Yes — and the effect looks larger than how long you sleep. In the two largest UK Biobank accelerometer studies on sleep timing, people with the most irregular sleep schedules had 53% higher all-cause mortality than the median (Sleep Regularity Index 5th percentile vs median: hazard ratio 1.53 (95% CI 1.41–1.66; n=88,975, 3,010 deaths, 7.1 years)). Separately, the most regular sleepers had 20–48% lower all-cause mortality than the least regular group at matched sleep duration (n=60,977, 1,859 deaths, mean 6.3 years).
What “regular” means in these studies
Both studies used the Sleep Regularity Index (SRI), computed from wrist-worn accelerometer data. The SRI scores 0–100 and captures how likely you are to be in the same sleep-or-wake state at the same clock time across days. A person with a perfectly consistent schedule scores 100; a person whose sleep drifts around the clock scores low.
In practical terms from the same literature: the most regular group typically varied their sleep window by under an hour night to night, while the least regular group varied by around three hours — the difference between "in bed by 11, up at 7 every day" and "11 on weekdays, 2 on Saturday, 7 on Sunday".
How the Sleep Regularity Index is computed
The SRI is built from seven days of wrist actigraphy — the same accelerometer data your Apple Watch already collects. For every pair of clock times across the week, the score asks one question: were you in the same state (asleep or awake) at both? The index is the percentage of pairs that match, on a 0–100 scale.
That design is why the measure is hard to game: it does not care how long you sleep, only whether your body is doing the same thing at the same hour each day. In the Cribb et al. sample the median participant scored 60 (SD 10); the 5th percentile sat at about 41 and the 95th at about 75 — the at-risk tail is roughly an hour of nightly drift, not some exotic schedule.
The two studies at a glance
| Cribb et al. 2023 | Windred et al. 2024 | |
|---|---|---|
| Journal | eLife | Sleep |
| Participants | 88,975 | 60,977 |
| Deaths recorded | 3,010 | 1,859 |
| Follow-up | 7.1 years | mean 6.30 ± 0.83 years |
| Headline | SRI 5th percentile: HR 1.53 (95% CI 1.41–1.66) vs median | Most vs least regular fifth: 20–48% lower all-cause mortality |
Both drew on UK Biobank wrist-accelerometer recordings — independent teams asking adjacent questions of the same biobank's objective sleep data, which is why the two findings reinforce each other rather than duplicate.
The numbers
| Outcome | Most vs least regular sleepers |
|---|---|
| All-cause mortality | 20–48% lower risk (most regular vs least regular, matched duration) |
| Cancer mortality | 16–39% lower risk |
| Cardiometabolic mortality | 22–57% lower risk |
Source for all three rows: Windred et al., Sleep (2024), 60,977 UK Biobank participants, 7-day accelerometer monitoring, 1,859 deaths over a mean 6.30 ± 0.83 years.
The second study looked at the same question from the other direction and found the same shape: participants in the 5th percentile of the Sleep Regularity Index (an SRI of about 41) had a 53% higher hazard of all-cause death than the median group (hazard ratio 1.53, 95% CI 1.41–1.66; n=88,975, 3,010 deaths, 7.1 years), while the 95th percentile (SRI 75) sat 10% below the median. The median SRI in that sample was 60.
Regularity beats duration
This is the part most people get backwards. In the same analysis, when the researchers added sleep duration on top of regularity, the extra hours did not meaningfully change the association (p=0.14–0.20 across models). The signal was coming from the consistency, not the length of the night.
For context, a separate meta-analysis of 1.38 million people found that short sleep carried a relative risk of about 1.12 and long sleep about 1.30 for all-cause mortality — real effects, but smaller and less consistent than the regularity gap above.
What these studies can and can't tell you
Both papers are observational: they measured sleep timing once and followed people for years, so they show association, not proof that changing your schedule changes your risk. The authors say so themselves, and so should any page summarising them.
Three further caveats worth knowing. UK Biobank participants were predominantly older, white European adults who volunteered — healthier on average than the general population. The SRI came from a single 7-day window: a snapshot of a typical week, not a lifelong pattern. And a wrist accelerometer infers sleep from movement, which is good enough for timing studies but is not a sleep lab.
None of that erases the signal — two large, independent, objective-measurement studies pointing the same way is about as strong as observational sleep evidence gets — but it is why the honest summary is "consistency is associated with lower risk", not "an irregular schedule will shorten your life".
What actually moves the needle
- 1Anchor your wake time, not your bedtime. The single change with the strongest clinical support is a consistent wake-up time. This is also a core component of stimulus-control therapy, the first-line behavioural treatment recommended for chronic difficulty falling or staying asleep by medical societies.
- 2Keep the weekend within an hour of the weekday. "Social jetlag" — sleeping in by two or more hours on days off — is exactly the pattern the SRI measures as irregular.
- 3Let a late night be a late night, not a late morning. When you do go to bed late, still get up at your usual time. Chasing the missed sleep the next morning is what creates the drift.
Your Apple Watch already logs the raw material for this — bedtime and wake time every night. Comparing the last 7 to 14 days against your own median tells you which pattern you actually run, no extra hardware required.