Hi, I would like to challenge (positively hopefully) what should be considered as the standard concept for Race. There are 50 standard concepts for Race but I would argue when we define “standard concept” and encourage the adoption of OMOP at the international level, we should aim for international standard terminology at first instance, rather than the one that is only used in a particular country. In the UK, we have SNOMED CT as the NHS data standards, which is also an international terminology. It is debatable for Race, we are not aligning that, maybe we should. this has created an extra layer of “translation” for our technical team because the “standard” we adopt at our source EHR system is SNOMED CT. Can this be reviewed please?
1 Like
Hi @LeileifromUK. We develop and support Race vocabulary, there are not only snomed concepts. If you need some specific race concepts, we can add it to the Race vocabulary.
Hi @LeileifromUK,
Thanks for pointing to this; you aren’t alone in thinking this. I have found a number of messages on the forums and in github issues with people who have called for a review of this domain. I have been trying to collect these into a recommendation for updating the vocabulary and standards, which you can find as a work-in-progress here. It includes updating the standard concepts to be inclusive of the SNOMED “Race” and “Ethnic group” categories (section 2.3 of the design). Please feel free to review, comment, and/or contribute. I’d also be happy to meet or receive any additional details on how this information is collected in the UK. I have most familiarity with the US data, and want this update to reflect an international perspective.
2 Likes
Thanks Jake. The more we got into the OMOP work, the more we find this disconnection of the “standard” vocabulary considered by OMOP is not " data standard" in other countries. Vice versa. What constitutes a “data standard” in our countries, if I speak from UK perspective, is not “standard vocabulary” (for some CDM fields). Or even it is considered as “not standard” vocabulary, the acceptable values are either questionable or “missing” e.g. admitted_from_concept_id. However the issues I raised here is rather a principle issue, which is how does OMOP truly support organisations and align international data standards. I think this needs a dedicated discussion, moving away from the specifics even the specifics are affecting us as we speak.
There are no standard vocabularies. We only have standard concepts. Some vocabularies are used more to pick them. But it really shouldn’t matter which vocabulary a concept came from. For example, if “Myocardial Infarction” is used from ICD9CM, ICD10CM, SNOMED or Read is irrelevant. Only ONE of them can be standard, the others are source and get mapped to the standard.
To do that, we continuously map vocabularies to the standard. Do you have something you are missing?
Do you mind creating the list, @Jake? Take all the input ethnicities and races from the various sources including SNOMED, stir them, dedup them, and then we can take a look?
The other question is: right now we have a race_concept_id and an ethnicity_concept_id. Your mixed race/ethnicity vocabulary - where does it go? And what do we do with the other one?
Not sure this is the right thread for this discussion, seeing as Leilei appears to want to talk about something broader than the race and ethnicity vocabularies, and the answer to your question gets into the weeds of this design, but since you asked… 
Do you mind creating the list, @Jake? Take all the input ethnicities and races from the various sources including SNOMED, stir them, dedup them, and then we can take a look?
On it! @piper was one step ahead of us and had already started this work. We are in communication and hope to have something to look at soon. Before we spend too much time on this though, it would be good to have our general strategy validated. I suppose you asking for the output without /too/ much complaint serves as partial validation.
The other question is: right now we have a race_concept_id and an ethnicity_concept_id. Your mixed race/ethnicity vocabulary - where does it go? And what do we do with the other one?
I’m proposing the person table stays just as is, retaining the semantic distinction between a concept recorded as an ethnic identity and a racial identity. The target “valid answers” for the “questions” ‘race’ and ‘ethnicity’ will be one pool, the “race/ethnicity” domain, a union of our current “race” and “ethnicity” domains (plus additional concepts as in the design). Maybe I should put the conventions section up top in this design, because I hoped this section would lay that out clearly.
Any news?
The problem is that we want to get to the point where we have a one-to-one relationship between field and domain. In other words, any concept of a domain is a valid standard concept for a field. If you combine ethnicities and races into one domain you cannot have two different parallel fields. Makes sense?