Too short. How about you?
All of the cohorts in Charybdis are available in this CSV list. We have ATLAS and JSONs links in here: https://github.com/ohdsi-studies/Covid19CharacterizationCharybdis/blob/master/documents/CharybdisPhenotypeLibrary.csv
Charybdis thinks like this. We cut a count of the people in a Target cohort (e.g. the flavors of COVID-19 and the Flu). We then stratify the Target cohort by a number of custom stratum (all labeled as Stratum in the list). We then create a count of the Features that are within the Target-Stratum combination.
The cohort IDs you might be interested in are Cohort 132, 134 and 136. These are (1) COVID-19 diagnosed or tested positive with 365 days of prior history before their index, (2) hospitalized with COVID-19 diagnosed or tested positive with 365 days of prior history before their index and (3) hospitalized, receiving intensive services (aka proxy for ICU – it’s hard to find this reliably so we used procedure codes indicative of intensive care units) with COVID-19 diagnosed or tested positive with 365 days of prior history before their index.
There’s currently no Stratum for Atopic Dermatitis individually. So in the available results you could look at the CONDITION_ERA (and search ‘atopic dermatitis’ it will show up as a string and what that represents is the way atopic dermatitis is represented in the SNOMED hierarchy) and see the counts of people who are in that Target cohort with atopic dermatitis.
Now, in Charybdis we didn’t actually pull custom definitions for drugs. We just have DRUG_ERAs (derived from the drug_exposure table) and those are by the ingredient level. But there’s another study you could use. You can look at Scylla Characterization shiny app to see information characterizing these: https://data.ohdsi.org/ScyllaCharacterization/
The challenge you’ll see is that we don’t have the intersection of what you really want which is COVID-AD-drug exposure. I’m sorry to say that’s a gap at the moment. But maybe there’s a view here that’s still usable.