Sleep and 24-hour rhythm characteristics in preschool children born very preterm and full term

Abstract

Study Objectives:

Sleep impacts the quality of life and is associated with cardiometabolic and neurocognitive outcomes. Little is known about the sleep of preterm-born children at preschool age. We, therefore, studied sleep and 24-hour rhythms of preschool children born very preterm compared with full-term children.

Methods:

This was a prospective cohort study comparing sleep quality and quantity of children born very preterm (gestational age [GA] < 30 weeks) with full-term children at the (corrected) age of 3 years, using (1) 2 parent-reported questionnaires (Brief Infant Sleep Questionnaire and The Munich Chronotype Questionnaire) and (2) at least 3 days of triaxial wrist actigraphy combined with sleep diary. We performed regression analyses with adjustment for sex (corrected), age, and birth weight standard deviation (SD) score.

Results:

Ninety-seven very-preterm-born (median GA 27+5; interquartile range 26 + 3;29 + 0 weeks) and 92 full-term children (GA 39 + 3; 38 + 4;40 + 4 weeks) were included. Sleep problems and other reported sleep parameters were not different between groups. As measured with actigraphy, sleep and 24-hour rhythm were similar between groups, except for very-preterm born children waking up 21 minutes (4;38) minutes later than full-term children (adjusted P = .001).

Conclusions:

Based on parent reports and actigraphy, very-preterm-born children sleep quite similar to full-term controls at the corrected age of 3 years. Reported sleep problems were not different between groups. Actigraphy data suggest that preterm-born children may wake up later than children born full term. Further studies are needed to explore how sleep relates to cardiometabolic and neurodevelopmental outcomes after preterm birth and whether early interventions are useful to optimize 24-hour rhythm and sleep.

Citation:

Bijlsma A, Beunders VAA, Dorrepaal DJ, et al. Sleep and 24-hour rhythm characteristics in preschool children born very preterm and full term. J Clin Sleep Med. 2023;19(4):685–693.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Twenty-four-hour rhythm and sleep start to develop in fetal life, and are vital for health, neurocognitive development, and well-being in children. Preterm birth and neonatal intensive care may disturb normal development, leading to a disturbed rhythm and quantitative and qualitative sleep problems in childhood.

Study Impact: Parents of very-preterm-born children did not report more sleep problems or different sleep parameters at 3 years’ corrected age than those of children born full term. Actigraphy showed later wake-up times but no clear differences in 24-hour activity rhythms, nor in sleep quantity or quality, in children born very preterm as compared with those born full term.

INTRODUCTION

Sleep and 24-hour rhythm are vital for development and physiological function in children.¹ Approximately one-quarter of parents, however, report sleep problems in their children.² Sleep problems are most prevalent in infancy, declining toward middle childhood.³ Insufficient or disturbed sleep is not only associated with decreased neurocognitive functioning but also with an increased risk of obesity and cardiometabolic diseases.⁴ Preterm-born children are at increased risk for these adverse outcomes, potentially creating opportunities to improve long-term health and well-being by improving their sleep patterns.

Sleep problems may be more common in children born preterm, due to disturbance of the fetal development of the 24-hour rhythm and sleep.⁵ Possible mechanisms include premature disconnection to maternal circadian cues, impaired growth, neonatal morbidities, and adverse environmental factors, all in a critical period of development of the immature nervous system.⁶ Most studies on sleep patterns after preterm birth are based on parental reports and describe sleep at various ages between 3 and 18 years. They generally showed lower sleep quality, more nocturnal awakenings, and daytime sleepiness than after full-term birth.^7–11 The few studies using objective measurements, like actigraphy or polysomnography,¹² in children at school age, reported inconclusive results based on small populations.^13,14 Studies using polysomnography suggest that preterm-born children have more nocturnal awakenings at school age.¹⁰ Overall, very little is known about sleep problems and 24-hour rhythm in very-preterm-born children in the preschool period.

Therefore, the aim of this study was to compare sleep and 24-hour activity rhythm between very-preterm- and full-term-born children at the age of 3 years, using parent reports and actigraphy. We hypothesized that very-preterm birth is associated with more reported sleep problems and with lower quantity and quality of sleep, including shorter sleep duration, lower sleep efficiency, and more fragmented 24-hour activity rhythms than those born full term. We expected to find this both in parent reports and actigraphy data.

METHODS

Study population

This study was nested in 2 ongoing prospective observational birth cohort studies at the Erasmus MC Sophia Children’s hospital in Rotterdam, the Netherlands. The first study (the Bodycomposition and NeuroDevelopment in preterm infants study) included 142 very-preterm infants born at less than 30 weeks’ gestational age (GA) and admitted within 48 hours after birth to the level-IV neonatal intensive care unit between 2014 and 2017. This study excluded congenital anomalies, early severe brain injury (intraventricular hemorrhage [IVH] grade > II or posthemorrhagic ventricular dilatation), congenital infection, or perinatal asphyxia (umbilical cord pH < 7.00 and Apgar score < 5 after 5 minutes).¹⁵ The second study (the Sophia Pluto Study) provided the full-term (≥ 37 weeks’ GA) participants, which included 1,012 healthy infants born on several maternity wards in Rotterdam between 2013 and 2021 with an uncomplicated neonatal period. It excluded severe asphyxia (Apgar score < 3 after 5 minutes), sepsis, or respiratory ventilation in the neonatal period; confirmed intrauterine infection; and known congenital, postnatal, or maternal disease or medication that could interfere with the child’s growth and development, including maternal corticosteroid use.¹⁶

For the current study, participants of both studies, with a study visit at 3 years of (corrected) age between June 2019 and May 2022 were eligible. In the very-preterm cohort, sleep measurements were part of the general study protocol applying to all participants, while in the full-term cohort, parents were asked to opt-in for the additional sleep measurements during the study period. The Medical Ethics Committee of the Erasmus Medical Centre approved both studies (MEC-2014-379, MEC-2012-164). Written informed consent was obtained from all parents/caregivers.

Data collection

Prenatal and neonatal factors were collected from hospital and midwife records and parental questionnaires. Retrieved from questionnaires, ethnicity was classified as “Western European” or “non-Western” if 1 or both parents were born in a non-Western country, and level of parental educational level was based on both parents.¹⁷ Age- and sex-adjusted standard deviation scores (SDS) for birth weight were calculated with the Fenton Growth Chart Calculator. Small for GA (SGA) was defined as < 10th percentile for weight.¹⁸ Age- and sex-corrected SDS for weight and weight-for-height SDS were calculated using Dutch reference values.¹⁹

Sleep questionnaires

Parents were asked to complete the following paper sleep questionnaires:

1. Brief Infant Sleep Questionnaire (BISQ).²⁰ The BISQ aims to evaluate sleep patterns and habits and is composed of questions related to the following areas: (a) bedtime, (b) nocturnal sleep duration (between the hours of 7 pm and 7 am), (c) daytime sleep duration (between the hours of 7 am and 7 pm), (d) number of nightly awakenings, (e) sleep-onset latency (SOL) at night, (f) method of falling asleep, (g) location of sleep, (h) preferred body position, and (i) parental rating of sleep problems, in children aged 0–3 years old.

2. The Munich Chronotype Questionnaire (MCTQ).²¹ This questionnaire documents sleep times, sleep duration, SOL, and self-reported exposure to daylight on weekdays and weekend days. We computed midpoint sleep, defined as the middle time point between sleep-onset time and wakeup time. If midpoint sleep was different for week days and weekend days, sleep and wake data would be presented separately.

Actigraphy

Sleep was assessed using a triaxial actigraph (GENEActiv; Activinsights, Cambridge, UK) for at least 5 consecutive nights (3 week nights and 2 weekend nights). The actigraph is a wristwatch-like device that monitors activity levels for extended continuous periods.¹² As in preschool children the nondominant and dominant wrists yield similar results, children were free to use their preferred wrist.²² Additionally, parents were asked to complete a paper sleep diary daily for each day and night the actigraph was worn. In the sleep diary parents filled in the child’s nocturnal sleep duration, as well as any “daytime naps” if the child slept more than 15 minutes during daytime.

Actigraphs were set at a frequency of 50 Hz. To be included in the actigraphy analyses, a child should have worn the actigraph for 16 hours or more per day, capturing at least 4 hours sleep time per night, for a minimum of 3 days. Raw sleep data (.bin files) were analyzed with the R-package GGIR version 2.6.0. (R Foundation for Statistical Computing, Vienna, Austria), using an algorithm with 5-second epochs and the reported bedtime and wake-up times from the sleep diaries as guides.²³ Actigraphy sleep measures were calculated using this script, and defined as follows²³:

1. Twenty-four-hour sleep duration: total duration of estimated sleep per 24 hours, in hours:minutes. This measure was calculated by combining nocturnal sleep duration with the registered “accumulated sustained inactivity bouts during the day.”²⁴ Sustained inactivity bouts are periods labeled as sleep during the night but as “inactivity” during the day. These “inactivity periods” during the day may represent daytime sleep or wakefulness while being motionless for a sustained period of time. We only used sustained inactivity bouts that lasted at least 15 minutes.

2. Nocturnal sleep duration: total duration of estimated sleep between sleep onset in the evening and final wakening in the morning, in hours:minutes.

3. Daytime sleep duration: accumulated sustained inactivity bouts, based on bouts that lasted at least 15 minutes, in hours:minutes.

4. Sleep efficiency (percentage of time spent asleep between sleep onset at night and final waking time in the morning).

5. Wake after sleep onset (WASO; number of minutes scored as wake during the nightly sleep period).

6. Sleep-onset latency (SOL; time between bedtime and sleep onset at night, in hours:minutes [hh:mm]).

The following 24-hour activity rhythm parameters were calculated from the actigraphy data, using the GGIR script²³:

1. Intra-daily variability (indication of fragmentation of the sleep rhythm, ranging from 0 to 2, with higher scores indicating more fragmentation)

2. Inter-daily stability (indicating the stability of the 24-hour activity rhythm across days, ranging from 0 to 1, with higher scores indicating more stable rhythms)^24–26

Statistical analysis

Characteristics and outcome data were described for very-preterm and full-term-born children separately. Parametric and nonparametric tests were used for comparison of group characteristics between full-term and very-preterm-born children, as well as those who opted in and out in the full-term group, as appropriate. The primary analyses were based on linear and logistic regression models to compare sleep characteristics and 24-hour activity rhythms between the 2 groups, with a Poisson distribution for count data (number of awakenings). All models were adjusted for potential confounders selected based on the literature. These included sex, age, and birth weight SDS; as boys, younger age and low birth weight SDS were previously associated with lower quality or quantity of sleep.^27,28 Reported results refer to adjusted analyses unless stated otherwise; unadjusted results are shown in the supplemental material (Table S1 and S3 ^{(709.6KB, pdf)} ). P values < .05 (2-sided) were considered statistically significant. Data were analyzed using SPSS version 25.0 (IBM SPSS Statistics, Chicago, IL) and the R-package GGIR version 2.6.0.

RESULTS

In total, 97 very-preterm-born (42% female) and 92 full-term-born children (59% female) were included at 3 years of (corrected) age (Figure 1). In the very-preterm group, median GA was 27 + 5 (interquartile range 26 + 3;29 + 0) weeks and birth weight SDS was 0.14 (–0.40;0.70) vs 39 + 3 (38 + 4;40 + 4) weeks and –0.28 (–0.79;0.49), respectively, in the full-term group (Table 1). Comprehensive descriptions of the population characteristics have been provided previously.^16,29

Figure 1. Flowchart of study population.

*Actigraphy data were defined as “low quality” if the following criteria were met: < 16 hours wearing time per day capturing < 4 hours of sleep time with < 3 such days per participant. BISQ = Brief Infant Sleep Questionnaire, COVID = coronavirus disease, MCTQ = Munich Chronotype Questionnaire

Table 1.

Baseline characteristics of preterm and full-term group.

	Very Preterm (n = 97)	Full Term (n = 92)	P
Demographic Characteristics
Gestational age, wk	27 + 5 (26 + 3;29 + 0)	39 + 3 (38 + 4;40 + 4)	.00
Birth weight, g	1020 (828;1250)	3285 (2923;3708)	< .001
Birth weight SDS	0.14 (−0.40;0.70)	−0.28 (−0.79;0.49)	.02
SGA*	8 (8)	10 (11)	.54
Sex, female	41 (42)	54 (59)	.02
Apgar 5 min	8 (6;9)†	10 (9;10)†	.00
Missing	1 (1)	1(1)
Family Background
Educational level‡			.02
Low	12 (12)	5 (6)
Middle	28 (29)	13 (14)
High	51 (53)	59 (64)
Missing	6 (6)	15 (16)
Ethnicity			.01
Western European	76 (78)	58 (63)
Non-Western	21 (22)	29 (32)
Unknown	0 (0)	5 (5)
Neonatal Morbidity
IVH		NA
No IVH	77 (80)
IVH grade 1	11 (11)
IVH grade 2	9 (9)
BPD		NA
No BPD	59 (61)
Mild BPD	22 (23)
Severe BPD	16 (16)
3 Years Visit
(Corrected) age (in years)	3.22 (3.10;3.40)	3.05 (3.01;3.16)	< .001
Weight SDS	−0.99 (−1.84;−0.17	−0.02 (−0.90;0.52)	< .001
Weight-for-height SDS	−0.68 (−1.50;0.14)	0.14 (−0.63;0.96)	< .001

Data are presented as median (25th;75th percentile) or n (%). P values for comparisons using Mann Whitney U or chi-square tests. *SGA is defined as < 10th percentile for weight. † 1 missing. ‡ 6 (6%) missing in the Very Preterm group and 15 (16%) missing in the Full Term group. BPD = bronchopulmonary dysplasia, IVH = intraventricular hemorrhage, NA = not applicable, SDS = standard deviation score, SGA = small for gestational age.

Except for ethnicity (participants more often reported a non-Western ethnicity), no differences were found in population characteristics of the full-term group between those who opted in and out of the sleep measures (Table S4 ^{(709.6KB, pdf)} in the supplemental material).

Sleep questionnaires

Parents reported on the BISQ that very-preterm-born children slept 23 minutes (95% confidence interval [CI] 5–42) longer during the night than those born full term (P = .01) (Table S1 ^{(709.6KB, pdf)} ). After correction for sex, age, and birth weight SDS, this difference was not significant (19 minutes; 95% CI –1 to 39 minutes; P = .07) (Table 2). Sleep problems were reported in 26% of the very-preterm and in 20% of the full-term group (P = .44). Daytime sleep duration, 24-hour sleep duration, number of nighttime awakenings, method of falling asleep, sleep location, sleeping position, and sleep-onset time were not different between the very-preterm- and full-term-born children. In 20 (31%) and 31 (37%) of very-preterm and full-term children, respectively, parents reported no daytime sleep.

Table 2.

Parent-reported sleep characteristics in preterm and full-term children at the age of 3 years.

	Very Preterm (n = 97)	n	Full Term (n = 92)	n	β	95% CI	P
Nocturnal sleep duration (hh:mm)	10:56 (±00:58)	97	10:32 (±01:06)	85	00:19	−00:01 to 00:39	.07
Daytime sleep duration (hh:mm)*	00:51 (±00:58)	90	00:58 (±00:53)	85	00:03	−00:16 to 00:21	.76
24-Hour sleep duration (hh:mm)	11:45 (±01:06)	91	11:30 (±01:17)	85	00:20	−00:03 to 00:44	.09
No. of nighttime awakenings	1 (0;1)	97	0.5 (0;1)	86	−0.26	−0.69 to 0.18	.25
Sleep problem, yes, n (%)	25 (26)	97	17 (20)	87	NA	NA	.44

Shown are group means (± SD) or medians (IQR), numbers of participants with data, and effect estimates of the comparison between the very-preterm (1) and term group (0) based on linear and logistic regression analysis adjusted for sex, age, and birth weight SDS. Unadjusted analyses are shown in Supplemental Table S1. *n = 20 (31%) and n = 31 (37%) of parents reported zero daytime sleep in very-preterm and full-term children, respectively. BISQ = Brief Infant Sleep Questionnaire, CI = confidence interval, IQR = interquartile range, NA = not applicable, SD = standard deviation, SDS = standard deviation score.

Parent-reported sleep/wake times on weekdays and weekend days were not different between very-preterm and full-term children (Figure 2, Table S2 ^{(709.6KB, pdf)} ). In both groups, midpoint sleep was earlier on weekdays (mean 01:40 and 01:38 hh:mm) than on weekend days (mean 02:02 and 01:55 hh:mm) in very-preterm and full-term children, respectively (P < .001 in both groups).

Figure 2. Parent-reported sleep/wake timelines for the very-preterm and full-term groups.

Timeline with mean parent-reported sleep/wake time points based on the MCTQ for the very-preterm and full-term group, for weekdays and weekend days, after adjustment for sex, age, and birth weight SDS, as shown in Table S2 ^{(709.6KB, pdf)} . MCTQ = Munich Chronotype Questionnaire, SDS = standard deviation score.

Actigraphy

Actigraphy data were available for 69 (71%) of the very-preterm and 73 (67%) of the full-term children (Figure 1). Actigraph wearing time varied between 3 and 7 days and was comparable between the very-preterm (mean 4.94) and full-term children (mean 5.04; P = .50). The very-preterm-born children woke up 21 minutes (95% CI 4–38; P_adjusted = .01) later than the full-term-born children (Table 3 and Table S3 ^{(709.6KB, pdf)} ). Mean daytime sleep duration was 16 minutes (95% CI –00:31 to 00:00; P = .04) shorter in the very-preterm-born children compared with the full-term children (Table S3 ^{(709.6KB, pdf)} ). After correction for confounders, this difference was not significant (10 minutes; 95% CI –00:27 to 00:07; P = .26) (Table 3). No differences were observed in sleep-onset time, 24-hour and nocturnal sleep duration, SOL, WASO, or sleep efficiency between the very-preterm- and full-term-born children. Also, comparison of the 24-hour activity rhythm variables showed no differences in either intra-daily variability (mean 0.34 and 0.42; P = .10) or inter-daily stability (mean 0.71 and 0.73; P = .23) between the very-preterm- and full-term-born children (Table 3).

Table 3.

Sleep and 24-hour activity rhythm by actigraphy in very-preterm- and full-term-born children at the age of 3 years.

	Very Preterm (n = 69)	Full Term (n = 73)	β	95% CI	P
Sleep-onset time (hh:mm)	20:32	20:44	−00:22	−01:06 to 00:22	.33
Wake-up time (hh:mm)	07:11	06:50	00:21	0:04 to 0:38	.01
24-Hour sleep duration (hh:mm)	08:58	09:16	−00:07	−00:30 to 00:16	.53
Nocturnal sleep duration (hh:mm)	07:50	07:50	00:03	−00:19 to 00:25	.80
Daytime sleep duration (hh:mm)	01:10	01:25	−00:10	−00:27 to 00:07	.26
SOL (hh:mm)	0:41	0:44#	−0:03	−0:13 to 0:05	.44
Missing		1
WASO (minutes)	148	137	13	−2 to 28	.08
Sleep efficiency (%)	67	69	−1	−4 to 2	.50
Inter-daily stability*	0.71	0.73	−0.04	−0.11 to 0.03	.23
Intra-daily variability†	0.34	0.42	−0.08	−0.18 to 0.02	.10

Shown are group means, numbers of participants with data, and effect estimates of the comparison between the very-preterm (1) and term group (0) based on linear regression analysis adjusted for sex, age and birth weight SDS. *Inter-daily stability: range 0–1, with higher values indicating more stability. †Intra-daily variability: range 0–2, with higher values indicating more fragmentation. # 1 missing. Unadjusted comparisons are shown in Table S3 ^{(709.6KB, pdf)} . CI = confidence interval, SDS = standard deviation score, SOL = sleep-onset latency, WASO = wake after sleep onset.

Both in the very-preterm and full-term children, sleep data reported by parents were different from data measured by actigraphy. Specifically, the mean 24-hour sleep duration reported by parents was 02:46 (95% CI 02:27–03:05; P < .001) and 02:19 (95% CI 01:55–02:44; P < .001) hh:mm longer in the very-preterm- and full-term-born children, respectively, as compared with 24-hour sleep duration as measured by actigraphy. Likewise, parents reported shorter SOL, earlier sleep-onset time, and later wake-up time (Table S5 ^{(709.6KB, pdf)} ).

DISCUSSION

In this prospective observational study, we compared sleep and 24-hour activity rhythms between very-preterm (corrected) 3-year-olds and their full-term-born peers, while taking sex, age at assessment, and birth weight SDS into account. We observed a trend of longer parent-reported nocturnal sleep in children born very preterm, but this association was attenuated after adjustment. Contrary to our hypothesis, we found no large differences in sleep problems, quality of sleep, 24-hour sleep duration, or activity rhythms between the groups. With actigraphy, children born very preterm appeared to wake up 21 minutes later than their term-born peers. Although other actigraphy variables showed no significant differences, there may be a trend of lower sleep efficiency, higher WASO, and spending more time in bed in preterm-born children. The 21–26% incidence of parent-reported sleep problems in the groups was comparable to data from a large Dutch population–based cohort of children aged 2 to 14 years.² The 23-minutes-longer parent-reported 24-hour sleep duration in the preterm group was no longer significant after correction for confounders, nor was it observed with actigraphy. Based on previous literature, we expected parents of very-preterm children to report longer sleep duration.^7,30 Although not significant, the observed effect size of 23 minutes was in line with 3 earlier population-based cohort studies using BISQ, reporting that, in children of 3–36 months of age, each week of shorter GA was associated with 3 minutes longer sleep duration.³⁰ In another study, at a much later age of 11 years, parents reported 18 minutes longer sleep (9.6 vs 9.3 hours) in extremely preterm-born children (< 28 weeks of GA) compared with full-term-born children.⁷ Although the mechanisms are still unclear, longer sleep duration of preterm-born children in the first years of life might be explained as catch-up sleep needed for maturational processes after shorter gestation and a neonatal period complicated with persistent disturbed sleep.^30,31 However, the difference in sleep duration was not observed with actigraphy.

Our actigraphy data showed that very-preterm children woke up later than full-term-born children. On this topic, the literature is not consistent. In older preterm-born children aged 6 to 12 years, no difference was found in wake-up time using polysomnography,^32,33 whereas in adolescents, using questionnaires and actigraphy, an earlier wake-up time was found in those who were born preterm.^7,34 It is unclear whether the previous findings are explained by lack of power in these studies, the use of different techniques, selection of different study populations, or reflect real changes in sleep patterns and chronotype over time at increasing age. We cannot rule out that later wake-up time in our very-preterm group may (partly) be explained by unmeasured social factors influencing the family schedules and waking times. Also, data on family size, sleep problems of family members, behavior problems, and parenting style may be relevant.³⁵

This study provides unique actigraphy data on 24-hour rhythms in 3-year-old children born very preterm compared with peers born full term and found that inter-daily stability and intra-daily variability were not different between the groups. We found no previous studies with stability and variability data in preterm children in the preschool period. In a Dutch cohort of term-born children at preschool age, inter-daily stability was 0.72, which is comparable to the values in both our groups.³⁶ That earlier study showed a higher intra-daily variability of 0.60 (ie more fragmented 24-hour activity rhythms) than observed in our study groups.³⁶ This may be explained by more fragmentation of 24-hour activity rhythms in older children (6–12 years and 13–18 years) compared with younger children (aged 2–5 years). Their higher intra-daily variability may simply be the result of older children in their cohort, reflecting normal development of sleep-wake patterns during childhood.³⁶ To our knowledge, only 1 study compared sleep fragmentation in preterm vs term children, reporting more fragmented sleep patterns in children born preterm aged 5–12 years (mean age 9 years), based on polysomnography.²⁸ We could not confirm their findings, likely since we used different measurement techniques and our study cohorts consisted of children of a much younger age.

Parental reports and actigraphy showed large differences in sleep duration, which is in line with previous studies in older children and adults.¹² In BISQ, parents reported 2 to 3 hours longer sleep than was measured by actigraphy. A Portuguese study in children aged 3–6 years showed a 159-minute longer total sleep time reported by parents (Children’s Sleep Habits Questionnaire) as compared with actigraphy.³⁷ Possible explanations for parents reporting longer sleep may reflect parents not being aware of their children being awake for periods when lying quietly in bed.³⁷ Possible explanations for actigraphy measuring shorter sleep time may be found in falsely classifying movements during restless sleep episodes (typical in young children) as awakenings.^38–41 Similarly, daytime sleeping time may be misclassified as wake time in conditions with external movement—for example, when a child is sleeping in a stroller or car seat.⁴² As both methods have their strengths and shortcomings, and the superiority of 1 method over the other is not clear in this population, we considered both methods relevant.³⁷ Future studies using gold-standard methods, such as polysomnography, are needed to further evaluate both methods.

Quantity and quality of sleep are important for physical and mental health and child development. Sleep has been acknowledged as one of the most important newborn health outcomes by the International Consortium for Health Outcomes Measurement group of leading physicians, measurement experts, and patients.⁴³ The differences in sleep between very-preterm and full-term children found in our study may appear small. And needing more time in bed to reach the same amount of nocturnal sleep may not seem problematic. However, we cannot rule out that, in families who experienced preterm birth, even small differences may have a large impact on quality of life. We have no data available on the impact of sleep on the quality of life in our groups, except that sleep problems were not reported significantly more often by parents of very-preterm-born children and that the reported sleep duration in both of our study groups was within the recommended range of 10–13 hours per 24 hours for children aged 3–5 years.³⁰ Future studies would benefit from taking more patient-reported outcomes into account.

We found unexpectedly small or a lack of associations between very-preterm birth and sleep, which may have different explanations. The size of our study population limited power, which may have led to false-negative findings. For example, the reported sleep problems in 26% of very-preterm-born children may be significantly higher than the 20% in the full-term-born children, if measured in a large cohort. The sample size also limited the possibility of studying the role of socioeconomic factors, such as ethnic background (more Western), parental education (lower), or neonatal complications, in the very-preterm-born children. Assuming that our findings are valid, the very-preterm-born children in our study either had fewer sleep problems or the full-term-born children had more problems than expected. The latter is unlikely as the values found were all within normal ranges of previous cohort studies. Preterm-born children having unexpected normal sleep may cautiously suggest that the postnatal care and home environment were adequately adjusted to preserve or restore the development of the 24-hour rhythm and sleep. Possible protective factors may include our Newborn Individualized Development Care and Assessment Program (NIDCAP), which is associated with increased time in quiet and active sleep states.⁴⁴ Skin-to-skin contact or “kangaroo” care may positively affect the behavioral state organization, sleep and wake states, and brain maturation.⁴⁵ In addition, postdischarge interventions, such as the TOP program (transmural developmental support for very-preterm infants and their parents) may have had beneficial effects.⁴⁶

Strengths and limitations

As compared with previous studies, the strengths of this study include the relatively large number of very-preterm (< 30 weeks’ GA) and full-term-born children at the same age, born in the same period, and living in the same geographic area. We used both subjective and objective sleep measurement methods, including parental perspectives. We measured sleep with actigraphy for multiple days and included “daytime naps,” which are still relevant for the 24-hour sleep duration in the majority of preschool children. We were able to report unique data on intra-daily variability and inter-daily stability in children born very preterm at this age. In line with national policy and the literature, we adjusted for corrected age to account for a prolonged effect of low gestational age at birth.⁴⁷

One of the main limitations of this study is nonadherence, as 12 very-preterm (12%) and 17 full-term (16%) children refused to wear the actigraph. This may have resulted in selection bias if children with behavioral problems (related to sleep problems) were more likely to refuse participation. Furthermore, in a total of 12 children, there was a problem with the watch (broken or lost), resulting in loss of data. However, bias is likely limited, as the nonadherence percentages and material problems were comparable in both groups (child refusal in 12% and 16% and watch problems in 5% and 6% of very-preterm- and full-term-born children, respectively). If reporting and awareness differed between parents of very-preterm and full-term children, this may have biased our results. As all questionnaires were fully completed by the parents of the very-preterm-born children, but incomplete or missing in 16% of the controls, bias cannot be ruled out. Dilution of our results may have been introduced by the above issues, as well as by excluding infants with severe early brain injury (IVH grade > II or posthemorrhagic ventricular dilatation), who may have more severe sleep problems. This might also limit the external generalizability to other very-preterm cohorts.

To the best of our knowledge, no questionnaires on chronotype are validated for use below the age of 4 years. We used the short and simple MCTQ, which is validated starting from 6 years and can be used into adulthood.²¹ Using questionnaires and diaries always creates a risk of recall or response bias. We noticed that some parents perceived difficulties in reporting their child’s exact sleep and wake-up times. Parents also acknowledged that sleep may have been overreported as they may not be fully aware of their children’s behaviors during the night—for example, when lying quietly awake in bed. One of the main limitations of actigraphy compared with polysomnography is the poor agreement of total sleep time and WASO, while estimates of total bed time often show satisfactory agreement.⁴⁸ Furthermore, due to lack of access to (home) polysomnography and data on upper airway obstruction, we have no information on sleep-disordered breathing.⁴⁹ As preterm birth is associated with obstructive sleep apnea syndrome, this may have influenced sleep measures in this group. For future studies, we would recommend a multimodal sleep assessment to be able to study the duration of all sleep states and true awakenings. To disentangle the role of the earlier mentioned social and sociodemographic factors as well as intervention programs, larger cohorts are needed.

In conclusion, we found that very-preterm-born children at 3 years of age sleep quite similarly to their term-born peers while taking sex, (corrected) age at assessment, and birth weight SDS into account. Sleep problems were common, but not more prevalent than in full-term children. As sleep patterns evolve over a lifetime, this does not rule out that more serious sleep problems may occur later in life. Actigraphy data suggest that preterm-born children may wake up later than children who are born full term, although this was not reported by parents. Further studies are needed to explore how sleep relates to cardiometabolic and neurodevelopmental outcomes after preterm birth and whether interventions to minimalize disturbance of rhythm and sleep should start in the early neonatal and infant period.

DISCLOSURE STATEMENT

All authors had final approval of the version to be published and agree to be accountable for all aspects of the work. The authors report no conflicts of interest. The BOND study received no specific payments or services from a third party for this research. This publication is part of the project BioClock (with project number 1292.19.077) of the research program Dutch Research Agenda (NWA): Onderzoek op Routes door Consortia (NWA-ORC), which is (partly) financed by the Dutch Research Council (NWO). The Sophia Pluto study is an investigator-initiated cohort study, for which A.C.S.H.-K. received an independent research grant (number 120417) by Danone Nutricia Research. The sponsor had no role in the study design, collection, analysis or interpretation of the data; the writing of the manuscript; or the decision to submit it for publication.

ABBREVIATIONS

BISQ

Brief Infant Sleep Questionnaire

BOND study

Bodycomposition and NeuroDevelopment in preterm infants

BPD

bronchopulmonary dysplasia

CI

confidence interval

GA

gestational age

hh:mm

hours:minutes

IVH

intraventricular hemorrhage

MCTQ

Munich Chronotype Questionnaire

SDS

standard deviation score

SOL

sleep-onset latency

WASO

wake after sleep onset