Dr Rieko Takaya, Department of Human Development, Faculty of Human Development and Culture, Fukushima University, Kanayagawa, Fukushima 960-1296, Japan. Email: firstname.lastname@example.org
Developmental changes in the sleep–wake rhythm of preterm infants were compared with those of full-term infants, to clarify the timing of the developmental change of the sleep–wake rhythm and its dependence on either conception or birth timing. We obtained sleep log data for two preterm infants, and compared these with previous data gathered from 10 full-term infants over a period from 2 weeks after birth to 3 months after their expected delivery dates. Infant sleep logs were analyzed with the autocorrelation method to investigate the development of the circadian rhythm in each infant's sleep patterns during each weekly session. We then classified the patterns of correlograms into three groups using cluster analysis. The first group (pattern A) showed little evidence of any circadian component. The second (pattern B) had a prominent circadian component. The last (pattern C) was characterized by its prominent circadian peak and a negative peak in about a 12-h cycle, which represented very little daytime sleep. There were no differences in the sleep-wakefulness patterns of full-term infants and preterm infants when we compared their sleep log data using postmenstrual age. Almost all of the infants in both groups changed their patterns from A to B around the 46th week. Considering these findings, these changes in the circadian rhythm of sleep and wakefulness were viewed as showing dependence on conception rather than on birth timing.
Young infants show little or no evidence of circadian variations in either sleep or waking states. The typical newborn sleeps approximately 16–17 h per day, with periods of sleep and wakefulness evenly distributed between day and night.1 Sleep becomes more concentrated during the night hours and wakefulness increases during the day. A diurnal pattern of sleep and wakefulness is clearly established by the 12–16th week of age, with daytime sleep consolidated into well-defined naps.2
In a previous study, we found a discontinuous change in the sleep–wake rhythm that appeared in almost all of the 10 full-term infants (seven male and three female) around 7 weeks after their birth.3 In the study, the sleep log data were analyzed with the autocorrelation method to investigate the circadian rhythm of each infant's sleep and wakefulness during each week, and the authors classified the patterns in the correlogram taken from the 2nd to 26th weeks into three groups using cluster analysis. The first group (pattern A) showed little evidence of a circadian component. The second group (pattern B) exhibited a prominent circadian component. The last group (pattern C) was characterized by its prominent circadian peak and negative peak in about a 12-h cycle which represents very little daytime sleep. Almost all the infants changed their patterns from A to B around the 7th week, then some of the infants changed their patterns from B to C around the 12th week after their birth. However, because the participants of the research were all full-term infants (they were born at the 39th week of gestational age on average), we could not determine whether the first discontinuous change of the sleep–wake rhythm is triggered by the date of birth or the date of conception. In other words, it had still not been confirmed when the discontinuity of the circadian rhythm of sleep and wakefulness appears. Therefore, we conducted a longitudinal study of preterm infants from the preterm to post-term period, and compared the developmental changes of the sleep–wake rhythm of preterm infants with those of the full-term infants in the previous study in order to confirm the timing of emergence of the circadian rhythm of sleep and wakefulness in early infants' development.
Two preterm infants (both male) and 10 normal full-term infants (seven males and three females) participated in this study. Informed consent was obtained from their mothers. The data of the full-term infants were the same as those used in the research conducted by Fukuda and Ishihara.3 Two preterm infants (infant A and B, both born at the 29th week of gestational age) were born with the weight of 1230 and 1360 g, respectively. They had been admitted to the neonatal intensive care unit (NICU) of the Nagano Children's Hospital in Japan after their birth. In the NICU, the illumination level was lowered to 20 lux measured at the level of the bed during night-time from 20:00 to 7:00 hours, while the mean illumination used during the daytime was 140 lux. Infant A was transferred from the Nagano Children's Hospital to another hospital near the infant's home where the illumination level was not lowered during the night at the postmenstrual age (PMA) of 33 weeks, and was discharged from the hospital at the PMA of 38 weeks. Infant B was also discharged from the hospital at 38 weeks of PMA without changing hospitals. The development of both preterm infants was normal at 3 years old; their total developmental quotients using the Kyoto Scale of Psychological Development at 3 years of corrected age were 116, and 92 respectively.
The mothers of the 10 full-term infants were asked to fill out sleep logs for their babies after returning home. For the two preterm infants, sleep logs were filled out by nurses during their hospitalization and by their mothers after returning home. The sleep log data for the 10 full-term infants were obtained during the 2nd to 26th week after their births and the data from 2 to 16 weeks after their birth (the 39th to 53rd week of PMA) were utilized for this study, and the data for the two preterm infants was obtained from 3 to 24 weeks after their births (the 32nd to 53rd week of PMA).
The method of analysis of the sleep log data was the same as the previous research by Fukuda and Ishihara.3 The authors scored each 15-min unit of sleep as “one” and that of wakefulness as “zero” from each infant's sleep log data. The autocorrelation function was calculated from 288 units (72 h) of sleep log data in order to investigate the possible existence of rhythms. There was some lack of sleep log data during the preterm period of preterm infants: the 36th–39th weeks for infant A, and the 33rd, 36th–38th, and 49th weeks for infant B were not analyzed with the autocorrelation function, because we could not get continuous sleep log data.
To compare the sleep and wakefulness patterns, we classified the patterns of the correlograms from the 39th to the 53rd weeks of PMA, using cluster analysis with the Ward method. Numbers in the rectangles indicate the total amount of data allotted to the clusters. Numbers in parentheses are the data from the preterm infants that were included in the clusters. In the present research, PMA, which is calculated by gestational age plus chronological age, was employed for the cluster analysis to compare the patterns of the correlograms of preterm infants with those of full-term infants.
The raster plots of the sleep logs of the two preterm infants are shown in Figure 1. Each autocorrelogram pattern from the sleep log of preterm infants in each week is shown in Figure 2. These figures show that a peak at the period of 24 h becomes apparent around the 46th week of PMA in both of the preterm infants, not at approximately 7 weeks after their births (chronological age). Therefore, we compared preterm infants' sleep log data with those of full-term infants after the 39th week of PMA.
Figure 3 shows the results of the cluster analysis that overlaid the multiple sweeps of original patterns of correlograms that were divided into the three groups, and each pattern is shown in Figure 4. The first group (pattern A) showed little evidence of a circadian component. The second one (pattern B) had a prominent circadian component that had a peak at the period of 24 h. The last one (pattern C) was characterized by its prominent circadian peak at the period of 24 h and a negative peak around the period of 12 h. In the weeks when the correlograms were classified as pattern B, infants slept at night; however, they still had daytime naps, which can be confirmed by their raster plots. Therefore pattern B does not have a negative peak around 12 h. By contrast, pattern C has a negative peak around 12 h; in addition to the positive peak at 24 h, this suggests a decrease in the amount of daytime sleep.
The sleep pattern of each infant for each week of PMA is shown in Table 1, and the percentage of the infants showing each pattern in each week is shown in Figure 5. Both the preterm infants' correlograms change their patterns from A to B around the 45th week, and from B to C around the 48th week of PMA. Almost all the full-term infants changed their patterns from A to B by the 46th week; some of the infants changed their patterns from B to C around the 52nd week of PMA.
Table 1. Sleep patterns of each infant (preterm and full-term) in each week of postmenstrual age
The emergence of infants' sleep and wakefulness rhythm
The differences in sleep and wakefulness rhythm between preterm and full-term infants in the early period of life have been the focus of a number of studies.4–6 For example, Korte et al.5 clarified the differences in the activity–rest behavior between preterm and full-term infants during their first weeks of life. These researchers concluded that preterm neonates need more time to adapt to their environmental day–night cycles than full-term neonates, because their data indicated that a circadian rhythm in activity–rest behavior was nearly absent in the first weeks of life in the preterm neonates. On the other hand, Shimada et al.6 came to a different conclusion when comparing the developmental courses of sleep and wake rhythm in two groups of preterm and full-term infants. These researchers used both corrected and postnatal ages and recorded sleep and wakefulness states at home for more than 14 days. In the two groups, there were no significant differences in sleep and wakefulness behavior when the corrected ages were considered; therefore, they concluded that the sleep and wakefulness rhythm in preterm infants depend on their corrected ages. However, the observation periods of these studies was too short to confirm the developmental changes in the sleep–wake rhythm, because, in a previous study,3 we found a discontinuous change in the sleep–wake rhythm around 7 weeks after the birth of full-term infants.
In our study, we had access to the sleep log data for the 10 full-term infants during the 2nd to 26th week after their births, and the data for the two preterm infants for the 3rd to 24th week after their births. In the longitudinal study, our results clearly show that the correlograms of sleep log data for both the full-term and preterm infants could be classified into three pattern groups. These patterns are quite similar to ones shown in the previous study with full-term infants; namely, pattern A has little evidence of circadian rhythms, pattern B shows a prominent circadian rhythm with one peak at 24 h, and pattern C is characterized by its prominent circadian peak and a negative peak with approximately 12 h periodicity. Whether preterm or full-term, almost all of the infants change their sleep patterns from A to B by the 46th week of PMA. These results indicate that the developmental changes in sleep and wakefulness rhythm depend on conception not on birth timing.
The discontinuous change from 45 to 52 weeks of PMA
We found two discontinuous changes in the circadian rhythm, one from pattern A to B around the 46th week and another from pattern B to C around the 52nd week of PMA. Other investigators have reported developmental changes in infant behavior during the same period. Takaya et al.7 found U-shaped patterns of change in some spontaneous movements that disappear and reappear after some period; for example, hand–mouth contacts, which were observed frequently in the neonatal period, were not observed temporarily after the 45th week but re-emerged after the 50th week of PMA. Developmental changes in reaching and auditory localization also show the U-shaped pattern in the same period.8,9 Von Hofsten9 explained the U-shaped pattern of changes in early infancy on the basis of a neurological hypothesis that neonatal movements are restrained because of the immature nervous system, but as the cerebral cortex matures, the earlier types of movement are replaced by voluntary movements. It is not clear whether these changes coincide with the discontinuance of the sleep-wakefulness rhythm; however, we speculate that nervous system maturation is also involved in the discontinuance of the sleep-wakefulness rhythm. The well-established wakefulness of pattern C may promote the acquisition of voluntary movements after the U-shaped changes.
We express our gratitude to the children and their mothers who participated in this study. The authors are also grateful to the staff of the NICU of Nagano Children's Hospital where data was collected during infants' hospitalizations.