Field note
Sleep regularity beats sleep duration, on the biggest study we have
Almost all sleep advice is about how many hours. The largest analysis of this question found that when you sleep predicts mortality more strongly than how long. Windred and colleagues tracked 88,975 UK Biobank participants wearing accelerometers and scored how consistently each person was asleep or awake at the same time from one day to the next. People in the most regular fifth had roughly 30% lower all-cause mortality risk than the least regular fifth, and that relationship was stronger than the one for sleep duration. It is observational, so it cannot prove that irregularity causes the deaths, and the honest reading is more interesting than the headline.
- 88,975 people, measured with accelerometers rather than asked.
- Around 30% lower mortality risk in the most regular fifth.
- Regularity outperformed duration as a predictor.
- Observational. Illness and shift work disrupt sleep too.
What the study did
Windred and colleagues published Sleep regularity and mortality: a prospective analysis in the UK Biobank in eLife in 2023. The scale is what makes it worth your attention.
| Feature | Detail |
|---|---|
| Participants | 88,975 |
| Measurement | 7 days of wrist accelerometry, not self-report |
| Mean age | 62 |
| Women | 56% |
| Follow-up | Mean 7.1 years |
| Deaths recorded | 3,010 |
| Median regularity score | 60 out of 100 |
The measure is the sleep regularity index. It asks: pick any moment, then look at the same moment 24 hours later. What is the probability you are in the same state, asleep or awake, both times? Average that across a week and score it from 0 to 100, where 100 is perfectly regular.
Two things I like about that. It captures naps, lie-ins and shifted weekends automatically, without anyone deciding what counts. And because it came from accelerometers rather than a questionnaire, it does not depend on people accurately reporting their own sleep, which they famously do not.
What it found
Participants in the highest fifth of sleep regularity had approximately 30% lower risk of death from any cause over the follow-up than those in the lowest fifth, after adjustment for demographic and clinical variables. The relationship was non-linear rather than a straight line.
And the comparison that gives the paper its significance: sleep regularity was a stronger predictor of mortality risk than sleep duration was.
If you have spent years being told to hit seven or eight hours, that is a genuinely awkward finding. It does not say duration is irrelevant. It says that in this dataset, the consistency of your schedule carried more predictive weight than its length, and consistency is the thing nobody puts on the front of a sleep app.
What it cannot tell you
I wrote a page recently arguing that the long-nap mortality studies cannot support the causal claim made of them, because ill people nap more. It would be dishonest to drop that standard because this result is one I like.
So, the same treatment.
- This is observational. Nobody randomised anyone into an irregular schedule. It shows association.
- Reverse causation is live here too. Serious illness disrupts sleep timing. Pain wakes you at odd hours. Depression fragments schedules. All of those also raise mortality. An undiagnosed condition in year one could produce both the irregular week of accelerometry and the death in year five.
- Irregular sleep is not randomly distributed. It concentrates in shift workers, carers, people with insecure or multiple jobs, and people in poor health. Those things carry their own mortality risk, and adjustment only handles what was measured.
- One week of measurement. Seven days of accelerometry standing in for a life. A bad week is not a pattern.
- Mean age 62, UK cohort. UK Biobank participants are healthier and less deprived than the general population. Whether this transfers to a 25-year-old is not established, and I have leaned on American sources elsewhere on this site, so I should say plainly that this one is British.
Why I still find it persuasive
Two reasons, and they are about mechanism rather than statistics. Regularity has a plausible causal path that long napping does not: circadian timing is a real physiological system, light and behaviour entrain it, and disrupting it has documented metabolic and cardiovascular consequences. The shift-work literature exists independently and points the same way.
And unlike the nap findings, the intervention here is nearly free. Getting up at the same time costs nothing, has no known downside, and is worth doing on much weaker evidence than this.
Why this changes the advice
Nearly every sleep tool in existence, including the one on this site, is built around duration. You tell it when you must wake, and it tells you when to go to bed to accumulate enough hours. Duration is the input, the output and the whole framing.
If regularity is the stronger signal, that framing is at best incomplete. A person who sleeps 7 hours 15 minutes at wildly different times every night is doing something the calculator scores as a success, and this study suggests it may not be.
I do not think that makes sleep calculators useless, mine included. It does mean the honest thing for a calculator to say is: the number matters less than hitting it on a Tuesday and on a Saturday. The homepage already says something close to that, and now I know how much weight is behind it.
What regularity means in practice
- Anchor the wake time, not the bedtime. The wake time is the one you can actually control with an alarm, and it is the stronger signal to your clock. Bedtime follows it once sleep pressure does its job.
- Weekends count. The regularity index has no idea it is Saturday. A three-hour weekend lie-in is exactly the pattern it penalises.
- Consistent and slightly short beats variable and long. That is an uncomfortable trade, and on this evidence it is the direction I would pick.
- If you work shifts, this is not a personal failing. Your regularity index is being set by your roster. The occupational literature is the relevant one, not general sleep hygiene.
This is the top item on my own list of five things I changed, and it is the one I would keep if I could only keep one.
Questions people actually ask
Is sleep regularity more important than sleep duration?
In the largest analysis of the question, yes as a predictor of mortality. Windred and colleagues found across 88,975 UK Biobank participants that sleep regularity predicted all-cause mortality more strongly than sleep duration did, with the most regular fifth showing around 30% lower risk than the least regular. It is observational, so it establishes association rather than cause.
What is the sleep regularity index?
A score from 0 to 100 giving the probability that you are in the same state, asleep or awake, at any two time points 24 hours apart, averaged over a week. One hundred is perfectly regular. In the UK Biobank sample the median was 60.
Does sleeping in at weekends matter?
For regularity, yes. The sleep regularity index does not know what day it is, so a substantially later wake time at weekends registers as irregularity. Whether catching up on sleep offsets that trade is not settled by this study.
Should I fix my bedtime or my wake time?
The wake time. It is the one an alarm controls directly, it is a strong signal for your circadian clock, and bedtime tends to follow it once sleep pressure has built through the day.
Does this study prove irregular sleep causes early death?
No. It is observational. Illness, shift work, caring responsibilities and deprivation all disrupt sleep timing and independently raise mortality risk, and one week of accelerometry cannot separate those from a causal effect of irregularity itself.
References
- Windred DP, Burns AC, Lane JM, et al. Sleep regularity and mortality: a prospective analysis in the UK Biobank. eLife. 2023. Source for the sample size, the sleep regularity index, the 30% figure, the non-linear relationship, and the finding that regularity outperformed duration. Figures here are taken from the abstract and summaries rather than a full read, and that is noted in the research log.
- Watson NF, Badr MS, Belenky G, et al. Recommended Amount of Sleep for a Healthy Adult. J Clin Sleep Med. 2015;11(6):591–592. The duration recommendation this finding sits alongside.
Pick a wake time and keep it
The calculator counts backwards from your wake-up time. The evidence above suggests the more valuable half is keeping that time steady, including on Saturday.
Open the sleep calculator