 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* Pediatric Neurology and Clinical Neurophysiology
Pediatrics, Turku University Hospital, Turku, Finland
Department of Pediatrics, Hospital for Children and Adolescents, Helsinki, Finland
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods. A 24-hour sleep polygraphy study was conducted at home for 24 excessively crying infants and 23 control subjects at the age of 6 weeks. In addition, parental diaries were kept for 4 days.
Results. In sleep polygraphy recordings, no major differences between study groups were observed in either the duration or the structure of the 24-hour sleep. In the diaries, the parents overestimated the amount of sleep in both study groups. The parents of the control infants overestimated the amount of sleep more than the parents of excessively crying infants (69.8 minutes [standard deviation: 79.3] compared with 27.1 minutes [standard deviation: 65.4], respectively). In excessively crying infants, the proportion of rapid eye movement sleep was higher during the 3-hour period from the beginning of the first long sleep in the evening and lower during the preceding 3-hour period compared with the control group.
Conclusions. The results of this study suggest that diary-based studies are prone to be biased as the parents of the control infants are more likely to overestimate the amount of infant’s sleep and, therefore, report more sleep than the parents of the crying infants. Although no differences in the total amount of sleep or proportions of sleep stages were observed, excessively crying infants may be characterized by a disturbance that affects rapid eye movement and non–rapid eye movement sleep stage proportion during evening hours.
Key Words: cry • infant • infantile colic • sleep • polysomnography • rapid-eye-movement sleep • behavioral states
Abbreviations: REM, rapid eye movement • NREM, non–rapid eye movement • EEG, electroencephalogram • EMG, electromyogram
Crying in early infancy has a tendency to increase after birth.1–3 When the amount of crying is at its maximum, a proportion of infants cry more than is tolerated by the caregivers. Infants who cry >3 hours a day for 3 days or more a week are referred to as colicky infants.4 No organic cause can be demonstrated in the vast majority of these infants, and they recover from the crying period without any obvious consequences.5
Crying episodes tend to cluster in the evenings, especially at the age of high amounts of crying.2 Simultaneously, the development of sleep goes through rapid changes. At the age from 2 weeks to 3 months, infant’s sleep-wake rhythm changes from regular 4-hour cycles to a well-defined day-night rhythm. The longest awake period tends to occur during evening hours6,7 at the time when infants also cry the most.3 During the same time period, the structure of sleep changes. Neonatal sleep characteristics such as sleep onset with rapid eye movement (REM) periods rapidly subside after the age of 4 to 5 weeks, and the variable length of REM and non–rapid eye movement (NREM) sleep cycle length stabilizes at 4 months of age.8 During the night, the REM and NREM ratio decreases by increasing age.9 The coincidence of these phenomena suggests that excessive crying in infancy may be associated with disturbances in the developing sleep structure or sleep-wake rhythm.
An association between colic and later sleep problems originally received support from questionnaire studies in which mothers who reported sleep problems also recalled more colic in their children retrospectively,10,11 and mothers who reported more colic reported more sleep problems prospectively in their children at 3 years of age.12 These results are confronted by our later prospective study of colicky infants using daily diaries to document the amount of crying. It did not show any differences in the duration of night sleep, amount of night awakenings, duration of night awakenings, or day time sleep at either 8 or 12 months of age.13 This finding has been confirmed by a recent study by Canivet et al.14
Diary-based studies during colic have suggested that the total daily sleep time is shorter in colicky infants compared with control subjects. St James Roberts et al15 reported that the total daily sleep time was 77 minutes shorter in the group of excessively crying infants compared with control subjects at 6 weeks of age. When the 24-hour day was split into 6-hour quartiles, the difference was observed only during daytime (between 6 AM and noon). Prudhomme et al16 showed that 2-month-old colicky infants slept 2 hours less per day than control infants. In a study by Papou
ek and von Hofacker,17 excessively crying infants slept 93 minutes per day less than control subjects at the age of 3.6 months. In our earlier study, the diary data at 5 weeks of age showed 108 minutes less sleep a day in the group of excessively crying infants compared with the control subjects. There was less reported sleep during each quartile of the day, but the difference was statistically significant only during evening hours and nighttime. However, this difference was not observed in later overnight sleep polygraphy recordings of the same infants in a sleep laboratory at 2 or 7 months of age.18
Wolff19 proposed that crying is 1 of 5 behavioral states in infancy. According to his observations, fussing is a state of transition. Excessive proportion of fussing and crying may be taken off from either awake or sleep states. The diary-based studies have suggested that excessive fussing or crying is at least partly taken "off" from sleep time and therefore will be reflected also in the development of infants’ sleep-wake rhythm. However, excessive fussing and crying could also be a consequence of sleep deprivation, which may lead to difficulty in behavioral control.
Excessively crying infants may have characteristics that disturb the sleep structure. Cow milk allergy as a distressing factor has been described to cause a disruption of sleep and an increase in the proportion of undetermined NREM sleep in infants.20 Because REM sleep is easily disturbed in infants, a REM sleep disturbance would also be expected.21–23
In this study, we performed an ambulatory 24-hour sleep polygraphy in a home environment in excessively crying and control infants at the age of 6 weeks. The setting was designed to compare parental diary reports and sleep polygraphy during a whole 24-hour sleep-wake cycle and to asses whether excessive crying is associated with changes in sleep structure.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
All parents kept a diary for 4 successive days during a study at the age of 6 weeks. In the diary, they recorded whether the infant was awake and content, fussing, crying, feeding, or sleeping. They also recorded their own behavior in respect to the infant: holding the infant, moving the infant around (eg, in a stroller), or taking care of the infant.24 The diaries were kept with the accuracy of 5 minutes, but the codes were filled in by parents as convenient. The sleep polygraphy recording was performed during 1 of the diary days. The diary data during sleep polygraphy were not included in the baseline diary data; thus, the diary data were averaged across the 3 days.
The follow-up of the infants was organized as a health questionnaire at the age of 8 months. In the questionnaire, the following questions were asked in dichotomous manner (when the answer was yes, parents were asked to describe the problem in detail): "Has your child problems falling in sleep?" "Has your child problems in sleeping?" "How many times your child wakes up during a night?" "Has your child any allergies?" "Has your child eczema?" "Has your child any limitations in diet?" "Please, give the number of antibiotic treatments; the number of otitis media." "Has your child been diagnosed to have a disease?"
A written consent for the study was obtained from the parents of each infant. The study protocol was approved by the Joint Commission on Ethics of the University of Turku and the Turku University Hospital.
Sleep Polygraphic Recordings
The sleep polygraphic recordings were performed during 1 of the diary days. The preparations for ambulatory sleep polygraphic recording were done at the hospital. After mounting the electrodes at the hospital, the recording was continued at home in the natural environment of the infant. The recording was started daytime (between 9:30 AM and 4:30 PM) and continued for the next 25 hours. The first 30 minutes to 1 hour at the beginning of the recording were excluded from the analysis to exclude the time needed to return home from the hospital.
The recordings were made with EMBLA sleep recording polygraph and analyzed by Somnologica 3 software (Medcare, Reykjavik, Iceland). The following channels were recorded: 2 electroencephalograms (EEGs; C4/A1 and C3/A2), 2 electro-oculograms, chin electromyogram (EMG), and an abdominal strain gauge for respiration movements. Two electrocardiogram electrodes were placed on the chest area, and 2 electrogastrography electrodes were placed on the abdominal area. Disposable electrodes were used (Blue Sensor; Medicotest, Ølstykke, Denmark) on the hairless skin area, and silver cup electrodes were used on the hairy scalp area. The sampling frequency was 100 Hz for all signals, but respiration belt was recorded using 10-Hz data sampling.
The 24-hour ambulatory sleep recordings were scored in 30-second epochs. The infant state was scored as REM sleep, light NREM sleep, deep NREM sleep, movement time, or awake. Sleep scoring was based on the criteria published by Guilleminault and Souquet25 with some modifications. All sleep states that did not meet the criteria for REM sleep were scored as NREM sleep. The deep NREM sleep was scored when the criteria for scoring stages 3 to 4 were met without the strict amplitude criterion of EEG >150 µV or whenever EEG showed a trace alternate pattern. All NREM sleep epochs that did not meet the criteria of deep NREM sleep were scored as light NREM sleep. Scoring of awake included also modified criteria, because no observational data were available, ie, direct information about body movements or whether eyes were open or closed. Awake was scored when 1) low-voltage, mixed-frequency EEG was associated with moderate or high chin EMG tone with eye movements ± irregular respiration or 2) in the presence of high EMG tone ± eye movements ± movement artifacts and 3) when 2 or more successive movement time epochs occurred. The epochs of movement time were ignored in the determination of the length of sleep stages and in the number of sleep-stage transitions. The parameters analyzed from the sleep recordings were reported for each quartile of the day. The first incomplete quartile at the beginning of the recording was summed together with the last (corresponding) quartile at the end of the recording.
Statistical Analysis
As a way to report statistical data, the expression mean (standard deviation, range) was used. The estimation of the size of the study groups was based on our previous study.18 According to a power analysis, 18 study infants were needed to observe a 10% difference in a whole-day sleep time at the significance level of .05. For statistical analysis, 1-way analysis of variance or analysis of variance for repeated measures was used for the continuous and normally distributed variables and 
2 test for the categorical variables. The SPSS 10.0 for Windows statistical software was used (SPSS Inc, Chicago, IL).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
As shown in Fig 1, the duration of fussing and crying was longer in the group of excessively crying infants than in the control group during all quartiles of the day. The fussing and crying clustered in the evening hours in both groups. Caregivers used a longer time to hold the infant in the group of excessively crying infants than in the control group during the day (308 minutes [142; 107–702] compared with 231 minutes (93; 58–445); P = .03). The time used to move the infant in a stroller or in a car did not differ between the study groups (66 minutes [43; 0–155] in the group of excessively crying infants compared with 90 minutes [68; 0–330] in the control group [P = .25]); neither did the time used for caregiving (69 minutes [26; 30–138] compared with 66 minutes [24; 25–117; P = .62], respectively).
|
The duration of fussing and crying correlated inversely with the duration of reported sleep time. Within the group of excessively crying infants, the Pearson correlation coefficient was for the whole day (r = –0.34 (P = .1) and for quartiles of a day: nighttime (midnight to 6 AM; r = –0.84; P < .001), morning (6 AM to noon; r = –0.69; P < .001), afternoon (noon to 6 PM; r = –0.66; P < .001), and evening (6 PM to midnight; r = –0.27; P = .2). Within the control group, the duration of fussing and crying correlated with sleep during the nighttime (r = –0.50; P = .02) but not other quartiles of day (6 AM to noon: r = –0.25; P = .3; noon to 6 PM: r = –0.15; P = .5; 6 PM to midnight: r = –0.23; P = .3) or the whole day (r = 0.09; P = .7).
The diary data of the 24-hour polygraphy period showed shorter nighttime sleep in the group of excessively crying infants compared with the control subjects (Table 2), but the total reported duration of sleep did not differ between the study groups during the recording hours (P = .3). The reported amount of fussing and crying during the sleep polygraphy was 202 minutes (73; 110–395) in the group of excessively crying infants and 76 minutes (54; 0–170) in the control group (P < .001). The reported amount of holding or carrying did not differ between the study groups during the recording hours (276 minutes [165; 15–595] in the group of excessively crying infants compared with 201 minutes [116; 40–545] in the control group; P = .08). The reported amounts of sleep (r = 0.61), fussing and crying (r = 0.89), carrying (r = 0.84), moving (r = 0.59), and caregiving (r = 0.63, P < .001 for all) correlated with the averaged 3-day diary data. The reported amount of sleep was higher during the polygraphy recording than during the 3-day diary period (P < .001), and the reported amounts of carrying and caregiving were lower (31 minutes [81; –136 to 263) for carrying [P = .01] and 18.5 minutes [21; –30 to 77) for caregiving [P = .001]) in both study groups. No group differences were found (P = .1, P = .9, and P = .6, respectively).
|
|
|
Individual variation occurs between infants in the circadian sleep-wake periodicity. That is why a rigid division of a day into 4 quartiles may hide group differences. We analyzed also the proportions of sleep stages using the longest sleep period in the evening as a starting point. The study infants usually had a long sleep period after a long evening waking period typical for their age. This procedure allowed comparison of the infants "in the same physiologic circadian rhythm." The beginning of the long sleep period was at 10:42 PM (1:51; 8:06 PM–4:02 AM) in the group of excessively crying infants and at 10:37 PM (1:29; 7:26 PM–1:50 AM) in the control group (P = .8). The reported duration for fussing and crying during the 3-hour period before a long sleep period was 54 minutes (28; 10–110; 40% [19; 8–73] from awake time) in the group of excessively crying infants and 25 minutes (27; 0–90; 20% [20; 0–60] from awake time) in the control group (P = .001). The 2 subsequent 3-hour periods before and after the starting point were analyzed. The observed duration of sleep did not differ between the study groups in any of these 3-hour periods in polygraphic recordings (Fig 3A). The reported amount of fussing and crying correlated negatively to the observed amount of sleep in polygraphy in the group of excessively crying infants during the 3-hour period before a long sleep (Pearson correlation: –0.52 [P = .01] for the group of excessively crying infants and –0.25 [P = .2] for the control group). However, no such a correlation was found when total 24-hour observed sleep was compared with fussing and crying (P = .6).
|
The temporal relation between diaries and sleep polygraphy was evaluated by comparing the beginning of the long sleep period. The time difference between these 2 methods for all infants was 8 minutes (25; –65 to 182), and there was no group difference (P = .6).
Follow-up Questionnaire
To follow the general health of the studied infants, we sent a questionnaire to the parents after 6 months of age. The questionnaire was returned by the parents of 20 excessively crying infants and 22 control infants. The mean age of infants at the time of the questionnaire was 8.7 months (1.5; 6.4–12.6) in the group of excessively crying infants and 8.6 months (1.9; 6.1–12.7) in the control group (P = .9). No differences were found in suspected (by the parents or physician) or diagnosed illnesses (Table 4). One excessively crying infant had received cisapride medication for clinical suspicion (without pH recording) of gastroesophageal reflux after the study.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Parental perception of crying was used to determine the groups of colicky and control infants, because parents may reach characteristics of infant behavior that are not quantifiable by using simple measurement of crying amounts by diaries.26,27 Parental perception correlated well with the amount of fussing and crying according to the diary records, and only a little overlapping occurred in the amount of fussing and crying between the groups. According to follow-up questionnaires, the excessively crying infants did not differ from the control subjects in the amount of atopic illnesses or the amount of received antibiotic therapies. The reported amount of sleep was higher during the polygraph recordings than the averaged 3-day diaries (P < .001; text and Table 2), suggesting that the polygraphy recordings did not disturb the sleep of infants. However, fewer parental activities with their infants (holding and caregiving) were reported during polygraphic recording hours, suggesting that recordings affected infant–parent interaction, which may be reflected as increased sleep of their infant. No difference was observed in the total reported amount of sleep between the study groups during the polygraphic recordings, which may be attributable to the day-to-day variability of the amount of sleep and too low power of the study to find a difference in 1-day data.
The diary data suggesting less sleep in excessively crying infants is consistent with previous studies based on parental reports.15,16,18 A negative correlation between parental reports of the amount of sleep and fussing and crying also was found earlier.16 However, polygraphic recordings suggest that the total duration of daily sleep was not affected by excessive fussing and crying. This finding is consistent with our earlier sleep laboratory study.18 Compared with our first study,18 the strengths of this study had some advantages. The recordings were performed in infants’ natural home environment and lasted throughout a whole 24-hour sleep-wake cycle. Our second finding was that parental reports overestimated infant sleep more in the control group than in the group of excessively crying infants. The previous results suggesting shorter sleep in excessively crying infants were based on parental reports that, according to this study, may be biased by parents’ tendency to report more sleep than actually occurs. It is likely that parents miss quiet periods of wakefulness. It is logical that waking periods are easily missed in infants who do not fuss or cry when awake. It has been shown that 3-month-old infants who cry at night receive more interventions by mothers and are removed from the crib when awake more often than the self-soothers who do not cry and receive maternal intervention when awake.28
With this study setting, the observational component was not included in the scoring of sleep-wake states; therefore, we were not aware of whether eyes were open or closed and did not differentiate short manipulation periods by caregivers from active wake periods during polygraphy recordings. This may have caused some inaccuracy in the comparison between diary and polygraphic data. We have compared a behavioral scoring method (parental diaries) with a polygraphic method in assessing infants’ behavioral states instead of comparing parental reporting with another objective behavioral scoring method such as video recording. Sleep polygraphy has been used as a gold standard to determinate the behavioral states, although there could be inconsistency between polygraphic signals and sleep states in 6-week-old infants. In our scoring system, we relied on appearance of eye movements and chin EMG activity and basal tone in distinguishing between quiet awake and light NREM sleep and also between REM sleep stage and awake. It has been shown that there may be a lack of concordance between EMG tone and other sleep polygraphic signals in infants younger than 3 months. This may be attributable to immaturity of infants’ physiology or may reflect ambiguity in scoring infants’ states generally.29 Because of these characteristics of infant sleep, it is possible that some quiet awake states (behaviorally eyes open) were scored as sleep (light NREM) in the present study; therefore, overestimation of the amount of sleep compared with behavioral method can be possible in the sleep recordings. This kind of error would only strengthen our results as we already scored less sleep compared with the reported diary data. However, eye movements would be expected when eyes are open. Similarly, awakenings as a result of carrying or manipulation of the infants cannot be differentiated from the spontaneous awakenings. The study groups did not differ in carrying time during the polygraphic recordings, although the difference was found in 3-day diary data. The observed overall sleep time as well as total REM and NREM proportions were similar to what was described by Guilleminault and Coons6 in their 24-hour sleep polygraphy study of 10 infants at the age of 6 weeks in a sleep laboratory. Also, the results were similar to our earlier sleep polygraphy study in which the behavioral observation (video recording) was included in sleep-stage scoring.18
Our earlier overnight sleep polygraphy study suggested that colicky infants have normal sleep structure.18 Therefore, it seemed unlikely that colic crying would be caused by a physical, continuously interfering factor such as pain. Also in this study, a rigid time-related analysis with a 6-hour quartile approach did not reveal major differences between the study groups, although more REM sleep was observed during 0 to 6 am period in excessively crying infants than in control subjects. However, when the sleep-wake cycle variation was taken into account, excessively crying infants showed relatively less REM sleep and more NREM sleep during sleeping periods in the evenings. After falling into a long sleep period after an evening waking period, they presented with relatively more REM sleep and less NREM sleep compared with the control infants. However, because no differences were observed in the number or length of REM periods or sleep-stage transitions between the study groups, this study did not give any additional evidence of disturbed REM sleep. During the 3-hour period before the long sleep, excessively crying infants also showed relatively more deep NREM sleep. This low REM and high deep NREM sleep pattern may be caused by a factor that interferes with sleep during the evening period. The interference was strong enough to prevent easily disturbed REM sleep but not deep NREM sleep.21–23 One possible explanation for the observed difference could be that excessively crying infants were held and carried more than control subjects by their caregiver because of fussing and crying. This may prevent them from falling into REM sleep but not into NREM sleep. After falling asleep, the infants could have been placed in their cribs, where they may sleep undisturbed. However, this explanation is unlikely because body rocking would have disturbed polygraphic signals. Hence, rocking episodes would have been scored as wakefulness and thus would not have altered the proportion of REM and NREM sleep. This study supports the theory that excessive crying may be caused by a characteristic that disturbs REM sleep during the evening period. However, this changes after they fall into a long sleep.
Because there was no other evidence for REM sleep disruption than decreased proportion of REM sleep, it is also possible that this was attributable to an increase of NREM proportion of sleep. During the 3-hour period preceding the long sleep, excessively crying infants showed more deep NREM sleep than control subjects. During that 3-hour period, excessively crying infants slept 
27% of time, and fussing and crying occurred 40% of wake time. Thus, the sleep during that time period was associated with a stressful awake time. It has been described in 4-month-old infants that allergy to cow milk can cause a disturbance of sleep structure, frequent awakenings, and an increase in the proportion of undetermined NREM sleep.20 Sleep deprivation, which was not supported in our study in excessively crying infants, has been described to cause a similar type of change by increasing both S4 and REM sleep stages in adults.30
An association between REM sleep and crying spells has been proposed by Weissbluth,31 especially during fussy REM state defined by Emde and Metcalf32 in which the infant is fussing according to behavioral criteria but is in REM sleep according to polygraphic criteria. In this study, the recording of infant fussing and crying was based on parental reports, and accuracy of time matching between diary reports and sleep recordings was too low to allow direct comparison between cry onset and infant state.
The finding that there was no difference in the beginning time of the long evening sleep period between the study groups suggests that the diurnal sleep-wake cycle did not differ between the study groups. However, both observed and reported sleep time during polygraphy was longer during morning hours in excessively crying infants than in control subjects, which suggest a short sleep-wake cycle shift. This may be attributable to the recording procedure as there was a difference only during polygraphy (Fig 2) but not in the averaged 3-day data (Fig 1). Because one third of the sleep recordings were started during morning hours, that quartile of day is a mixture of 2-day data and therefore the data are less clear.
The length of uninterrupted REM sleep was longer in the group of excessively crying infants than in the control group. It may be concluded that the excessively crying infants had fewer short (30 seconds) interruptions of REM sleep, because no difference in the length of REM periods between the study groups was observed when 1-minute interruptions were allowed in the middle of REM periods. In our previous sleep laboratory study, we found a tendency that infants with colic had a fewer number of movements during sleep. The results of this study suggest that sleep of excessively crying infants may be more consolidated during nighttime and morning hours than in control subjects.
The parents of excessively crying infants perceived that their infants had more problems in falling asleep at 8 months of age than the parents of control infants. However, the number of night waking was not different between study groups. This is consistent with earlier findings. Despite a lack of evidence of objective differences in later child behavior in colicky infants, parental perception of child behavior is different.5,12–14
In conclusion, this study, together with our earlier study, shows that the duration of sleep and the overall structure of sleep are similar between normal and excessively crying infants at the time when crying amounts are shown to peak during early infancy. These results suggest that crying time does not reduce the sleeping time. That most of the crying is clustered in the evening hours may reflect normal maturing of the sleep-wake rhythm with the wakefulness occurring in late afternoon or evening.7,33 However, the present study suggests that excessively crying infants may have characteristics that lead to a relative REM sleep deficiency during evening hours, at the time when the most crying and fussing occurs. Later in night, a REM sleep rebound was observed.
|
ACKNOWLEDGMENTS |
|---|
We thank Helena Haapanen for informing parents of newborn infants about the study. We also thank the parents and infants who participated in this study.
|
FOOTNOTES |
|---|
Reprint requests to (J.K.) Department of Pediatric Neurology, Central Hospital of Central Finland, Keskussairaalantie 19, 40620 Jyväskylä, Finland. E-mail: jarkko.kirjavainen{at}utu.fi
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  V Wiseman, L Conteh, and F Matovu Using diaries to collect data in resource-poor settings: questions on design and implementation Health Policy Plan., November 1, 2005; 20(6): 394 - 404. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2004 American Academy of Pediatrics. All rights reserved. |
||
|
||
, nighttime (midnight to 6 AM);
, morning hours (6 AM to noon);
, afternoon (noon to 6 PM); 
, evening hours (6 PM to midnight). On the left side, the combined duration of crying and fussing was higher in the group of excessively crying infants than in the control group during all quartiles of the day (P 
.001). In both groups, the highest amount of crying and fussing occurred during the evening hours (6 PM to midnight). On the right side, the duration of reported sleep was shorter during the nighttime in the group of excessively crying infants compared with the control group (P = .008). Sleep times did not differ between the study groups during any other quartiles of the day.
, control infants (standard error is shown as downward error bars). B, Proportions of REM sleep stage during 3-hour time periods.