Regular Sleep and Cardiovascular Events

A large prospective UK Biobank analysis linked less regular accelerometer-measured sleep with more later major cardiovascular events. The finding adds a timing pattern to the research conversation, but the cohort can show association, not that changing a sleep score prevents a heart attack or stroke.

What the UK Biobank Analysis Measured

Chaput and colleagues studied adults aged 40–79 in the UK Biobank accelerometer substudy. Participants wore a wrist-attached device for seven days; the investigators computed an SRI using a validated algorithm and classified the cohort as regular (above 87.3), moderately irregular (71.6–87.3), or irregular (below 71.6). These cut points reflected this sample’s distribution. They are not a clinical scale with a known boundary between safe and unsafe sleep.

The analysis included 72,269 people free of a prior MACE and without an event in the first year after measurement. Over an average 7.8 years, 4,887 participants had an event. Outcomes came from hospital and death records. MACE was a composite of heart failure, myocardial infarction, and stroke; its components were also considered. Following people forward and using records makes the temporal ordering clearer than a one-time survey, but it does not turn the comparison into a randomized test.

Against the regular category, the adjusted hazard ratio was 1.08 for moderately irregular sleep and 1.26 for irregular sleep. The second estimate’s confidence interval was 1.16–1.37. For the moderate category, the abstract reports a point estimate of 1.08 and a wide interval; it is better not to use it as a precise personal risk estimate. A hazard ratio compares event rates over follow-up conditional on the model—it is not the chance that any one person will experience an event.

Why a Prospective Association Matters

Sleep regularity has sometimes been inferred from self-report or assessed alongside metabolic markers at one point in time. This study had a different strength: device-derived sleep patterns came before the recorded events. The investigators also excluded people with known prior MACE and events in the first year, which reduces (but cannot eliminate) the possibility that an imminent or existing illness changed sleep before the event was recorded.

They adjusted for a broad set of measured characteristics, including demographics, deprivation, activity, diet-related behaviors, smoking, alcohol, medication, mental health, family history, shift work, and reported sleep problems. Adjustment helps compare groups that are more similar on observed factors. It does not guarantee that the groups are exchangeable. Unmeasured illness, occupational demands, access to care, neighborhood conditions, and measurement error may still shape both schedules and event risk.

The outcome also deserves care. A composite can capture more events and provide power, but heart failure, heart attack, and stroke have different pathways. The signal should not be translated into a claim about one specific event without looking at its component analysis and uncertainty. The cohort supports an association with later cardiovascular outcomes, not a mechanism proven in these participants.

Reading the Relative Estimates

The reported hazard ratio of 1.26 means that the irregular group had a 26% higher model-estimated hazard than the regular reference group over follow-up, under the study’s adjustments. It does not mean an extra 26 percentage points of absolute risk. Absolute event probability depends on age, existing cardiovascular disease, blood pressure, lipids, smoking, diabetes, and many other factors. The study’s display of 4,887 events across the cohort is not a prediction for a reader with a similar SRI.

Nor does the difference show what would happen if someone moved from one category to another. The reference group is a selected comparison, and participants were not assigned to improve their regularity. People with steadier sleep may also have more predictable shifts, caregiving responsibilities, housing, or health. This is a classic healthy-schedule problem: an exposure can carry information about life conditions as well as about sleep timing itself.

Researchers reported a near-linear association when they treated SRI continuously, rather than relying only on three categories. That makes the pattern less dependent on one arbitrary cutoff, but it still does not reveal a treatment dose, a threshold, or the amount of risk change achievable by routine adjustment. Continuous associations are still observational associations.

Duration Does Not Erase the Regularity Question

The same research group examined sleep duration alongside SRI. Meeting age-specific sleep-duration recommendations did not fully remove the association seen in the irregular group; in a joint analysis, the irregular-plus-recommended-duration group retained an elevated estimated hazard relative to regular sleepers. This helps separate two dimensions: adequate hours do not necessarily imply steady timing.

It does not follow that duration is optional, or that someone should shorten sleep to keep a clock pattern tidy. Both sleep amount and pattern can matter, and the joint model is not an intervention trial. It asks how measured groups compared; it cannot determine which practical change would improve an individual’s health. The linked regularity-versus-duration page discusses that distinction without treating the two dimensions as competing goals.

Timing questions already have a home on this site. The social-jetlag page explains weekday-to-free-day midsleep gaps; this page focuses on the prospective device cohort and cardiovascular endpoints. For recommended bedtime ranges and existing timing coverage, see the sleep-timing guide.

Where the Cohort May Not Travel

UK Biobank participants are volunteers from the United Kingdom, and the wearable substudy represents a subset who completed a week of measurement. Participants’ mean age was around 62, so findings may not transfer directly to adolescents, younger working adults, people in other countries, or workers with very different schedules. Device algorithms can also classify quiet wake as sleep or movement during sleep as wake, and a seven-day sample has limited ability to characterize a changing rota.

Cut points near the 25th and 75th percentiles label relative parts of one distribution; they do not indicate a biologically validated threshold. A participant’s SRI can also shift with nap handling, missing data, and the algorithm’s definition of state. Reproducibility across cohorts and devices matters before any cutoff becomes a portable clinical marker.

Most importantly, risk associations need not identify a modifiable cause. Irregular schedules can be imposed by night shifts, caregiving, unstable employment, or illness. Treating a person’s schedule as a freely chosen behavior obscures those constraints and overstates what a tracker or health message can achieve.

What a Sensible Response Looks Like

For an individual, a wearable trend can be a prompt to ask whether workdays and free days differ, whether sleep is sufficient, and whether symptoms warrant medical assessment. It is not a cardiovascular screening test. Major established risk factors should be assessed through routine care, not inferred from SRI. People with chest pain, sudden breathlessness, facial droop, or one-sided weakness need urgent emergency evaluation; do not wait for a sleep metric.

For researchers, the next steps include longer monitoring windows, repeat measurement, diverse populations, transparent scoring algorithms, and tests of whether schedule-focused interventions change intermediate measures and eventually meaningful clinical outcomes. Studies should report absolute event rates as well as relative estimates and distinguish bedtime consistency, duration variability, and full sleep–wake regularity.

The result earns attention because a large prospective sample used objective monitoring before events and found a graded association. The appropriate inference stops there: sleep regularity may be a cardiovascular risk marker worth studying. Whether deliberately improving it prevents events, how much change matters, and for whom remain open questions.

MACE is a composite label, not one disease. In this cohort it encompassed nonfatal or fatal myocardial infarction, heart failure, and stroke identified in records. Combining endpoints improves the number of events available for analysis, while each condition can have distinct causes and baseline rates. The paper recorded 4,887 MACE events among the analytic participants, but the relative hazard does not mean every component had an identical association. Readers should avoid transferring the overall estimate to a specific diagnosis or interpreting a composite as a uniform clinical pathway.

Selection of the analysis also deserves attention. Participants had no prior MACE and were required to remain event-free during the first follow-up year. That makes the result relevant to new events in a primary-prevention-like cohort, not necessarily to people already living with cardiovascular disease. The early-event exclusion can reduce reverse causation, since deteriorating health may disrupt sleep, but it also changes who is counted and cannot remove long preclinical disease. For someone with established disease, clinician-directed risk management—not this cohort’s SRI categories—remains the appropriate framework.

Related Topics