OHDSI Home | Forums | Wiki | Github

Duplicate ‘standard’ concepts for ‘Sexual Orientation’

Hi All,

Athena has the following ‘duplicate standard’ concepts for ‘Sexual Orientation’:

  • 46235214 76690-7 Sexual orientation – LOINC
  • 4283657 66621004 Sexual orientation – SNOMED

A few questions:

  1. Why are there two ‘standard’ concepts for the same thing, i.e., for ‘Sexual Orientation’?
  2. Which one should we use for OBSERVATION.observation_concept_id to create a ‘Sexual Orientation’ record?
  3. In a separate topic on the Forum, someone recommended creating multiple records in the target table if there are multiple ‘standard’ concepts for the same thing. Is that still the official recommendation?
  4. Can someone please share a query that helps identify all ‘duplicate standard’ CONCEPT records?

Thanks much in advance!

They are not the same, they have different concept_class_id’s.

Thanks, Mark. Which one should be used to create an OBSERVATION record?

What is your use case? as in, where does the data come from?

I need to convert a PCORNET dataset to OMOP.

I do not use PCORNET, so I will defer to others for specifics. I should have clarified that I was asking where the original data came from, as-in is this a actual observation or is it self reported? The best advice I can give is look to see if there are answers to a question, then LOINC, if it is an actual clinical observation, then SNOMED.

I am sorry, but without better knowledge of your specific data, that is as good as I can do.

1 Like

Thanks, Mark. That’s a good rule to keep in mind.

1 Like

We haven’t done an exhaustive deduplication job, yet. LOINC and SNOMED overlap all over the place. However, these concepts we should de-standardize entirely. They don’t represent any actual sexual orientation.

Except in Measurement and question-answer pairs of surveys, we should write pre-coordinated concepts representing facts, like heterosexual, gay, bisexual, etc. Pick from the list “Interprets of” in Athena.

No, and should have never been. If you really have two identical records representing fact (not like “sexual orientation”), report on the forum as an issue and pick one of them for your records.

Love that question!!!

Don’t love that question. That’s not a use case! A use case is a scientific question that can be answered using the concepts or relationships. Having and preserving existing data is called the “attic case”. :slight_smile:

How can self reporting be the same as a third party observation? I am missing some piece of logic as we have had a similar conversation about 3 times.

I am not a researcher nor am I a medical provider, for OMOP I am just a simple ETL’er, and like the OP, I find instances where I have to make a choice between two, or more, codes. The best I can do is look at the source and use deductive logic to figure out which one to use. I am not attempting to ‘attic case’ the data, I am attempting to do the correct mapping. It is either look at the source or use randomness to pick as I am under strict requirements NOT to use my knowledge to pick.

Perhaps this should be it’s own thread but the OQ seems done. With respect, thank you.

Oh, I see, @Mark. You need a order of precedence for source information when there are contradictions. It may be a case for @MPhilofsky’s EHR WG.

1 Like

Thanks for the clarification, @Christian_Reich. Much appreciated!