Hi peeps, any insight on how wide range of medications should be grouped for analysis? it’s not just dopaminergic meds, the data come from people being asked what meds they use for longer than 6 months so it could be anything really from vitamins to diuretics. i was checking out the general drug categories from FDA but not sure if it’s a good one or if there’s another more preferred categorization. I’m all ears!
Hi Ece, this is an excellent topic to bring up!
First thought: the dataset you’re working with might already contain a classification for drugs. For example, PPMI has a field for indication code for concomitant medications (non-levodopa equivalent meds), with ~25 different indications. It focuses on the typical PD comorbidities and related conditions, such as RBD, daytime sleepiness, bladder dysfunction, etc, and doesn’t include unrelated conditions such as cancers, dermatological conditions, etc. This could be either a pro or a con, depending on what you want to do with the medication class information!
My second thought is that if you are looking more broadly than the typical PD comorbidities, you would probably be better to use a more general ontology. I haven’t used the FDA drug categories myself, but a couple I have used are:
- Therapeutic Class from the Therapeutic Target Database https://db.idrblab.net/ttd/
- MeSH class ID for RxNorm drugs: https://mor.nlm.nih.gov/RxClass/search
For programmatic conversion of a large set of medications, I’d suggest starting with the TTD. The data can be downloaded and used locally for mapping, instead of having to call APIs for every search.
And, of course, nothing is as easy as it first sounds! There are some drugs with dual purposes, and if there is little prescribing context, it can be hard to definitively code an entry as one specific class versus another.
Hope this helps, but I’d be interested to hear more about what you plan to do with the drug class info!
Thank you so much Victoria! It’s a pilot survey study that I collected data for, and I switched institutions right as I wrapped up the data collection, so no readily done classifications and the only laborer available is moi 
I’m gonna dive into the databases you shared and appreciate the info!!!
That sounds great! Please let me know how you get on. I’m always interested in hearing about new databases or ontologies, and people’s experience with them!