OHDSI Home | Forums | Wiki | Github

ATC alterations missing in OMOP

(Javier Gracia-Tabuenca) #1

I notice few ATC codes listed here


are missing in the ATC vocabula ry. (attached list)
atc_modified_not_in_OMOP.xls (3.8 KB)

Is there any reason for this? Should not be difficult to add them ??

(Anna Ostropolets) #2

These are the ATC codes which are no longer included in the releases. Is there any specific reason you need them?

(Javier Gracia-Tabuenca) #3

Hey @aostropolets

Few of these codes appear on our data base.

What drug_source_concept_id should I the use ??

Do you mean in the OMOP relases or in the ATC releases ?

(Anna Ostropolets) #4

I meant ATC releases. They didn’t include those ATCs in the current release (deprecated them), and these codes subsequently do not end up in OMOP releases. ATC also provide reasons for such changes, e.g. “Split of ATC code. The different fixed combinations of ibuprofen and opioid analgesics are given new ATC 5th levels in N02AJ Opioids and non-opioid analgesics. M01AE51 ibuprofen, combinations remains for other combinations.”

Now, back to the drug_source_concept_id. ATC is a classification system as you know, so it corresponds to a list of drugs in RxNorm rather than a single drug. This is why we do not suggest using ATC to populate drug_exposure at all. For example, if you have a source drug vocabulary and ATC as a reference, source vocabulary should be used.
Did you map those ATC to RxNorm yourself or how did you come up with standard concepts to populate drug_concept_id?

(Vojtech Huser) #5

Seems like terminologies behave differently in Athena.

There are either:

  1. permanent terminologies - retired codes stay for ever and use status to indicate validity
  2. non-permanent terminologies - retired codes disappear from release

E.g., ICD10PCS is permanent (since events may have been recorded under them)

you just described ATC as non-permanent. We should keep a documentation of which class is each included terminology


(Anna Ostropolets) #6

All of them are permanent. Whenever something is in the system it either stays or gets deprecated.
Those codes simply never made it into OMOP, Vojtech.

(Hamed Abedtash) #7

Thanks @Javier for bringing this to our attention. Can you please tell us what time range the missing ATC codes were used in the data and what proportion of records are affected?

@aostropolets, we will eventually need to keep deprecated codes in OMOP vocab as they might have been used before in the data. I suspect this will affect the mappings to RxNorm, but Let’s talk to find the best approach to tackle this.

(Polina Talapova) #8

Dear @Javier, thank you for informing us! To let the vocabulary team take care of this, please, create a GitHub issue here and describe your needs (e.g. the addendum of new ATC concepts with or without mapping).

(Javier Gracia-Tabuenca) #9

thanks all,

@aostropolets : that is a good point. We do have an other source drug vocabulary and ATC as a reference (VNRO). However, we are working towards including it in OMOP. At the moment we were planing ot use ATC in drug_source_concept_id and the mapping to RxNorm made by @abedtash_hamed 's ATC-working-group. When VNRO is in OMOP we will update this.

@abedtash_hamed : In our database the missing ATCs only affect 358983 events wich are 0.792% of the total evens. Medicine data stars in 1995.

@Polina_Talapova: done

(Anna Ostropolets) #10

Sounds good. Please feel free to reach out if you need any help with incorporating your source vocabulary into OMOP. Especially if you want it to be incorporated as a vocabulary with OMOP concept_ids and so. Vocabulary team has an established process for this.
Re old ATCs: they are not in the mappings we created. So if you need both the concept_ids for ATC to be placed in drug_source_concept_id AND mappings to OMOP you should state so.