Objective. The Task Force of The American Academy of Pediatrics (1996) recommends the nonprone sleeping position for asymptomatic preterm infants to prevent sudden infant death syndrome. The mechanism by which the nonprone sleeping position reduces the rate of sudden infant death syndrome is unclear for full-term infants and the precise effect of sleeping position on sleep and cardiorespiratory characteristics has never been addressed in preterm infants. The purpose of the present study was to clarify the effect of sleeping position on sleep and cardiorespiratory characteristics in preterm infants at an age when they are ready for discharge.
Study Design. Sixteen asymptomatic preterm infants were studied in both supine and prone sleeping positions at 36.5 ± 0.6 weeks' postconceptional age using videosomnography. Sleep, respiratory, and heart rate characteristics were compared between the two positions using each infant as his/her own control.
Results. More awakenings (ie, arousals ≥60 seconds) were seen during all sleep states in the supine sleeping position but overall the total sleep and percent sleep state were not affected by sleeping position. After each feeding, the first quiet sleep was significantly shorter, with more heart rate variability and awakenings in the supine position. There were no significant differences in the occurrence of arousals (<60 seconds) or the incidence or severity of apnea and periodic breathing. No clinically significant apnea (≥15 seconds), bradycardia, or oxygen desaturations were seen.
Conclusion. In 36-week-postconceptional age preterm infants, the supine sleeping position had less quiet sleep and was associated with greater heart rate variability during the first sleep cycle after the feeding. More awakenings were seen during all sleep states in the supine position. These data support the American Academy of Pediatrics recommendation for “Back to Sleep” for asymptomatic preterm infants because more awakenings and lower threshold for arousal may provide some benefit for the infant responding to a life-threatening event. However, further studies are needed to address positional effect on the physiologic measures in preterm infants at older ages (later stages of development). Precisely what constitutes the most healthy or advantageous sleep for newborn infants remains an important question.
- preterm infant
- sleeping position
- heart rate variability
- sudden infant death syndrome
It is widely accepted that prone sleeping is associated with an increased risk of sudden infant death syndrome (SIDS).1–4 However, the mechanisms by which the supine sleeping position reduces the rate of SIDS are elusive.5,,6There are few reports that give quantitative data on the effect of sleeping position on sleep and heart rate characteristics in infants,7–12 particularly in preterm infants.9 In comparison, there are more reports on effects of body position on respiratory function in preterm infants during neonatal intensive care.13–20 Although the beneficial effects of the prone position on pulmonary function were variable among these reports, it is generally believed that the prone position has advantages greater than the supine position in preterm infants with respiratory illness.14,,16,19,20
The American Academy of Pediatrics recommended in 1992 that term infants, and in 1996 also asymptomatic preterm infants, sleep nonprone to reduce risk for SIDS. The latter recommendation was based on the viewpoint that the advantage of reduced SIDS risk was greater than any short-term clinical benefit.3,,421–23 Although the basis for the recommendation is valid, it presents a dilemma for neonatologists and pediatricians because of the lack of acute and long-term quantitative data on the effect of sleeping position on sleep and cardiorespiratory physiology in preterm infants, particularly at discharge.9,,13
One method of understanding the acute clinical effects of sleeping position would be to routinely place infants in both the prone and supine positions during sleep before discharge from the hospital to determine if there are any problems. This approach has been used to observe infants in a car seat to assess for problems with oxygen desaturation or apnea and bradycardia.24–26 In a similar manner, we chose to study infants at a postconceptional age (PCA) of ∼36 weeks, because most preterm infants are discharged from the hospital around this corrected age and the results are clinically applicable for discharge planning. Furthermore, at 36 to 37 weeks' PCA sleep states are better differentiated and the effect of sleeping position on sleep architecture could be quantified more precisely.
Sixteen asymptomatic preterm infants (4 girls, 12 boys), with a gestational age of 32.2 ± 3.0 weeks (range, 27–36 weeks) and birth weight of 1733 ± 135 g (range, 945–2868 g), were studied at 36.5 ± 0.6 weeks' PCA. The following exclusion criteria were used: 1) major congenital or cardiac anomalies; 2) abnormal neurologic findings including intraventricular hemorrhage >Grade II; 3) birth weight <10th percentile or >90th percentile; 4) maternal substance abuse; 5) chronic lung disease; 6) gastroesophageal reflux; and 7) intrauterine growth retardation or small for gestational age. The infants were clinically stable, were not on positive pressure, were receiving all nutrients enterally, and had no supplemental oxygen. The protocol was approved by the Human Subjects Institutional Review Board at Stanford University. Informed parental consent was obtained before the study.
Videosomnography (polysomnography combined with time-lapse video) recordings were made in the intermediate care nursery, usually after the infants were nursed. The infants were dressed in shirt and diaper, which is routine nursery care and were lying in their cribs. The recordings lasted 6 hours (two interfeed intervals), usually from 11 am until 5 pm. After the first feeding, each infant was randomly assigned to either supine or prone position and the position was reversed after the second feeding.
Polysomnographic recordings were obtained with an ambulatory digital sleep recorder (4.5 cm × 8.3 cm × 13.3 cm; Embla, Flaga hf. Medical Devices, Reykjavı́k, Iceland). The following variables were recorded simultaneously: four scalp electroencephalograms (C3/A2, C4/A1, O1/A2, O2/A1), two electrooculograms (ROC/A1, LOC/A2), a chin electromyogram, an electrocardiogram (ECG), thoracic respiratory movement (piezoelectric respiratory movement sensor (Pro-Tech Services Inc, Woodinville, WA), and oxygen saturation (Nellcor N-200, Hayward, CA). All parameters were continuously stored on a hard disk (PCMCIA card) by the Embla. After each recording, the PCMCIA card was plugged into a desktop computer (Power Tower Pro 200, Power Computing Corp, Round Rock, TX) in our laboratory. The data were viewed and analyzed using Embla and Somnologica software (Flaga hf. Medical Devices).
Each infant was recorded using a time-lapse videorecorder (Panasonic AG6730, Osaka, Japan) and (Charged Coupled Device) infrared camera (Panasonic WVCD810, Osaka, Japan) to observe the behavior, including startles, myoclonic twitches, vocalizations, general body movements, and eye movements of each infant. The time-lapse video tapes were replayed at 11 times the recording speed and were synchronized with the polygraphic recordings for later analysis.
Sleep State and Arousal
Each 30-second epoch of the recordings was analyzed and classified as either awake (W), active sleep (AS), indeterminate sleep (IS), quiet sleep (QS), or periods of out-of-crib (or out-of-view or intervention) according to the criteria in the literature.27,,28 A 1-minute moving average of states was used. Because we were not able to observe all body movements of each infant precisely, particularly small limb movements, we have focused on the visual presence of general body movements and the simultaneous existence of artifact on at least one channel in the polygraphic recording. The duration of each movement was defined by direct observation of movement in the video. In addition to the polysomnographic recording, the videotape recording was reviewed to backup sleep state scoring, arousal, and awakening. As described in other studies in infants,29,,30 arousal was defined as the occurrence of one of the following criteria: 1) body movement lasting for 10 seconds or more; 2) cry; or 3) eye opening lasting at least for 5 seconds or more. A minimum of 30 continuous seconds of intervening sleep was necessary before scoring a subsequent arousal occurrence. Arousal time lasting for 60 seconds or more was defined as an awakening. Sleep onset after feeding was defined as two consecutive epochs of AS, IS, or QS. Scoring was suspended if W lasted for 10 minutes or more.
After scoring sleep states visually using Curzi-Dascalova and Mirmiran's28 manual for sleep scoring in preterm infants, hypnograms were made with Somnologica software (Flaga hf. Medical Devices). Figure 1 is an example of a hypnogram for an infant. Total sleep time (TST) for each position was defined as the summation of sleep epochs (QS, AS, and IS) from sleep onset until the start of the next feeding. We obtained the following sleep measures in each sleeping position from each infant: 1) AS, IS, and QS percentages (of TST); 2) number of awakenings per 100 minutes; 3) ratio of total minutes W to TST minutes; 4) number of arousals per 100 minutes TST; and 5) number of arousals during AS + IS, and QS.
Figure 1 illustrates the differences in sleep characteristics between the first and second AS periods in both positions. The second AS was always interrupted by more awakenings. This observation led to our decision to analyze sleep characteristics of each infant, not only based on sleeping position, but also according to a given sleep cycle. Because all recordings included at least one sleep cycle that usually began with AS, followed by QS, then followed by a second AS (but no second QS), we measured the following values: 1) the mean length of the first AS, first QS, and second AS; 2) number of awakenings or arousals per each of these 3 conditions; and 3) the longest bouts of AS and QS. If two bouts of AS was interrupted by IS and/or awakening that lasted for <10 minutes without intervening QS, they were considered as belonging to the initial AS (see Fig 1 for further explanation).
Apnea and Periodic Breathing (PB)
Apnea durations were measured from the piezoelectric respiration signal starting at the end of an inspiration and lasting until the start of the subsequent inspiration. All apneas lasting ≥3 seconds were analyzed. These were divided into three subgroups according to their duration and the presence of bradycardia and/or desaturation episodes: 1) apnea lasting 3 to 15 seconds; 2) apneas lasting 15 seconds or longer; and 3) apneas with bradycardia and/or desaturation.
PB was defined as an episode of three or more respiratory pauses lasting for 3 seconds or longer with intervening periods of normal respiration <20 seconds.31 We calculated the following respiratory measures in each position for each infant: 1) number of apnea per 100 minutes of total recording time (apnea index); 2) number of PB per 100 minutes of TST; 3) percentage of PB per TST; and 4) duration of longest PB.32 Analysis of PB was not divided into AS + IS and QS periods (as was done for apnea analysis) because PB continued into the next state.
Heart Rate Characteristics
The ECG data were sampled at an effective accuracy of 1 millisecond and analyzed using Somnologica's heart rate variability program. Artifacts on ECG traces were identified by visual inspection and were not used in further analysis. Group mean, standard deviation, and minimal and maximal heart rates were calculated for each 64 seconds. Heart rate was only analyzed in the first AS and first QS periods.
All values were expressed as mean plus/minus standard deviation. Statistical analyses were done using paired t test, Wilcoxon signed rank test, and Mann-Whitney U test. Significance was accepted at P < .05.
Table 1 shows sleep characteristics of supine and prone positions. There were no significant differences in the percentage of each sleep state between the supine and the prone position. The number of arousals per 100 minutes of AS + IS, and QS in the prone position were not significantly different from these values in the supine position. However, the number of awakenings (ie, arousal ≥60 seconds) and the ratio of total awakening time to TST time were significantly higher in the supine compared with the prone position.
The sleep characteristics in the first AS, first QS, and second AS are shown in Table 2. The noticeable differences between the two positions were found in the first QS. The duration of the first QS was significantly longer in the prone position. In addition, the prone position was associated with the longest bout of QS and less awakening in the first QS in comparison with the supine position. Trends were found in less awakening in the prone position both in the first AS and the second AS.
There was no pathologic apnea, ie, lasting >15 seconds or accompanied by bradycardia and/or desaturation in either position, during the recording of all 16 infants. Apnea indexes of AS + IS and QS did not show any significant differences between the supine and prone positions, although a trend toward more apnea in AS + IS during prone position was found (Table 3).
There were no significant differences in PB between the two sleeping positions with respect to both the number and percentages of PB as well as the longest duration of PB (Table 3).
An example of change in minimal, maximal, and mean heart rate as well as the standard deviation of heart rate mean in each position in an infant is shown in Fig 2. Heart rate variability showed considerable change even in the same sleep state and sleeping position and was significantly higher in AS versus QS. Table 4 summarizes the differences in heart rate variability characteristics between the two positions. The mean heart rate and maximal heart rate in the supine position were significantly higher than in the prone position during the first AS. During the first QS, maximal heart rate and heart rate variability (mean of standard deviation of heart rate) were significantly higher in the supine versus the prone position. A trend toward higher mean heart rate in supine was also found in the first QS.
Previous studies, primarily in term infants, have shown more sleep and fewer arousals in the prone sleeping position. We hypothesized that preterm infants, by the time they are term corrected age, would show similar results. However, we found no significant differences in total sleep or in the percentage of each sleep state between the supine and the prone positions. There was no difference in the number of arousals during QS (trend toward fewer arousals during AS in prone). However, there were significantly more awakenings (arousals >1 minute) observed in each sleep state in the supine position. Moreover, an increase in QS in the prone position was found during the first sleep cycle. The longer sleep bouts and fewer awakenings associated with the prone position would support higher vulnerability for preterm infants to life-threatening events during prone sleeping.
The Kahn et al11 study in 3-month-old full-term infants showed that the prone position was associated with a significant increase in total sleep duration, QS, and fewer arousals. These findings were repeatable in the same infant and independent of whether the infant was a habitual prone or supine sleeper. Our data support that the prone sleeping position overall is associated with less awakening and particularly in the first sleep cycle after a feeding. However, the effect seems to be less than what has been reported in the term infant during the first few months of life.11 A greater sleep-maintaining effect and higher threshold for arousal during prone sleeping may, in part, explain the higher risk of SIDS.33,,34 The mechanism for how sleeping position influences sleep organization is still unclear.
Several factors should be considered when comparing our data to the literature. With the exception of one report, all the infants in the earlier studies were full-term infants. Masterson and coworkers9 were the only earlier group recording preterm infants, but the corrected ages of their infants at the time of recording were not stated. They demonstrated that the preterm infant with a gestational age ranging from 28 to 34 weeks recorded at 12 to 57 postnatal days spent more time in wakefulness, had less sleep, and had higher metabolic rates, measured by oxygen consumption and carbon dioxide production, in the supine position. They also suggested (data not shown in results) that less sleep was the result of a reduction in QS and increased wakefulness because there was no change in percentage of time spent in AS between the two positions. With respect to the number and total duration of awakenings, our findings agree with Masterson and coworkers.9 All other studies have examined the positional effect on sleep in full-term neonates.7–9Hashimoto et al8 reported that full-term infants at 3 to 4 days of age spent less time in wakefulness, more time in QS, and had fewer body movements (indicating fewer arousals) in the prone position. Brackbill and Amemiya7,,10 showed similar results in full-term neonates but found more time spent in AS associated with the prone position. Konishi,36 who did not record sleep, reported that the mean duration of a sustained posture was longer in the prone position both in preterm and full-term infants compared with supine, suggesting fewer state changes. In contrast, Casaer et al12 found no consistent differences in full-term infants between the two positions in total percentage of QS and W, although AS was less in the supine position.
Another difference among studies is whether the infants were clothed. In our study and Casaer's,12 the patients were clothed during sleep. In the four other studies,7–10 the infants were unclothed during the polygraphic recordings. In Fig 3, we have summarized the percentage of AS and QS in all these studies to better illustrate sleep results at 36 to 37 weeks' PCA preterm compared with full-term newborn infants.8,37–40 Differences in sleeping position and clothing may, in part, explain differences in sleep results.41–44
This study did not show that sleep positioning significantly influences the occurrence of apnea and PB. Heimler et al16 and Kurlak et al20 observed that the supine position was associated with more frequent central apnea in the preterm infants at an earlier PCA than the infants in our study and in those with clinically significant apnea, and also reported more PB in supine. They speculated that differences in lung function and occurrence of gastroesophageal reflux contributed to the genesis of the positional difference in apnea.16,,20 We recorded clinically stable, asymptomatic preterm infants at 36 to 37 weeks' PCA (just before discharge) without clinical apneic episodes. The absence of a positional effect on apnea seems to support that pulmonary function is unaffected by position in asymptomatic preterm infants.17 However, it should be acknowledged that Martin et al,13 by recording pulmonary function in younger healthy preterm infants, found more respiratory vulnerability in the supine position. A large variability in PB was found among the 16 infants, eg, some infants had episodes lasting longer than 51 minutes, but some infants had no PB at all. Similar data on variability have been published.45
In comparing heart rate variability during QS between the two positions, small but significant differences in minimal, maximal, and standard deviation from mean heart rate suggested that the prone position was associated with less heart rate variability. Our study did not support the findings of previous studies which reported that mean heart rate was higher in the prone than in the supine position.7,,8,10,16,17,19 Hashimoto et al8suggested that higher heart rate in the prone position during QS might be because of the increased effort of respiratory muscles. Because details of the method were not described in their study, we cannot critically compare our results with theirs. However, we observed that general body movements, short arousals, and sighs can have a significant effect on heart rate variability. We excluded these periods in our analysis so that sleep state effects on heart rate could be analyzed more precisely. A limitation of the earlier studies is that a short period of recording was analyzed and the effect of behavioral state influences were not evaluated. We found less heart rate variability during prone QS that correlates with longer and deeper QS in prone with fewer awakenings. This is also in accordance with the lower energy expenditure found in preterm infants in the prone position.12
It was interesting to find that the prone position was accompanied by less heart rate variability. This is one of the first reports on the effect of sleeping position on the spontaneous heart rate variability in the preterm infant. Franco et al46 showed that prone sleeping was associated with a decrease in cardiac responses to auditory stimulation and a possible increase in orthosympathetic activity in the full-term infants with a median age of 11 weeks. Spontaneous heart rate variability at any given time represents the net effect of the parasympathetic and sympathetic influences. Because vagal tone may be a significant contributor to the resting heart rate,47–50 we speculate that the balance of sympathetic and parasympathetic tone, both of which are influenced by numerous physiologic factors, is affected by sleeping position in preterm infants.
This study showed that sleeping position affected sleep architecture and heart rate variability in preterm infants predominantly in the first sleep cycle after a feeding. How or why sleeping position affects sleep architecture and arousal threshold7–12 has not been further elucidated. Why the effect of position on sleep state duration was limited to the first sleep cycle is also unclear. Harper et al51 has suggested that sleep state after a feeding was different from that which followed a nonfeeding period of prolonged wakefulness. It is known that the ingestion of food increases metabolic rate.52 This change in metabolic rate could influence the temporal pattern of sleep during interfeed periods. Less energy expenditure and a reduced ability to lose heat have been reported in prone sleeping infants.9,,53 Nonprone sleepers, on the other hand, have greater exposure of face and head compared with prone sleeping infants and more freedom to move limbs, all which could facilitate heat loss.53 Sleeping position and the presence or absence of clothing may further influence metabolic rate and sleep architecture.
The precise mechanism(s) by which prone sleeping increases the risk for SIDS remains controversial. Potential mechanisms that have been proposed include impaired cardiorespiratory control, hyperthermia, rebreathing of carbon dioxide, arousal deficiency, and increased orthosympathetic activity.5 This study does not replicate the more consistent finding of an overall increase in QS and fewer short arousals that have been reported in prone sleeping full-term infants. We have found no contraindications for supine sleeping in asymptomatic preterm infants; on the contrary, we saw some potential benefits such as shorter bouts of QS in the first sleep cycle after a feeding and overall more long arousals throughout the interfeed period.
Our study showed that sleep and heart rate variability in preterm infants just before discharge is influenced by sleeping position. The duration of QS is less in the supine sleeping position in the first sleep cycle after a feeding. More awakenings were seen during all sleep states in the supine position but overall percentages of sleep state and total sleep were not affected by sleeping position. No adverse effects of the supine sleeping position were seen. Although our results support the recommendation for the supine sleeping position the advantages and potential disadvantages of sleeping position deserve further study, particularly in older (2–3 months corrected age, the age of maximum vulnerability for SIDS) asymptomatic preterm infants.
This work was supported by National Institutes of Health Grant HD29732, by the Bloomingdale's Fund, and by NWO (Nederlandse Organisztie voor Wetenschappelijk onderzoek), The Netherlands.
- Received March 20, 1998.
- Accepted August 11, 1998.
Reprint requests to (R.L.A.) Department of Pediatrics, 750 Welch Rd, Suite 315, Stanford University School of Medicine, Stanford, CA 94304.
- SIDS =
- sudden infant death syndrome •
- PCA =
- postconceptional age •
- ECG =
- electrocardiogram •
- W =
- awake •
- AS =
- active sleep •
- IS =
- indeterminate sleep •
- QS =
- quiet sleep •
- TST =
- total sleep time •
- PB =
- periodic breathing
- Mitchell EA,
- Brunt JM,
- Everard C
- Oyen N,
- Markestad T,
- Skaerven R,
- et al.
- AAP Task Force on Infant Positioning and SIDS
- American Academy of Pediatrics Task Force on Infant Positioning and SIDS
- ↵Glotzbach SF, Ariagno RL, Harper RM. Sleep and the sudden infant death syndrome. In: Ferber R, Kryger MH, eds. Principles and Practice of Sleep Medicine in the Child. Philadelphia, PA: WB Saunders; 1995:231–244
- ↵Corwin MJ. Infant sleep position and the sudden infant death syndrome. In: Frantz ID, ed. Tufts University School of Medicine and Floating Hospital for Children Reports on: Neonatal Respiratory Diseases. Newtown, PA: Associates in Medical Marketing Co, Inc; 1997:1–7
- Masterson J,
- Zucker C,
- Schulze K
- Kahn A,
- Groswasser J,
- Sottiaux M,
- Rebuffat E,
- Franco P,
- Dramaix M
- Casaer P,
- O'Brien MJ,
- Prechtl HF
- Martin RJ,
- DiFiore JM,
- Korenke CB,
- Randal H,
- Miller MJ,
- Brooks LJ
- Heimler R,
- Langlois J,
- Hodel DJ,
- Nelin LD,
- Sasidharan P
- Wolfson MR,
- Greenspan JS,
- Deoras KS,
- Allen JL,
- Shaffer TH
- Lipsky CL,
- Gibson E,
- Cullen JA,
- Rankin K,
- Spitzer AR
- Malloy MH,
- Hoffman HJ
- Bass JL,
- Mehta KA,
- Camara J
- Bass JL,
- Mehta KA
- ↵Anders T, Emde R, Parmelee AH. A Manual of Standardized Terminology, Techniques and Criteria for Scoring States of Sleep and Wakefulness in Newborn Infants. Los Angeles, CA: UCLA Brain Information Service; 1971
- ↵Curzi-Dascalova L, Mirmiran M. Manual of Methods for Recording and Analyzing Sleep-wakefulness States in Preterm and Full-term Infant. Paris, France: Les Éditions INSERM; 1996:1–162
- Thoppil CK,
- Belan MA,
- Cowen CP,
- Mathew OP
- Skadberg BT,
- Markestad T
- Glotzbach SF,
- Baldwin RB,
- Lederer NE,
- Tansey PA,
- Ariagno RL
- ↵Dubowitz L, Dubowitz V. The Neurological Assessment of the Preterm and Full-Term Newborn Infant. London, England: William Heinemann Medical Books Ltd; 1981
- ↵Zotterman Y. Thermal sensations. In: Field J, Magoun HW, Hall VE, eds. Handbook of Physiology Section 1. Washington DC: American Physiological Society; 1959;431–477
- ↵Mountcastle V. Neural mechanisms in somesthesis. In: Mountcastle V, ed. Medical Physiology. St Louis, MO: The CV Mosby Company; 1980:348–390
- ↵Glotzbach SF, Ariagno RL. Periodic breathing. In: Beckman PC, Brouillette ST, Hunt FE, beds. Respiratory Control Disorders in Infants and Children. Baltimore, MD: Williams & Wilkins; 1992:142–160
- France P,
- Groswasser J,
- Sottiaux M,
- Broad Field E,
- Kahn A
- ↵Hainsworth R. The control and physiological importance of heart rate. In: Malik M, Camm AJ, eds. Heart Rate Variability. New York, NY: Futura Publishing Company, Inc; 1995;3–19
- ↵Ganong WF. Review of Medical Physiology. Norwalk, CT: Appleton & Lange; 1995;1:255–258
- Copyright © 1999 American Academy of Pediatrics