OHDSI Home | Forums | Wiki | Github

Drug Group Era under Characterization tool


(Shinyoung Ju) #1

Hi OHDSI enthusiasts,

I had a quick question regarding Drug Group Era features in Characterizaion tool.
I used this function to view the list of medications used by a cohort before the index dates (30 days, 180 days, and 365 days) which are groupped by ATC 1 through 4 levels.
However, I see all ATC levels in one single list, and found it difficult to interpret why some medications are groupped with ATC level 4 whereas the other medications are groupped with ATC leve 1. In order to make the list more harmonized, I wanted to know if there is a way for me to choose a specific ATC level.

It owuld be GREATLY appreciated if any of you could give me some insight on this – (maybe I do not fully understnad ATC grouppings?) We can have a call as well. Please let me know :wink:

Best regards,
Mariana


(Seng Chan You) #2

@byr0sxj
You can find how it works here
It seems like feature extraction package of OHDSI tool ecosystem classifies drug according to ATC group level 1, 3, 4, and 5. And I think they’re overlapped. You can make your own custom feature extraction function based on the code above.

Could you add some more comments? @schuemie


(Martijn Schuemie) #3

The drugs are simply rolled up the ATC hierarchy provided by the Vocabulary. A single drug will therefore be represented at multiple ATC levels.

I don’t know if there’s a way to limit your features to a specific ATC level in the Characterization tool. If you’re running FeatureExtraction in R you could write some SQL to select only those characteristics corresponding to a specific level.


t