OHDSI Home | Forums | Wiki | Github

Status of adding FHIR terminologies to OMOP (Athena) terminologies

It looks like FHIR has developed their own terminology systems and these are currently in use (for example by Synthea).

I’m not seeing these codes in Athena

Are these code systems being added to OMOP/Athena?

Who can I talk to for more information?

I’m seeing a FHIR Terminology subgroup meeting at the link below, but I’m not seeing the date/time for the next occurance.
OMOP + FHIR Terminologies Subgroup Meeting - Monday 6Dec21 - OMOP on FHIR - Confluence (hl7.org)

I’ve also added a stackoverflow question on this here:

@greshje.gmail:

Welcome to the family. You are in the right place. Davera Gabriel is running that Subgroup, meeting every Monday at 2pm EST.

Yes, we need to figure out how to make terminologies from FHIR available in OMOP and the other way around. That’s what this group is building.

1 Like

Thanks Christian Reich!

Hi @greshje.gmail , interestingly the one that you are referring to in your stackoverflow question (CDC Race & Ethnicity, FHIR v. 4.0.0) is not listed in the FHIR terminology page. You can find there however lots of those that are represented in Athena (not all, I have to admit). The harmonization between OMOP and FHIR terminologies is indeed a core part of the initiative lead by @DaveraG . What we are missing for this is a translation layer (and of course one or two of the terminologies that we do not yet have). The example you are giving us provides some evidence that this layer will need to be very complex and quite powerful.

Thans for the info Michael! It looks like the particular terminology (race/eth) is part of the US Core profile (is profile the correct term here). It is listed here (towards the bottom of the page): HL7.FHIR.US.CORE\Artifacts Summary - FHIR v4.0.1 and here: HL7.FHIR.US.CORE\Race & Ethnicity - CDC - FHIR v4.0.1. I guess it is not listed on the terminologies page as it is an internal terminology, but that’s just a guess.

@mik CDC Race & Ethnicity are not defined in FHIR V4.0.0 because they are not part of FHIR V4.0.0 but are required elements on the US Core Implementation Guide for FHIR V4 (HL7.FHIR.US.CORE\Index - FHIR v4.0.1). US Core extends core FHIR with these elements (and others).

While not what @greshje is seeking (he wants FHIR terminology in Athena), appended (can’t attach json files…) are the two trivial FHIR ConceptMaps that we created for a CDC project that maps OMOP race/ethnicity values into valid USCore coded terms.

{
“resourceType”: “ConceptMap”,
“id”: “Person.race.conceptid”,
“version”: “0”,
“group”: [
{
“source”: “omop”,
“sourceVersion”: “5.3”,
“target”: “urn:oid:2.16.840.1.113883.6.238”,
“targetVersion”: “4.0.0”,
“element”: [
{
“code”: “8515”,
“display”: “Asian”,
“target”: [
{
“code” : “2028-9”,
“display” : “Asian”
}
]
},
{
“code”: “8516”,
“display”: “Black or African American”,
“target”: [
{
“code”: “2054-5”,
“display”: “Black or African American”
}
]
},
{
“code”: “8527”,
“display”: “White”,
“target”: [
{
“code”: “2106-3”,
“display”: “White”
}
]
},
{
“code”: “8557”,
“display”: “Native Hawaiian or Other Pacific Islander”,
“target”: [
{
“code”: “2076-8”,
“display”: “Native Hawaiian or Other Pacific Islander”
}
]
},
{
“code”: “8657”,
“display”: “American Indian or Alaska Native”,
“target”: [
{
“code”: “1002-5”,
“display”: “American Indian or Alaska Native”
}
]
},
{
“code”: “38003613”,
“display”: “Other Pacific Islander”,
“target”: [
{
“code”: “2076-8”,
“display”: “Native Hawaiian or Other Pacific Islander”
}
]
}
]
},
{
“source”: “omop”,
“sourceVersion”: “5.3”,
“target”: “http://terminology.hl7.org/CodeSystem/v3-NullFlavor”,
“targetVersion”: “2.0.0”,
“element”: [
{
“code”: “0”,
“display”: “No matching concept”,
“target”: [
{
“code”: “UNK”,
“display”: “Unknown”
}
]
}
]
}
]
}

{
“resourceType”: “ConceptMap”,
“id”: “Person.ethnicity.conceptid”,
“group”: [
{
“source”: “omop”,
“sourceVersion”: “5.3”,
“target”: “urn:oid:2.16.840.1.113883.6.238”,
“targetVersion”: “4.0.0”,
“element”: [
{
“code”: “38003563”,
“display”: “Hispanic”,
“target”: [
{
“code” : “2135-2”,
“display” : “Hispanic or Latino”
}
]
},
{
“code”: “38003564”,
“display”: “Not Hispanic”,
“target”: [
{
“code”: “2186-5”,
“display”: “Not Hispanic or Latino”
}
]
},
{
“code”: “0”,
“display”: “No matching concept”,
“target”: [
{
“code”: “2186-5”,
“display”: “Not Hispanic or Latino”
}
]
}
]
}
]
}

1 Like

Thanks again for all of the great information everybody!
Yes, I am looking at a bunch of data that uses the FHIR terminology for race defined here:
https://www.hl7.org/fhir/us/core/CodeSystem-cdcrec.html
That is part of the US Core as defined here:
https://www.hl7.org/fhir/us/core/artifacts.html
(…continuing on next reply as I’m only allowed 2 links per post)

It looks like there is a code system called ‘Race’ (vocabulary_id=‘Race’) that has all of the standard concepts for the ‘Race’ domain.

I’m guessing this is an (OMOP/OHDSI) internally defined terminology?

I was able to find some documentation here:
https://www.ohdsi.org/web/wiki/doku.php?id=documentation:vocabulary

And the page for Race is here:
https://www.ohdsi.org/web/wiki/doku.php?id=documentation:vocabulary:race
But this page comes up as “This topic does not exist yet”

Are the above the best place to go for documentation to answer questions like "Hey, I see there’s a terminology in OMOP called ‘Race’. Where did it come from? What is its purpose (i.e. it looks like it is the standard terminology for race, is this correct)?

Thanks again for all of the help. It really is a pleasure and honor to be working with so many amazing people!

@greshje.gmail or @greshje:

The race subject has been discussed heavily at the end of 2019 (including your question of the origin), only for all the attention to be taken away by the upcoming pandemic. The documentation is lacking because of that lack of action. This year, we should finish up what we started 2 years ago. My strong preference is to reduce the races to a minimal set that can be upheld in international data, rather than focus on the US reality, and replace the awkward hispanic/not hispanic choice in ethnicity by some generally acceptable list of ethnicities. No solution will be perfect, or even good, as ethnicity and especially race are not objective attributes of people, but self-assigned social constructs only relevant in the concrete context of a specific society, which itself is not objective or properly defined. We will need real use cases to help with the decision making.

Are you looking for these generic artifacts as well? Or is this just part of the race question?

To expand on this topic, I am looking for any and all mappings of HL7/FHIR codes to OMOP concepts for FHIR core R4+ and HL7 terminologies from Internal CodeSystem list, starting with those with Normative maturity level.

For example:

Any pointers will be greatly appreciated.

PS: I already collected a few relevant links but they are not too helpful

@opatterson:

Thanks for bringing these up.

  • research-subject-status: you should ask the Clinical Trial Workgroup. They may have plans to address that. @mike_hamidi?
  • observation_status: OMOP doesn’t work that way. It’s a closed world system collecting facts that happened. When the information never goes beyond preliminary then that becomes the fact. But the status itself is not only irrelevant, it is dangerous: Each fact would have to be tested for its status, killing every single existing analytic (it expects a record of a fact to be definitive) and making querying god-awfully slow. So, the ETL should toss out everything that is not a fact, and write the remaining as is into the appropriate table.
  • Observation Category Codes: These are Domains in OMOP: Social History are Observations, Vital Signs are Measurements, Imaging itself is not used, but the findings could be Conditions, Observations or Measurements, Laboratory are Measurements, Procedure are Procedures (this one is easy), Surveys and Exams are are Observations, Therapy is Drug Exposure, Procedure or Device and Activity is Observation. Each concept knows which Domain it is. So, in OMOP these are not data, but reference data.

FHIR to OMOP transformation is not an exercise where you take an incoming code list and just dump it over the fence, unfortunately. It requires mapping, pivoting, complex logic.

Thank you very much for your thoughtful response!

t