Modifier_concept_id in PROCEDURE_OCCURRENCE table belong to other domains

[QI_omop (Qi Yang)]

I am trying to load modifier_concept_id column in PROCEDURE_OCCURRENCE table. Please note this is NOT procedure_concept_id but modifier_concept_id. The source procedure modifier code are CPT and HCPCS procedure modifier codes. When I map the source procedure modifier to concept table, I found many records belong to Observation, Measurement, and Device domain instead of Procedure domain. If these records are procedure code, then I know they should load them to their respective domain. However they are not procedure code but procedure modifier (for modifier_concept_id field in PROCEDURE_OCCURRENCE table). So how should I handle this situation?

I tend to think to not consider domain information at all. In another word, just load all of them to PROCEDURE_OCCURRENCE table regardless of their domain. The reason for this is because it will truly reflect description for the underlying procedures. For example, here is a source data record (sensitive information has been removed):

Prc_cd 00840 is a procedure (Anesthesia for intraperitoneal procedures in lower abdomen including laparoscopy; not otherwise specified). Prc2_modr_cd QK is a HCPCS code meaning MEDICAL DIRECTION OF TWO, THREE, OR FOUR CONCURRENT ANESTHESIA PROCEDURES INVOLVING QUALIFIED INDIVIDUALS, and it belong to Observation domain. So I would like to put the this procedure modifier in the modifier_concept_id column in the PROCEDURE_OCCURRENCE table even though it is in the Observation domain. I would like to know the opinion of the community. Thanks.

[Dymshyts (Dmytry Dymshyts)]

Yep, it’s seems to be fine to put modifier_concept_id regardless of what domain it belong to.

[Christian_Reich]

Friends:

I think this is more complicated. Currently, the modifier_concept_id is not the same thing as Modifier Concepts in HCPCS and CPT4. They are separate, get their own concept_id and domain, and may end up in another table, only connected to the procedure by person_id and date. That could be the right thing to do, or the wrong thing to do. The other problem is that you can have more than one modifier for CPT4 and HCPCS codes, and the CDM table won’t let you.

I think this is a subject for the THEMIS Working Group http://www.ohdsi.org/web/wiki/doku.php?id=projects:workgroups:themis:members . They are meeting on the 17th. If you are in DC, come and bring it on.

[nadavrap]

I know it’s an old thread, but my I think my question belongs here.
Why is modifier_concept_id a required field in CDM5?
I do understand that Claims data comes with modifiers, but I’m working with a different EHR, and I don’t see an easy way to add modifiers to each procedure.
Alternatively, is there a default modifier that should be used?

[DTorok (Don Torok)]

Modifier concept id is not required in CDMv5, but is in CDMv6. The default, if you do not have a modifier, is concept id 0, ‘No matching concept’ or commonly referred to as Unknown.

[Alexdavv (Alexander Davydov)]

The issue was addressed here. Please look.