Almost every sleep tracker shows you the same two numbers: how long you slept, and some flavor of sleep quality score. Both are useful. Both are also incomplete. A May 2026 prospective cohort study from the Northern Finland Birth Cohort 1966 — the third in a recent string of large analyses on sleep regularity — adds a third variable that most wearables compute internally but rarely surface clearly: the standard deviation of your bedtime, and the standard deviation of your sleep midpoint, from one night to the next.
The Finnish team followed 3,231 adults for 10 years, tracking major adverse cardiac events and cardiovascular mortality. Their finding: in participants whose total sleep was below the cohort median of 7 hours 56 minutes, irregular bedtimes and irregular sleep midpoints were associated with approximately twice the rate of major adverse cardiac events compared with regular bedtimes and midpoints — even after adjusting for total sleep duration, age, BMI, smoking, and the usual cardiovascular suspects.
Wake-up time variability did not show the same effect. The signal lived in when sleep started and where its midpoint fell, not where it ended.
That is the kind of result that is interesting to read about and more interesting to test in your own data.
The Study
Pesonen et al. published the analysis in BMC Cardiovascular Disorders in May 2026 (doi:10.1186/s12872-026-05762-4). The cohort: 3,231 participants from the Northern Finland Birth Cohort 1966, all the same age (46 at baseline assessment in 2012–2014), followed through 2024 for 10 years of incident cardiac events.
What separated this study from prior sleep-regularity work is the methodology: each participant wore a research-grade actigraph for a week, and the researchers computed the standard deviations of three timing variables — bedtime, wake-up time, and the sleep midpoint (the clock-time halfway between sleep onset and offset). They then split each variable into regular and irregular categories based on the within-cohort median.
The headline associations:
- Irregular bedtime: roughly 2x risk of major adverse cardiac events in the under-8-hour sleep subgroup.
- Irregular sleep midpoint: similar magnitude of risk amplification.
- Irregular wake-up time: no independent association with hard cardiac endpoints.
- Effects largely disappeared in participants sleeping ≥ 7h 56min on average.
Caveats matter here. The study is observational, not randomized — it shows correlation. The cohort is a homogeneous Finnish population born in 1966, which limits generalizability. Shift workers, parents of newborns, and people with sleep disorders may have schedules driven by circumstances rather than choice, and chronotype differences make a simple "go to bed earlier" prescription unrealistic. The data is suggestive, not deterministic.
That said, the Finnish 2026 result lines up with the 2024 Windred et al. analysis in SLEEP (UK Biobank, n=60,977) that established the Sleep Regularity Index as a predictor stronger than sleep duration for all-cause, cancer, and cardiometabolic mortality. The two studies attack the same question with different cohorts and converge on the same conclusion: regularity is a separate lever from duration.
What This Means If You Already Use a Wearable
The math you need is sitting in your phone or wrist device. Apple Health, Oura, Whoop, Garmin, and Fitbit all log sleep onset and offset timestamps. Some apps surface a "consistency" or "regularity" score derived from those timestamps. Most do not — and the ones that do rarely show you the underlying standard deviations.
The variable the Finnish study singled out is your sleep midpoint variability: the standard deviation, in minutes, of the midpoint clock time across your week. A few reference points:
- Under 30 minutes: high-regularity territory. Your circadian system is receiving consistent cues from one night to the next.
- 30 to 60 minutes: moderate variability. Likely the modal range for most knowledge-worker schedules.
- Over 60 minutes: approaching the lower-regularity range of the Windred 2024 high-mortality quintile.
A simpler proxy, if midpoint feels abstract, is the standard deviation of your sleep onset time alone. The Finnish data found bedtime variability — not wake-up variability — was where the cardiac risk lived, so bedtime SD is a defensible single number to track.
Be the first to try Prova
We're building an app to track whether sleep midpoint variability actually works. Join the waitlist.
A 30-Day Sleep Midpoint Variability Experiment
The point of this experiment is not to demonstrate that the Finnish cohort result reproduces in your individual data — it cannot, with n=1. The point is to see whether deliberately reducing your bedtime and midpoint variability correlates with measurable shifts in the recovery markers you can actually track.
Setup
Before you start, gather:
- A wearable that logs bedtime and wake time accurately (Oura, Whoop, Apple Watch with a third-party sleep app, Garmin, or Fitbit).
- A way to capture morning resting heart rate and heart rate variability (most of the above do this on the wrist).
- A subjective recovery or readiness score from your wearable, or a self-reported 1–10 morning energy rating logged in Prova.
Phase 1 (Weeks 1–2): Baseline Variability
For two weeks, do not change your sleep timing at all. Live your normal schedule. Each morning, in Prova or a notes app, log:
- Bedtime (sleep onset per your wearable)
- Wake time
- Total sleep in hours and minutes
- Resting heart rate (overnight average)
- HRV (overnight average)
- Recovery / readiness score from your wearable, or a 1–10 morning energy rating
At the end of the two weeks, compute:
- Mean bedtime
- Standard deviation of bedtime in minutes
- Mean sleep midpoint (the clock-time halfway between onset and wake)
- Standard deviation of midpoint in minutes
- Mean total sleep duration
This is your baseline. The point is to find out what your variability actually is — most people overestimate how regular they are.
Phase 2 (Weeks 3–4): Bedtime Anchoring
For the next two weeks, set a target bedtime range of 30 minutes centered on your Phase 1 mean bedtime. (If your average bedtime was 11:15 PM, target 11:00–11:30 PM.) Do not target a fixed wake-up time — the Finnish data suggests bedtime is the lever that matters.
Continue logging the same six variables each morning. Track adherence: count nights where bedtime fell inside the 30-minute window.
At the end of Phase 2, compute:
- Standard deviation of bedtime (should drop substantially if adherence was high)
- Standard deviation of sleep midpoint
- Mean RHR
- Mean HRV
- Mean recovery / readiness score
- Mean total sleep duration
What to Look For
The Finnish data predicts that reducing bedtime SD should be associated with shifts in cardiovascular autonomic markers — lower RHR, higher HRV, possibly better subjective recovery — independent of any change in total sleep time. In your individual data, look for:
- Did bedtime SD actually drop? (If you cannot constrain your schedule, the experiment did not run — note that and stop here.)
- Did total sleep duration stay roughly the same? You want to isolate the timing variable from the duration variable.
- Did RHR or HRV trend in the expected direction (lower RHR, higher HRV)?
- Did your subjective morning energy rating change?
The pre-specified hypothesis: bedtime SD will drop in Phase 2, and that drop will correlate with a small but visible improvement in HRV or recovery score — even if total sleep stays constant. Write down your hypothesis before Phase 2 starts. Going in with a prediction protects you from post-hoc storytelling about the data.
Common Confounders
A few things will muddy the data and are worth tracking as covariates:
- Alcohol in the 4 hours before bed reliably drops HRV and raises RHR, often more than bedtime variability does.
- Hard training within 6 hours of bed shifts HRV and RHR for 24–48 hours.
- Late caffeine shifts sleep onset latency and can push midpoint later.
- Sickness, travel, or jet lag dwarf the regularity signal — exclude affected nights from the variability calculation rather than letting them pull the mean around.
Log these in Prova during both phases so you can mark and filter them when comparing the two-week blocks.
What This Experiment Will Not Tell You
The experiment is honest only about what it can answer. It cannot tell you:
- Whether reducing bedtime SD causes better cardiovascular outcomes. The 10-year endpoints in the Finnish study cannot be replicated in 30 days.
- Whether you specifically are in the population subgroup where the effect operates (under-8-hour sleepers showed the signal; ≥ 8-hour sleepers did not).
- Whether the regularity effect is mediated by circadian alignment, glucose handling, blood pressure dipping, inflammation, or something else.
- Whether your wearable's bedtime detection is accurate enough at the minute level to detect a small effect.
What the experiment can tell you is whether bedtime regularity is something you can control in your current life, and whether tightening it produces detectable shifts in the autonomic markers you already track. That is a useful piece of personal information regardless of what the long-term cardiovascular literature eventually concludes.
The Bigger Picture
The shift the 2024–2026 literature is pointing toward is uncomfortable for the standard sleep-duration framing. For two decades, the message has been "get 7 to 9 hours." That message was not wrong, but it may have been incomplete. When you sleep, week to week — measured as the variability of your bedtime and sleep midpoint — appears to be a separable axis that the duration-only framing missed.
For self-trackers, this is good news. Bedtime regularity is a variable you can measure with the wearable you already own, manipulate without supplements or dietary changes, and test in your own data. The Finnish study cannot answer for you individually. The 30-day experiment described above can at least tell you whether tightening the lever moves the markers you care about. From there, what you do with that information is yours to decide.