I make a concept set with: the descendants of “Metronidazole, local oral” (J01XD01).
If a create a cohort with initial event in drug exposure with this concept set.
I get around 30k patients.
If i use the same but in drug era. I get 180k patients.
I believe, this is bcs the drug era table uses ingredients, and therefore it is including all drugs with metronidazole ingredients, event these that are not under the J01XD01 ATC group.
Is this the expected behaviour ? Few EU studies I have seen use the ATC groups.
Have we done something wrong ? We use the suggested script SQL Scripts
Correct. The era goes after the ingredients, assuming any administration route renders it inside the body. It makes no difference between local and systemic. If you need to make such difference, you have to stick with DRUG_EXPOSURE.
Thanks for the clarification, Christian. So, to capture only specific formulations like local oral metronidazole, it’s best to use the DRUG_EXPOSURE table with the appropriate concept set. The DRUG_ERA table groups exposures at the ingredient level, regardless of route or formulation, which explains the broader patient count. This distinction is helpful for studies needing more granular control over drug administration routes.
