Hi,
If EHR data contains “German” as the race. I am not sure what would be the right standard concept-id to use. Please help.
Thanks,
Priya
At a medical intake, one would most likely be asked about their ethnicity rather than their home country. Therefore I would take the answer German to indicate their ancestry and map that to OHDSI concept 8527, White.
These are your choices. It looks like concept_id = 38003614, is the closest choice.
@Christian_Reich When did nationalities/country of origin/ethnicity get inserted into the domain_id = Race?
We are debating this in the Health Equity Working Group and on the Forum.
@Jake? Any progress on this?
@Christian_Reich A little! Very rough draft (ready for help, but not for comments yet) in progress here: OMOP Race and Ethnicity. Anyone who wants to contribute to this, please hop in and help! I’m a bit dissapointed in how little time I have been able to put into this in the past couple of weeks, but I am still dedicated to putting this together, and happy to collaborate.
@priagopal, if this proposal goes through, CDC Race & Ethnicity vocabulary v1.2 will be in OMOP, so “German” will be an available standard race concept.
@MPhilofsky, these are considered valid races (as well as valid nationalities and countries of origin) for many collectors of data, HL7 FHIR US Core, and the CDC. The CDC and HL7 FHIR US Core only consider the ones that map to “Hispanic or Latino” as valid ethnicities, which is not the reality for how data is collected in the US or other parts of the world. I am proposing that we merge the “Race” and “Ethnicity” domains to create a “Race / Ethnicity” domain, from which any concept can be valid for person.race_concept_id or person.ethnicity_concept_id.
Good start. But probably on the over-engineered end of things. How about we:
- Collect all public standards and bring them in
- Create a new OMOP standard, which has the following attributes:
- Most common denominator between the other standards, i.e. we create a union (rather than a intersection) of all races and ethnicities we find.
- No differentiation between race and ethnicity, it is one continuum
- No de-duping except only the most obvious ones (Caucasian=White, etc)
- No hierarchical relationships, except in the source vocabularies as provided
- Nothing in the CONCEPT_ANCESTOR table
- We map the source to the standard. This should be easy, since we do a union + simple de-duping
- We call it a day. Every other question or issue gets kicked over to the analyst responsible for a concrete use case, like the “Asian or Pacific Islander” problem.
Would that work?
Anyone who wants to contribute to this, please hop in and help!
- Collect all public standards and bring them in
Here’s an initial list from work we did. We have this in Excel workbook with one tab for each set of concepts. Not sure how to share the spreadsheet here but happy to send the workbook @Jake and @Christian_Reich with the harmonization we’ve done.
We have been requesting concepts for all sociodemographic concepts in our EHR that don’t currently exist in any standard terminology from SNOMED. They’ve been excedingly responsive and turned requests around quickly for us.
Hi @piper, yes! Let’s connect on this. I have a work in progress recommendation for updating the vocabularies and standards here: https://github.com/Jake-Gillberg/OHDSI-RFCs/blob/race_ethnicity/rfc/0000-race_ethnicity.md.
I’d love any feedback, and would welcome a conversation to understand the work that y’all have done. Would you mind setting up a time to meet? Calendly - Jake Gillberg
Got it.
Created a meeting via calendly. Cool tool!
piper