Sleep

Sleep is one of the highest-leverage longevity levers most adults take for granted. Cohort data covering tens of millions of person-years consistently show that short sleep (under about seven hours), long sleep (over about nine), irregular timing, and untreated sleep apnea each independently predict all-cause mortality, cardiovascular disease (CVD), dementia, type 2 diabetes, and depression. The American Heart Association formally added sleep to Life's Essential 8, its core cardiovascular-health metric, in 2022 — putting it on the same line as blood pressure and lipids.^[1] A UK Biobank analysis of 385,292 adults found that those with a healthy sleep pattern — early chronotype, 7–8 hours, no insomnia symptoms, no snoring, no daytime sleepiness — had roughly 35% lower CVD risk than those with the worst pattern.^[2]

Despite that centrality, sleep is also heavily meddled with and poorly understood. The 8-hour rule is approximately right; regularity matters at least as much as duration; and most over-the-counter sleep aids are weak, with a few that are actively harmful.

What the evidence actually supports

Strong (from randomised trials):

• Cognitive Behavioural Therapy for Insomnia (CBT-I) is the first-line treatment for chronic insomnia. It outperforms hypnotic drugs in head-to-head trials and effects persist after treatment ends.^[3] See Treating chronic insomnia.
• A 20–30 minute nap reliably improves vigilance, working memory, mood, and physical readiness, with a moderate-to-large pooled effect on cognitive performance.^[4] Expect a 10–20 minute lag after waking before performance peaks. See Daytime naps.

Strong, but observational — large, consistent cohorts, not trials, so these are associations:

• 7–8 hours of sleep on a regular schedule is associated with the lowest cardiovascular and all-cause mortality risk. The dose-response is U-shaped: against a 7–8 hour reference, sleeping under seven hours is associated with about 14% higher mortality and nine or more hours with about 34%.^[5] The long-sleep arm partly reflects underlying illness — depression, untreated apnea, frailty — rather than the sleep itself.
• Sleep regularity — going to bed and waking up at consistent times — predicts mortality at least as strongly as duration. A 2024 UK Biobank wrist-accelerometer study of 60,977 adults found 48% lower all-cause mortality in the most-regular fifth of sleepers compared with the least-regular — falling to 30% once socioeconomic position, lifestyle and existing health were accounted for.^[6] Irregular timing also tracked with 26% more major adverse cardiovascular events — and adequate total hours did not offset it.^[7] See Circadian rhythms.
• Severe untreated obstructive sleep apnea (OSA) is associated with roughly double the all-cause mortality and nearly triple the cardiovascular mortality, plus higher hypertension, atrial fibrillation, and stroke risk. The signal is concentrated in severe disease — mild and moderate OSA are not statistically distinguishable from no excess mortality, a distinction worth holding onto given how often "you have sleep apnea" is delivered as a single verdict. OSA is also badly underdiagnosed in midlife: an estimated 936 million adults aged 30–69 have mild-to-severe OSA globally — a figure from a ResMed-funded analysis, so treat its precision, not its direction, with caution.^[8] See Sleep-disordered breathing.
• Light at night. Wrist-light-sensor data from ~89,000 UK Biobank adults found the brightest-night group had 21–34% higher all-cause mortality than the darkest,^[9] and a separate analysis of the same cohort found 53% higher risk of incident type 2 diabetes — dose-dependent and independent of genetic risk.^[10] These come from a single accelerometry subsample, so confounding and reverse causation remain live; notably, outdoor satellite-measured night light lost its mortality association entirely once air pollution and noise were adjusted for. Personal bedroom light is not neighbourhood light. Even so, it is among the cheapest changes on this page.

Moderate:

• Light exposure — bright morning light plus dim, warm evening light — strengthens circadian alignment and improves both sleep and mood.
• Cool bedroom supports the natural core-temperature drop that initiates sleep. The often-quoted 18–20 °C (65–68 °F) is convention; the one home-monitoring study in older adults found sleep most efficient at 20–25 °C (68–77 °F), with wide individual variation.^[11]
• Melatonin, dosed to the purpose. 0.3–1 mg to shift the clock — jet lag, delayed sleep-phase syndrome, insomnia in older adults whose own melatonin has fallen — and up to 5 mg of the immediate-release form when the goal is simply falling asleep on arrival. Prolonged-release products are the wrong choice for jet lag: the Cochrane review found slow-release 2 mg relatively ineffective.^[12] Weak for ordinary insomnia, and best used strategically rather than as an open-ended nightly habit — see the long-term safety signal below.
• Ashwagandha (300–600 mg standardised root extract) for stress-driven sleep disruption, with rare hepatotoxicity reports attached.
• Habitual long naps as a warning sign. A daily 90-minute nap in a well-rested midlife adult tracks with cardiovascular disease, earlier death, and — more weakly — cognitive decline, though much of that signal is the nap acting as a marker of underlying illness rather than its cause.

Weak / preliminary:

• Magnesium glycinate (about 250 mg elemental — both the trial dose and the EU upper limit for supplements), L-theanine, and glycine all have small effects and excellent safety. The best magnesium trial found a real but small improvement in insomnia severity, with roughly one additional meaningful responder per seven people treated.^[13] None are first-line for chronic insomnia. See Sleep supplements.

Caution:

• Chronic use of benzodiazepines, Z-drugs, and over-the-counter antihistamines (Benadryl, ZzzQuil) is associated with falls, fractures, cognitive impairment, and dementia. The formal guidance — the American Geriatrics Society Beers Criteria — lists first-generation antihistamines from age 65; below that there is no threshold in the evidence, and the reason to skip them earlier is that they work poorly rather than that a cut-off exists.^[14] Pooled observational data put hip-fracture risk 90% higher with Z-drugs and 52% higher with benzodiazepines, and more than double in the first weeks of a new prescription.^[15] The dementia signal comes from anticholinergics specifically, and is observational — early dementia may itself drive both the sleep problems and the drug use.^[16]
• Alcohol as a sleep aid fragments sleep architecture — it suppresses rapid-eye-movement (REM) sleep and increases wakefulness in the second half of the night, even at low doses. It does reliably increase deep sleep in the first half,^[17] which is why a sleep tracker may flatter you after a nightcap — and a good illustration of why the deep-sleep number is not a health score.
• Orthosomnia — anxiously chasing tracker metrics — has a clinical name because it makes sleep worse.^[18]
• Night-shift work is classified by the World Health Organization's cancer agency as probably carcinogenic to humans (Group 2A). The human evidence behind that is graded limited — credible but not immune to confounding — and covers breast, prostate, colon and rectum; the sufficient evidence is in animals.^[19]

Why sleep is a longevity lever

Sleep is not passive downtime. It's the window in which the brain clears its metabolic waste, the body releases the bulk of its growth hormone, autonomic tone resets, and several of the molecular machines that track biological age are at their most active.

• Glymphatic clearance. The brain's perivascular drainage network — the glymphatic system — clears interstitial waste including amyloid-β and tau during slow-wave (deep) sleep. A 2026 randomised crossover study tried to measure this directly and its head-to-head comparison came out null: the overnight change in blood amyloid-β and tau was statistically indistinguishable after a normal night and after a sleepless one. The paper's positive headline is a model result built on the sponsor's own device output, not a measurement.^[20] Separately, a Framingham follow-up found each percentage-point yearly decline in slow-wave sleep was associated with about 27% higher dementia risk.^[21] That association is observational and should not be read as causal: the authors note that preclinical dementia may disrupt the mechanisms regulating deep sleep, and carrying the APOE ε4 risk gene predicted faster slow-wave decline — the disease may be eroding the sleep rather than the reverse. Sleep architecture covers both studies and their limits.
• Hormones. In men, roughly 70% of daily growth-hormone output occurs during early sleep, tied to the first slow-wave episode; cortisol reaches its nadir during early-night deep sleep.^[22] The endocrine cost of bad sleep is measurable: restricting ten healthy young men to five hours in bed for a week dropped their daytime testosterone by 10–15%, against a normal age-related decline of 1–2% per year — roughly a decade of aging in eight nights.^[23] See Testosterone therapy.
• Epigenetic aging. Short sleep and insomnia track with faster "epigenetic clocks" — chemical marks on DNA that estimate biological age. Across 3,795 older adults in the Health and Retirement Study, short sleepers ran about 1.3 years older than their birthdays on one such clock and people with insomnia about half a year older.^[24] Whether treating the sleep reverses any of this is a genuinely open question — that study is cross-sectional. The one randomised probe is suggestive rather than settled: in a secondary analysis of a CBT-I trial in 231 adults over 60, expression of a cellular-senescence marker rose over two years in the control group and held flat under CBT-I, falling only in those who achieved sustained remission of their insomnia.^[25]

The mechanistic upshot: bad sleep doesn't just feel bad. It plausibly accelerates the cellular processes that produce the diseases filling the last decade of an average adult's life — and the evidence that it does is stronger than the evidence that fixing it winds them back.

Sleep architecture: the stages

A normal night runs four to five 90–110-minute cycles through light non-rapid-eye-movement sleep (NREM stages N1 and N2), deep slow-wave sleep (N3), and REM — roughly 5% / 45% / 25% / 25% of the night. Deep sleep is front-loaded into the early cycles; REM lengthens toward morning. Deep sleep answers to how long you have been awake; REM answers to what time it is.

Two things are worth knowing before you read anything into a wearable's stage breakdown. Deep sleep is homeostatically defended — across a week of 6-hour nights, slow-wave activity barely moved and REM absorbed the loss, and after 40 hours awake it rebounded to 140–152% of baseline.^[26] And consumer devices mis-stage it badly: against a sleep-lab recording, current wrist devices run anywhere from 25 minutes under to 44 minutes over on deep sleep depending on the brand, so two of them on the same wrist on the same night can disagree by more than an hour.^[27]

Sleep architecture covers the stages, the normative percentages, how the pattern shifts with age, what each stage appears to do, and why deep sleep is a poor thing to chase.

Regularity and timing matter as much as duration

The most consequential recent shift in sleep epidemiology is the recognition that when you sleep matters at least as much as how long. The Sleep Regularity Index, measured by wrist accelerometry, captures the probability of being in the same sleep/wake state at any two points 24 hours apart. In the UK Biobank study above, the most-regular fifth also had 57% lower cardiometabolic mortality and 39% lower cancer mortality than the least-regular — and regularity outperformed duration as a mortality predictor in head-to-head models. A 2025 state-of-the-art review concluded that sleep irregularity is independently a robust risk factor for cardiometabolic disease.^[28] A separate UK Biobank analysis found irregular sleep timing predicted incident dementia over a decade of follow-up.^[29]

Clock time matters too, not just consistency. In 88,026 accelerometer-monitored adults, cardiovascular risk was lowest for sleep onset between 10 and 11 pm and rose in both directions — about 25% higher for midnight or later, about 24% higher for before 10 pm — independent of how long people slept.^[30]

Mechanistically, the master clock in the hypothalamus — a cluster of roughly 20,000 neurons in the mouse, somewhat larger in humans, firing in synchrony — coordinates cellular clocks running in virtually every tissue, gating DNA repair, autophagy, nutrient sensing, and the immune response. The system runs on a slightly-longer-than-24-hour period (~24.2 hours) and needs daily re-entrainment — by light, meal timing, and routine — to stay aligned.^[31] When central and peripheral clocks fall out of phase, the consequences spread across metabolism, immunity, and cardiovascular tone. Catching up on weekends partially restores subjective alertness but does not reverse the weekday metabolic cost.

The practical upshot is short: anchor your wake time first, get morning light, dim the evening, and eat earlier. Circadian rhythms covers the full picture — light dosing, chrononutrition, jet lag, and shift work — and the surprisingly small set of habits that move all of it at once.

Sleep-disordered breathing: the underdiagnosed cardiovascular risk

OSA is the repeated partial or complete collapse of the upper airway during sleep, and the majority of cases are undiagnosed. Severity is graded by the apnea-hypopnea index — the average number of complete or partial airway obstructions per hour of sleep — and the mortality signal is concentrated at the top of that scale. A 27-cohort meta-analysis covering 3,162,083 participants found severe OSA associated with roughly double the all-cause mortality (hazard ratio 2.13, 95% confidence interval 1.68–2.68) and nearly triple the cardiovascular mortality. That bracketed range is where the true effect most plausibly lies; when it crosses 1.0, the result is compatible with no effect at all — which is exactly what happened for mild OSA (1.19, 0.86–1.65) and moderate OSA (1.28, 0.96–1.69).^[32] The mortality case for treating severe OSA is strong; for mild OSA it is not made by this data — symptoms remain a good reason to treat it, but that is a different argument. In a separate cohort of 1,022 patients, OSA was associated with roughly double the risk of stroke or death from any cause after adjustment.^[33] The damage is thought to come not from the airway collapse itself but from the intermittent-hypoxia pattern of hundreds of nightly oxygen drops and re-oxygenations, an injury cycle that drives inflammation and blood-vessel dysfunction — though how much of OSA's clinical risk that explains, versus shared drivers like obesity, is unsettled.

Screening is cheap and the diagnostic pathway has shortened. The validated STOP-Bang questionnaire — eight yes/no questions on snoring, tiredness, observed apneas, hypertension, body mass index, age, neck size, and sex — flags high-risk adults; three or more "yes" answers warrants a sleep study.^[34] Home sleep apnea testing is now first-line for most adults; a full in-lab overnight recording (polysomnography, the reference standard) is reserved for complex cases. Treatment works when actually used: continuous positive airway pressure (CPAP) at four or more hours per night on at least 70% of nights reduces daytime sleepiness and blood pressure — substantially so in resistant hypertension — and crash rates fall alongside.^[35] Atrial fibrillation is the exception: treating apnea after ablation did not reduce recurrence in the randomised trial that tested it, 57% in both arms.^[36] Whether it cuts hard cardiovascular events is unresolved: the SAVE trial randomised 2,717 adults and found no reduction over 3.7 years (17.0% versus 15.4%), though mean use was only 3.3 hours a night in patients selected for minimal sleepiness, and the adherence-adjusted analyses that suggest benefit above four hours are post-hoc and non-randomised. The defensible reading is that CPAP's symptomatic benefits are well demonstrated and its cardiovascular benefit remains plausible but unproven. For obesity-driven OSA, a 10% reduction in body weight predicted a 26% fall in the apnea-hypopnea index in a long-running cohort.^[37] GLP-1 (glucagon-like peptide-1) receptor agonists are now a documented option: tirzepatide cut the apnea-hypopnea index by about 20 events per hour in adults with moderate-to-severe OSA and obesity not using CPAP, and 24 in those who were, over 52 weeks in SURMOUNT-OSA,^[38] — see Ozempic-class drugs.

Sleep-disordered breathing covers the full diagnostic pathway, the alternatives when CPAP isn't tolerated, the honest assessment of mouth taping, and orofacial myofunctional therapy.

Chronic insomnia: behaviour outperforms pills

Roughly 10–15% of adults have chronic insomnia (difficulty falling or staying asleep, three or more nights a week, for at least three months). The treatment landscape flipped in 2017: CBT-I is now first-line, ahead of any drug. It outperforms hypnotic medications in head-to-head trials and the effects persist after treatment ends — unlike drugs, which lose effect when discontinued.^[39] The five components — stimulus control, sleep restriction (the most powerful single element), cognitive restructuring, sleep hygiene, and relaxation — are now delivered at scale by validated digital programmes. Those are clearly effective and the right first move when a therapist isn't available, but only one head-to-head trial has tested them directly against in-person therapy, so treat them as a good substitute rather than a proven equal.

When drugs are genuinely needed (as a bridge, for short-term acute stress, or after CBT-I has failed), the safety hierarchy has shifted. Newer dual orexin-receptor antagonists (suvorexant, lemborexant, daridorexant) — which block orexin, the brain's wake-promoting signal, rather than sedating you — cut sleep-onset latency, raise total sleep time, and lower insomnia severity — all by small margins — without tolerance, withdrawal or rebound insomnia on stopping. They remain Schedule IV controlled substances, so "no dependence" overstates it.^[40] They are not free of trade-offs: across 11 randomised trials and 7,703 patients they roughly tripled the rate of sleep paralysis and roughly doubled excessive daytime sleepiness versus placebo — uncommon events, but a real and narcolepsy-like side-effect profile.^[41] Z-drugs (zolpidem and relatives) are reserved for short courses of four weeks or less; counter-intuitively, their hip-fracture signal is larger than that of benzodiazepines, not smaller. Benzodiazepines and over-the-counter antihistamines like diphenhydramine sit on the Beers Criteria for avoidance in older adults. A 2025 microsimulation estimated that if the 15.3 million Americans over 50 who take prescription sleep medication all stopped, lifetime falls would drop by 8.5%, cognitive impairment by 2.1%, and life expectancy would rise by about 0.11 years — roughly six weeks.^[42]

Treating chronic insomnia covers the full CBT-I protocol, the pharmacotherapy tier system, and the special populations — menopausal vasomotor insomnia, men who should be screened for OSA before pursuing testosterone, and older adults for whom the prescribing margin is narrow.

Daytime naps: opposite ends of the longevity ledger

The same behaviour at different durations sits on opposite ends of the ledger. A 20-minute power nap measurably restores vigilance, working memory, mood, and athletic readiness. A 90-minute habitual nap in a sedentary middle-aged adult points the other way: a 2026 umbrella review of 16 meta-analyses covering 244 outcomes found naps over 60 minutes associated with roughly 30% higher coronary-heart-disease risk and elevated diabetes, obesity, and all-cause mortality risk.^[43] Objective actigraphy from the Rush Memory and Aging Project found each additional hour of daily napping associated with about 13% higher mortality.^[44] A Mendelian-randomisation study using genetic predictors of napping found about 52% higher odds of coronary artery disease, holding up after adjustment for body mass index, smoking, and diabetes.^[45] Read that as suggestive rather than settled: the genetic instrument and the coronary outcome data overlap in the people they draw on, which biases the estimate toward the observed association, and the instrument captures how often someone naps rather than for how long.

The harm signal is conditional, not universal, and it begins at about 30 minutes — which is also where the cognitive benefit tops out, so the practical target is a nap that ends just under it. It concentrates in people who are already sleeping enough at night: in a 21-country cohort of 116,632 adults, the napping-related excess in cardiovascular events and death appeared in people sleeping more than six hours a night and was absent in those sleeping less — an absence of harm rather than a demonstrated benefit.^[46] A synthesis of 44 cohorts and more than 1.8 million people confined the cardiovascular and mortality harm to naps of 30 minutes or longer, and found napping associated with lower risk of cognitive impairment and of age-related muscle loss.^[47] The most useful contemporary framing: in older adults, excessive daytime napping is more likely a biomarker of decline than its cause. Habitual nappers carry slightly higher C-reactive protein than non-nappers, a difference that all but disappears once health and lifestyle are accounted for — inflammation makes you tired, and the nap is the body responding to a problem upstream, commonly undiagnosed OSA, cardiometabolic disease, or early neurodegeneration.^[48]

Daytime naps covers the dose-response in detail, the sleep-switch mechanism, athletic versus sedentary context, and when "creeping nap length" is a reason to investigate rather than indulge.

Sleep supplements: modest at best

Most "natural" sleep aids deliver small effects with few side effects, but only a couple have real evidence and several popular ones are oversold. The honest summary across magnesium, L-theanine, melatonin, ashwagandha, glycine, valerian, and CBD (cannabidiol): none replace CBT-I for chronic insomnia and none produce clinically meaningful tolerance, dependence, or withdrawal — though "no withdrawal" is not the same as "harmless for years," a distinction that now matters most for melatonin.

The two with the best risk-benefit profile are magnesium glycinate and L-theanine (200 mg), both with excellent safety and small, real effects. Magnesium's evidence is weaker than it is usually presented: the best trial found a statistically significant but small improvement in insomnia severity, largest in adults with low dietary intake, and the wider trial base is rated low-certainty.^[49] Keep supplemental magnesium near 250 mg elemental — that is both the trial dose and the EU upper limit for supplements. Low-dose melatonin (0.3–1 mg) earns its place for narrow indications — jet lag, delayed sleep-phase syndrome, insomnia in older adults — but the 2017 American Academy of Sleep Medicine guideline gave a conditional recommendation against it for general insomnia.^[50] Higher doses (3–10 mg) are not more effective and often produce next-day grogginess, and US over-the-counter products have been found to range from 83% below to 478% above their labelled dose.^[51] The "melatonin is harmless" framing has also softened: a preliminary 2025 analysis linked long-term daily use to higher heart-failure and mortality rates.^[52] That analysis is a conference abstract rather than a peer-reviewed paper, and it drew on prescription records — so it describes people whose insomnia was severe enough to be medicated for over a year, not someone taking an occasional low-dose tablet. It is still reason enough to use melatonin strategically rather than indefinitely. Ashwagandha carries rare hepatotoxicity reports; CBD interferes with the liver enzymes that clear many common prescriptions, including statins and warfarin.

Sleep supplements covers dosing, mechanisms, safety, and tier ranking for each, plus the honest answer to the addiction question.

Practical sleep checklist (evidence-weighted)

1. Anchor your wake time first. Same time every day, weekends included. The central nervous system adapts to consistent waking; it does not adapt to consistent sleep onset.
2. Get morning light (10–30 minutes outdoors within an hour of waking). Outdoor midday light delivers 50,000–100,000 lux versus 500–1,000 indoors — the order-of-magnitude gap is the point.
3. Dim the evening. Reduce overhead lighting and screens 1–2 hours before bed. Over a five-hour evening exposure, the average adult loses half their melatonin at about 25 lux — dimmer than most living rooms — and sensitivity varies more than fiftyfold between individuals.^[53] A 2022 expert consensus set targets in melanopic lux — light weighted to the receptors that actually drive the body clock: at least 250 by day, under 10 in the three hours before bed, under 1 while asleep.^[54]
4. Aim to be asleep between 10 and 11 pm where your schedule allows. The cardiovascular risk curve is U-shaped in clock time, not just in duration.
5. Cool, dark bedroom. Cool enough that core temperature can fall — the observed optimum in older adults is 20–25 °C (68–77 °F) and varies between people; blackout curtains; remove glowing electronics. A warm bath or shower (40–42.5 °C) about 90 minutes before bed exploits the same mechanism in reverse — skin vasodilation speeds the core-temperature drop and shortens sleep-onset latency.^[55]
6. Size the caffeine cut-off to the dose. A single cup (about 100 mg) had no measurable effect on sleep even four hours before bed, while 400 mg — a full day's allowance at once — altered sleep architecture taken within 12 hours of bedtime.^[56] A heavy total load wants a 10–12 hour buffer; one late-afternoon cup is defensible for most people. See Coffee.
7. No alcohol within three hours of bed. It hastens sleep onset but fragments architecture and suppresses REM at low doses.
8. Eat earlier. Last meal 2–3 hours before bed. Late, large meals delay peripheral clocks and degrade sleep architecture.
9. Screen yourself for OSA if you snore loudly, have witnessed apneas, hypertension, atrial fibrillation, or a body mass index of 30 or above — take the STOP-Bang questionnaire described above and ask for a sleep study on three or more "yes" answers.
10. For chronic insomnia, start with CBT-I, not pills. Digital programmes are effective and are the practical route when a therapist isn't available.
11. Cap daytime naps at 20–25 minutes and place them in the early-to-mid afternoon — under the 30-minute mark where the cohort harm signal begins. If you regularly need more, investigate the cause rather than indulge it.
12. Don't mouth-tape without a clinical evaluation. Treat nasal obstruction first.

What's overrated

• Catching up on weekends. Partial subjective recovery; the weekday metabolic deficits don't reverse. Sleeping in is not itself harmful, though — it just doesn't repair the weekdays.
• Blue-light-blocking glasses worn all evening as a sole intervention. Evidence is very low certainty; dimming the room matters more.^[57]
• High-dose evening melatonin (3–10 mg). Often exceeds the dose that maximises sleep onset and produces next-day grogginess, with no added benefit.
• Chasing a deep-sleep score. Deep sleep is homeostatically defended, wearables mis-measure it by up to three quarters of an hour, and the one intervention that reliably raises it is alcohol. Chasing the number is also its own harm — see orthosomnia above. See Sleep architecture.
• Mouth taping as a viral biohack. Defensible only after an ear-nose-throat (ENT) evaluation has cleared nasal obstruction; otherwise risk-asymmetric.
• Consumer apps and watches as a substitute for a sleep study. Machine-learning analysis of breathing sounds now reaches around 90% sensitivity in research settings, and some smartwatch apnea notifications have regulatory clearance. But the watch feature is tuned to be almost never wrong when it does alert, at the cost of detecting only about 43% of moderate apnea. A silent watch is not a negative test.^[58]
• Alcohol as a sleep aid. It hastens onset but suppresses REM and fragments the second half of the night.
• "Don't exercise in the evening." Mostly a myth: a 2021 Sleep Medicine Reviews meta-analysis found evening exercise does not disrupt sleep overall, with only acute high-intensity sessions ending within a few hours of bed slightly trimming REM.^[59] Vigorous exercise within about an hour of bed may still delay onset in sensitive people.

---

For the full ordered action list drawn from every pillar, see the healthspan long list.