I’m interested in the self-report cognition scores from the PDAQ-15 and the metagenomic data from the PDMB project, but I’m having trouble getting a full set of cognition scores to match with the cohort of participants with stool samples:
From the metagenomic data, I get 607 participants after filtering out incomplete metadata entries
Downloading a .csv with the PDAQ-15 raw scores from the Data Exploration Network gives me 355 unique matches
From the Stagman survey data, there are 248 unique entries that have a Cognition.score entry and match to the IDs from the metagenomic dataset, but the variable has been reduced to 0, 1, or 2
So from my end it looks like 355/607 have raw cognition scores, and 248/355 have reduced cognition scores. This leaves 252 participants with metagenomic data but no cognitive metadata in these exports.
Could you point me toward any additional data exports to capture those missing 252? Also, could you clarify the specific thresholds or PDAQ item combinations used to reduce the raw scores into the 0, 1, and 2 categories in the Stagman TSV?
Thank you for the questions and I’m happy to hear you are exploring these data. I just wanted to let you know that I’m working on this and will respond as soon as I am able.
Thank you for your patience while I investigated these questions.
First, I want to clarify that the PDAQ-15 from Fox Insight (i.e., Your Cognition and Daily Activities survey) and the Cognition.score column in the Stagaman Survey Data are not the same. The scores that you see in this column (0, 1, 2) were derived from other 23andMe research surveys and used in exploratory analyses; therefore, these results and descriptions of this variable were not published in Stagaman et al. (2024). I’ll provide a brief description of these values here:
Cognition.score of 2 indicates participants that self-report experiencing distressing hallucinations (seeing or hearing things you are told do not exist) at night in the past month.
Cognition.score of 1 indicates participants that:
self-report a diagnosis of mild cognitive impairment
OR
self-report that confusion or memory loss has prevented them from doing house activities or chores that they are used to doing AND require assistance due to confusion or memory loss (both in the last month).
Cognition.score of 0 indicates participants that self-reported “no” to the above criteria.
NA values reflect missing or incomplete data. These scores were hierarchically structured such that classification criteria were prioritized for higher scores (2) before considering lower scores (1, then 0). For example, a participant qualifying for a score of 2 and 1 would be given a score of 2.
Second, given that the PDAQ-15 is a Fox Insight questionnaire, only Fox Insight participants will potentially have data. Therefore, when joining PDAQ-15 data with PDMB data, only participants with a “FOX_” ID prefix will potentially have data. Those PDMB participants without a “FOX_” prefix will not have PDAQ-15 data.
I hope this helps clarify some of the questions and explain some of the missingness when it comes to joining these separate data sources together.