Change non-standard concept_id to standard concept_id

[MelaniePhilofsky]

Instructions

0. Create the initial issue.

Per multiple community discussions and Themis conventions, concept_id = 44814659, Multiple race needs to be updated from non-standard to standard concept.

1. Read the documentation.

Please familiarize yourself with the documentation here prior to initiating the submission. Done

2. Choose the category of submission.

Choose the category of your submission and download the corresponding template:

• Template 7 Done

3. Fill the template.

Please dowload the corresponding template Done

Make sure you filled the checklist! There isn't a checklist tab on the template

4. Upload the template.

Please go to this folder, create a subfolder with your contribution (name it as GitHubIssueNumber_ContributionName_YourName) and upload your contribution there. Done

5. Reference the template in this issue.

Please add the link to your submission to this issue. LINK

6. Get the feedback.

Please see Themis' ratified convention here.

[aostropolets]

It seems that you want something with domain Race rather than Observation. I don't think we want to promote PCORNet concepts since it's an abandoned vocabulary but rather want to add a new concept that says Multiple race to Race domain (also thinking long-term here: if we expect an influx of other races there should be a common place to add them).

[MelaniePhilofsky]

Yes, we need to have a standard concept_id with domain_id = 'Race' and a concept_name or meaning = 'Multiple race'.

RE: vocabulary for the multiple race concept and other race concepts to be added via community contribution, some contributions might come from national / local code systems and some might come from non-coded, source value sets which are generally categorical with a data type of string text. I suggested the multiple race concept_id from PCORNet because I know some are already sites (and other CDM's, networks) are using it. The fact that the vocabulary isn't supported any more isn't a concern from the Themis point of view since this concept_id doesn't contain any standard to non-standard or hierarchical mappings. And won't contain any hierarchical mappings in the future.

Thanks for your help, @aostropolets!

[Alexdavv]

...and other race concepts to be added via community contribution, some contributions might come from national / local code systems and some might come from non-coded, source value sets which are generally categorical with a data type of string text. I suggested the multiple race concept_id from PCORNet because I know some are already sites (and other CDM's, networks) are using it.

We are happy to accept all these contributions, but concepts from multiple vocabularies (especially local) shouldn't themselves be promoted to Standard concepts when it comes to the "service" Domains. The reason for that is the fact that other users won't know which of multiple OMOP vocabularies contribute to the Race Domain, even though some people know "PCORNet" contains some good races. And since nobody downloads the entire set of 156, they'll end up creating their own Standard duplicates. You can argue that these vocabularies can be added to the Atnena default pre-selection, but we usually do that only for small and Domain-specific vocabularies, like "UB04 Typ bill" or "ABMS". If one contains other stuff, we don't want them to overwhelm every user's OMOP instance. The proper solution here is the creation of Standard counterparts in the "Race" vocabulary together with mappings from the source concepts.

[cgreich]

Correct, except this is probably an impossible fate. Races are not defined by any objective criteria (there are mostly socio-economical constructs with some loosely defined biological features), which only work inside a country. Therefore, it is not possible to dedup them. So we decided to open up the space. Everything goes.

[Alexdavv]

Correct, except this is probably an impossible fate. Races are not defined by any objective criteria (there are mostly socio-economical constructs with some loosely defined biological features), which only work inside a country. Therefore, it is not possible to dedup them. So we decided to open up the space. Everything goes.

Absolutely right. But you don't want to get one loosely defined “Eastern European” from each of OMOP’s datasets. Now the mapping is arbitrary to the fact that “Race” vocabulary has some and doesn't have some others. In the future we either map to the extent people want/can OR don't map at all and the “Race” vocabulary should be gone because everybody will have their own set of races. But we can’t do a Frankenstein of both solutions.

[MelaniePhilofsky]

@aostropolets

We would like to update our ETL in order to utilize multiple race data for a person. Will concept_id = 44814659 be updated to standard concept with a domain_id = Race? Or is the vocabulary team going to create a concept with meaning of multiple race?

[MelaniePhilofsky]

Help! I can't find the standard concept_id for multiple race in the Race domain.

[cgreich]

Wait. Didn't we want to do a vocabularathon with all races and ethnicities?

[MelaniePhilofsky]

Multiple race does not need a vocabularathon to create a new concept or make a non-standard concept a standard concept. The ask for this has been well recorded in the forums for many years. The documentation on when and how to use this concept_id is documented in the Themis convention library. Albeit was a bit pre-mature, but the ETL folks need lead time to plan, design, build and test before deploying a big change, which we thought would go live with the Feb 2025 vocabulary release.

The impetus for modifying the conventions for race_concept_id was to house the idea of a person identifying as > 1 race or 'multiple race'. When someone brings an issue to the Themis WG, we gather all the use cases to ensure we have a comprehensive convention. Adding additional standard concept_ids for all the race values, moving >1 race to the Observation table, enabling provenance and recording the date the data were recorded are all additional use cases the community brought forth. Not only is multiple race one of these, but it is the most requested and loudest use case.

Having a standard concept_id for multiple race and allowing for > 1 race per person in the OMOP CDM helps eliminate the following potential biases: granularity bias, representational bias, omitted variable bias, and forced choice bias.

Collaborators have been knocking on my door since the OHDSI community ratified these in the well attended meeting held December 2023. I followed the community contribution process and submitted this request in November, the beginning of this thread.

When will the vocabulary hot fix be released? :)