Sleep position alters arousal processes maximally at the high-risk age for sudden infant death syndrome


Associate Professor Rosemary S.C. Horne, Ritchie Centre for Baby Health Research, Level 5, Monash Medical Centre, 246 Clayton Road, Clayton, Victoria, Australia 3168. Tel.: 61-3-9594-5100; fax: 61-3-9594-6811; e-mail:


An impaired ability to arouse from sleep may play an important role in the pathogenesis of sudden infant death syndrome (SIDS). This study aimed to investigate the effects of prone sleeping on the nature of both induced and spontaneous arousal responses in infants. Thirteen healthy term infants were studied longitudinally at 2–4 weeks, 2–3 months and 5–6 months postnatal age. A pulsatile jet of air to the nostrils was used to induce arousal from both active sleep and quiet sleep in both prone and supine positions. For each stimulus, arousals were classified as sub-cortical activations and cortical arousals, scored using physiological and electroencephalogram changes and expressed as a percentage of the total number of arousals. Spontaneous arousals were similarly analysed. Increased proportions of cortical arousals, hence decreased proportions of sub-cortical activations, were observed in the prone position at 2–3 months. This distinct peak in the proportion of cortical arousals occurred regardless of sleep state and regardless of whether the arousal occurred spontaneously or was induced by air-jet stimulation. The nature of arousal responses in healthy term infants is altered in the prone sleeping position at 2–3 months after birth, the age where SIDS incidence is highest. We postulate that a greater propensity for cortical arousal may be a protective mechanism to promote complete arousal in a vulnerable sleeping position and/or a vulnerable period of maturation. Inadequate or incomplete cortical arousals may explain the increased risk of SIDS associated with the prone position at this age.


Despite a dramatic decline in recent years, sudden infant death syndrome (SIDS) is the leading cause of mortality in infants aged between 1 month and 1 year in the western world (Krous et al., 2004). Sleeping in the prone position has been identified as a major risk factor for SIDS; however, the exact mechanisms behind why this position makes infants so vulnerable to sudden death remain unclear. It has been proposed that a failed arousal response from sleep is involved. In support of this, autopsy studies have revealed brainstem abnormalities in SIDS victims in areas responsible for cardiorespiratory control and arousal (Paterson et al., 2006).

Cortical arousal (CA) is thought to involve multiple direct and indirect pathways of activation from the brainstem to the cerebral cortex (Dringenberg and Vanderwolf, 1998). However, activation responses to internal or external stimuli during sleep do not require activation at the cortical level to effect physiological changes. Studies have shown that in both adult and infant subjects, ‘incomplete’ arousals can occur, consisting of autonomic responses such as respiratory, heart rate (HR) or blood pressure changes, in the absence of a full CA (Lijowska et al., 1997; McNamara et al., 1998; Sforza et al., 2000). Furthermore, larger degrees of stimulation are required to elicit an electroencephalogram (EEG) change than are required for a brainstem (heart rate) response (McNamara et al., 1998; Franco et al., 2002), supporting the concept that arousal consists of a hierarchical sequence of responses.

In infants between 1 and 6 months of age, full CA consists of a stereotypical sequence of sub-cortical events consisting of a breathing pause or sigh, startle and ‘thrashing’ behaviour before awakening with eyes open and crying (Lijowska et al., 1997; McNamara et al., 1998). Based on these events, the International Paediatric Work Group on Arousals has defined criteria for the scoring of two distinct types of arousal for infants of this age; sub-cortical activation (SCA) which involves HR, respiratory and behavioural changes; and CA, which also includes electroencephalogram (EEG) changes (The International Paediatric Work Group on Arousals, 2005).

In a recent study, more than 20 000 infants in Belgium underwent polysomnographic studies in their normal sleeping position; 16 died of SIDS soon after (between 4 days and 7 weeks) (Harper, 2003; Kato et al., 2003). Using the new arousal definitions (The International Paediatric Work Group on Arousals, 2005), Kato et al. found that the future SIDS victims spontaneously aroused from sleep less often than a group of age-matched control infants; additionally, SIDS victims exhibited a decreased frequency of CA with an increased frequency and duration of SCA (Kato et al., 2003). This suggests that despite appearing well and physiologically normal, future SIDS victims may have subtle pre-existing abnormalities in their arousal pathways which prevent the progression to full CA (Harper, 2003; Kato et al., 2003).

Previous studies have shown that infant arousability is altered by sleep state and postnatal age (Parslow et al., 2004a, 2004b), and can be depressed by known SIDS risk factors such as prone sleeping (Franco et al., 1998; Galland et al., 2000; Horne et al., 2001), exposure to maternal cigarette smoking (Horne et al., 2002; Parslow et al., 2004a, 2004b) and preterm birth (Horne et al., 2000). However, it is still unknown if these environmental factors which increase the risk for SIDS, also impair the pathways of arousal in otherwise healthy infants. The aim of this study was to investigate the effects of a major SIDS risk factor, the prone sleeping position, on the nature of both induced and spontaneous arousal pathways in sleeping infants. We hypothesized that healthy term infants, sleeping in the prone position, would have increased SCA and decreased CA compared with the supine position and that this effect would be most marked at 2–3 months of age when SIDS risk is highest.



Ethical approval for this project was obtained from the Southern Health (Monash Medical Centre) and Monash University Human Research Ethics Committees. Study participation was voluntary and no monetary incentive was provided. Written informed parental consent was obtained prior to commencement of the study.

Thirteen healthy infants (seven females/six males) born at term (38–41 weeks gestation) were recruited from Jessie McPherson Private Hospital and Monash Medical Centre, Melbourne, Australia. Infants were born at 39.7 ± 0.2 weeks gestation (mean ± SEM) with normal birth weights (3498 ± 102 g, range of 2765–4015) and Apgar scores at both 1 (5–10, median 9) and 5 (7–10, median 9) min after birth. None of the mothers smoked during pregnancy or after delivery, and all infants routinely slept in the supine position at home. Infants were studied longitudinally using daytime polysomnography on three occasions at 2–4 weeks, 2–3 months and 5–6 months postnatal age. Arousability data (probability of arousal and arousal threshold) from these infants have previously been reported (Horne et al., 2002).


Two scalp electroencephalogram (EEG) channels were continuously recorded (C4/A1 and O2/A1) at a sampling rate of 500 Hz; high-pass filters were applied at 0.3 Hz and low-pass at 30 Hz for both EEG signals. In addition, left and right electrooculograms, submentalis electromyogram (EMG), electrocardiogram and instantaneous heart rate (HR) were also recorded. Piezo-electric sensors (Resp-ez, EPM Systems, Midlothian, VA, USA) were used to monitor both thoracic and abdominal breathing movements. Oxygen saturation was measured at the ear, with the pulse oximeter averaging time set to 3 s (BIOX 3700e, Ohmeda, Louisville, CO, USA). Rectal and abdominal skin temperature was also measured (YSI 400 Series Thermistor, Mallinckrodt, Melbourne, VIC, Australia). All physiological variables were recorded using a 16-channel polygraph (Model 78A, Grass Instrument Co., Quincy, MASS, USA). Data were simultaneously stored on two computer systems for later analysis; MacLab/8e (ADInstruments, Sydney, NSW, Australia) and S-series Sleep System V5.2 (Compumedics, Melbourne, VIC, Australia).

Study protocol

Daytime polysomnography was performed in a sleep laboratory. Environmental temperature was maintained at 21–24 °C throughout the studies, whilst noise and light levels were minimal. Electrodes for recording were attached while infants fed. During each study, infants slept both supine and prone; starting positions were randomized for the first study then alternated for subsequent studies. Infants generally had both a morning and afternoon sleep, interrupted by a midday feed when sleep position was changed. The study began when the infant was in a stable sleep state, determined as AS, QS or indeterminate sleep according to standard criteria (Anders et al., 1971).

During both AS and QS, in both supine and prone positions, arousal was induced by nasal air-jet stimulation. As previously described, a pulsatile jet of air was delivered (at 3 Hz, for 5 s duration) through a specially designed cannula held at the infants’ left and right nostrils alternately (Horne et al., 2001, 2002). Infants received a series of stimuli at increasing driving pressures until arousal criteria were met, pressures were then decreased (Horne et al., 2001, 2002). This staircase protocol was repeated throughout each epoch of sleep. Changes in the driving pressure between consecutive stimuli were usually 100 cmH2O; the maximum stimulus driving pressure was 950 cmH2O (equivalent to 1.6 cmH2O received at the infant end of the cannula). Full awakenings of the infants with eyes open and crying were infrequent as the objective of our protocol was to induce arousal responses without compromising the infants’ natural sleep cycles.

Data analysis

From the polysomnographic recordings and cot-side observations, infant responses were identified visually as non-arousal, SCA or CA, by a single investigator (HLR), in accordance with the recently published consensus for scoring arousals in infants (The International Paediatric Work Group on Arousals, 2005). SCA was defined by the presence of at least two out of three criteria: gross body movement detected by artefact visible in all channels or simply observed at the bedside; a HR increase ≥10% above baseline (10 s prior to air-jet stimuli). The third criterion was sleep state dependent; in QS, a respiratory change in either amplitude or frequency, and in AS, an increased amplitude of the EMG signal. Full CA was defined using the SCA criteria, with the addition of an abrupt change in EEG background frequency for a minimum of 3 s. To allow HR changes to reach a peak and to reduce the probability of spontaneous arousal, these changes were only scored if they first occurred within 7 s of stimulus initiation. All SCA and CA observed during uninterrupted periods of sleep, that were not elicited by air-jet tests or other external stimuli, were classified as spontaneous arousals and were analysed separately.


All statistical analyses were performed using SigmaStat for Windows (Systat Software Inc, Richmond, CA, USA). Non-arousals, SCA and CA were expressed as percentages of total air-jet stimuli. Chi-square statistics were used to evaluate the proportions of these responses between different positions, sleep states and postnatal ages (Fig. 1). To control for any changes in total arousability (SCA + CA) and to further elucidate where differences lay within the data, secondary analyses investigated the proportions of SCA and CA as a percentage of total arousals, excluding the non-arousal responses (Figs 2–4). Similarly, for spontaneous arousals, SCA and CA were expressed as a percentage of the total number of arousals. The proportions of each arousal type were compared with chi-square statistics to assess the effects of sleeping position, sleep state and postnatal age.

Figure 1.

 Non-arousals (white bars), sub-cortical activations (grey bars) and cortical arousals (black bars) in response to air-jet stimulation in supine (solid) and prone (cross-hatched) positions during both AS and QS. Data are expressed as proportions of total number of stimuli delivered (indicated within histograms). Symbols represent statistically significant differences; *P < 0.05 between prone and supine positions; P < 0.05 between AS and QS; P < 0.05 between postnatal ages.

Figure 2.

 Cortical arousals induced by air-jet stimulation, expressed as percentage of total induced arousal responses, in (a) AS and (b) QS in both supine (solid bars) and prone (striped bars) positions, throughout the first 6 months of life. Symbols represent statistically significant differences; *P < 0.05, ***P < 0.001 between prone and supine positions; P < 0.05, ‡‡‡P < 0.001 between postnatal ages.

Figure 3.

 Individual proportions of cortical/total arousals (CA/CA + SCA) for each infant during both sleep states and in both positions; (a) AS-Supine; (b) QS-Supine; (c) AS-Prone and (d) QS-Prone, throughout the first 6 months of life. Note the consistent increase in CA proportions observed at 2–3 months when infants slept prone.

Figure 4.

 Spontaneous cortical arousals, expressed as percentage of total spontaneous arousal responses, in (a) AS and (b) QS in both supine (solid bars) and prone (striped bars) positions, throughout the first 6 months of life. Symbols represent statistically significant differences; *P < 0.05 between prone and supine positions; P < 0.05 between postnatal ages.

To address the issue of habituation with repeated arousal responses, paired t-tests were used to compare the first and last two arousals for every sleep epoch studied. In addition, by means of a two way analysis of variance for each age studied, driving pressures of air-jet stimuli were compared between each sleep state/position and arousal type. These are presented as mean ± SEM, with significance taken at P < 0.05.


Demographic and sleep data for the three studies are presented in Table 1. Two infants were not available for follow-up after the first study, three infants did not complete the third study and data from one infant’s first study was corrupted by technical complications. Thus, data were analysed for 12 infants at 2–4 weeks, 11 infants at 2–3 months and 8 infants at 5–6 months.

Table 1.   Demographic and sleep data for the three studies (mean ± SEM)
 2–4 weeks2–3 months5–6 months
Number of infants12118
Postnatal age (days)14 ± 174 ± 2189 ± 5
Body weight (g)3709 ± 1295453 ± 2137865 ± 359
Total sleep time (min)157 ± 10136 ± 1296 ± 9
Active sleep (%)62 ± 353 ± 444 ± 5
Quiet sleep (%)38 ± 347 ± 456 ± 5

At 2–4 weeks, 2–3 months and 5–6 months, the mean total sleep time was 157 ± 10, 136 ± 12 and 96 ± 9 min, respectively. There were no significant differences in baseline respiratory rate, oxygen saturation or core temperature between prone and supine sleeping positions in either sleep state or at any of the three ages studied.

A total of 2419 air-jet stimuli were administered which produced 902 arousal responses (503 in AS, 399 in QS) for analysis. During the uninterrupted periods of sleep between air-jet stimuli, a total of 526 spontaneous arousals (456 in AS, 70 in QS) were observed.

Responses to air-jet stimulation

Fig. 1 shows arousal and non-arousal responses expressed as proportions of the total number of air-jet stimuli associated with different sleep states, positions and postnatal ages.

Effects of sleep state and position

Although there was a trend for increased total arousals during AS compared with QS, this was only significant at 5–6 months (P < 0.01). Positional differences were observed between prone and supine at 2–4 weeks in QS (P < 0.05) and 2–3 months in both AS (P < 0.01) and QS (P < 0.05); in each case, total arousability (SCA + CA) was depressed in the prone position. There were no positional differences observed at 5–6 months.

Maturational changes

Age-related differences were observed in both sleep states when infants slept prone; the proportions of non-arousals, SCA and CA were significantly different at 2–3 months when compared to both 2–4 weeks (P < 0.05 in AS, P < 0.01 in QS) and 5–6 months (QS only, P < 0.01).

Stimulus intensity

The mean driving pressures of air-jet stimuli which induced arousals are presented in Table 2.

Table 2.   Mean driving pressures (in cmH2O) of air-jet stimuli which induced sub-cortical activations (SCA) and cortical arousals (CA) at 2–4 weeks, 2–3 months and 5–6 months postnatal age in prone and supine positions during each sleep state (AS and QS)
 2–4 weeks2–3 months5–6 months
  1. Values in brackets were excluded from statistical analysis.

  2. *Indicate differences between sleeping positions in a given sleep state; *P < 0.05.

  3. Indicate differences between sleep states in a given position; P < 0.05, ††P < 0.001.

AS-Prone309 ± 32374 ± 38254 ± 47316 ± 54180 ± 33138 ± 60
AS-Supine200 ± 26234 ± 33131 ± 36162 ± 32140 ± 52(104 ± 46)
QS-Prone375 ± 64465 ± 81404 ± 74*481 ± 58*657 ± 92††575 ± 106
QS-Supine272 ± 46331 ± 69211 ± 34211 ± 25490 ± 126(350 ± 132)

Effects of sleep state and position

When compared with the supine position, it was found that higher pressures were required to induce both arousal types in the prone position during QS at 2–3 months (SCA and CA, P < 0.05). Sleep state differences were also observed, with arousals being induced at higher pressures in QS-Prone at 2–3 months (both SCA and CA, P < 0.05) and 5–6 months (SCA only, P < 0.001) when compared with AS-Prone; also in QS-Supine at 5–6 months (SCA only, P < 0.05) when compared with AS-Supine.

Sub-cortical activation versus cortical arousal

There were no significant differences between the stimulus pressures that elicited SCA and those which led to CA in either sleep state or sleeping position at any of the ages studied.

Proportions of arousal types: air-jet induced

Both SCA and CA were observed in response to repeated air-jet stimulation in each infant at each study. There were no significant differences in the distribution of arousal types observed between the beginning and end of each sleep epoch, in either position or either sleep state at any ages studied.

Effects of sleep state and position

Regardless of sleeping position, there was no effect of sleep state on the proportion of SCA and CA of total arousals, at any of the ages studied.

Fig. 2 shows that at 2–3 months, the prone position was associated with a higher proportion of CA, in both AS (supine 32%, prone 51%, P < 0.05) and QS (supine 11%, prone 55%, P < 0.001), hence there was a decreased proportion of SCA. No significant differences were found between positions at either 2–4 weeks or 5–6 months.

Maturational changes

Individual infant data for the proportions of CA are plotted in Fig. 3. In six out of eight infants, a distinct peak in the proportion of CA was observed in the prone position, at 2–3 months, during both sleep states. This peak was absent when infants slept supine. Group changes in the proportions of CA across the first 6 months of life are shown in Fig. 2. When infants slept in the supine position, the proportions of CA remained similar across all three ages studied, regardless of sleep state. In contrast, when infants slept prone, in both AS (Fig. 2a) and QS (Fig. 2b), a significant increase in the proportion of CA (and hence a decreased proportion of SCA) was observed at 2–3 months when compared with both 2–4 weeks and 5–6 months age.

Proportions of arousal types: spontaneous

Effects of sleep state and position

There were no effects of sleep state on the proportions of spontaneous SCA or CA, regardless of sleeping position.

In periods of uninterrupted sleep, both SCA and CA were observed to occur spontaneously. At 2–3 months, there were significant increases in the proportions of CA observed in the prone position during both AS (supine 30%, prone 70%, P < 0.001) and QS (supine 20%, prone 69%, P < 0.05) and these are shown in Fig. 4. The proportions of arousal types did not differ significantly between prone and supine positions at the other two ages studied.

Maturational changes

Changes in the proportions of spontaneous CA across the first 6 months of life are presented in Fig. 4. When infants slept in the supine position, there were no maturational changes in the proportion of arousal types observed across the three ages studied. However, when slept in the prone position, in AS CA proportions were significantly increased at 2–3 months compared with 5–6 months (P < 0.05, Fig. 4a). A similar trend was also evident during QS (Fig. 4b); however this failed to reach statistical significance.


This study was the first to examine the effects of sleep state, sleeping position and postnatal age on both induced and spontaneous arousal responses in infants in accordance with the recent definitions of SCA and CA (The International Paediatric Work Group on Arousals, 2005). Increased proportions of CA, hence decreased proportions of SCA, were observed in the prone position at 2–3 months of age, the age where SIDS incidence is highest. This distinct peak in the proportion of CA occurred regardless of sleep state and regardless of whether the arousal occurred spontaneously or was induced by air-jet stimulation. Importantly, individual data reflected similar trends in the majority of infants studied. Most previous studies have demonstrated that in both AS and QS, infant overall arousability is depressed in the prone position compared with supine (Galland et al., 2000; Horne et al., 2001). This study supports these findings; and expands our current knowledge of normal healthy infant arousal responses by demonstrating that sleeping position can also alter the activation of arousal pathways.

Our findings appear to be contrary to those of Kato et al. who reported that at approximately 11 weeks of age, future SIDS victims exhibited a decrease in spontaneous CA and an increase in SCA compared with controls, suggestive of incomplete arousal processes (Kato et al., 2003). However, it is important to note that the Kato et al. study subjects were subsequent SIDS victims, who had been studied because they were at increased risk for SIDS (2 siblings of SIDS, 12 for obstructive sleep apnoea syndrome and 2 for breath-holding spells) (Kato et al., 2003). In contrast, infants in this study were healthy, with non-smoking mothers and no family history of SIDS; furthermore, no infant succumbed to SIDS. Taken together, the findings of our study of normal healthy infants and those of Kato et al. (Kato et al., 2003) provide further evidence to support the notion that SIDS victims have inadequate or incomplete arousal processes.

Our findings of increased CA (both induced and spontaneous) as a proportion of total arousals observed at 2–3 months when infants slept prone are in contrast to a recent study which described a decreased frequency of spontaneous CA per hour of AS in the prone position, in infants of a similar age (Kato et al., 2006). These differing results may be attributed to design differences between the two studies. Kato et al. analysed spontaneous arousals from two groups of infants, one group who regularly slept supine and another who regularly slept prone. Our study was of infants who all routinely slept supine at home and who were studied sleeping in both positions. As all infants routinely slept supine at home, infants in our study were ‘naïve’ prone sleepers at the time of each study. It has been suggested that infants who are inexperienced in prone sleeping are at increased risk of SIDS if they are placed prone (Cote et al., 2000; Paluszynska et al., 2004). In the Kato study, infants were accustomed to sleeping in the prone position (Kato et al., 2006); whereas in our study, it is possible that unfamiliarity with the prone position may have promoted the increase of CA in these infants. This seems unlikely however, as the arousal change was only evident at 2–3 months of age. Rather, these changes may be due to a protective mechanism employed during a vulnerable developmental period and/or in a vulnerable sleeping position. At both 2–4 weeks and 5–6 months of age, sleeping position did not alter the proportion of arousal types, consistent with the lower risk of SIDS at these ages.

At each age studied and in both positions, we observed more arousal responses during AS than QS; a finding that we had anticipated as it is consistent with previous studies of arousal from sleep in both term and preterm infants (Horne et al., 2000, 2001; McNamara et al., 2002; Kato et al., 2006). However, an unexpected finding of our study was the absence of a sleep state effect on the proportions of each arousal type, both spontaneous and air-jet induced. It has previously been shown that infant arousal thresholds are higher during QS compared with AS, to both respiratory and somatosensory stimuli (Galland et al., 2000; Horne et al., 2001; Parslow et al., 2004a, 2004b). On these grounds we had anticipated increased CA and less SCA during AS. By contrast, the proportions of SCA and CA did not differ between AS and QS throughout the first 6 months of life. This suggests that factors determining arousal thresholds, and the sleep-state related threshold differences, are equivalent for both CA and SCA.

During both AS and QS, with the exception of the prone position at 2–3 months, there were generally more SCA observed in response to air-jet stimuli than full CA (approximately 70% versus 30%). Similar to our findings, fewer CA than purely autonomic responses have previously been observed in both infants and children in response to auditory stimuli (Franco et al., 2002). A predominance of SCA was also found by Lijowska et al., who reported that isolated sighs were the most common response to hypercapnia in infants aged between 2 weeks and 6 months; sighs with startles were less common and the entire sequence leading to full arousal was least common (Lijowska et al., 1997). Previously, it has been suggested that SCA in normal infants can adequately modify respiration, HR or BP or evoke a behavioural response to stimuli such as hypoxia or hypercapnia, without requiring the progression to full CA, hence avoiding unnecessary sleep disruption (Carley et al., 1997; Halasz et al., 2004).

The maintenance of sleep integrity may also involve a gradual diminishing of the arousal response to repeated stimuli habituation. A previous study reported that infants readily habituated to repeated tactile stimuli (McNamara et al., 1999). McNamara et al. analysed responses to tactile stimulation performed at the same pressure, every 5 s. In some infants, this led to the habituation of arousal responses which occurred serially in the reverse order to the typical infant arousal sequence, that is, cortical responses were eliminated first, then brainstem responses, and finally spinal responses (McNamara et al., 1999). To address the issue of habituation with repeated arousals in our study, the first and last two arousals observed during each sleep epoch were compared for all infants studied. If habituation was indeed occurring, we postulated that more CA would be observed at the beginning of each epoch with more SCA at the end. During the course of our study, both SCA and CA were observed, though each arousal type appeared to be distributed stochastically throughout each sleep epoch, regardless of sleep state, position and postnatal age. This is not surprising as the experimental protocol was designed to prevent any habituation of arousal responses, with air-jet stimuli being delivered to the left and right nostrils alternately at various driving pressures and with varied time intervals between each stimulus.

It is important to acknowledge that quantitative analyses of spontaneous arousals in this study were limited by low numbers, as sleep periods were interrupted by the application of air-jet stimuli; particularly during QS, a state where spontaneous arousability is already decreased. Nevertheless, our data show that the changes in proportions of arousal types with different sleeping positions and ages followed similar trends for both spontaneous and air-jet induced arousals. This provides support for the contention that both spontaneous and tactile-induced arousal responses are generated by the same neural pathways (McNamara et al., 1998).

The EEG changes that occur during arousal from sleep are traditionally thought to be produced by an ascending pathway beginning with stimulation of the reticular activating system in the brainstem, progressing to the midline nuclei of the thalamus and finally resulting in activation of the cerebral cortex (Castro-Alamancos and Oldford, 2002; Thach, 2002). Early concepts saw the thalamus as providing a gateway for communication between peripheral receptors and the cortex; during sleep, synaptic inhibition at the thalamus minimizes sleep disruption by blocking all ‘non-essential’ signals from entry into consciousness (Steriade et al., 1993; Bastien et al., 2000; Halasz et al., 2004). Evidence from more recent studies suggests that the process is much more complex and the role of the thalamus and its importance is being questioned. Previous studies have shown that sections through the midbrain eradicated thalamus-induced cortical activation, though without interrupting direct thalamo-cortical communication (Schlag and Chaillet, 1963). Karen Ann Quinlan, a coma patient who lived in a persistent vegetative state for 10 years, had intact sleep stages and arousal activity, despite extensive damage to the thalamus (Kinney et al., 1994). Contrasting with earlier concepts, arousal may be mediated by multiple pathways, both thalamic and extrathalamic, in parallel. It has been suggested that direct cholinergic and serotonergic inputs from the basal forebrain and raphe nuclei, respectively, are essential for cortical activation; and that additional non-essential pathways contribute indirectly, though are still finally mediated by one or another of these cholinergic and serotonergic systems (Dringenberg and Vanderwolf, 1997, 1998).

Although the exact mechanisms leading to SIDS remain elusive, it is becoming more apparent that an impaired ability to arouse from sleep in response to a potentially life-threatening stimulus may play an important role. Epidemiological studies have revealed a distinct peak in the incidence of SIDS between 2 and 3 months of age (Hoffman and Hillman, 1992); the same age that modified arousal pathways were observed in the prone position. This is a period of dramatic maturational changes in the infant brain, including the development of sleep spindles (Wulbrand et al., 1998). Sleep spindles, generated by synaptic interactions in the reticular nucleus of the thalamus, are associated with a loss of perceptual awareness hence are said to play an important role in maintaining sleep (Steriade et al., 1993; Steriade and Amzica, 1998). First appearing in the second month of life, spindle amplitude increases to a maximum by 3 months of age before decreasing with postnatal age (Hughes, 1996). We have previously shown that at 2–3 months, spindle density was decreased when infants slept in the prone position compared with supine (Horne et al., 2003). This change in spindle density may indicate alterations in the thalamus, perhaps related to the increased proportion of CA of total induced and spontaneous arousals which we observed at the same age in the prone position.

In conclusion, this study has demonstrated that in healthy term infants, the prone position appears to promote full cortical activation at 2–3 months of age. We postulate that this may normally be a protective mechanism to promote complete arousal in a vulnerable sleeping position and/or a vulnerable period of maturation. Abnormalities or impairment of these pathways may leave an infant unable to appropriately compensate for various respiratory or cardiovascular challenges during sleep, and perhaps render them more susceptible to SIDS.


The authors wish to thank the infants and parents who participated in this study.