Dear community,
when using the Levodopa medication log from PPMI I noticed something strange about the start and stop dates (STARTDT & STOPDT in Concomitant_Medication_Log_*.csv). With the newest medication log downloaded today (16th June 2025), there is one specific date (07/2020) that occurs 960x as a STARTDT while the next most frequent STARTDT occurs 98x. The same is true for the STOPDT where one date (06/2020) occurs 1813x with the next most common date occurring 71x.
Does anyone have an idea why this is the case? Are these dates still valid or is there some mistake in the medication log?
Many thanks in advance for your help!
Kind regards,
Julia
3 Likes
Hi @juschuma, thanks for the question!
Wondering if @jodiefm or others who work with PPMI data (@pbchan, @stnava, @MiaFeng, @alberto.imarisio) would have any insight re: the start and stop dates (STARTDT & STOPDT) and why there would be such a discrepancy between the most frequent and the second most frequent dates?
It looks like a lot of these dates get defaulted, presumably because the participant can’t recall exactly when they started a particular medication, but the June/July 2020 date does stand out as anomalous. We probably need to do some digging to get to the bottom of this.
2 Likes
Thanks, @jodiefm!
@mcbrumm, wondering if this is something you might have insight into? Not sure if this has anything to do with PPMI 2.0 data harmonization?
The PPMI 2.0 data harmonization was my first thought, too. I don’t have any insight other than what I have observed from the data, but you can tell whether an entry is pre-2.0 or post-2.0 (which occurred in 2020) based on its EVENT_ID. The EVENT_ID is LOG for pre-2020, and ED for post-2020.
Another difference to be aware of is that the pre-2020 data records LEDTRT (drug name) and LEDD (levodopa equivalent daily dose) only, while the post-2020 data records 5 fields of LEDTRT, LEDDSTRMG, LEDDOSSTR, LEDDOSE, and LEDDOSFRQ.
Since 2.0 started in 2020, my assumption is that if a patient enrolled saying they were on a given medication, but couldn’t remember the start date precisely, it may have been recorded as July 2020 (the beginning of PPMI 2.0). This would explain the over-enrichment of this as a starting date!
Hope this helps, and I would love to get confirmation from someone who knows the inside details!
3 Likes
Right, I was going to ask if it was a default date based on enrollment, either:
- A new enrollment from PPMI 2.0 (say they were already on levodopa and this was a default start date for PPMI 2.0) OR
- An enrollment from PPMI prior to PPMI 2.0 that defaulted with the data harmonization to the 2020 date. OR
- Something related to the first two.
Let me see if I can contact the statistics core and verify; I’m in touch with them for another project right now.
4 Likes
Hi everyone,
thanks a lot for your replies, this is all really helpful information!
I have checked the enrollment dates for those participants with the 07/2020 STARTDT and all except two were enrolled before 2020, so it seems that this is more related to the data harmonization maybe?
I wonder if this means that the actual dates could still be retrieved somehow.
It would be great if you could check with the statistics core to get a better idea of what is going on!
Thanks everyone!
Best wishes,
Julia
3 Likes
Hi All,
I checked in with my team here in the Statistics Core and confirmed that this is indeed related to data harmonization efforts during the transition from 1.0 to 2.0. This transition coincided with the implementation of a new EDC. For participants enrolled during the 1.0 (pre-2020) time period who remained active in PPMI and transitioned into the 2.0 EDC, some steps were taken to avoid duplication of their LEDDLOG data. Specifically, it was necessary to set an end date for these records in the 1.0 EDC (STOPDT =06/2020) and then “restart” these records in the 2.0 EDC (STARTDT=07/2020). In sum, these data are valid and reflect having to close out these records in the 1.0 dataset and restart the same records in the 2.0 dataset. I hope this helps to clarify!
Best regards,
Michael
4 Likes
Wow, that makes perfect sense! Thanks for investigating Michael, and for explaining so clearly.
It’s great to get the confirmation that the underlying data represents the situation correctly, meaning that the LEDD records are continuous for those patients straddling the 2.0 EDC transition (albeit with an artificial extra “stop” and “start” at the time of the transition). Very helpful, thanks!
1 Like
Thanks, @mcbrumm ! Appreciate you clearing this up. Also tagging @juschuma to make sure she sees this update.
2 Likes
Thanks a lot @mcbrumm and everyone for the explanation! Great to know where these extra stop and start dates come from.
I really appreciate everyone’s help and insights, many thanks!
2 Likes