Sleep physiology

The two-process model, first set out by Alexander Borbély in 1982, remains the main framework for explaining when sleep occurs. It divides the drive to sleep into two separate processes whose effects combine [1].

Process S is homeostatic sleep pressure. It rises the longer a person stays awake and falls during sleep, in a way that resembles hunger building between meals. A chemical messenger closely tied to this pressure is adenosine, a by-product of the brain's energy use, which builds up during waking and acts on receptors that dampen wake-promoting brain circuits. Caffeine blocks adenosine receptors. It does not remove the accumulated adenosine; it prevents the brain from responding to it.

Process C is the circadian rhythm, a roughly 24-hour cycle generated by the suprachiasmatic nucleus (SCN) of the hypothalamus. The SCN tracks biological time rather than time awake, and it promotes wakefulness during the biological day. Brain circuits in the hypothalamus act as a switch between the sleep and wake states, and other hypothalamic neurons stabilise that switch; loss of these stabilising neurons causes the abrupt state changes seen in narcolepsy [2].

Sleepiness is greatest when high sleep pressure coincides with the biological night. Jet lag, shift work and irregular sleep schedules separate the two processes, so that a person may be under strong sleep pressure while the clock still promotes wakefulness, or the reverse.

The SCN sits just above the point where the optic nerves cross, and it receives direct input from a class of light-sensitive cells in the retina (intrinsically photosensitive retinal ganglion cells) that are separate from the rods and cones used for vision [2]. These cells respond most strongly to blue, short-wavelength light.

Light is the strongest time cue the clock receives. Light in the morning shifts the clock earlier, and light in the evening and at night shifts it later. In a study of adults who spent a summer week camping with only natural light, their internal clocks moved earlier and lined up with sunrise and sunset, while ordinary life with electric lighting was associated with less daytime sunlight, more light after sunset and a later clock [7].

The SCN passes its time signal to the rest of the body partly through melatonin, a hormone released by the pineal gland in darkness and suppressed by light. Melatonin signals biological night rather than directly forcing sleep. In a randomized crossover study, reading a light-emitting e-reader for about 4 hours before bed on five evenings suppressed evening melatonin by 55%, delayed the circadian clock by 1.5 hours and lengthened the time taken to fall asleep by nearly 10 minutes compared with reading a printed book [6]. Melatonin taken as a supplement is described in melatonin.

Sleep is classified into stages by brain-wave pattern, eye movements and muscle tone, measured together in a sleep study (polysomnography). The stages are NREM stages N1, N2 and N3, and REM sleep. A healthy adult passes through several NREM to REM cycles each night; the first lasts about 70 to 100 minutes and later cycles about 90 to 120 minutes [3].

Stage N1 is the brief transition from wakefulness into sleep. Stage N2 takes up the largest share of a typical night and contains sleep spindles and K-complexes, short bursts of brain activity associated with memory processing and with keeping sleep stable through minor disturbances. Stage N3, slow-wave sleep, is the deepest stage: brain waves slow into large, synchronized delta waves, and waking the sleeper is hardest [3].

In REM sleep the brain is nearly as electrically active as in wakefulness, but most skeletal muscles are paralysed, which normally prevents dreams from being acted out. Body temperature regulation is weakest during REM sleep. REM and NREM sleep have both been linked to the consolidation of memories and to emotional memory processing [8].

The balance of stages shifts across the night. Slow-wave sleep is concentrated in the first third of the night, while REM periods grow longer toward morning [3]. As a result, going to bed late mainly shortens slow-wave sleep, and waking early mainly shortens REM sleep.

Core body temperature follows a circadian rhythm, falling to its lowest point in the early morning and rising to a peak in the late afternoon or evening. Falling asleep is accompanied by a drop in core temperature, as blood flow to the hands and feet increases and heat is released through the skin.

A meta-analysis of 17 studies found that a warm shower or bath at 40 to 42.5 °C taken 1 to 2 hours before bedtime, for as little as 10 minutes, shortened the time taken to fall asleep and improved sleep efficiency (the share of time in bed spent asleep) and self-rated sleep quality. The authors attributed the effect to increased blood flow to the palms and soles, which speeds heat loss and the fall in core temperature [9].

Room temperature also affects sleep. In a study of older adults monitored at home with wearable sleep trackers and room sensors, sleep was most efficient and restful at night-time bedroom temperatures between 20 and 25 °C, and sleep efficiency fell by 5 to 10% as the temperature rose from 25 to 30 °C [10].

One proposed function of sleep is clearing metabolic waste from the brain. In mice, natural sleep and anaesthesia were associated with a 60% increase in the interstitial space (the fluid-filled space between brain cells), which increased the exchange of cerebrospinal fluid with interstitial fluid and raised the rate of beta-amyloid clearance [5]. This fluid-exchange system is called the glymphatic system.

Human imaging studies are consistent with this. In 20 healthy adults scanned with positron emission tomography (PET), one night without sleep increased beta-amyloid burden in the right hippocampus and thalamus by about 5% compared with a night of rested sleep [11].

Sleep also supports memory. Research reviewed by Walker and Stickgold links sleep to the encoding of new memories, their consolidation after learning, brain plasticity and the reprocessing of existing memories [8].

Poor sleep is associated with inflammation. A meta-analysis of 72 studies with more than 50,000 participants found that sleep disturbance was associated with higher levels of the inflammatory markers C-reactive protein and interleukin-6, and that long sleep was also associated with higher levels of both; experimental sleep deprivation in the laboratory did not raise them [12]. The wider role of chronic inflammation in ageing is described in inflammation.

Large prospective studies show a U-shaped relationship between habitual sleep duration and mortality, with higher risk at both ends of the range. In a 2010 meta-analysis of 16 studies (27 cohort samples, 1,382,999 participants, 112,566 deaths, 4 to 25 years of follow-up), short sleep was associated with 12% higher all-cause mortality (relative risk 1.12, 95% CI 1.06 to 1.18) and long sleep with 30% higher mortality (1.30, 1.22 to 1.38). Relative risk here means the death rate in one group divided by that in the reference group. The reference was most often 7 hours a night; short sleep was defined between 4 hours or less and under 7 hours, and long sleep between more than 8 hours and 12 hours or more, depending on the study [4].

The same group found similar patterns for heart disease and stroke across 15 studies and 474,684 participants. Short sleep was associated with 48% higher risk of developing or dying of coronary heart disease (relative risk 1.48, 1.22 to 1.80), and long sleep with 38% higher coronary heart disease risk (1.38, 1.15 to 1.66) and 65% higher stroke risk (1.65, 1.45 to 1.87) [13].

Short sleep in midlife has also been linked to dementia. In the Whitehall II study of 7,959 British civil servants followed for 25 years, sleeping 6 hours or less at age 50 was associated with 22% higher dementia risk (hazard ratio 1.22, 1.01 to 1.48) and at age 60 with 37% higher risk (1.37, 1.10 to 1.72) than sleeping 7 hours; persistent short sleep from 50 to 70 was associated with 30% higher risk [14]. A hazard ratio is the rate of an event in one group relative to another.

The two ends of the curve are interpreted differently. Short sleep has plausible direct effects through inflammation, blood sugar regulation and brain waste clearance. The long-sleep association is more disputed and may partly reflect reverse causation, in which an existing illness both lengthens sleep and shortens life. In a Swedish study of 43,880 adults followed for 13 years, people under 65 who slept 5 hours or less on both weekdays and weekends had 65% higher mortality (hazard ratio 1.65, 1.22 to 2.23) than those who consistently slept 6 to 7 hours, while short weekday sleep combined with long weekend sleep was not associated with higher mortality; no association was seen in people aged 65 or older [15].

Sleep habits appear to combine. In 172,321 US adults from the National Health Interview Survey followed for a median of 4.3 years, people with all five low-risk sleep factors had 30% lower all-cause mortality (hazard ratio 0.70, 0.63 to 0.77) than those with none or one, corresponding to an estimated 4.7 more years of life expectancy at age 30 for men and 2.4 more years for women [16].