BACKGROUND. Adequate sleep optimizes children's learning and behavior. However, the natural history and impact of sleep problems during school transition is unknown.
OBJECTIVES. To determine (1) the natural history of sleep problems over the 2-year period spanning school entry and (2) associations of children's health-related quality of life, language, behavior, learning, and cognition at ages 6.5 to 7.5 years with (a) timing and (b) severity of sleep problems.
METHODS. Data were drawn from the Longitudinal Study of Australian Children. Children were aged 4 to 5 years at wave 1 and 6 to 7 years at wave 2. Parent-reported predictors included (1) timing (none, persistent, resolved, incident) of moderate/severe sleep problems over the 2 waves and (2) severity (none, mild, moderate/severe) of sleep problems at wave 2. Outcomes included parent-reported health-related quality of life and language, parent- and teacher-reported behavior, teacher-reported learning, and directly assessed nonverbal (matrix reasoning) and verbal (receptive vocabulary) cognition. Linear regression, adjusted for child age, gender, and social demographic variables, was used to quantify associations of outcomes with sleep-problem timing and severity.
RESULTS. Sleep data were available at both waves for 4460 (89.5%) children, of whom 22.6% (17.0% mild, 5.7% moderate/severe) had sleep problems at wave 2. From wave 1, 2.9% persisted and 2.8% developed a moderate/severe problem, whereas 10.1% resolved. Compared with no sleep problems, persistent and incident sleep problems predicted poorest health-related quality of life, behavior, language, and learning scores, whereas resolving problems showed intermediate outcomes. These outcomes also showed a dose-response relationship with severity at wave 2, with effect sizes for moderate/severe sleep problems ranging from −0.25 to −1.04 SDs. Cognitive outcomes were unaffected.
CONCLUSIONS. Sleep problems during school transition are common and associated with poorer child outcomes. Randomized, controlled trials could determine if population-based sleep interventions can reduce the prevalence and impact of sleep problems.
Adequate sleep quantity and quality are important for the developing child's learning, memory, and behavior.1–3 Although practitioners may predominantly associate children's sleep problems with infancy, it is now clear that they are extremely common at least to the age of school entry. For example, in the 2004 Longitudinal Study of Australian Children (LSAC), 19.8% of parents of 4- to 5-year-olds reported a mild sleep problem and 13.8% reported a moderate or worse sleep problem, with specific issues including difficulty falling asleep (12.8%), night waking (18.1%), snoring (9.7%), and morning fatigue (9.1%).1
Recent cross-sectional and longitudinal studies suggest that sleep difficulties in school-aged children have serious consequences. In 449 children aged 9 to 14 years, quality of sleep was directly related to the child's self-image, achievement motivation, and control of aggression,4 whereas in 635 children aged 6 to 8 years, poor sleep was associated with teacher ratings of poor academic performance and behavioral problems.3 In Australian children,1 moderate/severe preschool sleep problems are associated with a 12-fold increase in the diagnosis of attention-deficit/hyperactivity disorder, and sleep problems of any degree (including mild) are associated with hyperactivity/inattention and conduct problems and poorer quality of life, particularly in the psychosocial domain. Chervin et al5 found that sleep-disordered breathing and daytime tiredness was associated with poor teacher ratings of academic performance.
Poor transition into school has been shown to have a significant negative effect on a child, with respect to their overall school achievement, completion rates, and social and emotional development.6 A child's successful transition to school is influenced by many factors, including their own intrinsic academic, social, emotional, and behavioral competencies. 6,7 Children's positive, communicative relationships with teachers and their friendships and social status with their peers are important factors in children's early school and classroom adjustment.7 If sleep problems affect a range of these competencies, they could be responsible for significantly impaired school transition.
Many childhood sleep problems are amenable to brief treatment.2,8–12 Therefore, if their population impact is substantial, identifying and managing sleep problems at school transition could represent an important population health opportunity. However, the natural history of sleep problems and their outcomes over this important period have not yet been quantified in longitudinal population studies.
This article addresses these issues using data from the first 2 waves of a large, nationally representative, Australian population cohort. Specifically, we aimed to determine:
the natural history of moderate/severe sleep problems over a 2-year period spanning school entry (ie, proportions of those with persistent, resolving, incident, and no sleep problems); and
associations of children's health-related quality of life (HRQoL), language, behavior, learning, and nonverbal and verbal cognition at ages 6.5 to 7.5 years with:
severity of current sleep problems at wave 2 (cross-sectional analyses); and
timing and persistence of moderate/severe sleep problems from wave 1 (longitudinal analyses).
Study Design and Sample
Data were drawn from the first and second waves of the nationally representative LSAC. The sampling design and field methods for the first wave have been described elsewhere.13 Briefly, the LSAC employed a 2-stage cluster sampling design. In the first stage, postcodes (except the most remote) were sampled after stratifying by state of residence and urban versus rural status to ensure proportional geographic representation. The sampling frame for the second stage comprised all children born between March 1999 and February 2000 and enrolled in the Australian Medicare database, with which 98% of all 4-year old Australian children are registered. Children were randomly selected within each postcode to achieve a cohort aged between 4.3 and 5.2 years at the wave 1 interview with all birth months represented. Of the 10596 children selected, 8391 were still resident within that postcode and could be contacted and, of these, 4983 (59%) took part during 2004. The second wave of the LSAC took place during 2006; 4464 of the 4983 wave 1 children participated in wave 2 (89.6%).
Procedures and Inclusion Criteria
At both waves 1 and 2, trained researchers administered a face-to-face interview in the child's home with the primary caregiver, as well as brief direct assessments with the children. In addition, written questionnaires were completed by the primary caregiver and, wherever possible, the child's primary school teacher or day-care provider. The study was approved by the Australian Institute of Family Studies Ethics Committee, and a parent provided written informed consent for every participant.
Predictor Measures (Child Sleep: Waves 1 and 2)
In both waves, the primary caregiver reported at interview whether they considered their child to have a sleep problem (no, mild, moderate, or severe problem). For aims 1 (natural history) and 2(b) (longitudinal associations with timing and persistence of sleep problems), sleep problems at each wave were dichotomized (moderate/severe = sleep problem, no/mild = not a sleep problem). These responses were then combined across the 2 waves to create the following sleep-problem timing/persistence categories: (1) never (no sleep problem at either wave); (2) resolved (sleep problem at wave 1, but not at wave 2); (3) incident (sleep problem at wave 2, but not at wave 1); and (4) persistent (sleep problem at both waves). In contrast, for aim 2(a) (associations with current sleep problems), we wanted to be able to ascertain dose-response relationships in keeping with research at younger ages,1 and therefore, we trichotomized wave 2 sleep problems into none, mild, and moderate/severe.
Outcome Measures (Wave 2)
The Wechsler Intelligence Scale for Children IV (WISC-IV) matrix reasoning subtest was directly administered as a proxy for nonverbal cognition (mean score in norming population: 10, SD: 3).14 As a proxy for verbal cognition, receptive vocabulary was assessed by using an adapted version of the Peabody Picture Vocabulary Test III (PPVT-III), shortened with publisher permission to 40 items for the LSAC.15 In the LSAC wave 2 pilot study of 421 children aged 67 to 95 months, the Pearson product-moment correlation between the full PPVT-III and the adapted PPVT-III was predominantly in the range 0.93 to 0.97 for each item, with the lowest item correlation being 0.89.16
Language was assessed by summing the 28 items comprising the speech, syntax, semantics, and coherence subscales of the parent-reported Child Communication Checklist 2 (CCC-2), a validated questionnaire for children aged 5 to 18 years with a possible raw score range of 28 to 112, with higher scores indicating greater communication competency.17
HRQoL was measured by the parent-proxy Pediatric Quality of Life Inventory 4.0 (PedsQL-4), a 23-item validated questionnaire for children aged 2 to 18 years yielding a total score with a possible range of 0 to 100, with higher scores representing better quality of life.18
Behavior was assessed by the parent-reported and teacher-reported Strengths and Difficulties Questionnaire (SDQ), a 25-item validated measure of behavioral and emotional problems for children aged 4 to 16 years19; 20 items contribute to the total problems score used here (possible range is 0–40, with higher scores representing worse behavior/mental health).
Learning was assessed by the mean raw score of adapted versions of the teacher-reported language and literacy (11 items) and the mathematical thinking (9 items) subscales of the academic rating scale from the Early Childhood Longitudinal Study.20 Each subscale has a possible range of 1 to 5, with higher scores indicating greater proficiency.
The following potential sociodemographic confounders were chosen a priori to be included in multivariable analyses, because each has been associated both with sleep problems and the outcomes of interest in previous studies1,21,22: child's gender, age in months at wave 2, primary caregiver's highest level of education at wave 1, and equivalized household income at wave 1 (annual gross household income equivalized to household size by taking the midpoint of the 15 income brackets reported in the LSAC data set and dividing by the square root of the number of people residing in the house).23
Child and primary caregiver characteristics and the natural history of sleep problems (aim 1) were described by using standard statistical summary measures.
Cross-sectional associations between wave 2 sleep problems and outcomes (aim 2a) were determined using linear regression to estimate the mean differences between those with no sleep problems and those with each of mild and moderate/severe problems, adjusting for the above covariates. Effect sizes for the adjusted mean difference were calculated by dividing the adjusted mean difference by the SD for those with no sleep problem.24
The extent to which timing and persistence of sleep problems from wave 1 predicted outcomes at wave 2 (aim 2b) was determined by using linear regression to estimate the mean difference between those with no sleep problems at either wave and each of the resolving, incident, and persistent sleep-problem groups, adjusting for the above covariates.
All analyses were implemented by using Stata 10.0 (Stata Corp, College Station, TX). Linear regression assumptions were checked and found to be upheld.
Sleep data were available in both waves for 4460 children (89.5%) (Table 1). Data were largely complete for measures obtained in the face-to-face interview (SDQ, n = 4335) and via direct assessments (matrices, n = 4398; adapted PPVT-III, n = 4405). However, there were substantial missing data for the measures completed in the leave-behind parent questionnaire (PedsQL-4, n = 3471; CCC-2, n = 3440) and the teacher-completed SDQ (n = 3606) (language and literacy [n = 3616] and mathematical thinking [n = 3603]). At wave 2, children were aged between 75 and 94 months, 51.0% were boys, and 62.1% had a primary caregiver who had completed at least high school or equivalent.
Prevalence and Natural History of Sleep Problems (Aim 1)
In wave 1, 20.6% of parents reported mild, 8.7% moderate, and 4.3% severe sleep problems. In wave 2, this fell to 7.0% mild, 3.8% moderate, and 1.9% severe sleep problems. Thus, 13% of children were classified as having a moderate/severe sleep problem in wave 1 and 5.7% in wave 2; this translated into 2.9%, 10.1%, and 2.8% being classified with persistent, resolving, and incident sleep problems, respectively.
Impact of Current Sleep Problems (Aim 2a)
The concurrent presence of sleep problems at wave 2 was adversely associated with all outcomes except for the verbal and nonverbal proxy cognition measures (Table 2). Effect sizes for moderate/severe problems ranged from small (language, effect size: 0.25) to large (behavior by parent report, effect size: 1.04). Effect sizes for mild problems were intermediate to no and moderate/severe sleep problems. Conclusions were similar in posthoc analyses across the full gradient of sleep problems (ie, considering moderate and severe problems separately) with a single exception: severe sleep problems were associated with impaired child nonverbal cognition (WISC-IV matrices mean score: 8.9 [SD: 3.1]) compared with no problem (mean: 10.4 [SD: 2.9]), with an effect size of 0.40 and mean difference of −1.27 (95% confidence interval [CI]: −1.92 to −0.61; P = .002).
Impact of Sleep-Problem Status Over Time (Aim 2b)
Longitudinal moderate/severe sleep-problem status was not associated with either the nonverbal (WISC-IV) or verbal (PPVT-III) proxy cognition measures (Table 3). The largest effect sizes were for persistent sleep problems and poorer child language (CCC-2), behavior (SDQ), and learning (Early Childhood Longitudinal Study subscales) mean scores, and for incident sleep problems and child PedsQL-4 scores. Scores for children with resolving problems were generally intermediate to those with persistent or incident problems (worst) and those with no problems at either wave (best).
Sleep problems during the school-transition period are both common and important to recognize, because they are strongly associated with poorer child HRQoL, language, behavior, and learning. A causal role is supported by the dose-response relationship between increasing severity of sleep problems and worsening child outcomes and by the temporal relationships (with child outcomes worsening from never having sleep problems, through resolving and incident, to persistent sleep problems).
Although these findings are novel, it is not surprising that child outcomes are related to duration and severity of sleep problems, both of which can contribute to total sleep debt. Problems such as frequent night waking and delayed sleep onset lead to sleep deprivation,25 which is believed to decrease a child's ability to control emotions, and increases behavioral problems and daytime fatigue.2,5,26–30 This is consistent with research showing that sleep deficits affect the frontal lobe, which is responsible for control of emotions, spontaneity, language, and social behavior.31 However, cognition seemed to be largely preserved, suggesting that sleep problems at the population level in this age group may have less effect on a function that is largely determined by intrinsic factors such as genetic make up.
This is the first study, to our knowledge, to examine the natural history of sleep problems in children over the transition from preschool to formal education in a population study. Its large sample size and high retention rate supported precise estimates of the natural history and prevalence of sleep problems. Because the same sleep measure was used in both waves, it was possible to examine the adverse effects of persistent, incident, and resolving sleep problems on a diverse range of important outcomes, each assessed using a validated measure. We are not aware of any other studies that have examined the natural history of sleep problems with such a wide range of outcomes in children of this age.
The study had some limitations. First, missing data for some outcomes (eg, HRQoL, language) and some potential confounders (eg, household income) may have affected the associations between sleep problems and child outcomes. However, findings for outcomes with and without missing data were similar, and the analyses adjusted for confounders were very similar to the unadjusted analyses (available on request). In addition, families with missing outcome data reported a similar prevalence of sleep problems (23.6%) but had worse mean SDQ behavior scores (8.5) than families with complete data (prevalence 22.0% and behavior 7.5, respectively). Therefore, if anything, the missing data would likely underestimate the effect of sleep problems in the population. Second, sleep was measured by subjective parent report rather than a validated measure. However, previous studies have established that parent report is an important marker of child sleep problems, there are strong indications of the reliability of parent reporting, and it is parent perceptions that would be most likely to drive uptake of any interventions to improve sleep in this age group. Third, the study examines sleep problems and their associations over a relatively narrow time window. This study addresses neither the relative impacts of sleep problems from earlier in childhood, nor longer-term impacts as children age and sleep patterns alter, although this will be possible in subsequent waves. Finally, an effective randomized, controlled trial would be required to rule out the possibility of reverse causation (ie, behavior problems leading to sleep problems).
This study's results are consistent with, but go beyond, the cross-sectional studies noted previously.1,3,4 None of these studies examined the effect of the timing and natural history of sleep problems in children during or after the school-transition period. In a randomized trial of 47 children with obstructive sleep apnea, improved sleep after adenostonsillectomy resulted in improved child quality of life, general health perception, and behavior and reduced emotional impact on parents compared with children who did not undergo adenostonsillectomy.32 However, obstructive sleep apnea only affects around 2% of children in the primary school years,33 and most sleep problems are not due to obstructive sleep apnea.2 Therefore, the longer-term effects of poor sleep through the primary school years remain to be addressed in this and other studies. If sleep problems reduce the child's ability to successfully adapt to school, then their treatment could prevent the negative consequences associated with poor school transition.6,7,34
Sleep problems among Australian children entering school are common and are associated with poorer outcomes at the population level. Many common sleep problems are amenable to simple behavior strategies. The systematic management of sleep problems at or around school entry could represent a feasible way of reducing child behavior problems and improving quality of life and academic performance in these crucial early years. If rigorous trials show that population strategies can effectively reduce sleep problems, it could have important public health ramifications.
This article uses confidential unit record files from the LSAC survey. The LSAC project was initiated and funded by the Commonwealth Department of Families, Housing, Community Services, and Indigenous Affairs and was managed by the Australian Institute of Family Studies. The findings and views reported in this article are those of the authors and should not be attributed to either the Commonwealth Department of Families, Housing, Community Services, and Indigenous Affairs or the Australian Institute of Family Studies. We thank all the parents and children who took part in waves 1 and 2 of the LSAC.
- Accepted August 22, 2008.
- Address correspondence to Jon Quach, BSc, Royal Children's Hospital, Centre for Community Child Health, Flemington Road, Parkville, Victoria 3052, Australia. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
Sleep problems are prevalent in preschool-aged children and are adversely associated with child behavior, HRQoL, and maternal well-being. However, the natural history and impact of sleep problems over the important school-transition period are unknown.
What This Study Adds
Across school transition, children with sleep problems display substantially poorer behavior, communication, HRQoL, and learning. Outcomes are worst among children with the most persistent and/or severe sleep problems.
- ↵Hiscock H, Canterford L, Ukoumunne OC, Wake M. Adverse associations of sleep problems in Australian preschoolers: national population study. Pediatrics.2007;119 (1):86– 93
- ↵Smits MG, Nagtegaal EE, van der Heijden J, Coenen AML, Kerkhof GA. Melatonin for chronic sleep onset insomnia in children: a randomized placebo-controlled trial. J Child Neurol.2001;16 (2):86– 92
- Hiscock H, Wake M. Randomised controlled trial of behavioural infant sleep intervention to improve infant sleep and maternal mood. BMJ.2002;324 (7345):1062– 1065
- ↵Soloff C, Lawrence D, Johnstone R. LSAC technical reference paper No. 1: sample design. Available at: www.aifs.gov.au/growingup/pubs/techpapers/tp1.pdf. Accessed May 22, 2008
- ↵Wechsler D. The Wechsler Intelligence Scale for Children-4th ed (WISC-IV)—Technical and Interpretive Manual. San Antonio, TX: Psychological Corporation; 2003
- ↵Rothman S. An Australian Version of Adaptive PPVT-III for Use in Research. Melbourne, Australia: Australian Council for Educational Research; 2003
- ↵Australian Institute of Family Studies. Longitudinal Study of Australian Children Data User Guide. Version 3.0. Melbourne, Australia: Australian Institute of Family Studies; 2007
- ↵Bishop D. The Children's Communication Checklist: 2nd ed. Manual. London, United Kingdom: Psychological Corporation Limited; 2003
- ↵Rock DA, Pollack JM. Early Childhood Longitudinal Study—Kindergarten Class of 1998–99 (ECLS-K), Psychometric Report for Kindergarten Through First Grade. Working Paper. Washington, DC: US Department of Education, National Center for Education Statistics; 2002
- ↵Lam P, Hiscock H, Wake M. Outcomes of infant sleep problems: a longitudinal study of sleep, behavior, and maternal well-being. Pediatrics.2003;111 (3). Available at: www.pediatrics.org/cgi/content/full/111/3/e203
- ↵Martin J, Hiscock H, Hardy P, Davey B, Wake M. Adverse associations of infant and child sleep problems and parent health: an Australian population study. Pediatrics.2007;119 (5):947– 955
- ↵Australian Bureau of Statistics. AusStats 6523.0, Household Income and Income Distribution, Australia 2003–04 (Appendix 3). Canberra, Australia: Australian Bureau of Statistics; 2005
- Copyright © 2009 by the American Academy of Pediatrics