OHDSI Home | Forums | Wiki | Github

Bug or feature of procedure hierarchy?

DQD’s plausible gender check is failing on one of our claims databases due to a large number of male patients with concept 2102639. This concept represents CPT code 17311, which is a code for Mohs surgery not specific to any gender.

However, concept 2102639 is a child of concept 4041261 (“Procedure on female genital system”) as well as several other parent concepts (including…“Procedure on male genital system”). This is why it was flagged by the DQD check.

My question is - is this expected? Is it normal that this CPT code, which might be used to code a procedure on a female genital organ, but also might not, would map up to a female specific parent code?

cc @Alexdavv since we’ve been chatting about this (this is a generic OMOP vocab example of the topic we were discussing)

This link in CONCEPT_ANCESTOR looks like it comes from explicitly declared link ‘CPT4 - SNOMED cat’, which is more likely than not just taken as is from UMLS MRREL table (Code line).

Looking briefly in detail:
CUI C1742725, corresponding to the original term and full synonym, does not seem to have this link; is it attached to one of the ETCLIN synonyms? Needs more investigation, maybe the way we consider UMLS-given relationships is outdated, or maybe this is just a one off mistake.

Hey @katy-sadowski, thank you for finding this issue and @Eduard_Korchmar thank you for saving time investigating it.

We have a related issue on github. We receive these relationships from the UMLS and they work as ‘sort of classification’. However, as you and @Dymshyts showed us, there are lots of flaws in this hierarchy.

We will assess the impact and likely make these relationships non-hierarchical during the next vocabulary release. It means there will no longer be any links in the concept_ancestor table, but the relationships will stay.

Makes sense?

Yes this makes sense! Glad you’re on the case :slight_smile: thanks!