Sleep and obesity among children: A systematic review of multiple sleep dimensions

Summary The objectives were to systematically investigate the multiple dimensions of sleep and their association with overweight or obesity among primary school‐aged children. CINHAL, PsycINFO, SPORTDiscus, Medline, Cochrane, Embase, and PubMed databases were searched for papers reporting on an association between children's sleep and weight status. Studies on clinical populations, published in languages other than English, without objectively measured weight status, or where weight status was reported outside the outlined age bracket (5‐13 years) were excluded. A total of 34  248 citations were extracted from our systematic search protocol, of which 112 were included for detailed review. Compared with sleep duration, of which 86/103 articles found a significant inverse association between sleep duration and measured weight status, few studies examined other dimensions of sleep, such as quality, efficiency and bed/wake times, and relationship with weight status. Where studies existed, variation in defining and measurement of these dimensions restricted comparison and potentially influenced discrepancies across results. Overall, the findings of this review warrant the need for further research of the outlined dimensions of sleep. Future research would benefit from clarity on definitions across the different dimensions, along with the use of valid and reliable tools.

quality, and sleep timing (onset/off-set) should also be considered as part of the sleep-obesity relationship. 15 Definitions and categorization of these dimensions differ slightly across studies; however, as guided by definitions from the previous literature, [15][16][17]  Beyond the prominent focus of children's sleep duration on the sleep-obesity association of previous reviews, [9][10][11][12][13][14] there is emerging literature for each of these additional dimensions. For example Jarrin et al studied 240 Canadian children and adolescents (aged 8-17 years) and found that, independent of self-reported sleep duration, children with delayed sleep timing (late to bed and late to wake) related to a higher risk of being affected by overweight and obesity relative to early to bed/wake counterparts. 18 An Australian study found later bedtimes were linked with higher body mass index (BMI) z-scores and lower diet quality scores regardless of self-reported wake time. 19 Sleep quality and sleep efficiency have also been highlighted as important dimensions of sleep associated with overweight and obesity risk among children. [20][21][22] Lui et al used polysomnography (PSG) to analyze sleep quality by recording stages of sleep among a sample of American children and adolescents (7-17 years). It is suggested that nonrapid eye movement (REM) sleep could be an important stage of sleep for endocrine and metabolic regulation as reduced REM was associated with higher BMI z-scores. 23 Furthermore, the authors also reported that higher sleep efficiency (determined by the percentage of time spent asleep between sleep onset to wake time, measured via PSG) was a significant dimension of sleep associated with reduced risk of overweight and obesity. 23 While PSG is considered the gold standard for measuring multiple sleep components, it is quite invasive and costly and therefore not always practical for use in larger samples. 24 Self-reported proxies for sleep quality and efficiency have therefore often been used, with results appearing to support those from more objective measures. Studies have reported that those with lower self-perceived sleep quality (ie, less likely to report sleeping "well") or lower perceived sleep efficiency (reported issues around waking up during the night or issues falling asleep) are more likely to experiences poorer weight status outcomes. 25 With emerging empirical evidence of associations between the specific dimensions of sleep and weight status, a more nuanced understanding of these associations is now possible and needed to adequately inform future obesity prevention initiatives. 11,12 A systematic understanding of the current conceptions and measurements of the dimensions of sleep is needed; and an examination of variability of results across the different dimensions of sleep will help determine the importance of these on the sleep-obesity association, among population samples.
As sleep requirements notably vary across the life span, with quite considerable differences in optimal sleep duration recommendations for primary school aged children (5-13 years; 9-11 hours per night) compared with those for adolescents (14-18 years; 8-10 hours per night), the association between sleep habits and obesity needs to be unpacked across the age groups. 26,27 Furthermore, there is a particular need to better understand the etiology of obesity among early primary school-aged children. Data from the National Child Measurement Program in England found that of children with obesity at the beginning of primary school only 10% were of a healthy weight by the end of primary school, while more than two thirds remained in the category of being with obesity or severe obesity. 26 Therefore, with strong evidence for the tracking of obesity and health behaviors from this age through to adolescence and beyond, [26][27][28][29] along with the differences in sleep needs, there is a need to better understand the sleep-obesity nexus of this particular at risk age group. This article presents a systematic review of the peer-reviewed literature with data on the association between different dimensions of sleep and weight status among primary school-aged children (5-13 years old).

| Inclusion/exclusion criteria
Included studies were restricted to (a) peer-reviewed original research; (b) contain some measure of sleep; (c) had objectively measured weight status; (d) report on the association between the two variables (sleep and weight status); (e) participants were primary school aged children (aged 5-13 years); and (f) sample was nonclinical/free-living population.
Reviews, meta-analyses, dissertations, expert opinions, conference abstracts, unpublished studies, and studies published in a language other than English were excluded from the review. Studies were also excluded if the main outcome variable (children's weight status) was reported outside the age bracket (5-13 years).

| Recording and synthesis of findings
For each included article, the two reviewers (B.M. and C.S.) independently utilized a developed data extraction tool (available in Table S2) to obtain relevant information for each study (eg, study design, sample characteristics, dimension[s] of sleep analyzed, association of each dimension, and the main findings). The study quality was also assessed for each article utilizing an adjusted version of the Newcastle-Ottawa Scale (NOS), 30,31 with a 10-point star scale (10 as the highest quality) to critic studies based on the selection of the study groups (two points), the comparability of the groups (three points), and the ascertainment of either the exposure or outcome of interest (five points).
The two data sets were then cross checked and discrepancies amended, using a third reviewer (S.A.) if required. Once included in the review, a narrative synthesis of the data was conducted. Studies were grouped by dimension of sleep and then measurement type categories, enabling comparison of results and measurements across and within each of the different dimensions of sleep.

| RESULTS
A total of 34 248 citations were extracted from the outlined databases using the key search terms, and duplicate citations (n = 11 001) were removed (see Figure 1). Following the inclusion/exclusion criteria, abstracts were screened and a total of 22 852 irrelevant articles were excluded. The remaining 393 articles were read in full, with 281 excluded: 120 did not meet the age criteria (5-13 years); 8 were of clinical or nonfree-living population groups; 8 did not measure weight status objectively; 68 did not analyze the association between sleepobesity; and 77 were excluded either due to being non-English, not peer-reviewed, review/meta-analysis, duplicate data, or a conference abstract/other unpublished data. This left 112 articles deemed relevant and included in the analysis (Table 1). 8,[20][21][22] The included studies examined sleep across four dimensions (duration, timing, efficiency, and quality). Sleep duration was the most frequent dimension analyzed, with all but nine of the 112 studies

| Sleep duration
Overall, there was strong evidence in support of an inverse associa-    with four reporting a significant negative association between higher sleep efficiency scores and poorer anthropometric outcomes among children, while three found no significant association (Table 3). Abbreviations: Key: ✓ = significant association reported; NS = no significant association with weight status; Pos = positive association (higher % or high problems or high awakenings leads to higher weight status); Neg = negative association (lower % or low problems or low awakenings leads to higher weight status); CS = cross-sectional analysis.

| Sleep quality
Of all 112 studies reviewed, 13 analyzed the association between quality of sleep and children's weight status (Table 4) the two studies indicating lower sleep quality to be associated with poorer weight status, 21,33 one found the association as significant among girls only. 21 El-Sheikh et al, 21 found that those who reported higher sleep problems at T1 (9 years old) had higher BMI scores at T3 (11 years old), although this was only significant among girls.

| DISCUSSION
This systematic review highlights a strong importance on assessing sleep as a behavioral factor associated with increased risk of Ow/Ob among primary school aged children across multiple studies set in multiple countries. Out of the 112 reviewed studies, 98 reported a significant association with a dimension of sleep and increased weight status. Supporting findings from previous reviews, the current review found 86 of 103 (83%) studies found a significant negative association between sleep duration and overweight and obesity among children aged between five to thirteen years old. As previously reported, [9][10][11]13,14 this association has been demonstrated cross sectionally as well as longitudinally, using both subjectively and objectively measured duration of children's sleep.
Of particular interest to the current review, is the in-depth inves- This review identified several timing factors that can be measured, including bedtime, wake time, sleep midpoint, and sleep timing patterns. Across the reviewed articles bedtime appears to be more influential than wake time, with a much stronger evidence for the association of later bedtimes and increased weight status among school children, 36,45,94,121,122 than that of later wake times. 121 Considering the target population, this could mostly be due to the commonality of the daily school routine, acting as a regulator on morning wake times/ schedules, and more individual variability on bedtime routines. 142 One of the other two measures of sleep timing minimally outlined included sleep midpoint, which used as a single marker (the midpoint) to assess the timing of sleep, taking duration completely out of the equation. 22,99,131 The second measure was sleep wake cycles, almost a combined sleep timing/duration variable, categorising children as late to bed/wake early (shortest duration), early to bed/wake early or late to bed/wake late (mid duration), and early to bed/wake late (long duration). 44,72,115,118  questionnaire items were included across these. The variation in measurement approach and definition of these dimensions could be impacting findings. There is also a risk of error associated with potential recall bias from self-report among young children, or reporter bias through proxy report by parents. 143 The Delayed sleep initiation and poor sleep maintenance not only potentially reduce overall sleep duration but can also, as with delayed sleep timing, present available time for snacking. As the length of wake episodes is reportedly more influential on weight status outcomes, rather than simply the number of episodes, 20  In agreeance with a recent review that evaluated the association between sleep quality and obesity among children and young adults, 145 more research is required to explore how these factors impact each other and the sleep-obesity relationship, along with the need for clarity on defining and assessing these dimensions.

| Strength and Limitations
To the authors' knowledge, this is the first systematic review that has