Webinar: Exploring UK Biobank's Sleep Data

This session provides  insights on our newly released sleep data, including how the sleep questionnaire was built, the process of gathering responses from our community of participants, and how we turn these responses into accessible, research-ready data. 

The panel also share what’s been done with the data since they were made available to UK Biobank researchers, key early findings, and insights on how this could impact the future of sleep science.

Watch here

This webinar was streamed live on 19 May 2025

The sleep questionnaire data were made available to researchers as part of data release v19. Full details here.

Speakers

Martin Rutter - Deputy Chief Scientist, UK Biobank
Jo Holliday - Scientific Study Development Group Lead, UK Biobank
Simon Kyle - Professor of Experimental and Clinical Sleep Research, University of Oxford
Hang Yuan - Early Career Research Fellow, University of Oxford
 

Q&A

Here’s a quick recap of the Q&A section of the session:

Were there any inclusion or exclusion criteria for participants?

There were no specific exclusions beyond those who opted out of contact. Participation was open to all eligible UK Biobank participants.

How did the response rate compare to other questionnaires?

The sleep questionnaire had a 54% response rate, consistent with recent UK Biobank surveys.

Is the questionnaire still open?

Yes, but we're only actively inviting people to complete it if they newly register an e-mail address with us. Otherwise it's just available on our participant portal for people to find.

What results surprised the team?

• Respondents reported high use of electronic devices before bed.
• Sleep medication use was higher than expected. Around 4% of the total sample reported using sleep medication, higher than in the baseline UK Biobank cohort. This could be due to ageing since baseline.
• Some participants linked their sleep problems to specific causes like worry or pain. These kind of insights often aren't captured in large studies.
• Self-reported sleep efficiency closely matched accelerometer data, this was unexpectedly consistent.

How reliable are self-reported sleep measures vs. wearable data?

Both are useful but capture different aspects of sleep. They’re often correlated, but not interchangeable - wearables provide objective data, while self-reports reflect personal experience. Sleep is multidimensional so no single measure tells the whole story. When there are discrepancies between subjective and objective measures, these can still offer valuable insights into broader patterns across different populations.

Can sleep data be linked to other UK Biobank datasets like genetics?

Yes - this is a major strength of UK Biobank. Sleep data can be linked to genetics, imaging, proteomics, metabolomics, mental health, and wearable data for rich, multidimensional analysis.

Was the questionnaire deliberately released in winter to capture seasonal effects?

No, the timing was operational. However, some questions do explore seasonal variation in sleep and light exposure.

(from the live chat section) Do these questionnaire data overlap with the previous 100k accelerometer collected data?

Yes, there is about a 72k overlap between the two datasets.