Sleep Duration

Abstract

Sleep duration, which MeSH classifies under sleep, is the total time a person spends asleep, measured per night or per 24-hour day. It is the most studied of the quantitative sleep parameters because it is simple to state, varies systematically with age, is partly heritable, and tracks health outcomes in a way that is now measured in millions of person-years of follow-up. This article defines sleep duration and separates it from the quantities it is often confused with, surveys how it is measured and why self-report and objective estimates diverge, sets out the normative values across the lifespan and the expert recommendations built on them, explains how the two-process model regulates how long a person sleeps, reviews the genetics of natural short sleep, and lays out the U-shaped association between sleep duration and mortality that anchors its public-health importance.

Keywords: sleep duration, sleep need, two-process model, sleep restriction, mortality

Of all the numbers that describe a night's sleep, the plainest is how long it lasted. Sleep duration is the total time asleep, and it is the variable a person means when they say they got six hours or eight. Its very simplicity is what makes it powerful: unlike the fine architecture of sleep stages, duration can be asked about in a single survey question, recorded by a wristband, and pooled across hundreds of thousands of people, and when it is, it turns out to predict obesity, diabetes, cardiovascular disease, and death (Cappuccio, D'Elia, Strazzullo, & Miller, 2010; Grandner, Hale, Moore, & Patel, 2010). The same plainness hides real complexity, because how long a person sleeps is set by two biological clocks, drifts with age, differs by genotype, and is reported with a systematic error that measurement studies have had to untangle.

What Sleep Duration Is

Sleep duration is the total amount of time spent in the state of sleep, counted either over a single main sleep period or summed across a 24-hour day to include naps. Stated that simply, it is one number, but it is easy to confuse with three neighbouring quantities that it is not. It is not time in bed, which includes the periods of wakefulness before, during, and after sleep; the difference between time in bed and sleep duration, expressed as a fraction, is sleep efficiency. It is not sleep latency, the time taken to fall asleep after lights out. And it is not sleep quality, the subjective and architectural sense of how restorative the sleep was, which can be poor even when duration is ample. Sleep duration is purely a quantity of time, and the other parameters describe how that time is entered, how continuous it is, and how good it feels.

The reason duration is singled out among the sleep parameters is that it behaves like a dose. A great deal of research treats the number of hours slept as an exposure and asks what varying it does—to alertness the next day, to metabolism, to long-term disease risk—and the resulting dose-response relationships are the empirical core of the field (Van Dongen, Maislin, Mullington, & Dinges, 2003). The sections that follow build the subject up in that spirit: first how the quantity is measured, then how much of it people of different ages actually get and are advised to get, then the biological control system that sets it, its genetic basis, and finally its consequences for health.

Measuring Sleep Duration

How long someone slept seems like it should be easy to establish, but every method of measuring it trades accuracy against practicality. The reference standard is polysomnography, which records brain, eye, and muscle activity overnight and scores sleep second by second, yielding an exact duration; it is also expensive, is usually confined to a single night in a laboratory, and may itself disturb the sleep it measures. At the other extreme is the self-report question—a respondent is simply asked how many hours they usually sleep—which costs nothing and scales to whole populations, and which is therefore the basis of almost every large epidemiological study of sleep duration. Between the two sits actigraphy, a wrist-worn movement sensor that estimates sleep from stillness over many nights in the home.

The trouble is that these methods do not agree. Lauderdale and colleagues compared self-reported habitual sleep against actigraphy in the same people and found that self-report overstated sleep duration by about 0.8 hours on average, with wide scatter between individuals, so that the rank order of who sleeps most is only loosely preserved (Lauderdale, Knutson, Yan, Liu, & Rathouz, 2008). People, in short, believe they sleep longer than they do, and by an amount that is neither small nor constant. This matters directly for the health literature: a study that defines short sleep as a self-reported six hours is capturing people whose true duration may be well under that, and comparisons across studies that used different instruments have to be read with that bias in mind.

How Much Sleep People Get, and Are Advised to Get

The single largest determinant of sleep duration is age. Ohayon, Carskadon, Guilleminault, and Vitiello assembled quantitative sleep parameters from healthy people across the whole lifespan and produced the normative values against which any individual's sleep is now judged: total sleep time falls steeply through childhood and adolescence and then declines gently across adult life, so that the hours that are normal for a teenager would be excessive for an older adult (Ohayon, Carskadon, Guilleminault, & Vitiello, 2004). Normative, here, means typical and healthy—what people of a given age actually sleep when nothing is wrong—and it is the empirical foundation on which recommendations are built.

Those recommendations are the second layer. The National Sleep Foundation convened an expert panel that translated the normative evidence into recommended ranges by age, settling on 7-9 hours for young and middle-aged adults and a slightly narrower 7-8 hours for older adults (Hirshkowitz et al., 2015). The American Academy of Sleep Medicine and the Sleep Research Society, working jointly, reached the compatible conclusion that adults should sleep 7 or more hours per night on a regular basis to avoid the health risks of insufficient sleep (Watson et al., 2015). Table 1 gives the National Sleep Foundation's recommended ranges, and the first demonstration lets the reader move through the age bands and read off the recommendation.

Chaput, Dutil, and Sampasa-Kanyinga reviewed how the ideal number of hours shifts with age and stressed that the recommendations are population ranges, not individual prescriptions: within any age band there is a spread of healthy sleep needs, and a figure outside the recommended range is a reason to look more closely, not a diagnosis (Chaput, Dutil, & Sampasa-Kanyinga, 2018).

The Regulation of Sleep Duration

How long a person sleeps on a given night is not chosen; it is set by a control system. The dominant account is the two-process model, in which sleep is regulated by the interaction of a homeostatic process and a circadian process (Borbely, Daan, Wirz-Justice, & Deboer, 2016). The homeostatic process, called Process S, is a sleep pressure that accumulates during wakefulness in proportion to how long one has been awake and dissipates during sleep; the longer the prior waking, the higher the pressure and the more sleep is needed to discharge it. The circadian process, Process C, is a roughly 24-hour oscillation driven by the internal clock that sets a time-of-day window in which sleep is permitted and a window in which it is opposed, independent of how tired one is.

Figure 1

The two-process model across a 24-hour day.

Sleep duration emerges from the interplay of the two. Sleep begins when accumulated pressure meets a falling circadian threshold and ends when the dissipating pressure crosses a rising one, so the length of a sleep episode depends both on how much pressure was built up beforehand and on when, in the circadian cycle, sleep is taken. This is why a nap in the early evening is short and unsatisfying though pressure is high—the circadian signal opposes sleep then—and why sleeping in after a late night does not extend proportionally to the lost hours. The second demonstration traces the exponential dissipation of Process S across a night and shows how a curtailed sleep leaves residual pressure as a carry-over debt.

The homeostatic side of the model is what the study of sleep deprivation probes directly. When sleep is restricted night after night, the pressure that normal sleep would have discharged instead accumulates, and Van Dongen and colleagues showed that the resulting neurobehavioral deficits grow in a dose-dependent way with the size of the accumulated debt—a chronic six hours in bed produced deficits that mounted steadily across two weeks, to a degree the sleepers themselves badly underestimated (Van Dongen et al., 2003). Sleep duration, in this light, is not a preference but the output of a pressure that must eventually be paid.

The Genetics of Short Sleep

If sleep duration were set entirely by the environment and by sleep pressure, everyone kept under identical conditions would need the same amount. They do not, and the clearest evidence comes from rare families of natural short sleepers—people who sleep far less than average, feel fully rested, and show none of the deficits that curtailed sleep inflicts on others. He and colleagues identified a mutation in the gene DEC2 (also called BHLHE41) that segregates with a natural short-sleep phenotype of around six hours, and showed that engineering the same mutation into mice reduced their sleep too, the first demonstration that a single gene can set sleep need (He et al., 2009).

A decade later Shi and colleagues extended the picture with a mutation in the gene for the beta-1 adrenergic receptor (ADRB1), again found in a short-sleeping family and again reproduced in mice, where the altered receptor changed the activity of brainstem neurons that regulate sleep and wakefulness (Shi et al., 2019). Together the findings establish that habitual sleep duration has a specific, heritable genetic component, and that the natural short sleepers are not simply chronically sleep-deprived people who have adapted but a biologically distinct group whose sleep need is genuinely lower. The genetics therefore caution against reading any single recommended number as right for everyone.

Sleep Duration and Health

The reason sleep duration draws the attention of epidemiologists is that both too little and too much of it are associated with worse health. Cappuccio and colleagues pooled prospective studies following healthy people over years and found that short sleep duration predicted higher all-cause mortality, and that long sleep duration predicted it too, so that the risk plotted against hours slept forms a U, lowest in the middle and rising toward both ends (Cappuccio et al., 2010). The same group showed that short sleep was associated with obesity in both children and adults (Cappuccio, Taggart, et al., 2008) and with the incidence of type 2 diabetes (Cappuccio, D'Elia, et al., 2010), and Yin and colleagues, pooling still more cohorts, traced a dose-response curve for cardiovascular events and mortality with the same U-shape and a nadir near seven hours (Yin et al., 2017). The third demonstration lets the reader move across the range of nightly hours and read the associated relative risk off the U-shaped curve.

The U-shape raises an interpretive problem the field takes seriously. That short sleep harms is mechanistically plausible—Grandner and colleagues laid out the candidate pathways, from metabolic and endocrine disruption to inflammation and raised blood pressure (Grandner et al., 2010). But that long sleep harms is harder to credit as a direct effect, and the leading reading is reverse causation: long sleep is less a cause of disease than a symptom of it, so that illness lengthens sleep rather than the reverse. The association of long sleep with mortality may therefore mark sickness already present, while the short-sleep arm is more likely to be causal—a distinction that matters for whether lengthening or shortening anyone's sleep would actually change their risk.

Worked Example

The homeostatic side of the two-process model can be worked numerically, and the second demonstration reproduces the arithmetic. Represent the homeostatic sleep pressure, Process S, as a quantity that decays exponentially during sleep from its value at sleep onset toward a floor, so that after t hours of sleep the fraction of the onset pressure still remaining is e raised to the power of negative t divided by τ, where τ is the decay time constant. A representative value for the decay of Process S during sleep is about 4.2 hours.

Put numbers to it. Normalize the sleep pressure at sleep onset to 1. After a full eight-hour night, the residual pressure is e raised to the power of negative 8 divided by 4.2, that is e to the power of negative 1.905, which is about 0.149: a full night discharges roughly 85 percent of the accumulated pressure, leaving only 15 percent. Now curtail the night to six hours. The residual is e to the power of negative 6 divided by 4.2, that is e to the power of negative 1.429, which is about 0.240—so a six-hour night leaves 24 percent of the pressure undischarged, over half as much again as the eight-hour night leaves. Cut sleep to four hours and the residual is e to the power of negative 4 divided by 4.2, about 0.386, nearly 39 percent left unpaid.

The lesson of the arithmetic is that the relationship between lost sleep and residual pressure is not linear. Because the discharge is exponential, the last hours of a normal night remove only a little pressure, while the first hours remove a great deal; conversely, the hours cut from the end of a shortened night leave a disproportionately large residual, and that residual carries forward. Repeated night after night, as in the Van Dongen restriction study, the small nightly surplus compounds into the accumulating debt that drives the mounting daytime deficits (Van Dongen et al., 2003).

Discussion

Sleep duration occupies an unusual position among the constructs of sleep science: it is at once the crudest measure and the most consequential. It throws away everything about the structure of sleep—the cycling of stages, the timing within the night, the depth—and keeps only a scalar, and yet that scalar carries most of the weight in the epidemiology, because it is the one quantity that can be gathered at the scale of whole populations (Cappuccio et al., 2010; Yin et al., 2017). The tension between crudeness and consequence runs through the whole subject. The measurement literature shows that even this one simple number is reported with a large, systematic error (Lauderdale et al., 2008), and the genetics show that the single recommended figure masks real biological variation in how much sleep different people need (He et al., 2009; Shi et al., 2019).

What keeps sleep duration from collapsing into a mere number is that it is the readout of a well-characterized control system. The two-process model explains why duration is defended, why it cannot be banked or borrowed freely, and why its curtailment produces a debt that compounds (Borbely et al., 2016; Van Dongen et al., 2003). Duration is the behavior; the homeostatic and circadian processes are its cause. The practical upshot—sleep around seven to nine hours as an adult, judged against one's age and measured honestly—rests on that chain of reasoning, from the control system that sets duration, through the normative values that describe it, to the mortality curves that give it its public-health stakes.

Current Directions

The most active recent work has been genetic. The discovery that mutations in DEC2 and the beta-1 adrenergic receptor produce natural short sleepers has turned sleep need from a soft, environmental quantity into a trait with identifiable molecular causes, and the search is now on for the fuller set of genes and neural circuits that set how long a person sleeps (Shi et al., 2019). Large genome-wide association studies of self-reported and accelerometer-measured sleep duration are mapping the common variation that underlies the normal range, complementing the rare high-effect mutations found in short-sleeping families.

A second front is methodological and concerns exactly the measurement problem this article has stressed. As wearable actigraphy and consumer sleep trackers put objective, multi-night duration estimates within reach of population studies, the field is beginning to replace the self-reported hours that anchor the classic mortality meta-analyses with measured ones, which both sharpens the exposure and tests how much the known self-report bias has distorted the dose-response curves (Lauderdale et al., 2008). Whether the U-shaped mortality association survives when duration is measured rather than reported—and in particular whether the long-sleep arm is confirmed or revealed as a reporting and reverse-causation artefact—is among the open questions the better measurement is meant to settle.

Common Misconceptions

Everyone needs eight hours of sleep.

Eight hours sits inside the recommended adult range, but the range is 7-9 hours and it shifts with age, and genuine individual variation—including genetic short sleepers—means no single figure is right for everyone (Hirshkowitz et al., 2015; He et al., 2009).

The more sleep, the better.

Mortality plotted against sleep duration is U-shaped: long sleep is associated with worse outcomes as well as short sleep, though the long-sleep arm is widely read as a marker of existing illness rather than a cause (Cappuccio et al., 2010).

Reported sleep equals actual sleep.

Self-report overestimates measured sleep by about 0.8 hours on average, with large individual scatter, so reported and actual duration can differ substantially (Lauderdale et al., 2008).

Lost sleep can be fully repaid by sleeping in.

Because sleep is gated by the circadian clock as well as by homeostatic pressure, recovery sleep does not extend in simple proportion to the hours lost, and chronic restriction builds a debt that accumulates faster than it is repaid (Borbely et al., 2016; Van Dongen et al., 2003).

Glossary

Actigraphy.

The estimation of sleep and wakefulness from a wrist-worn movement sensor worn over many nights, intermediate in accuracy between self-report and polysomnography.

Circadian process (Process C).

The roughly 24-hour oscillation, driven by the internal clock, that sets the time-of-day windows in which sleep is promoted or opposed, independent of accumulated sleep pressure.

Homeostatic process (Process S).

The sleep pressure that accumulates during waking in proportion to time awake and dissipates during sleep; one of the two processes that set sleep duration.

Natural short sleeper.

A person who habitually sleeps far less than average without ill effect, in known cases because of a specific heritable mutation such as those in DEC2 or ADRB1.

Normative sleep values.

The typical, healthy sleep durations observed at each age, against which an individual's duration is judged age-appropriate.

Polysomnography.

The reference-standard overnight recording of brain, eye, and muscle activity that scores sleep second by second and yields an exact sleep duration.

Sleep debt.

The cumulative shortfall of sleep relative to need that builds up under chronic restriction, reflected in residual homeostatic pressure and mounting daytime deficits.

Sleep efficiency.

The fraction of time in bed actually spent asleep; it distinguishes sleep duration from the longer period spent in bed.

Sleep latency.

The time taken to fall asleep after attempting to; a distinct sleep parameter, not part of sleep duration.

Sleep quality.

The subjective and architectural sense of how restorative sleep was, which can be poor even when sleep duration is ample; distinct from duration.

Sleep.

The reversible, recurring state of reduced responsiveness and distinctive brain activity whose total nightly amount is sleep duration.

Time in bed.

The total time spent in bed including wakefulness before, during, and after sleep; sleep duration as a fraction of time in bed is sleep efficiency.

Two-process model.

The account in which sleep timing and duration arise from the interaction of the homeostatic sleep pressure (Process S) with the circadian clock (Process C).

U-shaped association.

The relationship in which both short and long sleep duration carry raised risk, so that plotted risk falls to a minimum at an intermediate duration and rises on either side.

Key Researchers

Alexander A. Borbély

. University of Zurich (Professor Emeritus); originated the two-process model of sleep regulation, in which a homeostatic sleep pressure interacts with the circadian clock to set sleep timing and duration. ORCID - Wikipedia

Francesco P. Cappuccio

. University of Warwick; led the landmark prospective-cohort meta-analyses that established the U-shaped association of sleep duration with all-cause mortality, and the links of short sleep with obesity and type 2 diabetes. ORCID - Faculty Page

Ying-Hui Fu

. University of California, San Francisco; discovered the human familial natural short-sleep mutations in DEC2 and the beta-1 adrenergic receptor, showing that habitual sleep duration has a specific heritable genetic basis. ORCID - Faculty Page - Wikipedia

Michael A. Grandner

. University of Arizona; reviewed the epidemiology and candidate mechanisms linking short sleep duration to mortality and co-authored the joint consensus on the recommended amount of sleep for adults. ORCID - Faculty Page

Nathaniel Kleitman

(1895-1999). University of Chicago; founder of modern sleep research, whose systematic studies of the sleep-wake cycle, sleep need, and sleep deprivation established sleep duration as a measurable physiological variable. Wikipedia

Maurice M. Ohayon

. Stanford University; produced the large normative meta-analysis of quantitative sleep parameters across the human lifespan, the reference against which an individual's sleep duration is judged age-appropriate. ORCID - Faculty Page

Frequently Asked Questions

What is sleep duration?

It is the total amount of time spent asleep, measured either over the main nightly sleep period or across a full 24-hour day to include naps. It is a pure quantity of time, distinct from time in bed, how long it takes to fall asleep, and how restorative the sleep feels (Ohayon, Carskadon, Guilleminault, & Vitiello, 2004).

How much sleep should an adult get?

Expert panels recommend 7-9 hours per night for younger and middle-aged adults and 7-8 hours for older adults, with 7 or more hours regarded as the threshold below which health risks rise (Hirshkowitz et al., 2015; Watson et al., 2015).

Why does recommended sleep duration fall with age?

Normative sleep time declines across the lifespan, steeply through childhood and gently through adult life, so the amount that is healthy and typical for a given age falls as well; the recommendations track those normative values (Ohayon et al., 2004; Chaput, Dutil, & Sampasa-Kanyinga, 2018).

Is it true that some people need very little sleep?

Yes. Rare natural short sleepers carry mutations—in DEC2 and in the beta-1 adrenergic receptor—that let them thrive on far less sleep than average without the deficits curtailed sleep inflicts on others, showing sleep need has a genetic basis (He et al., 2009; Shi et al., 2019).

Does sleeping longer always improve health?

No. Both short and long sleep durations are associated with higher mortality, giving a U-shaped curve; the raised risk at long durations is widely interpreted as a sign of existing illness rather than a direct harm of the sleep itself (Cappuccio et al., 2010; Grandner, Hale, Moore, & Patel, 2010).

Why can't I just catch up on lost sleep at the weekend?

Because sleep is gated by the circadian clock as well as by homeostatic pressure, recovery sleep does not lengthen in simple proportion to the hours lost, so a large chronic debt cannot be fully repaid by occasional long nights (Borbely, Daan, Wirz-Justice, & Deboer, 2016).

How accurate is it when people report their own sleep duration?

Not very. On average people overstate their sleep by about 0.8 hours relative to objective measurement, and the error varies widely between individuals, so self-reported figures must be read with that bias in mind (Lauderdale, Knutson, Yan, Liu, & Rathouz, 2008).

What sets how long a person sleeps on a given night?

The two-process model: a homeostatic sleep pressure that builds during waking and discharges during sleep, interacting with a circadian signal that permits sleep only within a time-of-day window, together determine when sleep begins and ends and therefore how long it lasts (Borbely et al., 2016).

References