Stream: terminology
Topic: Race and Ethnicity
Jake Gillberg (Jan 05 2022 at 17:05):
Hello, does anyone know of another profile / extension / IG that includes race and/or ethnicity value sets outside of US Core / CDC race and ethnicity code system?
David Pyke (Jan 05 2022 at 17:14):
Here's a Canadian example using SNOMED ethnicity codes: https://build.fhir.org/ig/HL7-Canada/ca-baseline/StructureDefinition-profile-patient.html
Is there something specific you're looking for?
Elliot Silver (Jan 05 2022 at 17:30):
@David Pyke , current CA Baseline content is at https://build.fhir.org/ig/HL7-Canada/ca-baseline/index.html or https://simplifier.net/canadianfhirbaselineprofilesca-core.
David Pyke (Jan 05 2022 at 17:31):
Simplifier? Lazy!
Thanks for the pointer
John Moehrke (Jan 05 2022 at 17:32):
in theory, you should be able to search in simplifier for any use of them.
David Pyke (Jan 05 2022 at 17:32):
I searched in google...
Jake Gillberg (Jan 05 2022 at 17:59):
Thanks. I'm trying to get a feel for what terminologies are out there for race and ethnicity coding, contemplating proposing changes within OHDSI to the race and ethnicity vocabularies to 1- provide better support for research on data collected in health systems internationally (with a focus on data useful for health equity research) and 2- align with FHIR for the HL7 FHIR + OHDSI OMOP collaboration.
David Pyke (Jan 05 2022 at 18:25):
Pretty much, every country has their own Race/Ethnicity terminologies but most don't differentiate between the two. Working with the SNOMED set which is probably the most comprehensive is likely your best bet.
Lloyd McKenzie (Jan 05 2022 at 19:33):
Tracking race and ethnicity tends to be driven by what groups have historically been subject to discrimination in that area and is tracked to help counterbalance those issues. What characteristics are relevant and the granularity that's relevant varies considerably from place to place.
Mareike Przysucha (Jan 06 2022 at 08:07):
Please keep also in mind when giving feedback to OHDSI that in some countries, e.g. Germany, it is forbidden to capture race and ethnicity, thus allowing a code to capture this. OHDSI already allows that, I just want to be sure. Thanks in advance.
Ward Weistra (Jan 06 2022 at 09:48):
Searching in the Registry indeed mainly gives ValueSets/CodeSystems from FHIR Core and the US:
- https://registry.fhir.org/results?query=%22race%22&latestFilter=true
- https://registry.fhir.org/results?query=%22ethnicity%22&latestFilter=true
Richard Townley-O'Neill (Jan 07 2022 at 00:16):
HL7 AU has an extension http://build.fhir.org/ig/hl7au/au-fhir-base/StructureDefinition-ethnicity.html but no preferred value set. Nothing is clearly appropriate. One issue is that values useful for statistics of social dis/advantage are different to values useful for selecting the best care.
Peter Jordan (Jan 07 2022 at 00:57):
Likewise, NZ has an ethnicity extension, bound to our unique set of ethnicity concepts
Elliot Silver (Jan 07 2022 at 17:38):
I know it's far too late, but it seems like there might be value in standard race/ethnicity extensions in FHIR, but only providing an example vocabulary.
David Pyke (Jan 07 2022 at 17:44):
Then the question would become if so many countries are using that data, then why not add it directly to Patient?
Lloyd McKenzie (Jan 07 2022 at 18:12):
The problem is that the 'meaning' of those things varies. Even the notion of those concepts as distinct things isn't globally understood. In Canada, we're interested in first nations status. That's at least as much a political/legal concept as it is a concept of ancestry. It's serving similar purposes to race/ethnicity in the U.S., but we wouldn't want to map it to the same elements, I don't think.
David Pyke (Jan 07 2022 at 18:14):
Perhaps not, but it has the same general categorization as Latino/not that exists in the US. So for many countries, there are primary groups that would be in one element and then an overall set in another.
Jens Villadsen (Jan 07 2022 at 18:53):
(FYI - If I'm not mistaken it is illegal to register ethnicity in a range of european countries )
David Pyke (Jan 07 2022 at 19:04):
Yep, so it/they would need to be 0..*
David Simons (Jan 10 2022 at 12:29):
My gut feeling is these may end up on specific national realms, like us, nl, uk - not so much the inclusion of race and ethnicity attributes - but the binding to specific valuesets and codes.
Creating an universal codesystem for ethnicity/race - plus mappings to existing ones - is going to be an enormous effort... not sure who can drive/prioritize this, and if feasible. Hence my recommendation to select and endorse an appropriate set per realm.
E.g. in the NL the etnicity is uses in maternal care, afaik:
https://simplifier.net/packages/nictiz.fhir.nl.stu3.geboortezorg/1.0.0/files/316963
https://simplifier.net/packages/nictiz.fhir.nl.stu3.geboortezorg/1.0.0/files/317201
And this is the NL-core patient, compare to US-core patient, it does not have race/ethnicity.
https://simplifier.net/nictizstu3-zib2017/nl-core-patient
Also, colleagues tell me that German university hospitals use this https://simplifier.net/guide/GermanCoronaConsensusDataSet-ImplementationGuide/Ethnicgroup-duplicate-2 extension+valueset, which is largely based on WHO ISARIC eCRF https://isaric.org/wp-content/uploads/2021/02/ISARIC-WHO-COVID-19-CORE-CRF_EN.pdf.
New Zealand base IG http://igs.clinfhir.com/nzbasev1/ValueSet-ethnicity.html on the other hand has a totally different idea based on their existing national standards.
David Pyke (Jan 10 2022 at 14:06):
You wouldn't create a universal codesystem (beyond SNOMED), you'd have each national profile set have their own codesystems and/or valuesets that address their needs.
David Simons (Jan 10 2022 at 14:30):
David Pyke said:
You wouldn't create a universal codesystem (beyond SNOMED), you'd have each national profile set have their own codesystems and/or valuesets that address their needs.
Agreed - that was my point too :) But it is unfortunate. A universal valueset, using universal codes, would address the question of OP (and others). Bringing data together for data science across multiple realms in a semantically-interoperable way is a big puzzle - separate from data-privacy challenges, of course.
Lloyd McKenzie (Jan 10 2022 at 16:00):
The challenge is that even if we were to draw from a single value set, the data wouldn't actually be comparable. That's because the intention and use of the data - and the rules around how the element is populated would vary widely. None of this information is intended for use for clinical purposes. (Though it might occasionally be used for that purpose when there's no other more appropriate source.)
Peter Jordan (Jan 11 2022 at 01:13):
David Simons said:
New Zealand base IG http://igs.clinfhir.com/nzbasev1/ValueSet-ethnicity.html on the other hand has a totally different idea based on their existing national standards.
New Zealanders can be assigned up to 6 ethnicities and the related Code System contains unique concepts such as 'NZ European' that are unlikely to be fully understood in other domains. From July 1, 2021 we will also have a separate Maori Health Authority - an illustration of the importance attached to ethnic monitoring in this country.
Davera Gabriel (Jan 18 2022 at 18:15):
Hello @Jake Gillberg ... are you gathering information for the OHDSI / FHIR solution for Race & Ethnicity?
Jake Gillberg (Jan 18 2022 at 19:04):
Thanks for the information and discussion everyone! I'm interested in seeing where this discussion goes within this community. @Davera Gabriel, the OHDSI / FHIR perspective for race and ethnicity is certainly on my mind, but the primary purpose of my investigation is to develop a proposal for changes within the OHDSI OMOP common data model to support upcoming health equity research work taking place in the OHDSI Health Equity Workgroup, some of which seems to necessitate an update to the race/ethnicity vocabulary in OHDSI, or at least develop recommendations for data mapping for data that does not appear to easily conform to the OMOP data model.
Davera Gabriel (Jan 21 2022 at 23:20):
Jake Gillberg said:
Thanks for the information and discussion everyone! I'm interested in seeing where this discussion goes within this community. Davera Gabriel, the OHDSI / FHIR perspective for race and ethnicity is certainly on my mind, but the primary purpose of my investigation is to develop a proposal for changes within the OHDSI OMOP common data model to support upcoming health equity research work taking place in the OHDSI Health Equity Workgroup, some of which seems to necessitate an update to the race/ethnicity vocabulary in OHDSI, or at least develop recommendations for data mapping for data that does not appear to easily conform to the OMOP data model.
@Jake Gillberg agree with your assessment. I noted the OHDSI community is still mulling an universal solution. Do you think the emergence of the V3 of the USCDI which specifies the Gravity SDOH Value Sets may get some traction with OHDSI and the OMB + CDC Race & Ethnicity value sets can go along along for the ride with all of that? If not: I am happy to support you / a proposal for a value set that works for you in that CDM Vocab working group in addition to adoption into the harmonized OMOP + FHIR proposal(s). Happy to do so.
Jake Gillberg (Jan 25 2022 at 22:31):
@Davera Gabriel Based on what I know, but I do need to know more, I think the Gravity SDOH Value Sets would be a reasonable and useful addition as "classification concepts" to the OHDSI Vocabulary (they group underlying SNOMED / LOINC / ICD codes, right?). Probably worth separating this from a race / ethnicity proposal update, which has some more internationalization concerns and potentially some other data model changes as well. I'd be happy to connect about how any proposals in these two areas would fit into the OMOP + FHIR proposals, and how to move them forward in both communities.
Grahame Grieve (Jan 27 2022 at 05:53):
it doesn't seem likely that there's going to be a universal solution to me. The concept of race/ethnicity has a lot more to do with the culture of the jurisdiction than facts about actual individuals
Last updated: Apr 12 2022 at 19:14 UTC
