Article Text

Download PDFPDF

The physiology of sleep in infants
  1. J L Heraghty,
  2. T N Hilliard,
  3. A J Henderson,
  4. P J Fleming
  1. Institute of Child Life and Health, University of Bristol, and Bristol Royal Hospital for Children, Bristol, UK
  1. Professor Peter Fleming, Level D, St Michael’s Hospital, Southwell Street, Bristol BS2 8EG, UK; peter.fleming{at}


Despite the fact that infants spend more time asleep than awake, an understanding of the importance and effects of sleep on the pathophysiology of illness in infancy is a relatively recent development, and is commonly overlooked in paediatric training. In this review we describe some of the characteristics of sleep in infancy, with particular reference to normal developmental physiology and its relevance to the signs, symptoms and pathophysiology of illness in this age group.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

By the age of 3 years, the average child will have spent more time asleep than in all wakeful activities.1 Sleep is an essential component of mammalian physiology and is characterised by periods of reduced mobility and responsiveness, rapid reversibility and the need for “catch-up” following enforced sleep deprivation. However, despite the recognition that sleep in humans is more than simply a passive process governed by cyclical changes in environmental stimuli, the precise functions of sleep, and in particular the need for two different types of sleep (rapid eye movement (REM) and non-REM), still give cause for some speculation. Before considering features of sleep that are particular to the infant, we will briefly review the control and function of sleep in humans (reviewed in detail elsewhere).24 REM sleep is characterised by high levels of neuronal activity and is the phase of sleep most commonly associated with vivid dreams. Although in adults it occupies a relatively small proportion of total sleep time, the regulation of REM sleep appears to be considerably more complex than the equivalent for non-REM sleep, and has been localised to groups of neurons within the brain stem that control different aspects of the REM sleep state, including REM-on neurons located in the pontine reticular formation and REM suppressive (REM-off) neurons in the dorsal and other raphe nuclei. Other neurons in the pontine reticular formation generate the lateral saccades that result in rapid eye movements, which define this behavioural state. In adults non-REM sleep can be divided into stages I–IV. Slow wave sleep (SWS) (stages III and IV) is considered the most quiescent state of the brain and is preceded by reduced activity in the wake-promoting regions of the brain. Gating of sensory inputs to the cortex is achieved by hyperpolarisation of relay neurons by burst activity of GABA-ergic neurons in the thalamus, and SWS is generated by activation of GABA containing neurons in the preoptic area of the hypothalamus.

As stated above, the functions of sleep remain speculative and our understanding has been largely based on observations following periods of sleep deprivation. There is a strong inverse correlation between body size and total sleep duration in mammals, leading to the suggestion that sleep may have a role in the defence against oxidative stress, as reactive oxygen species accumulate more rapidly in association with the relatively higher metabolic rate of small mammals. Neocortical maintenance and energy conservation are potential functions of SWS, the latter being a particularly important adaptive response in newborns with high surface area to body mass ratios. There is also evidence for a role of SWS in immune function and regulation.5 Sleep is believed to have an important role in memory and learning6 and REM sleep has been hypothesised to have a fundamental role in memory consolidation, although this is not universally accepted.7 The increased proportion of REM sleep at birth (see below) compared with adulthood is also linked in mammalian species to immaturity at birth8 and there is experimental evidence to support a role of REM sleep in the regulation of neuronal development and brain plasticity.9


In newborn infants, the signature EEG features that characterise different sleep stages in adults are not discernible and a combination of EEG and behavioural criteria is often used to assign sleep state during this period of development.10 As in adults, sleep can be categorised into two main patterns, active sleep (an immature form of REM sleep)11 and quiet sleep (analogous to SWS), but a proportion of sleep time is not attributable to either of these patterns and is accordingly classed as indeterminate. In preterm and term newborns, the predominant sleep state is active sleep which gradually falls as a proportion of total sleep time from about 60% at 34 weeks postconceptional age to around 50% at full term. The proportion of active sleep diminishes rapidly over the first few months after birth to reach 25% of total sleep time (approaching adult values) at the age of 6 months.12 The organisation of sleep in infants also changes from birth, when active and quiet sleep periods are of approximately equivalent duration over a sleep cycle of around 50–60 min. Sleep periods are shorter (2–4 h) and more frequent than in older children. At around 3 months of age, REM sleep starts to become organised into later sleep cycles and non-REM sleep dominates the earlier parts of sleep. By 6 months of age, sleep is entered through a non-REM stage and typical inhibition of muscle tone occurs in REM sleep (in contrast to the earlier activity seen during active sleep). Periods of continuous sleep gradually lengthen and become consolidated into a predominantly night time pattern over the first year, so that by the age of 12 months, total sleep time has fallen from 16–18 h/day at term to 14–15 h/day, with most occurring at night and one or two daytime naps.

The sleep–wake cycle in humans is governed by two main processes: homeostatic and circadian. The latter is a mechanism for entraining the rest–activity cycle to environmental cues, which in humans is bound to the light–dark cycle; the suprachiasmatic nucleus is recognised as having an important role in the setting of this biological clock. The endogenous (or free-running) circadian rhythm of humans is actually closer to 25 h13 and is entrained to a 24 h period by cyclical changes in the environment (so-called zeitgebers). Anyone who has ever looked after a newborn infant will immediately recognise that this system is not in full working order from birth. There is a gradual maturation of day–night cycles of behaviour and hormone secretion from about 1–3 months after birth.14 However, there is also emerging evidence that fetal functions have circadian rhythms synchronised to maternal rest–activity cycles.15 16 These are believed to prepare the fetus for adaptation to light–dark cycles after birth.17 As discussed in detail below, body temperature and cortisol 24 h rhythms emerge in term infants between 8 and 16 weeks of postnatal age.18 19 The development of circadian rhythms in preterm infants is linked to postconceptional age and also to intrauterine growth, leading to later acquisition of clear day–night patterns of behaviour and physiological function. However, preterm infants are more likely to be exposed to constant illumination in the nursery and the institution of light–dark cycles in this setting has been associated with the advancement of circadian rest–activity patterns in preterm infants compared with those exposed to constant dim light.20


The infant human brain develops rapidly during the first 3 months of life with a slowing in development thereafter; during this period REM sleep predominates and, as mentioned previously, also progressively reduces as a proportion of sleep. The decline in REM sleep as the brain rapidly develops can also be seen in other mammals such as rats, which undergo most of their development postnatally. Conversely, mammals which have most of their CNS development before birth (such as the guinea pig) have high levels of REM sleep prenatally, which falls to low adult levels at birth.17 In 1966, Roffwarg first proposed that the primary purpose of active sleep was as an inducer of CNS development in the neonate, stimulating the brain in a period when waking life is limited.21

In newborn rats it has been noted that an enriched environment increases the size of the cerebral cortex and produces more synaptic connections and better problem solving ability. This effect is negated when the rats are deprived of REM sleep by medication.22 It is difficult to know whether this effect is solely due to REM sleep deprivation or is the consequence of the medication altering the sleep pattern. It has also been shown that these neonatally REM sleep deprived animals have increased anxiety, reduced sexual activity and disturbed sleep as adults.23 Hence, long-lasting effects occur after a disturbance in the infant sleeping pattern.

In the early part of life the brain has the plasticity to develop functions which cannot be achieved in later life. Kittens deprived of visual stimulus and light in one eye have a reduced number of cells in the lateral geniculate nucleus on the contralateral side, a reduction which persists into adulthood long after the visual stimulus has been restored. This effect is amplified if the kitten is deprived of REM sleep by keeping it awake during this critical period of brain plasticity.24 25 This would suggest that the activity of the visual system during REM sleep is essential to the developing brain. This effect can probably be expanded to all areas of development in the infant.

Quiet sleep may also be important for developing animals. The maturation of quiet sleep coincides with the formation of thalamocortical and intracortical patterns of innervation and periods of heightened synaptogenesis. During quiet sleep waking patterns of neuronal activity are reactivated, suggesting that information acquired during wakefulness is further processed during sleep.4 Learning enhancement following sleep has been noted in baby chicks learning a mother’s song. Baby chicks were divided into two groups; half the chicks were rested but were awake following a period of listening to the song, while the remainder were allowed to sleep. The group that had slept during the rest period learnt the mother’s song faster and more accurately, suggesting that sleep had allowed the chicks to commit the song to memory by rerunning or practicing the song in sleep.26

Most of the neurodevelopmental effects of sleep can only be presumed after noting the effect of sleep deprivation in animals or in older humans. This has the obvious disadvantage of making assumptions using results from other mammals which have different sleep patterns, proportions of time spent in each sleep state and rates of development to the newborn infant. However, interference with REM sleep during critical brain development is likely to have long term effects as the infant matures into adulthood.


There are marked differences in the physiology of the cardiorespiratory system between different sleep states, and between sleep and waking. During REM sleep, metabolic rate is higher than during non-REM sleep, partly due to the higher metabolic rate of the brain. In infancy functional residual capacity is lower in sleep (particularly REM) than when awake. The ventilatory responses to mild hypoxia or hypercarbia are more variable, the respiratory and heart rates are higher and more variable, and arterial oxygen saturation is more variable in REM than non-REM sleep.27 28

Over the first year of life the respiratory rate and heart rate during sleep initially rise over the first 4–6 weeks28 and then fall. Breathing is generally regular in quiet sleep and is interrupted only by episodic sighs followed by an oscillatory pattern of breathing.29 30 In active sleep, breathing is irregular in both rate and depth, with a variable continuing oscillatory pattern. In newborn infants, pauses in breathing of up to 10 s are common, particularly in active sleep, and decrease in frequency with advancing postnatal age.29 Longer apnoeic pauses are more common in preterm infants and may persist until around 52 weeks postmenstrual age (ie, 12 weeks past term) but also occur in term babies in the first few weeks after birth.

Oscillatory patterns of breathing including periodic breathing (three episodes of apnoea lasting greater than 3 s in a period of continuous respiration of 20 s or less) are common in early infancy, decreasing in incidence and duration after 4–5 months of age, and are more marked in a warm environment.27 30 31

Neither the presence nor absence of apnoeic pauses or periodic breathing has been shown to be of any predictive value for the risk of sudden infant death syndrome.29

The inhibition of skeletal muscle tone (including intercostal muscles) during REM sleep leads to paradoxical inward inspiratory rib cage movement, which may (particularly in the youngest infants) be accompanied by intercostal recession as a normal feature of REM sleep in infants and young children up to age 3 years, with increased diaphragmatic work of breathing. Diaphragmatic fatigue particularly in REM sleep has thus been suggested as a factor leading to apnoea in respiratory illness in early infancy,32 but other studies have cast some doubt upon this.33 Reduced muscle tone in upper airway and tongue muscles (particularly genioglossus) may lead to airway obstruction in REM sleep in infants with relative macroglossia and micrognathia (eg, Robin anomalad).34


In most mammals thermoregulation is less effective during REM sleep than non-REM sleep, but in humans at all ages the reverse is true, with more active and effective thermoregulation in REM sleep.35

In adults a characteristic fall of 0.5–0.6°C in core body temperature occurs in the first 2 h after night time sleep onset, followed by a slow rise during the latter part of the night. This pattern develops between 2 and 4 months of age, occurring later in low birthweight or preterm infants and those who are bottle fed rather than breastfed.36 37 A corresponding change in peripheral skin temperature shows a reciprocal pattern of rise and fall which precedes and possibly induces the respective fall and rise in core temperature during night time sleep.38 39 Periodic oscillations of core temperature during sleep correspond to sleep states, with a slight rise in temperature during REM and a fall during non-REM sleep throughout the night.40 Close physical contact between mothers and infants has effects upon the diurnal fall in core temperature, with lower core temperatures observed when infants are bedsharing compared to when sleeping alone, despite a warmer environment when bedsharing.38


There are complex interactions between the development of diurnal rhythms and hormonal secretion during sleep in infancy.14 Growth hormone is secreted exclusively during sleep with peak production in SWS during the first third of the night.41 Melatonin passes through the placenta, so that maternally derived melatonin can drive fetal diurnal rhythms.42 As well as influencing the sleep–wake cycle, melatonin has antioxidant and free radical scavenging properties,43 and it has been proposed that it can function as a tissue protector as well as having a role in brain development.4246 Maturation of the diurnal rhythm of melatonin production takes place from about 2 months, although maturation is delayed in premature infants.46 This has led to the suggestion that melatonin might help establish circadian rhythms in preterm infants.44 45 The development of a diurnal pattern of cortisol production varies between infants but is commonly present from 2 months of age.47 It is not clear when this occurs in preterm infants, but one small study found onset at a similar postnatal age to the development of a diurnal sleep–wake cycle.48 However, a delay in the overnight reduction of cortisol production was found in infants with intra-uterine growth retardation, with the possibility that this reflects fetal programming that continues into adulthood.49


Sleep has profound effects upon all aspects of developmental physiology in the human infant, but little is known about how cultural or social practices (eg, the place or duration of sleep or common sleep care practices such as bedsharing) affect these processes in healthy term or preterm infants. Any attempt to investigate how such practices or illnesses affect infants50 must take account of normal developmental sleep physiology.


The authors wish to acknowledge grant support from The Foundation for the Study of Infant Deaths and The James Tudor Trust.



  • Competing interests: None.