NSRR staff
Boston, MA
0000-0002-0506-8368
Top Topics
Recent Topics
Unfortunately not. We only have the limited set of channels from the Unicorder, so no EEG/staging.
Are you looking to download the sleep study EDF files? The total file size of SHHS is ~350 GB, so that will probably take a few days for most people. We recommend using the NSRR gem for downloading full datasets.
Note that IRB review is not a firm requirement (though it's preferred) for SHHS data access. There is a second option to select for NSRR Internal Review when completing the Data Access and Use Agreement process. I look forward to reviewing your submission!
Aek,
You must have an approved Data Access and Use Agreement before you can download data from the Sleep Heart Health Study (SHHS). We have received your four (4) DAUA requests and will notify you of our decision next week.
I cannot speak with absolute certainty since I wasn't involved in those analyses, but here is my take on your other questions:
Prevalent CVD at Baseline
I believe the variables form the SHHS1 Health Interview Form (e.g. STROKE15 and MI15) are the ones that have typically been used to gauge prevalence at the baseline visit. That said, from Punjabi 2009 it looks like they might have used a combination of the self-report medical history and the adjudicated outcomes data (variables in the "CVD Outcomes" folder).
Prevalent cardiovascular disease was defined as history of physician-diagnosed angina, heart failure, myocardial infarction, stroke, and coronary revascularization, and was determined by adjudicated surveillance data provided by the parent cohorts or by self-report at enrollment.
AHI Variable of Interest
'ahi_a0h4' certainly will have been used regularly since it's a pretty standard AHI definition. I noted in Redline 2010 that "OAHI" is referenced, and we have a variable named OAHI in the dataset. The paper doesn't mention central apneas, so I'm guessing 'OAHI' was used. 'OAHI' mirrors with the more recently created 'ahi_o0h4'. (Slight sample size differences due to filtering on PSG signal quality in 'OAHI' at SHHS1.)
Hope this is helpful. Please keep us up to date on your progress!
Phil,
Great questions. Thanks for taking such care in your handling and analysis of the NSRR data. I will look into the variable questions and get back to you next week.
As for the sample sizes, the difference of ~600 (6,400 -> 5,800) arose from the removal of participants from the Strong Heart Study cohort. We include this note in the dataset introduction:
Note: Due to sovereignty issues, Strong Heart Study participants are not included in the shared SHHS data. Data from a total of 5804 participants (1915 ARIC, 1230 CHS, 688 Framingham Offspring and 1971 from other studies) consenting to share data are available.
Thanks again for using the site!
The full-night lab studies are from the PSG baseline, not the follow-up titration night.
Good question, I will make a note in the documentation about it soon.
Jennifer,
I emailed this response, but it would be useful to have on the Forum as well:
Hi Jennifer, Sleep architecture and respiratory event scoring annotations are available in XML files (separate from the EDFs). You can read more about the XML types on the Polysomnography Introduction page: https://sleepdata.org/datasets/cfs/pages/polysomnography-introduction.md You will find epoch-by-epoch staging information in the XML files, which you can use to reconstruct the hypnogram. Thanks for using the site!
Hi Jennifer,
Sleep architecture and respiratory event scoring annotations are available in XML files (separate from the EDFs). You can read more about the XML types on the Polysomnography Introduction page: https://sleepdata.org/datasets/cfs/pages/polysomnography-introduction.md
You will find epoch-by-epoch staging information in the XML files, which you can use to reconstruct the hypnogram.
Thanks for using the site!
Benjamin,
You are correct - height and weight are not included in the NSRR MESA datasets. This was done at the behest of the MESA Coordinating Center, who already have a well established manuscript/presentation approval process for investigators who wish to use the MESA data. Height and weight data from Exam 5 (concurrent with MESA Sleep) are available from BioLINCC and directly from the Coordinating Center, although those other sources will have their own barriers to entry (e.g. submitting an analytic proposal).
NSRR is the only source for the polysomnography and actigraphy raw data collected in MESA Sleep.
Mike
Thanks for raising the issue. In SHHS, gender has two levels: 1 = Male and 2 = Female.
Values for categorical variables are described in two places.
1. Online variable pages
E.g. https://sleepdata.org/datasets/shhs/variables/gender
2. 'domains.csv' file
Each categorical (type: 'choices') variable has a domain associated with it. These domains are fully described in the 'domains.csv' file, which is part of the data dictionary.
You mentioned the data dictionary CSV, so I'm guessing you were looking in the 'variables.csv' file. We have discussed possibly creating a combined version of 'variables' and 'domains' so as to not have to jump from one file to the next (i.e. 'variables.csv' references the 'gender12' domain, which is spelled out in 'domains.csv').