OHDSI Home | Forums | Wiki | Github

Piloting Chemo infusion capture with the PCORnet CDM 4.0 draft for MED_ADMIN: quick OMOP question about drug_type_concept_id for chemotherapy infusion


I’ll also post this on the OHDSI forum but didn’t know if
you monitor that and those in our project are not in that community.

We have a PCORnet rapid cycle research project for cancer and
one element is to incorporate or link chemotherapy for targeted therapeutics across
sites in MidSouth, GPC, and OneFlorida.

We plan to push the infusions into the PCORnet CDM4.0 draft
for MED_ADMIN table and pilot a couple columns to resolve some ambiguities.

  1.  One field we want to add is a concept code ala OMOP to distinguish the type of administration and we were inspired by the OMOP drug_type_concept_id

It looks like if I browse Athena, I can find your current
allowable values

but it’s not totally clear to me what your community would call chemotherapy infusion in a cancer treatment room.

A OHDSI thread at one point indicates inpatient but then in other ways isn’t clear to me.
I don’t consider that to be an inpatient unit but it’s also not administered by the physician. It’s definitely not based on the Jcode procedure as that kind of data will be placed in the PROCEDURE table for the CDM

But, I don’t know if your community just always says they are physician administered based on an EHR order (581373), physician administered from an EHR observation (43542358), or takes a big tent healthsystem
view of inpatient administration (38000180). Same would go for emergency rooms as well I suppose.

I’d call it something like “clinic infusion room administration” as a new concept and then perhaps have other concepts for med admin not done by a physician in other procedural areas but since you don’t choices like that in your vocabulary I’d love to know why before we create concepts orthogonal to your work.

Thank you for your guidance as we’d like to be aligned with
as broad a part of PCORnet as possible across these data models.
2. The other fields we see as valuable to pilot would be

a. PRESCRIBING_ID (a link back to the PRESCRIBING table which should contain the parent medication order from the physician for the chemotherapy)

b. MEDADMIN_DOSE_CALCULATED – using the same field to store strength or dose is suboptimal in terms of
clarity of the current 4.0 draft (MEDADMIN_STRENGTH_CALCULATED). The actual dose received is more important to
us than the strength of the infusion bag/pb.


  •     Potentially (not critical to this study but we see some use cases where having the start and stop of the infusion session could be useful):





Russ Waitman, PhD
Director of Medical Informatics
Associate Vice Chancellor for Enterprise Analytics
Professor, Department of Internal Medicine
University of Kansas Medical Center, Kansas City, Kansas
913-945-7087 (office)
http://www.gpcnetwork.org – a PCORnet collaborative

Type concepts can be difficult to decide. If you need new concept, that is OK to pursue.

Ad other questions: Can you link to what the specs are for MED_ADMIN table.
Consider convention to use datetime (see newest 5.2 model).

Linking orders to administration is a great topic to get feedback. (other EHR users may have such data and want to do research questions about those). Inpatient medication administration is a gap that needs more community nudges.

Thank you for the reply Vojtech. I’d hate to make a new one up if there’s existing OHDSI guidance on which one should be used.

The CDM 4.0 spec is a draft but we’ve put a copy here


1 Like

Hi @rwaitman,

We as a community haven’t tackled the complex world of oncology data, but we have established a working group and would love your participation! We need more use cases, like the one you posted, to robustly model the data. @rimma is our fearless leader! You should join :smile:

If you need this done now, I would use drug_type_concept_id = 38000180 inpatient administration. It’s not completely accurate, but the drug is being administered on site. The visit_type_concept_id = 9202 outpatient. You have a person being seen for an outpatient visit, but being administered an inpatient medication.

@Vojtech_Huser is correct:

We also have a new sub-work group (part of THEMIS) to clean up the type_concept_ids. I will add your suggestion to our list. Would you like to join the next call? I can send you an invite when we schedule it.

This data goes in the OMOP Drug_Exposure table with a type_concept_id = 38000177 prescription written

Would the quantity field work for this?


Hello @rwaitman,

I concur with @MPhilofsky and would like to invite you to join our Oncology workgroup: http://www.ohdsi.org/web/wiki/doku.php?id=projects:workgroups:oncology-sg

Also, if you are not aware, a group of PCORnet OMOPers created a crosswalk between OMOP CDM 4 and 5 and PCORnet CDM 3 (we stopped at version 3). There is a set of documentation and, if I am not mistaken, some ETL code, in GitHub. @DTorok is still actively developing ETL for PCORnet sites. He should also point you to the documentation in GitHub.


Hi Rimma and Melanie,
Our main work is in i2b2. So, we currently dump from i2b2 -> CDM.

For the specific work at hand for PCORnet I am not trying to fully implement OHDSI.

But I had a hope that if we could make the piloting of the MED_ADMIN table include a field like the OMOP drug_type_concept_id, it would make it easier for people to harmonize as opposed to we do our work in complete isolation from the OHDSI community.

It sounds like for chemo infusions, OHDSI doesn’t yet have clear guidance with the current types. If it’s anything like PCORnet, it probably depends on if you come from a clinical versus a claims world view what you’d assign.

A claim centric person probably feel it’s a procedure because it may generate a J Code HCPCS for the chemo bill

A EHR clinically focused person would view it as the drug is being infused as the main activity itself and not in conjunction with say a cath placement so it looks more like an inpatient administration in a non-inpatient setting. They wouldn’t want to call it inpatient administration as it doesn’t happen during an inpatient encounter most of the time; there are some cases of inpatient chemo treatment.

I’m also going to look at see what the HL7 people think in this space