be an Observation (which has an identifier)
Last updated: Apr 12 2022 at 19:14 UTC
