OHDSI Home | Forums | Wiki | Github

[2022 US Sympoisum] #70 - ETL convention for DRUG_ERA/DRUG_EXPOSURE

Please use this spot to document:

  1. Describe the issue/topic?
  2. What do we know about this topic? What has been discussed?
  3. What are recommendations for how to handle this issue/topic?
  4. What next steps should be taken?

Related posts:

https://ohdsi.github.io/CommonDataModel/cdm53.html#DRUG_EXPOSURE

https://ohdsi.github.io/CommonDataModel/cdm53.html#DRUG_ERA

1 Like

1. Describe the issue/topic?
DRUG_ERA.DRUG_EXPOSURE_COUNT - there is no ETL convention listed.

2. What do we know about this topic? What has been discussed?
There is standardized code that is already building the DRUG_ERAS.
http://ohdsi.github.io/CommonDataModel/cdm53.html#DRUG_ERA
https://ohdsi.github.io/CommonDataModel/sqlScripts.html#drug_eras

3. What are recommendations for how to handle this issue/topic?
We recommend add the following summary text to the ETL convention
The count of grouped DRUG_EXPOSURE records that were included in the DRUG_ERA row.

4. What next steps should be taken?
Open a GitHub issue so Clair can add it to GitHub.io site for 5.3 & 5.4.




1. Describe the issue/topic?
The description for DRUG_EXPOSURE.QUANTITY is hard to understand.

2. What do we know about this topic? What has been discussed?
What is currently written on the DRUG_EXPOSURE.QUANTITY column

3. What are recommendations for how to handle this issue/topic?
First bring tablets to the top of the description:
For clinical drugs with fixed dose forms (tablets etc.) the quantity is the total number of units/tablets/capsules prescribed or dispensed. The quantity can be a partial number, but then only 1/2 or 1/3, not 0.01.

We could not come to an understanding of what to do when the dose form is not a tablet and we think the current description is not sufficient in that. Also including the link to the drug dose calculations isn’t really helping understanding what the quantity is.

4. What next steps should be taken?
We can start by reordering the text that is currently in the column so that tablets is at the top.

There is more discussion needed on how to handle non-tablet. Maybe leveraging the Darwin work around DRUG_EXPOSURE Characterization.

1 Like

Excited to work on this with you Erica! We should continue to look at how to improve the value of the quantity field. There seems to be a gap in certain formulations where we would not be able to calculate the total quantity of a drug injected/consumed by a patient if we are only ever including the TOTAL quantity handed to the patient, the day’s supply, and the strength – think about a tube of Benadryl lotion – not all will be consumed/used by the patient over a 7 day period…

2 Likes

Appreciated the chance to participate in this discussion at the OHDSI symposium working group session.

  • I concur with the recommendations for documenting the DRUG_ERA.DRUG_EXPOSURE_COUNT field.
  • The current guidance for DRUG_EXPOSURE.QUANTITY requires further discussion.
2 Likes

Sorry I had to bounce early, but it was a great discussion and I also agree with the recommendations above!

Hang on. Don’t remember where that is described, but wasn’t the quantity field supposed to be the number of units if the dose form is discrete (tablets, syringes), and the amount if the form is continuous or divisible (solution, but also component, ingredient)? So you have quantity=6 for 6 tablets or 6 mg of ingredient. (Whether the unit is mg or mL is indicated in the DRUG_STRENGTH table).

This works. And is used for DOSE_ERA.

But: It is a violation of the OMOP CDM principle, by which the meaning of the content of any field must not depend on the content of another field. We need to split quantity into unit_quantity and substance_quantity. Version 6.1?

t