OBJECTIVE. Prematurely born compared with term born infants are at increased risk of sudden infant death syndrome, particularly if slept prone. The purpose of this work was to test the hypothesis that preterm infants with or without bronchopulmonary dysplasia being prepared for neonatal unit discharge would sleep longer and have less arousals and more central apneas in the prone position.
METHODS. This was a prospective observational study in a tertiary NICU. Twenty-four infants (14 with bronchopulmonary dysplasia) with a median gestational age of 27 weeks were studied at a median postconceptional age of 37 weeks. Video polysomnographic recordings of 2-channel electroencephalogram, 2-channel electro-oculogram, nasal airflow, chest and abdominal wall movements, limb movements, electrocardiogram, and oxygen saturation were made in the supine and prone positions, each position maintained for 3 hours. The duration of sleep, sleep efficiency (total sleep time/total recording time), and number and type of apneas, arousals, and awakenings were recorded.
RESULTS. Overall, in the prone position, infants slept longer, had greater sleep efficiency (89.5% vs 72.5%), and had more central apneas (median: 5.6 vs 2.2), but fewer obstructive apneas (0.5 vs 0.9). The infants had more awakenings (9.7 vs 3.5) and arousals per hour (13.6 vs 9.0) when supine. There were similar findings in the bronchopulmonary dysplasia infants.
CONCLUSIONS. Very prematurely born infants studied before neonatal unit discharge sleep more efficiently with fewer arousals and more central apneas in the prone position, emphasizing the importance of recommending supine sleeping after neonatal unit discharge for prematurely born infants.
The incidence of sudden infant death syndrome (SIDS) is higher in preterm compared with term infants; the increased incidence is inversely related to gestational age.1 The association of prone compared with supine sleeping and a higher sudden infant death rate is well documented. The odds ratio for prone sleeping and SIDS in term born infants is 13.9 and in preterm infants is 48.4.2 A possible explanation for the increased risk of SIDS in the prone position is that infants arouse less easily and sleep more deeply in that position. In a case-control study3 polygraphic sleep recordings of 16 infants who subsequently died of SIDS suggested that arousals were less common in infants who later died of SIDS than matched control infants. Healthy term-born infants, regardless of their usual sleeping position, were demonstrated to sleep longer, with a greater duration of nonrapid eye movement sleep and less arousals in the prone compared with the supine position when studied at 3 months of age,4 and arousal thresholds were significantly higher in both active and quiet sleep when term-born infants were studied sleeping prone at both 2 to 3 weeks and 2 to 3 months.5 In addition, healthy preterm infants born between 30 and 35 weeks postmenstrual age (PMA) have been demonstrated to have higher arousal thresholds in both active and quiet sleep when slept prone at 36 to 38 weeks (PMA).6 Among infants born before 32 weeks of gestational age studied before neonatal unit discharge, however, no position dependency was noted in either the arousal or apnea/hypopnea indexes, but only 15 infants were studied.7 As a consequence, we felt it important to test the hypothesis that very prematurely born infants being prepared for NICU discharge would have decreased spontaneous arousals and more central apneas and sleep longer in the prone position. Our study design also allowed us to determine whether there were interactions among sleeping position, type of sleep, and apnea, arousal, or awakening.
Infants born at <33 weeks of gestational age being prepared for neonatal unit discharge were eligible for entry into this study. When the infants were tolerating 3-hourly feeds, their parents were approached and the study design explained. If informed, written consent was obtained, infants were studied on 2 successive days on the neonatal unit. On each day, infants were examined in both the supine and prone position each for 3 hours, the examination in each position beginning immediately after the infant had been fed. Infants were studied during both a morning and an afternoon sleep interrupted by a “midday” feed when sleep position was changed from prone to supine or vice versa. The order in which the positions were examined was randomized between infants; on the following day, the positions were examined in the reverse order. The results are expressed as the mean for each position from the 2 study days.
Video-polysomnographic recordings were made using an Alice 4 sleep study system (Alice Recording System, Respironics, Carlsbad, CA). Recordings were made with 2 scalp electroencephalograms ([EEGs] central and occipital leads [C4/A1 and C3/A2]), 2 electro-oculograms, and 1 electrocardiograph lead. Thoracic respiratory movements were measured by impedance. Chest and abdominal bands were used, and chest and abdominal wall movement was assessed using a piezoelectric transducer. Airflow was measured with thermistors taped under both of the infant's nostrils. Gross body movements were assessed using 2 activity meters attached to the upper and lower limbs, as well as by video. The recording period was fully monitored, hence, any adjustments could be made immediately if any of the leads or the thermistor became displaced. Oxygen saturation was measured continuously with a pulse oximeter (Datex Omeda 3900, GE Healthcare, Chalfont St Giles, United Kingdom). The data were collected using the computerized sleep recorder. The recordings were examined to determine whether the infant had apneas, arousals, and/or awakenings during the period of sleep. Apneas were defined as a pause in respiration indicated by a lack of nasal airflow of ≥5 seconds. Apneas were classified as obstructive if, despite chest and abdominal wall movements, there was no nasal airflow; central if there was no nasal airflow and an absence of chest and abdominal wall movements; and mixed if there was a combination of central and obstructive apneas. An apnea index (the number of apneas per hour of sleep) was calculated for each type of apnea. Arousals were defined as spontaneous body movements of ≥10 seconds8 and an awakening as an arousal of ≥60 seconds or crying. Sleep efficiency was calculated as the total sleep time divided by the recording time. Each recording was divided into 30-second epochs and the type of sleep during each epoch determined manually.9 Sleep was divided into 3 types: active, quiet, or indeterminate. Active sleep was diagnosed if there was a continuous low voltage EEG with ≥1 eye movement per epoch (30 seconds9), quiet sleep if the EEG was discontinuous with no eye movements, and indeterminate sleep if there was discordance between the EEG and eye movements.
To determine the reliability of the sleep staging, 5-minute epochs were randomly selected from the first 8 studies and were independently scored by 2 researchers with a third comparing their scores. Cohen's κ was used to estimate the agreement between the categorization of sleep staging by the 2 observers. We independently scored 230 epochs. There was a good level of agreement (a κ value of 0.73).
Differences between sleep positions were tested for statistical significance using the Wilcoxon signed rank test. Multiple regression analysis, which allowed for serial measures in the infants, was used to investigate the effects of position and type of sleep on duration of sleep and whether there was an interaction between sleep position and type of sleep. Negative binomial regression, which allowed for serial measures, was used to investigate possible effects of both position and type of sleep on numbers of apneas, arousals, or awakenings and to test for any interaction effects. Gestational age and postnatal age were included in the models as confounders where their effect was statistically significant. The analyses of effects of position and type of sleep were repeated excluding episodes of periodic breathing. Periodic breathing was defined as an episode of ≥3 respiratory pauses of ≥3 seconds of duration with intervening periods of respiratory movements of ≤20 seconds. These represented only 1.18% of the total duration of the studies, and the occurrences of periodic breathing did not differ significantly between sleeping position (P = .36). Additional analysis was undertaken for apneas associated with desaturation alone, that is, a drop in the oxygen saturation level to <90% for ≥6 seconds during or within 10 seconds of an apnea. (There were no apneas associated with bradycardia alone and very few associated with both.) Administration of supplementary oxygen to preterm infants has been shown to be associated with an increase in the overall duration and percentage of total sleep time spent in quiet sleep.10 Hence, we have further analyzed our results according to whether or not the infants were receiving supplementary oxygen at the time of study, that is, those who were developing or had bronchopulmonary dysplasia ([BPD] BPD group) and those that did not have BPD (non-BPD group).
Recruitment of 24 infants allowed detection of differences between the prone and supine positions of 0.66 SDs of the difference between prone and supine measurements within the subjects, with 90% power at the 5% level.
Twenty-four infants with median gestational age of 27.9 (range: 24.9–32) weeks and birth weight of 995 (range: 660–1614) g were studied at a median PMA of 36.4 (range: 34.6–40.7) weeks. Fourteen infants developed BPD (that is, they were oxygen dependent beyond 36 weeks' PMA; BPD group), their median gestational age was 26.9 (range: 24.9–29.6) weeks and birth weight was 814 (range: 660–1370) g, which differed significantly from that of the 10 non-BPD infants (median gestational age: 31 [26.6–32] weeks and median birth weight 1244 [828–1614] g; P = .003 and P = .003, respectively). The 2 groups were studied at similar PMAs (median: 37.2; range: 35.4–40.7 weeks for BPD infants; median: 36.4; range: 34.6–40.1 weeks for non-BPD infants; P = .28). The infants had all received a methylxanthine to facilitate weaning from the ventilator11 and remained on caffeine until 32 weeks' PMA or longer if they had had significant apneas. The BPD infants were studied while receiving supplementary oxygen to maintain their oxygen saturations between 92% and 95%, but none of the infants were receiving medications at the time of study, although 3 were receiving antireflux medication but were asymptomatic. The routine policy of the neonatal unit was that prematurely born infants were slept supine ≥2 weeks before discharge from the neonatal unit, thus, the usual sleeping position for all of the infants examined was the supine position. This study was approved by King's College Hospital Research Ethics Committee.
There was no significant difference in the recording time in the 2 positions. In the prone position, the infants slept significantly longer (P = .0001) and more efficiently (P <.0001), and they spent less time in active sleep (P = .003) but longer in quiet sleep (P <.0001; Table 1). In the supine position, the infants had more awakenings (P <.0003) and arousals (P <.0001) and less central apneas (P <.04) but more obstructive apneas (P = .007; Table 1). Comparison of findings in the supine to the prone position demonstrated that the ranges of durations of apneas were similar (central apneas: 5–11, 5–10 seconds; obstructive apneas: 5–12, 5–13 seconds; mixed apneas: 7–15, 7–17 seconds, respectively).
Multiple regression analysis revealed a significant effect of both sleep position and type of sleep on the duration of sleep (P <.0001; P <.001, respectively). In addition, a significant interaction between position and type of sleep was demonstrated (P = .002). The difference in the duration of sleep between active and quiet sleep was greater when the infants were sleeping supine (mean: 87 [SD: 22] minutes of active sleep; mean: 18 [SD: 7] minutes of quiet sleep) rather than prone (mean: 101 [SD: 18] minutes of active sleep; mean: 35 [SD: 12] minutes of quiet sleep).
Position (P = .008) but not type of sleep (P = .63) was significantly related to the number of central apneas. Infants were 1.4 (95% confidence interval [CI]: 1.1–1.7) times as likely to have a central apnea when sleeping prone rather than supine. Similarly, only position (P <.001) and not type of sleep (P = .41) was related to the number of obstructive apneas. Infants were 2.3 (95% CI: 1.5–4.1) times as likely to have an obstructive apnea when sleeping supine rather than prone. For mixed apneas, there were significant effects of both position and type of sleep (P <.001; P = .001, respectively). Infants were 1.8 (95% CI: 1.3–2.5) times as likely to have a mixed apnea when sleeping supine rather than prone after allowing for type of sleep. Mixed apneas were less common during both quiet and indeterminate sleep than during active sleep (rate ratios [95% CI]: 0.41 [0.25–0.66]; 0.76 [0.40–1.45], respectively) after allowing for sleep position. Analysis allowing for gestational age and PMA (where significant) and exclusion of episodes of periodic breathing demonstrated similar findings. There were no significant effects of sleeping position on central apneas with desaturation (rate ratio: 1.8 [95% CI: 0.6–5.7]; P = .30, prone/supine) or type of sleep (P = .37) or on obstructive apneas with desaturation (rate ratio: 0.9 [95% CI 0.3–2.6]; P = .83) or type of sleep (P = .20). There was, however, a significant increase in the risk of mixed apneas with desaturation for infants sleeping supine (rate ratio: 5.4 [95% CI: 1.9–15.6]; P = .002) but no significant association with type of sleep (P = .12).
Position (P <.001) and sleep type (P <.001) were each significantly related to the number of arousals. Infants were 1.4 (95% CI: 1.3–1.6) times as likely to have an arousal in the supine rather than the prone position. They were less likely to have an arousal in quiet compared with active sleep (rate ratio: 0.19 [95% CI: 0.15–0.25]) and indeterminate compared with active sleep (rate ratio: 0.73 [95% CI 0.57–0.94]). Similarly, position (P <.001) and sleep type (P <.001) were significantly related to the number of awakenings. Infants were 2.0 times as likely to have an awakening in the supine compared with the prone position (95% CI: 1.7–2.3). They were less likely to have an awakening in quiet compared with active sleep (rate ratio: 0.09 [95% CI: 0.05–0.15]) and indeterminate compared with active sleep (rate ratio: 0.81 [95% CI: 0.58–1.12]). There were, however, no significant interactions between position and sleep type and either arousals or awakenings.
In the supine position, both the BPD (Table 2) and the non-BPD infants (Table 3) had more awakenings and arousals, and the infants spent a lower proportion of sleep time in quiet sleep. In the BPD infants only, a significant effect of sleep position on the occurrence of apnea was seen (Table 2). Comparison, however, of the differences seen between the 2 positions in the BPD and non-BPD infants did not reveal any statistically significant differences in the magnitude of the differences in the 2 positions, that is, there was little evidence of an interaction between the effect of sleep position and BPD status.
We have demonstrated that, in the prone compared with supine position, very prematurely born infants studied at a median PMA of 36 weeks had fewer spontaneous arousals and had more central apneas. In addition, they slept longer and spent a greater time in quiet sleep regardless of their oxygen dependency status. Our findings of a lower spontaneous arousal rate when sleeping prone in very prematurely born infants is consistent with the finding of a higher arousal threshold in the prone position in infants born at 30 to 35 weeks of gestational age studied at a similar PMA.6 Our data are also consistent with findings in term-born infants.5,12 In healthy term-born infants studied by daytime polysomnography, arousal thresholds to air-jet stimulation were significantly higher in prone compared with supine sleeping at both 2 to 3 weeks and 2 to 3 months.5 In addition, spontaneous arousals from sleep were less frequent in term-born infants slept prone.12 The results of previous studies13,14 examining the effect of sleep position on preterm infants are consistent with our finding of greater sleep efficiency in the prone position in more immature infants when examined at a PMA of 36 weeks. Materson et al13 reported that healthy preterm infants with a gestational age of 28 to 34 weeks studied at 12 to 57 days (none were requiring supplementary oxygen) spent more time in wakefulness and slept less in the supine position.13 In the second study,14 asymptomatic infants with a mean gestational age of 32 weeks were studied at a PMA of 36 weeks and were also demonstrated to have less quiet sleep and more awakenings in the supine compared with the prone position.
We found that arousals were less likely to occur in quiet or indeterminate than in active sleep. The arousal thresholds of term-born infants have been demonstrated to be elevated in quiet compared with active sleep.5,15 In a previous study of prematurely born infants,16 however, no sleep state-related difference in arousal thresholds was noted when the infants were examined at 36 weeks' PMA, but a difference was highlighted when they were studied at 2 to 3 months after term. The infants examined16 were born at a higher gestational age (range: 31–35 weeks) than those currently studied. In comparison with the earlier study,16 we were able to examine the infants for a longer time in each position and on 2 successive days. We did not see a significant interaction among sleep position, type of sleep, and the occurrence of arousals or awakenings. This is consistent with results from term-born infants, with arousal thresholds being found to be higher in the prone compared with the supine position, regardless of whether the infants were studied in quiet or active sleep.5
Our finding of significantly more central apneas in the prone position differs from that of previous reports.7,17–19 Previous studies have found either no effect of sleeping position7,17,18 or that central apnea was more common in the supine position.19 Three of the studies,17–19 however, examined infants with apnea and bradycardia, whereas the present study population was asymptomatic. In the study7 in which infants ready for discharge from the neonatal unit were also examined,7 no position dependency was found on an apnea/hypopnea index. It is possible that inclusion of hypopneas, defined as a 50% reduction in the amplitude of 2 of the respiratory channels for ≥3 seconds, may have masked differences in the frequency of apneas between the 2 positions. An alternative explanation is that the numbers studied7 were too small to confidently detect differences in arousal and apnea/hypapnea index, particularly because the authors report a wide range in their results. In addition, the infants were only studied once in each position, and, thus, time of day may have influenced their results.7 Our findings that central apneas were significantly more common in the prone position further emphasize the importance of supine sleeping after neonatal unit discharge.
Obstructive apneas were significantly more common in the supine position regardless of type of sleep. Key to the diagnosis of obstructive apnea is accurate detection of airflow. We, therefore, used a thermistor, which was securely taped in position. We did not use diaphragmatic electromyogram to assist in determining obstruction, but the recording period was fully monitored so that adjustments could be made to the thermistor if it became displaced, and the infants were videoed throughout; hence, we were confident that we did not miss obstructive apneas because of impaired monitoring. We found that ∼12% of the total apneas were obstructive, which is similar to the 6% and 10% found in previous studies.20,21 The number of obstructive apneas per hour was low in each position, but this likely reflects that our study population was asymptomatic prematurely born infants being prepared for discharge home. Mixed apneas, which contain both central and obstructive components, were also more common in the supine position. Mixed apneas usually commence with central apnea, and then airway occlusion occurs22 because of loss of tone of the muscles of the upper airway. Obstructing the airway, however, can lead to central apnea in prematurely born infants,23 which may explain our findings of more mixed, as well as obstructive apneas, in the supine position.
It has been anticipated that infants would sleep better in their usual sleep position.24 Healthy term infants studied at 3 months of age, however, slept longer prone regardless of whether they usually slept prone or supine.12 All of the infants in this study were usually slept supine and yet also slept longer when slept prone. It is not clear why the prone position favors sleep, although several explanations have been proposed. Factors known to interfere with sleep maintenance, such as environmental temperature, infant handling, and feeding,25 were kept constant in the 2 sleeping positions in our study. It is possible that placing the infants prone with their head to the side reduces their exposure to both light and sound, thus lessening their sensory input. The face-down position with the infant's cheek resting on a firm warm surface may have a soothing effect and, hence, favor better sleep maintenance. Another potential mechanism of the prone position favoring sleep is that there may be less acid gastroesophageal reflux. Acid reaching the proximal position of the esophagus is a potent arousal stimulus in infants.26 Gastric reflux has been shown to be lower in the prone compared with the supine position in term-born infants.27 None of the infants in the current study, however, had symptomatic acid gastric reflux. Other potential mechanisms may be the development of hypercarbia, because sleeping prone increases the resistance to air flow, and rebreathing exhaled gas could occur.12 Measurements of transcutaneous gas tensions, however, have demonstrated lower transcutaneous carbon dioxide levels when infants were sleeping prone rather than supine.28 Changes in core temperature can also lead to sleep disruption. We did not measure the infants' temperatures in the 2 positions, thus, we cannot exclude that there were differences in temperature because of differences in heat losses from the skin resulting from the changes in body position,13 but in another study,6 there was no difference in the rectal temperature of term infants in the prone compared with the supine position. In adults, sleep can be disrupted by recurrent airway obstructions. We found obstructive apneas to be significantly less common in the prone compared with the supine position, as has been noted in term-born infants.17
The effect of sleeping position on state distribution in preterm infants has been shown to vary with time after feeding,29 with increases in quiet sleep in the prone position being found only within the first hour after a feed and again near the end of the interval feed. In this study, we studied the infants either supine or prone immediately after a feed and for the subsequent 3 hours. To further exclude any bias regarding time of feeding or time of day, the infants were studied in each position on 2 successive days, with the starting position reversed on the second day and the results on the 2 days in each position meaned.
All of the infants had previously received caffeine. Caffeine administration is associated with a reduction in apnea and has been suggested to have an effect on sleep. In adult cats, the initial effect is a marked increase in wakefulness,30 but, when the cats were habituated, their total sleep returned to normal, although differences spent in different sleep stages persisted. The observed modifications persisted ∼20 days after caffeine withdrawal, suggesting that our population may still have been affected by their previous medication. Ten-hour polysomnographic recordings in 15 neurologically normal and clinically stable neonates of 33- to 34-week PMA, 10 of whom had been treated for ≥3 days with caffeine, however, demonstrated no significant difference in sleep organization between those who were and were not receiving caffeine.31
Regression analysis demonstrated that the differences between positions were similar in the BPD and non-BPD infants, yet all of the BPD infants were receiving supplementary oxygen. In a previous study,10 administration of low flow supplementary oxygen was associated with a reduction in apneas and an increased duration of sleep. The infants were of a similar gestational and PMA to the present population, but all were in air before examination. As a consequence, administration of supplementary oxygen resulted in their oxygen saturations increasing from a mean of 96.4% to 98.7%.10 In contrast, in the present study, supplementary oxygen was given to the BPD infants, because they would have been hypoxic in air, and the supplementary oxygen was given to maintain their oxygen saturation levels between 92% and 95%. Thus, the difference in oxygen saturation levels in the 2 studies may explain the differences in the results.
We have demonstrated that prematurely born infants, when studied before neonatal unit discharge, sleep longer with less arousals and more central apneas in the prone compared with the supine position. Arousal from sleep is an important survival response to life-threatening events, such as prolonged apnea.6 By arousing from sleep, ventilation is increased32 and a behavioral response evoked.33 Any impairment in arousal could then contribute to SIDS. The lower arousal rate and greater number of central apneas that we demonstrate in the prone position emphasize the importance of counseling parents to ensure that their prematurely born infant sleeps supine when discharged from the hospital.
Dr Bhat was supported by King's College Hospital Joint Research Committee, which, with the Foundation of Sudden Infant Death, provided the equipment used in this study.
We thank Deirdre Gibbons for secretarial assistance.
- Accepted January 18, 2006.
- Address correspondence to Anne Greenough, Department of Child Health, King's College Hospital, London SE5 9PJ United Kingdom. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Malloy MH, Hoffman HJ. Prematurity, sudden infant death syndrome, and age of death. Pediatrics.1995;96 :464– 471
- ↵Oyen N, Markestad T, Skjaerven R, et al. Combined effects of sleeping position and prenatal risk factors in sudden infant death syndrome: the Nordic epidemiological SIDS study. Pediatrics.1997;100 :613– 621
- ↵Curzi-Dascalova L, Mirmiran M. Manual of methods of recording and analysing sleep-wakefulness states in preterm and full term infants. Les Editions INSERM.1996:160
- ↵Simakajornboon N, Beckerman RC, Mack C, Sharon D, Gozal D. Effect of supplementary oxygen on sleep architecture and cardiorespiratory events in preterm infants. Pediatrics.2002;110 :884– 888
- ↵Kahn A, Grosswasser J, Sottiaux M, Rebuffat E, Franco P, Dramarx M. Prone or supine body position and sleep characteristics in infants. Pediatrics.1993;91 :1112– 1115
- ↵Materson J, Zucker C, Schulze K. Prone and supine positioning effects on energy expenditure and behaviour of low birthweight neonates. Pediatrics.1987;80 :689– 902
- ↵Goto K, Mirmiran M, Adams MM, et al. More awakenings and heart rate variability during supine sleeping in preterm infants. Pediatrics.1999;103 :603– 609
- ↵Orr WC, Stahl ML, Duke J, et al. Effect of sleep state and position on the incidence of obstructive and central apnea in infants. Pediatrics.1985;75 :832– 835
- ↵Kurlak LO, Ruggins NR, Stephenson TJ. Effect of nursing position on incidence, type and duration of clinically significant apnoea in preterm infants. Arch Dis Child Fetal Neonatal Ed.1994;71 :F16– F19
- ↵Upton CJ, Milner AD, Stokes GM. Upper airway patency during apnoea of prematurity. Arch Dis Child.1992;67 :419– 424
- ↵Weissbult M Ed. Healthy sleep habits. Happy child. New York NY Fawcett Colubmine.1987:35– 36
- ↵Eaton-Evans J, Dugdale AE. Sleep patterns of infants in the first year of life. Arch Dis Child.1987;63 :647– 649
- Copyright © 2006 by the American Academy of Pediatrics