Nonpharmacological Interventions to Lengthen Sleep Duration in Healthy Children

Key Points

Question

Are nonpharmacological interventions effective in lengthening sleep duration in healthy children, and if so, what are the key elements?

Findings

In this systematic review and meta-analysis of 45 trials, interventions to improve sleep in healthy children were associated with a small increase in sleep duration (by 10 minutes per night). Face-to-face delivery was an important component of interventions, but interventions that included earlier bedtimes were associated with 47 minutes longer sleep duration per night.

Meaning

Supporting children to go to bed earlier can increase sleep duration in healthy children, and interventions to encourage earlier bedtimes should be included.

This systematic review and meta-analysis evaluates whether nonpharmaceutical interventions to improve sleep duration in healthy children are effective and identifies the key components of these interventions.

Abstract

Importance

Adequate sleep duration is necessary for many aspects of child health, development, and well-being, yet sleep durations for children are declining, and effective strategies to increase sleep in healthy children remain to be elucidated.

Objective

To determine whether nonpharmaceutical interventions to improve sleep duration in healthy children are effective and to identify the key components of these interventions.

Data Sources

CENTRAL, MEDLINE, Embase, PsycINFO, Web of Science Core collection, ClinicalTrials.gov, and WHO trials databases were searched from inception to November 15, 2021.

Study Selection

Randomized clinical trials of interventions to improve sleep duration in healthy children were independently screened by 2 researchers. A total of 28 478 studies were identified.

Data Extraction and Synthesis

Data were processed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) reporting guideline. Random-effects meta-analytic models were used to estimate pooled effect sizes.

Main Outcomes and Measures

Difference in sleep duration, measured in minutes.

Results

A total of 13 539 child participants from 45 randomized clinical trials were included. Of these, 6897 (50.9%) were in the intervention group and 6642 (49.1%) in the control group, and the mean age ranged from 18 months to 19 years. Pooled results indicate that sleep interventions were associated with 10.5 minutes (95% CI, 5.6-15.4) longer nocturnal sleep duration. There was substantial variation between trials. Sources of variation that were not associated with the study effect size included age group, whether the population was identified as having a sleep problem or being at a socioeconomic disadvantage (eg, coming from a low-income family or area), method of assessment of sleep duration (objective vs subjective), location of intervention delivery (home vs school), whether interventions were delivered in person or used parental involvement, whether behavioral theory was used, environmental change, or had greater or lower intensity. Interventions that included earlier bedtimes were associated with a 47-minute sleep extension (95% CI, 18.9-75.0; 3 trials) compared with remaining studies (7.4 minutes; 95% CI, 2.9-11.8; 42 trials) (P = .006 for group difference). Trials of shorter duration (6 months or less) had larger effects.

Conclusions and Relevance

Interventions focused on earlier bedtimes may offer a simple, pragmatic, effective way to meaningfully increase sleep duration that could have important benefits for child health.

Introduction

Early childhood experiences and development can have a significant impact on subsequent health, and adequate sleep during childhood is a critical component of well-being. Healthy sleep involves a combination of appropriate sleep timing, quality, and duration.^1,2 Sleep in children is vital for appropriate growth, cognitive processing, mental well-being, effective social interactions, and metabolic health.^3,4,5,6

Marked differences in sleep patterns are observed through childhood, with declines in sleep duration with increasing age. Guidance from the US National Sleep Foundation recommends that toddlers age 1 to 2 years should sleep 11 to 14 hours per night, preschool-aged children age 3 to 5 years should sleep 10 to 13 hours, school-aged children age 6 to 13 years should sleep 9 to 11 hours, and teenagers age 14 to 17 years should sleep 8 to 10 hours,¹ with similar decreasing sleep durations with increasing child age in the UK,⁷ reflecting the changes in circadian rhythm and homeostatic sleep processes during child development. While these recommendations are not derived from systematic reviews and meta-analysis of the evidence, they represent the best available consensus from a multidisciplinary expert panel.¹ Unfortunately, fewer children are achieving these recommendations, with a trend toward shorter sleep durations,^8,9,10,11 where currently one-third or less of younger children and adolescents are achieving adequate sleep durations.^8,9,10 These short sleep durations have been attributed to increased screen time, electronic media use, stimulant intake (particularly caffeine),^{12,13,14,15,16,17,18,19} and increasing emergence of unhealthy sleep behaviors, including poor bedtime routines (such as irregular sleep timings and unsupportive environments).^20,21

Effective strategies are therefore needed to encourage healthy sleep and optimize sleep behavior in children. Recent systematic reviews and meta-analyses largely of intervention studies have focused on those with sleep problems and/or medical issues^22,23,24 restricted to narrow age groups,^{21,25,26,27,28} including infants and preschoolers where circadian sleep patterns are not yet fully established,^26,27,28 have also considered suboptimal nonexperimental studies²⁶ and have lacked sufficient evidence to unravel the role of single component vs multicomponent interventions as part of broader health improvement agendas.²¹ Most reviews to date have also been narrative, which may reflect a lack of standardization in reporting sleep-related outcomes, particularly sleep duration.²⁵ Of the few meta-analyses that have quantified overall intervention effects, pooled estimates have been modest at best.²⁶ Increasingly, objective measurements are being used, which limits comparisons with previously reported self-reported or parent-reported sleep.²⁶

Interventions in school-aged children may offer an ideal age group to alter nocturnal sleep behavior, given circadian sleep patterns are well established, and scope to deliver interventions within both school and home settings. Hence, we have carried out an up-to-date systematic review of intervention studies to promote optimal sleep patterns in all children, including those of school age (aged 1 to 18 years).

The main aim of this systematic review and meta-analysis was to estimate the effectiveness of sleep-related interventions in healthy children. Secondary aims included subgroup exploration to identify which sleep-related intervention components are most or least effective and to estimate the effect size of these different elements.

Methods

This systematic review and meta-analysis was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline,²⁹ aligned with the EQUATOR reporting guidelines. The protocol was registered on PROSPERO (CRD42019160089).

Eligibility Criteria

Eligible studies were both individual and cluster randomized clinical trials (RCTs) at both the class and school level investigating interventions to affect sleep duration in children aged 1 to 18 years without a known medical condition or disability. Studies were excluded if they included infants younger than 1 year, those with a formal diagnosis of a sleep disorder, or use of pharmacological interventions. Pharmacological interventions are not recommended as first-line therapy for sleep improvement and differ from behavioral interventions.³⁰

The sleep-related intervention could be stand-alone or part of a multicomponent health promotion strategy, eg, addressing obesity. The comparator group could be delayed intervention, alternative child health or safety advice, or usual care. The primary outcome was change in mean self-reported, parent-reported, or objectively measured daily total sleep time (in minutes) in children receiving the intervention compared with controls. Non-English language studies were excluded at the full-text screening stage.

Additional proposed outcomes were change in measures of sleep quality (based on study-specific subjective scales) and sleep hygiene (based on study-specific sleep hygiene scales). However, due to the heterogeneity of these measures and lack of consistency between studies, it was not possible to meaningfully synthesize these outcomes.

Search Strategy

We carried out searches using both text words and Medical Subject Headings/index terms in 7 databases: CENTRAL, MEDLINE (Ovid), Embase (Ovid), PsycINFO (Ovid), Web of Science Core Collection, ClinicalTrials.gov, and World Health Organization trials databases. Searches were carried out from database inception to November 15, 2021. The search strategy used is provided in eTable 1 in the Supplement.

We used the Cochrane RCT Classifier³¹ for an initial assessment of the results. Following this, all remaining titles and abstracts were screened independently by 2 reviewers using Covidence version 1.0,³² with discrepancies resolved by a third reviewer (L.M., U.A.R.C., A.S.D., C.W., C.M.N., A.R.R., C.G.O.). Full-text screening was carried out independently by 2 reviewers in the research group, and disagreements were resolved with a third reviewer (L.M., A.S.D., C.W.).

Data Extraction

Study information was extracted using the Covidence data extraction form, edited to include relevant fields specific to this review, including details about the study population, intervention design, and delivery. Use of theory in the interventions was assessed using a framework³³ that has been used in previous systematic reviews,^34,35 assessed against 4 levels: informed by theory, applied theory, testing theory, and building theory. Data were extracted by 2 reviewers independently, with discrepancies discussed and resolved with a third reviewer (L. M., A. S. D., and C. W.), as necessary.

Quantitative data were extracted using a structured data extraction sheet in Microsoft Excel version 2205. The data extraction sheet was piloted on 6 trials by 2 of us (U. A. R. C. and C. G. O.). Extraction included data in different forms (eg, means at end point or change-on-change), using different measurement methods (eg, actigraphy or self-report/parent-report) related to sleep at different times of the week (eg, weekend or weekday) and at different follow-up periods. Quantitative data extraction was undertaken independently by 2 reviewers, with discrepancies discussed and resolved with a third reviewer where necessary (U. A. R. C., L. P. G., and A. R. R.). Due to reporting inconsistency of intervention length and follow-up period, these were combined under study period.

Study Quality

Cochrane’s Risk of Bias tool³⁶ was chosen over alternative instruments, as it is sensitive to variation in the quality of RCTs. Two reviewers independently scored the domains (selection bias, performance bias, detection bias, attrition bias, reporting bias, and other biases) prior to data analysis, with discrepancies resolved by a third reviewer (L. M., C. W., and A. S. D.).

Statistical Analysis

We used Cochrane’s Review Manager version 5.4 to calculate effect estimates (in minutes) and associated 95% CI, using a random-effects mean difference model. Trials were excluded if the number of participants at follow-up could not be established. The comparator group was split to avoid double counting in 3-arm trials, and where a study did not report pooled results for the entire sample, effects reported were combined using a fixed-effects meta-analysis. Adjusted effect estimates were preferred over unadjusted estimates, and the longest follow-up period was selected for inclusion where multiple follow-up periods were reported. Data for weekday sleep were used in preference to weekend sleep.

Further analyses were performed in Stata version 17 (StataCorp) using metan and metareg commands. Subgroup analyses, examining differences in effect estimates for different populations (eg, by age group, continent, reported social deprivation, self-identified sleep problem), and study design features (eg, method of measuring of sleep duration, length of intervention, location, and mode of delivery) were performed using 2-tailed z tests with statistical significance defined as P < .05. Statistical heterogeneity was assessed using the I² statistic and P values. An I² value of 50% to 75% suggested moderate heterogeneity and of 75% or greater suggested high heterogeneity.³⁷ We assessed publication bias by visual examination of funnel plots, and using Begg and Egger tests, overall and by subgroups.

Results

A total of 28 478 studies were identified, of which 8854 were duplicates. The RCT Classifier excluded 5460 records and a further 436 duplicates were identified, resulting in 13 728 studies to be screened. After screening of titles and abstracts, 411 potentially relevant full-text articles were screened. Figure 1 shows the article selection process. In total, 13 539 child participants from 45 RCTs with 47 estimates of mean differences in sleep duration between intervention and control groups were included in the analyses. Of these, 6897 (50.9%) were in the intervention group and 6642 (49.1%) in the control group, and the mean age ranged from 18 months to 19 years. The key characteristics of selected trials and interventions are shown in Table 1; more detailed characteristics can be seen in eTable 2 in the Supplement and study estimates can be seen in eTable 3 in the Supplement. Of 45 trials, 10 were in preschool-aged children, 11 in primary school–aged children, and 26 in secondary school–aged children.

Figure 1. Study Selection Process.

Table 1. Study Characteristics.

Source	Age	Target population	Settinga	Inclusion criteria	Mode of delivery	Intervention overview b,c	Dose	Intervention duration	Study duration
Beijamini and Louzada,38 2012	Range, 13-14 y	School students in public high schools	School	General population	Face-to-face	Educational program of 4 themed lessons on sleepb	4× 50-min Lessons over 5 d	<1 wk	3 wk
Blake et al,39 2016d	Range, 12-16 y	Students in year 7-10 at risk of major depressive disorder	Secondary schools	High levels of anxiety and sleeping difficulties but without a depressive disorder	Face-to-face	Sleep education program, including 7 sessions and information book	7× 90-min Sessions	7 wk	9 wk
Delli Bovi et al,40 2021	Range, 6-14 y	Obese children	Pediatric obesity clinic in university hospital	BMI >95th percentile for age and sex according to the CDC 2000 growth curves	Face-to-face group sessions and WhatsApp texts	Clinician-led group sessions on health, including nutrition and physical activity	Group session with 3 messages a week for 24 wk or as above plus monthly in-person visits (4× over 6 mo)	24 wk	9 mo
Cain et al,41 2011	Range, 16-17 y	Psychology students	Secondary schools	Students with discrepant weekday-weekend out-of-bed times (>2 h) and insufficient sleep on school nights (<8 h)	Face-to-face	Sleep education classes and motivational interviewing frameworkb	4× 50-min Sessions	4 wk	6 wk
Cepni et al,42 2021	Range, 12-36 mo	Parent-toddler dyads	Home environment	Low-income and ethnically diverse general population	Face-to-face, text reminders	Trained facilitators and health workers delivered health information and positive parenting coaching	10× 90-min Sessions, delivered weekly	10 wk	12 wk
Dewald-Kaufmann et al,43 2013	Range, 12-19 y	Adolescents with chronic sleep reduction	School with home phone calls	Adolescents with a Chronic Sleep Reduction Questionnaire (CSRQ) score ≥40	Face-to-face and telephone	Personal sleep schedule to gradually extend sleep by 10 min per night	Daily sleep schedule provided for 2 wk	2 wk	3 wk
Dong et al,44 2020	Range, 10-18 y	Adolescents with eveningness chronotype	University-based clinic	Reported eveningness and 7-d sleep diary with late sleep time	Face-to-face	Promoting behavior change via 4 cross-cutting modules (functional analysis, goal setting, motivational interviewing, and education)	6× 50-min Sessions, delivered weekly	6 wk	6 mo
Haines et al,45 2013	Range, 2-5 y	Ethnic minority and low-income families	Community health centers in deprived areas	Low-income and ethnically diverse general population with a television in their room	Face-to-face, phone calls, texts, and printed information	Motivational coaching to adopt healthy routines to reduce obesity	4× Home visits, 4× phone calls, weekly text messages	24 wk	24 wk
Hammersley et al,46 2019	Range, 2-5 y	Parent-child dyads	Home environment	General population (healthy weight and overweight)	Online	Online modules targeting multiple behaviors, including sleep	6× 30-min Modules to complete over 11 wk, weekly email reminders	23 wk	24 wk
Hammersely et al,47 2021	Range, 2-6 y	Children aged 2-6 y	Home environment	General population	Telephone and online	Modules targeting multiple behaviors including sleep via telephone calls or online modules	6× 30-min Telephone calls or 6 online modules	3 mo	9 mo
Hart et al,48 2017, and Hart et al,49 2016	Range, 8-11 y	Children aged 8-11 y	Not stated	General population	Face to face, telephone	Behavioral strategies (eg, goal setting, stimulus control, positive reinforcement) to increase time in bed by 1.5 h	4 Sessions over 8 wk (week 1, 60 min; week 2, 30 min; telephone week 4 and week 6)	8 wk	8 wk
Hiscock et al,50 2019	Range, 5-6 y	Children with sleep problems in the first year of schooling	Primary schools	Parent report of moderate or severe sleep problems	Face-to-face, follow-up phone call	Nurse-led consultation, providing information on good sleep practice and new sleep routine	1× 45-min Consultation, 10-min follow-up phone call, optional 20-min consultation	4 wk	52 wk
Kalak et al,51 2012	Range, 17-19 y	School students	Secondary school	General population	Face-to-face	Daily morning run at school in small groups	30-min Run weekdays for 3 wk	3 wk	3 wk
Kira et al,52 2014	Range, 13-16 y	School students	Secondary school	General population	Face-to-face plus workbook	A sleep educational program, including small group discussionsb	4× 50-min Sessions	4 wk	10 wk
Knebel et al,53 2020	Range, 11-15 y	School students	Secondary school	General population	Face-to-face	Program to increase physical activity and reduce sedentary time, teacher training, environmental improvements, and health educationc	Structures created and equipment provided to encourage physical activity and teachers encouraged to use	39 wk	39 wk
Lin et al,54 2018	Median, 15 y	School students	Secondary school	General population	Face-to-face	A sleep education program, encouraging behavior change with individual plan and parental involvement	4× 1-h Sessions every 2 wk	8 wk	24 wk
Lufi et al,55 2011	Range, 13-14 y	School students	Secondary school	General population	Restructure environment	School start time delayed for 1 hb	Daily for 1 wk	1 wk	2 wk
Marsh et al,56 2020	Range, 2-4 y	Parent-child dyads	Clinic and home	Children whose daily screen use exceeded recommendations (≥1 h per day)	Face-to-face and online	Workshop and online materials aimed to promote positive parent-child interactions at bedtime, mealtimes, and when restricting screen time	A half-day workshop and subsequent access to the study website over a 6-wk period	6 wk	12 wk
Mindell et al,57 2009	Range, 18-36 mo	Mother-child dyads	Home	Mother reported small to severe sleep problem, but not a sleep disorder	Face-to-face	A nightly 3-step bedtime routine (bath, massage, quiet activities), lights out within 30 min of end of bath	Everyday	2 wk	3 wk
Mindell et al,58 2016	Range, 2-12 y	Socioeconomically disadvantaged children	Home	Children from low income households without proper bedding	Face-to-face and restructured environment	Sleep education messages for parents	Messages provided twice: week 1 and week 4	4 wk	4 wk
Mitchell et al,59 2021	Range, 10-12 y	Children aged 10-12 y	Home	Parent/guardian report of typically 8-9 h in bed	Online	Mobile health platform used to send goal achievement messages around bedtimes	Daily messages over a 7-wk period	7 wk	9 wk
Moore et al,60 2019	Range, 11-12 y	Low-income children, racial minority children, and children with overweight or obesity	Middle school	Children with BMI ≥85th percentile, entering 6th grade	Face-to-face and telephone	2 Active arms: healthy change, behavioral strategies used in CBT and motivational interviewing; system change, emphasis on restructuring family daily routines	25 Face-to-face sessions in year 1, alternating monthly face-to-face and phone sessions year 2, 4 face-to-face sessions and 8 phone sessions in year 3	156 wk	3 y (annual data collection)
Morell-Azanza et al,61 2019	Range, 7-16 y	Children with abdominal obesity	Clinical	Waist circumference >90th percentile, according to national data	Face-to-face	Family-based lifestyle program with education sessions around diet	6× 30-min Sessions	8 wk	8 wk
Moseley and Gradisar,62 2009	Median, 15.6 y	Year 11 psychology students with delayed sleep	Secondary school	Adolescents with discrepant out-of-bed times (school vs weekend mornings >2 h) and insufficient sleep on school nights	Face-to-face	Education sessions promoting and maintaining a healthy lifestyle using a CBT frameworkb	4× 50-min Sessions	4 wk	6 wk
Moula et al,63 2020	Range, 7-10 y	Primary school children	Primary school	Children with mild emotional and behavioral difficulties	Face-to-face	Art therapies	1-h per wk for 8 wk	8 wk	8 wk
Mousarrezaei et al,64 2020	Range, 7-12 y	Mother-child dyads	Home	Children with sleep inadequacy based on sleep duration	Mobile phone	Educational messages sent to mothers via phone about sleep	Daily messages for 4 wk	4 wk	13 wk
Pablos et al,65 2018	Range, 11-12 y	Students attending Spanish primary schools in eastern Valencia	Primary school	General population	Face-to-face	Education sessions around diet, physical activity, sleep, and hygiene plus a practical physical activity sessionb	Twice weekly sessions lasting 150 min	32 wk	32 wk
Puder et al,66 2011	Median, 5.1 y	Children at school in high migrant population areas in Switzerland	Preschool	General population with >40% migrant children	Face-to-face and take-home resources and restructured environment	Multidimensional lifestyle intervention, focused on physical activity, nutrition, media use, sleep, and adaptation of the built environment of the preschool classb	Children: 4× 45-min sessions a week plus 22 additional sessions; teachers: 2× workshops; parents: 3× evening lessons	42 wk	42 wk
Quach et al,67 2018	Range, 6-7 y	Students in the second year (grade 1) with low working memory	Primary schools	Children with low verbal and/or visuospatial working memory scores relative to their peers	Face-to-face	The Cogmed Working Memory training tasks on temporary storage and manipulation of verbal and/or visuospatial information	20-25 Sessions, 25 min each, over 5-7 wk	5-7 wk	26 wk
Rigney et al,68 2015	Range, 11-13 y	Students in year 6 and 7 from 12 south Australian schools	Secondary schools	General population	Face-to-face and take-home resources	Education sessions on sleep physiology and sleep hygienec	4× Weekly lessons then a group project on sleep, which students then presented to parents	4 wk	18 wk
Santiago et al,69 2022	Range, 14-19 y	Students aged 14-19 y	Secondary schools	Adolescents reporting poor sleep quality or sleep disturbance and daytime sleepiness	Face-to-face	Strength training sessions	55-min Sessions 3× times a week for 12 wk	12 wk	12 wk
Sousa et al,70 2013	Median, 16.8 y	12th-Grade students	Secondary schools	General population	Face-to-face	Education program about physiological and behavioral processes of sleep and healthy lifestyle	5× 50-min Classes	5 d	4 wk
Sundgot-Borgen et al,71 2020	Median, 16.8 y	12th-Grade students	Secondary schools	General population	Face-to-face	Interactive workshops, with 3 themes related to body image, social media literacy, and lifestylec	3× 90-min Interactive workshops	13 wk	52 wk
Tamura and Tanaka,72 2014	Range, 9-11 y	Children in year 4-6	Elementary schools	General population	Face-to-face	Education session with quiz about sleep habits, then monitored practicing target sleep behavior for 2 wkb	1× 40-min Session, then performing sleep behavior for 2 wk	2 wk	2 wk
Tamura and Tanaka,73 2016	Range, 12-13 y	Adolescents in grade 7 public junior high schools	Junior high schools	General population	Face to face	Education session, then monitored practicing a target sleep behavior for 2 wkb	1× 45-min Session, then the sleep behavior for 2 wk	2 wk	2 wk
Taylor et al,74 2015	Median, 6.5 y	Families with children aged 4-8 y with overweight and obesity	Clinical setting	Children with BMI ≥85th percentile	Face to face, phone calls	Families attended 1 multidisciplinary session to develop specific goals, then met with a mentor	Initial session (1-2 h), then meet mentor monthly in year 1 and 3 visits in year 2	104 wk	104 wk
Tomayko et al,75 2018	Range, 2-5 y	American Indian parent-child dyads living on tribal reservations	Home	General population from American Indian communities	Mail, text reminders, social media	12 Healthy lifestyle lessons (1 mailed each month)	12 Lessons (1 each mo for 1 y)	52 wk	52 wk
Uhlig et al,76 2019	Range, 8-13 y	Adolescents in 8th grade	Public primary school	General population	Face-to-face	Rap and sing music therapy sessionsb	16× 45-min Sessions weekly over 4 mo	16 wk	16 wk
Van Dyk et al,77 2017	Range, 14-18 y	High school students who sleep 5-7 h on school nights	Home	General population with reduced sleep	Face to face	Increase time in bed on school nights	Increase time in bed by 1.5 h/night for 2 wk	5 wk	5 wk
van Rijn et al,78 2020	Median, 14.0 y	8th-Grade secondary school students from an all-boys school in Singapore	Secondary school	General population	Face-to-face	Education program about the importance of sleep, barriers to getting enough sleep, and improving their opportunities for sleepb	4× 1-h Lessons delivered over 4 wk	4 wk	11 wk
Walton et al,79 2016	Range, 2-5 y	Parents of children age 2-5 y attending early-years centers	Ontario early-years centers	General population	Face to face	Group discussions about weight-related topics, sometimes stimulated by videos of vignettes	9× 120-min Weekly group sessions	9 wk	36 wk
Wilson et al,80 2014	Median, 4.3 y	Low-income preschool children and their families	Head Start preschools	General population	Face to face	Education sessions for children with a 20-min daily lesson followed by 20 min of related activityb	8× 40-min Sessions for children; 1× 45-min presentation for parents	2 wk	4 wk
Wing et al,81 2015	Median, 14.86 y	7th- to 11th-Grade students aged 12-18 y	Secondary schools	General population	Face to face	Didactic teaching with sleep hygiene tips, followed by small group discussions with casesb	1× 60-min Town hall seminar, 2× 40-min small workshops, 1 mo apart	12.2 wk	17.2 wk
Wolfson et al,82 2015	Median, 12.5 y	7th Graders and their parents	Public middle schools	General population	Face to face	Didactic information on adolescent sleep and specific strategies for establishing healthy sleep hygiene practicesc	8× 40-min Sessions over 4 wk, then 4× booster sessions at 2 mo and 12 mo postintervention	52 wk	12 mo (full results at 4 wk only)
Yoong et al,83 2019	Range, 3-6 y	Parents of children aged 3-6 y at a convenience sample of childcare centers	Home	General population	Video, phone, text message	Video targeting key behavioral beliefs about sleep practices and a telephone call using motivational interview techniques	1× 12-min Video, 1-3× phone calls 2-4 wk later, 2× text messages 1 wk after call	12 wk	12 wk

Abbreviations: BMI, body mass index; CBT, cognitive behavioral therapy.

^a Setting includes where participants were recruited and tested and not necessarily where the intervention has taken place.

^b Study reported cluster randomization at the class level.

^c Study reported cluster randomization at the school level.

^d Blake 2016³⁹ data was supplemented by Blake 2018⁸⁴ data as both report the same study.

The primary analysis found that sleep interventions were associated with 10.5 minutes (95% CI, 5.6-15.4; P < .001) longer nocturnal sleep duration compared with the control condition. Figure 2 shows the primary meta-analysis forest plot, with quality assessments.^{38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83} There was marked heterogeneity across study estimates (I² = 87.2%). All subgroup comparisons by population characteristics, study design features, and intervention features are summarized in Table 2.

Figure 2. Forest Plot Showing Mean Differences in Sleep Duration in 45 Studies, Comparing Sleep Intervention Groups With Control.

Mean differences in sleep duration (in minutes) for each study, comparing duration in those receiving a sleep intervention vs control. Point estimates are shown by blue solid squares, with horizontal lines showing the 95% CI. The pooled estimate based on a random-effects model is shown with a blue diamond. In the risk of bias column, category A indicates random sequence generation (selection bias); category B, allocation concealment (selection bias); category C, blinding of participants and personnel (performance bias); category D, blinding of outcome assessment (detection bias); category E, incomplete outcome data (attrition bias); category F, selective reporting (reporting bias); and category G, other bias. Risk of bias criteria graded from low risk (green), unclear risk (yellow), or high risk (red).

Table 2. Results of Subgroup Comparisons.

Subgroup comparison	Studies, No.	Pooled estimate for subgroup, min (95% CI)	P value	Heterogeneity		Difference in subgroup estimates, P value
				I2, %	P value
Combined estimate, no subgroups	47	10.49 (5.56 to 15.42)	<.001	87.2	<.001	NA
Population characteristics
Age group
Preschool (1-5 y)	10	5.73 (−5.81 to 17.26)	.33	69.1	.001	.68
Primary school (6-11 y)	11	11.56 (3.34 to 19.78)	.006	77.9	<.001
Secondary school (>11 y)	26	11.36 (4.45 to 18.27)	.001	90.5	<.001
Continent
North America	15	15.63 (3.76 to 27.50)	.01	90.4	<.001	.02
Australasia	12	−0.56 (−5.82 to 4.70)	.83	28.9	.16
Europe	9	9.97 (2.06 to 17.88)	.01	49.0	.047
Asia	7	16.93 (4.47 to 29.38)	.008	95.5	<.001
South America	4	4.96 (−1.15 to 11.02)	.11	0	.76
Socioeconomic disadvantagea
Population reported to be disadvantaged	6	7.90 (−3.28 to 19.09)	.17	69.7	.006	.67
Population not reported to be disadvantaged	41	10.64 (5.23 to 16.04)	<.001	87.9	<.001
Population has a sleep problem
No	37	9.73 (4.46 to 15.00)	<.001	85.5	<.001	.75
Yes	10	12.26 (−2.20 to 26.73)	.10	91.9	.10
Study design features
Measure of sleep duration
Objective (actigraphy, EEG)	19	11.00 (1.42 to 20.58)	.02	88.1	<.001	.86
Subjective (child/parent report or sleep diary)	28	9.95 (3.98 to 15.91)	.001	86.9	<.001
Subjective measures
Child self-report	11	7.76 (−2.31 to 17.65)	.13	92.8	<.001	<.001
Parent report	10	0.43 (−3.41 to 4.27)	.83	9.3	.36
Sleep diary	7	26.44 (17.69 to 35.19)	<.001	13.5	.33
Nights used for data collection
Weekend nights	2	2.86 (−0.10 to 5.83)	.06	0	.56	.04
Weekday nights	19	10.34 (2.50 to 18.19)	.01	88.1	<.001
All weeknights	26	11.86 (4.04 to 19.68)	.003	87.2	<.001
Study period for participants (including intervention length and follow-up)
≤1 mo	12	15.22 (7.55 to 22.89)	<.001	39.6	.08	<.001
1-<6 mo	24	14.61 (6.21 to 23.01)	.001	91.6	<.001
6-<12 mo	6	1.64 (−4.65 to 7.93)	.61	41.0	.13
≥1 y	5	0.65 (−2.80 to 4.10)	.71	0	.90
Intervention features
Location of intervention delivery
Home	13	16.42 (−1.21 to 34.05)	.07	89.9	<.001	.49
School	23	9.85 (3.50 to 16.20)	.002	88.1	<.001
Combination of locations/other/unspecified	11	5.84 (−1.57 to 13.25)	.12	72.9	<.001
Mode of delivery
Face-to-face	31	12.86 (6.16 to 19.55)	<.001	89.1	<.001	.046
Telephone, text message, and/or online	14	4.00 (−1.67 to 9.59)	.17	66.9	<.001
Parental involvement in delivery
No	19	10.61 (5.20 to 16.02)	<.001	69.5	<.001	.96
Yes	28	10.39 (2.71 to 18.09)	<.001	90.5	<.001
Physical activity component
No	36	13.39 (7.14 to 19.64)	<.001	88.7	<.001	.001
Yes	11	1.64 (−1.21 to 4.49)	.26	0	.56
Intervention aims to improve sleep only
No	18	1.25 (−2.54 to 5.03)	.52	30.3	.11	<.001
Yes	29	15.76 (8.90 to 22.62)	<.001	90.2	<.001
Intervention focused on obesity
No	35	12.82 (6.97 to 18.68)	<.001	88.5	<.001	.06
Yes	12	3.29 (−4.78 to 11.35)	.43	73.6	<.001
Behavior models used
No	29	11.33 (5.78 to 16.88)	<.001	83.8	<.001	.65
Yes	18	8.79 (−0.62 to 18.19)	.07	87.8	<.001
Environmental change
No	41	11.06 (5.55 to 16.58)	<.001	88.3	<.001	.22
Yes	6	5.34 (−2.01 to 12.68)	.15	36.8	.16
Intervention involves an earlier bedtime
No	44	7.37 (2.93 to 11.82)	.001	82.8	<.001	.006
Yes	3	46.97 (18.94 to 75.00)	.001	89.6	<.001
Earlier bedtime in younger vs older children
Primary school (6 to 11 y)	1	44.00 28.74 to 59.26)	<.001			.88
Secondary school (>11 y)	2	48.01 (−0.48 to 96.52)	.05	94.4	<.001
Intensity
Low/unclear	23	11.13 (4.78 to 17.48)	.001	87.9	<.001	.64
Medium	11	11.83 (0.74 to 22.93)	.04	77.3	<.001
High	13	7.01 (−0.11 to 14.12)	.05	54.4	.01

Abbreviations: EEG, electroencephalogram; NA, not applicable.

^a Socioeconomic disadvantage was defined differently in different studies, but primarily includes individuals from low-income areas or families.

Differences in Sleep Duration by Study Population Characteristics

There were no significant differences in sleep duration estimates across preschool, primary school, and secondary school age groups, presence of social disadvantage (eg, coming from a low-income family or low-income area), or whether the study population included those reporting sleep problems. Most trials were carried out in high-income countries in children of largely White European ancestry (ie, 15 trials were carried out in North America, 12 trials in Australasia, 9 in Europe); the remaining trials were carried out in Asia (7 trials) and South America (4 trials). There were significant differences in the effect size estimates by study geography; pooled estimates from North American (15.6 minutes; 95% CI, 3.8-27.5) and Asian (16.9 minutes; 95% CI, 4.5-29.4) trials appeared to show the largest difference in sleep duration compared with pooled trials from Australasia (−0.6 minutes; 95% CI, −5.8 to 4.7) and South America (5.0 minutes; 95% CI, −1.2 to 11.0), which showed no change with wide CIs (P = .02) (Table 2).

Differences in Sleep Duration by Study Design Features

There were no overall differences in sleep duration when measured using objective (accelerometer or electroencephalography) vs subjective (sleep diaries, child or parental report) methods. However, there were considerable differences in sleep duration among the trials relying on subjective reports, where those using parent report were associated with very small changes in sleep duration (0.4 minutes; 95% CI, −3.4 to 4.3) and those relying on child report were associated with a nonsignificant difference in sleep duration (7.8 minutes; 95% CI, −2.3 to 17.7), whereas trials using sleep diaries were associated with a large difference in sleep duration (26.4 minutes; 95% CI, 17.7-35.2) (P < .001). Estimates from trials recording data on 7 nights per week (all weeknights; 11.9 minutes; 95% CI, 4.0-19.7) ) were similar to those recording data 5 nights per week (weekday nights; 10.3 minutes; 95% CI, 2.5-18.2). Two trials only reported data for weekend nights, reporting a lower effect estimate (2.9 minutes; 95% CI, −0.1 to 5.8); the difference between subgroups was significant (P = .04). Trials that reported final follow-up data within 6 months of the baseline reported significantly greater effect sizes than those reporting final follow-up beyond 6 months.

Differences in Sleep Duration by Intervention Features

There were no statistically significant differences in sleep duration estimates by the location of intervention delivery (eg, home or school), with parental involvement in the intervention, or in whether the intervention was delivered face-to-face. Trials that focused on lengthening sleep alone showed significantly more change in sleep duration (15.8 minutes; 95% CI, 8.9-22.6) than interventions with broader aims (1.2 minutes; 95% CI, −2.5 to 5.0; P < .001). There were no statistically significant differences in the effects of trials by intensity of intervention delivery, use of environmental change to support longer sleep, or whether behavioral theory was used to inform intervention design.

Trials that focused on obesity had a smaller effect (3.3 minutes; 95% CI, −4.8 to 11.3) compared with other trials (12.8 minutes; 95% CI, 7.0-18.7; P = .06). Similarly, trials with a physical activity component (1.6 minutes; 95% CI, −1.2 to 4.5) had a significantly smaller effect size compared with those with no physical activity component (13.4 minutes ;95% CI, 7.1-19.6; P = .001). Trials that involved an earlier bedtime had a significant and substantially greater effect estimate (47.0 minutes; 95% CI, 18.9-75.0) compared with the effect of trials that did not encourage participants to go to bed earlier (7.4 minutes; 95% CI, 2.9-11.8; P = .006). The effect of early bedtimes was in fewer studies (3 trials), but the effect appeared similar in studies carried out in primary school–aged children, and 2 studies solely in secondary school–aged children. Heterogeneity in subgroup analyses varied between 0% and 95.5%.

Possible Association of Bias and Other Factors With Differences in Estimates of Sleep Duration

Visual inspection did not reveal any clear pattern of small study or publication bias overall (Egger test = 0.85); there was no clear pattern of small study bias by subgroups. Equally, risk of bias assessments, including selection bias, performance bias, detection bias, attrition bias, reporting bias, and other biases, did not provide notable patterns in the data (Figure 2). There was no evidence that differences in sleep duration between intervention and control differed by quartile groups of baseline levels of sleep duration. Moreover, there was no evidence of a systematic difference in unadjusted vs adjusted sleep duration analyses, both at a study level (n = 45) or estimate level (n = 58).

Discussion

This systematic review found that sleep interventions in children aged 1 to 18 years were associated with 10.5 minutes longer sleep duration per night. Many intervention components have small effects on sleep duration. Substantial increases in sleep duration were observed in a subgroup of trials encouraging earlier bedtimes, with objectively measured sleep durations increasing by 47 minutes. Interventions that have a broader focus (eg, interventions that focus on obesity) were ineffective in increasing sleep duration, extending sleep by 1.2 minutes compared with 16 minutes in interventions focused on sleep only.

The pooled effect estimate of 10.5 minutes longer sleep duration per night is similar to previous reviews that included fewer trials with correspondingly fewer participants²⁶ or narrower age groups (eg, infants, or infants and preschool-aged children only).^26,27,28 While a 2020 meta-analysis⁸⁵ of older children and young adults (12 to 25 years) found a larger intervention effect (35 minutes), the pooled estimate was associated with wide CIs (95% CI, 8.70-61.14), inclusive of our overall finding. Notably, the pooled effect estimate from that review is similar to the estimate reported in the present review for trials that encouraged earlier bedtimes and importantly used objective actigraphy measurement of sleep duration (with 2 trials being used in both reviews^43,77), which limits well-known biases associated with self-reported or parent-reported sleep times.²⁶ While prior qualitative reviews have suggested that multicomponent interventions underpinned by behavioral theory might be more effective,²⁵ our findings (in agreement with another previous review²⁶) demonstrate that simple sleep-focused interventions are more effective.

We included a large number of RCTs (45 trials with 13 539 participants), which provide the best available evidence of potential intervention. This review spans the entirety of childhood (excluding infants younger than 1 year) and finds no clear evidence of differences in overall intervention effects by age. Most trials included in this review were carried out in primary school–aged and secondary school–aged children, where circadian sleep patterns and nocturnal sleep behavior are well established. The trials included delivered interventions in both school and home settings, and no difference was found in the effect sizes of interventions in either setting. These findings suggest that reinforcement of interventions in both settings could be used for future implementation. Differences in individual study results were reflected by high levels of heterogeneity in both the main and subgroup analyses.

There were inconsistencies measuring sleep-related outcomes across trials. Sleep duration was the most commonly reported outcome, but there was variation in the analytical approach, with some trials reporting mean sleep durations by intervention and control groups and others reporting mean differences in sleep duration between groups at follow-up or differences-in-differences (with differing levels of adjustment); notably, there was no evidence of systematic differences across these groups. Other important sleep-related outcomes beyond duration, such as sleep quality and latency (not reported here), were inconsistently and less frequently reported, which presents challenges for appropriately pooling results. Similarly, other intervention characteristics were not routinely reported, and while interventions underpinned by theory did not appear to be associated with longer sleep durations, more detailed reporting of the use of theory (including behavioral models underpinning proposed interventions) and study fidelity (including uptake and compliance with a given intervention) could usefully contribute to the evidence by allowing effective components to be more readily identified, which would support the development of novel interventions.

While an overall 10-minute increase in sleep duration would have a modest effect in addressing declining sleep durations in children,^8,9,10,11 a 47-minute increase associated with earlier bedtimes would substantially affect this population-level problem. However, while the difference in sleep duration among trials using earlier bedtimes was much larger, this was only based on 3 studies (1 trial in primary school–aged children⁴⁸ and 2 trials in secondary school–aged children^43,77), largely in the US,^48,77 and these findings need to be replicated in further studies and settings. If confirmed, earlier bedtimes are needed given wake-up times are less changeable, especially on schooldays. While one RCT from Israel⁵⁵ included in our review showed 55 minutes longer sleep duration by delaying school start times by 1 hour within the first week, this was not sustained. Further high-quality experimental evidence is needed,⁸⁶ especially as interventions to shift wake-up times by delaying school start times in a UK context have been shown to be impractical.⁸⁷ However, recent observational evidence from the US has shown that increased sleep duration (by as much as 40 minutes) could be achieved by delaying school start times, suggesting that such an approach might be feasible.^88,89 In our study, control group children achieved only 8.6 hours of sleep overall, with 10.4 hours in 10 trials of preschool-aged children, 9.2 hours in 9 trials of primary school–aged children, and as little as 7.6 hours in 22 trials of secondary school–aged children. These durations are much shorter than current recommended levels of 10 to 11 hours in those aged 4 to 12 years and of 8 to 10 hours in those aged 3 to 18 years.¹ Lengthening time asleep by 47 minutes among those with inadequate sleep duration could not only address these shortfalls but could also have important benefits for child health. Observational evidence in those aged 8 to 10 years suggests that such increases could have an appreciable effect on cardiometabolic health (eg, 0.1 lower body mass index [calculated as weight in kilograms divided by height in meters squared] and 2% reduction in insulin resistance),⁹⁰ as well as having potential benefits for mental health and well-being.⁶ Patterning sleep behavior from early age could also track into adulthood, with potential benefits for health in later life. Currently, the only studies encouraging earlier bedtimes within the age ranges of this study were conducted largely in adolescents. Further research is urgently needed to establish whether similarly large effect sizes are observed in sleep interventions encouraging earlier bedtimes in younger children, and, moving beyond this, to establish whether the effects can be sustained. Additional research is also needed to clarify the long-term benefits of such interventions on health outcomes.

Limitations

This study has limitations. Reasons for high levels of heterogeneity in effect sizes remain unclear, and further studies confirming the large increase in sleep duration associated with earlier bedtimes are needed in different age groups and population settings.

Conclusions

Typically, nonpharmacological interventions increase sleep by a small amount (10.5 minutes). Intervention elements, such as a narrow focus on sleep including earlier bedtimes, should be taken forward in terms of study design to help children achieve adequate sleep to support growth, cognitive development, mental well-being, and metabolic health. Based on the best available evidence to date, earlier bedtimes offer a simple, pragmatic strategy to meaningfully increase sleep duration.

Supplement.

eTable 1. Search Strategy Used in Medline

eTable 2. Detailed Table of Study Characteristics

eTable 3. Spreadsheet of Study Estimates Used