Blog

Sleep regularity vs sleep duration: what a 60,000-person study found

A large UK Biobank study compared how consistent sleep times are with how long people sleep. What it measured, what it reported, and what it cannot show.

Published by Intus ยท 6 min read

Most sleep numbers we track are about how long: hours in bed, hours asleep, a nightly total that goes up or down. A large study published in the journal Sleep asked a different question. What if the more useful number is how consistent your sleep times are from one day to the next?

This post walks through what that study did, what it found, and what it cannot tell us. It is a summary of published research, not advice about your own health.

What the researchers did

The paper comes from a team at Monash University in Melbourne with colleagues in Boston and Manchester, with Daniel Windred as first author. It was published online in September 2023 and appeared in the January 2024 issue of the journal. It used data from the UK Biobank, a long-running research project that follows hundreds of thousands of volunteers in the UK.

A subset of participants wore a wrist accelerometer, a motion sensor a bit like a basic fitness band, for seven days. The authors used that movement data from 60,977 people, with an average age of about 63 and just over half of them women, to work out when each person was asleep and awake. They then followed the group for about six years on average and counted deaths from any cause.

The metric: Sleep Regularity Index

The measure they used is called the Sleep Regularity Index, or SRI. The idea is simple. Take any moment in the week and the same moment 24 hours later. How often is the person in the same state, asleep both times or awake both times? Average that over the whole recording. A score of 100 means a perfectly repeating pattern, and 0 means a random one.

The metric was introduced in a 2017 study in Scientific Reports of 61 university undergraduates who kept sleep diaries for 30 days. Comparing the most regular and least regular fifths of students, the less regular sleepers also had later circadian timing and a weaker daily light pattern. It also found a modest link between regularity and academic performance. It was small, and it was in students, but it is where the index comes from.

The middle half of people in the Biobank sample scored between about 74 and 86, with a median of 81.

What the study reported

The authors split people into five equal groups by SRI and compared the most regular groups with the least regular one. After adjusting for age, sex, ethnicity and a range of lifestyle and health factors, the four more regular groups all had a lower rate of death from any cause than the least regular group. The reported difference ranged from 20% to 48% depending on the group. The paper reports similar patterns for deaths from cancer and for cardiometabolic causes.

The headline in the title is the comparison with sleep duration. When the authors built matching models, regularity predicted all-cause mortality more strongly than the number of hours slept. They also report that adding sleep duration to a model that already included regularity did not improve it significantly. They are careful to say duration still matters. Their point is that, of the two, regularity was the stronger predictor.

One detail from the discussion helps make the score concrete. People in the most regular fifth tended to fall asleep and wake within roughly a one-hour window on most days. People in the least regular fifth had windows closer to three hours. So the difference is about spread in bed and wake times, with night-time waking and fragmented sleep also pulling a score down.

What it cannot tell you

The authors list the limits themselves, and they matter.

  • It is observational. People were not assigned a sleep pattern. The paper notes that irregular sleep could be a cause of poor health, a marker of it, or both.
  • One week of data. Each person wore the sensor for seven days, which is a snapshot, and the authors say longer recordings would be better.
  • A specific group. Participants were mostly older and mostly of similar ethnic background, so the findings need to be repeated in other groups.
  • Adjustments are imperfect. Some of the factors the authors corrected for may themselves sit on the path between sleep and outcomes, so the true size of the link is uncertain.
  • A wrist sensor is not a sleep lab. Accelerometers infer sleep from movement. Consumer wearables use their own methods, and their sleep scores are not the SRI.

None of this is a reason to dismiss the finding, and none of it is a reason to treat one study as settled. It is one large, careful dataset pointing at a measure that many people with wearables already have the raw material for.

Why this is interesting for people who track

If you wear a watch or ring, you probably already have bed and wake times for many weeks. Many apps show them night by night, and the focus tends to fall on the total. The regularity idea is a reminder that the shape of those times across weeks is also data: how wide the spread is, whether weekends look different, whether a stretch of travel or a new job shows up as a change.

It also shows why context helps. A change in your bed and wake times looks different next to a change in training, in how much you ate or when, in a supplement you started or stopped, or in a lab result a few weeks later. None of that proves anything on its own. But it is much easier to ask a good question when those inputs are on the same timeline instead of in five apps and a folder of PDFs.

That is what we are building at Intus. It takes inputs such as wearables, nutrition, supplements and workouts, and shows how they relate to outputs and biomarkers over time in one place. It does not diagnose, interpret your results as good or bad, or suggest doses. It just puts your own data side by side.

If that sounds useful, you can join the waitlist at intus.fit. The iPhone app is coming soon.

Intus is not a medical advisor. It does not diagnose conditions or suggest dosages. This article describes published research and is not advice about your health. If you have a health concern, talk to a qualified professional.

Sources

  1. Windred et al. (2024). Sleep regularity is a stronger predictor of mortality risk than sleep duration: a prospective cohort study. Sleep, 47(1), zsad253 (2024-01-11)
  2. Phillips et al. (2017). Irregular sleep/wake patterns are associated with poorer academic performance and delayed circadian and sleep/wake timing. Scientific Reports, 7, 3216 (2017-06-12)

What Intus is not

Intus is a personal health data tracker, not a medical advisor. It does not diagnose conditions, recommend treatments or suggest doses. For questions about your own health, talk to a qualified clinician.

Intus brings your wearable data, lab reports and meals together in one place, in plain numbers.

Join the waitlist