A systematic review and meta-analysis of the overall effects of school-based obesity prevention interventions and effect differences by intervention components

A systematic review and meta-analysis of the overall effects of school-based obesity prevention... Background: Childhood obesity is a serious public health concern. School-based interventions hold great promise to combat the rising trend of childhood obesity. This systematic review aimed to assess the overall effects of school-based obesity prevention interventions, and to investigate characteristics of intervention components that are potentially effective for preventing childhood obesity. Methods: We systematically searched MEDLINE, CENTRAL and Embase databases to identify randomized- or cluster randomized- controlled trials of school-based obesity interventions published between 1990 and 2019. We conducted meta-analyses and subgroup analyses to determine the overall effects of obesity prevention programs and effect differences by various characteristics of intervention components on body mass index (BMI) or BMI Z- score of children. Results: This systematic review included a total of 50 trials (reported by 56 publications). Significant differences were found between groups on BMI (− 0.14 kg/m (95% confidence interval: − 0.21, − 0.06)) and BMI Z-score (− 0.05 (− 0.10, − 0.01)) for single-component interventions; significant differences were also found between groups on BMI (− 0.32 (− 0.54, − 0.09) kg/m ) and BMI Z-score (− 0.07 (− 0.14, − 0.001)) for multi-component interventions. Subgroup analyses consistently demonstrated that effects of single-component (physical activity) interventions including curricular sessions (− 0.30 (− 0.51, − 0.10) kg/m in BMI) were stronger than those without curricular sessions (− 0.04 (− 0.17, 0.09) kg/m in BMI); effects of single-component (physical activity) interventions were also strengthened if physical activity sessions emphasized participants’ enjoyment (− 0.19 (− 0.33, − 0.05) kg/m in BMI for those emphasizing participants’ enjoyment; − 0.004 (− 0.10, 0.09) kg/m in BMI for those not emphasizing participants’ enjoyment). The current body of evidence did not find specific characteristics of intervention components that were consistently associated with improved efficacy for multi-component interventions (P > 0.05). Conclusions: School-based interventions are generally effective in reducing excessive weight gain of children. Our findings contribute to increased understandings of potentially effective intervention characteristics for single- component (physical activity) interventions. The impact of combined components on effectiveness of multi- component interventions should be the topic of further research. More high-quality studies are also needed to confirm findings of this review. Keywords: Childhood, Obesity, Prevention, Systematic review, Meta-analysis * Correspondence: whjun@pku.edu.cn Zheng Liu and Han-Meng Xu contributed equally to this work. Department of Maternal and Child Health, School of Public Health, Peking University, Beijing, China Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 2 of 12 Introduction components could be used in both effective and non- Childhood overweight and obesity are global public effective trials, so that the exact components uniquely health issues. The prevalence has increased from 16.9 to related to intervention effectiveness were still unknown. 23.8% in boys and from 16.2 to 22.6% in girls from 1980 Second, a previous review, focusing on the specific role of to 2013 in developed countries, while in developing coun- behavior change techniques, summarized “effectiveness ra- tries, the prevalence has also increased from 8.1 to 12.9% tio” which was determined by the ratio of intervention in boys and 8.4 to 13.4% in girls [1]. Childhood obesity is components used in effective trials relative to those used in associated with a variety of adverse consequences [2, 3], both effective and non-effective trials [19]. However, the which often persist into adulthood [4]. Therefore, preven- trials included in the review were weighted equally by this tion of childhood obesity has become one of the import- approach regardless of the sample size and standard error ant public health priorities. of the outcomes. Third, another review compared sub- The main cause of childhood overweight and obesity group differences in effect sizes between trials with and is an energy imbalance between calories consumed and without the intervention characteristics by using meta- calories expended. Children spend half of their waking analytic technique [20]. However, to our knowledge, this hours and consume at least one-third of their daily calo- approach has not been used in specifying the effective ries at school, and thus schools are being recognized as intervention components in school-based obesity preven- ideal vehicles for delivering obesity interventions to most tion interventions. children [5]. To fill the research gaps in this field, we conducted Based on the Environmental Research framework for a systematic review and meta-analysis of the best weight Gain prevention as well as an energy balance ap- available evidence from randomized controlled trials proach [6], the goal of obesity prevention might be (RCT). This review aimed to firstly summarize the achieved by improvement of energy balance-related behav- overall effect size of school-based obesity prevention iors (physical activity (PA)), dietary improvement (DI)), interventions, and secondly to explore characteristics which can be influenced by environmental influence of intervention components that were associated with (school policy (SP)) directly or indirectly. The direct influ- the improved intervention efficacy. ence reflects the “automatic, unconscious” influence of the SP on behavior. The indirect mechanism reflects the medi- Methods ating role of knowledge, cognitions related to behavior Literature search (health education (HE)) in the influence of the environ- We systematically searched three databases including ment on behavior. As such, a range of intervention compo- MEDLINE, CENTRAL and Embase to identify RCTs of nents (PA, DI, SP, HE) have been widely used in childhood school-based obesity interventions. We included publica- obesity prevention interventions. tions between January 1990 and July 2019. Our searching Notably, a great deal of variability existed in the fre- strategy primarily contained terms in relation to partici- quency, duration and content of intervention components pants, interventions, body weight and study design. The [5, 7, 8]. For instance, some school-based interventions fo- full search strategy was attached in the online supporting cused on increasing students’ daily physical activity [9, 10], document. The reference lists of all retrieved full text re- while others only increased the frequency of physical ac- views were further searched for additional relevant publi- tivity by 2–3 times/week [11, 12]. Topics of health educa- cations. The date for our final search was July 8th, 2019. tion interventions also varied. Some focused primarily on Inclusion criteria for this review were: (1) individual- nutrition education with few physical activity or sedentary or cluster-RCT, (2) interventions implemented among behaviors education [13, 14], some mainly on physical ac- students of elementary or secondary schools (aged tivity or sedentary behaviors education with few nutrition 5~18 years), (3) studies assessing students’ body mass education [15, 16], while others covered both physical ac- index (BMI) or BMI Z-score, (4) anthropometric data tivity and nutrition education [9, 17]. The variety of char- being collected by physical examination, (5) interven- acteristics of intervention components raises the question tions lasting for at least 3 months, (6) intervention of what is specifically associated with intervention efficacy. groups aiming for promoting healthy weight or preven- Previous reviews attempted to address question of this tion of overweight or obesity rather than treatment of kind and revealed some general findings. That is, interven- overweight or obesity, (7) comparison groups being tions covering multiple components and involving families active controls, usual practice controls (maintaining tended to be effective [5, 6]. Three issues remained yet. “normal” school activities) or wait-list controls, (8) the First, some reviews only summarized intervention compo- English version of full-text publications available (for nents that were commonly used in previous trials [5, 18], pragmatic reasons), as well as(9) studies providing data but they did not compare various components used in ef- for meta-analyses (means, standard deviations (SDs) or fective or non-effective trials. In other words, the identified 95% confidence intervals (CIs)). Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 3 of 12 Exclusion criteria included (1) studies only using ques- Data synthesis tionnaires to collect the adiposity outcomes, and (2) We calculated differences in means of BMI and BMI Z- studies specifically designed for the treatment of obesity- score between intervention and control groups that were related diseases (e.g., type 2 diabetes or hypertension). reported change from baseline or follow-up BMI indices controlled for baseline measures. If the trials reported Screening and data extraction data at both immediately post-intervention and subse- First, two reviewers (HMX; YZP) independently screened quent follow-ups, only the data at immediately post- the titles and abstracts of publications obtained by the intervention was included in the meta-analyses (as most searches. Second, full texts were further identified for their of the included studies did not report the sustained eligibility. Reference lists of reviews were additionally effect of interventions). checked for their eligibility. Discrepancies between the As we expected considerable heterogeneity across two reviewers (HMX; YZP) were discussed by themselves studies, the random-effects model was used to pool or with a third reviewer (ZL) and resolved with consensus. the weighted results by inverse variance methods. We The first reviewer (ZL) developed a detailed coding used the I statistic to provide a measure of hetero- scheme, and the extraction items included authors, year geneity. Results with P < 0.05 are reported as signifi- of publication, study design, sample size, age of partici- cant. The level of heterogeneity across studies were rated 2 2 2 pants, percentages of female participants, components as low (I = 25%), moderate (I =50%) or high (I =75%). and characteristics of interventions, outcome measures We used Stata/SE 15.0 (StataCorp) for all analyses. and assessment of risk of bias. The components and characteristics of interventions were extracted from both Subgroup analyses the main papers and the intervention protocols [20–30]. To identify the characteristics of interventions poten- Authors were further contacted for details that were not tially contributing to the improved effects, we first cate- reported in the publications (in three cases). A second gorized interventions into those having the specified reviewer (HMX) independently extracted data from all intervention components (i.e., SP, HE, PA and DI) and the included studies, and 20% of the extracted data were those without these. Then, we classified interventions double checked by the first reviewer (ZL). Disagreements into those using single or multiple components, as their in relation to data extraction were resolved by a brief effect sizes were detected as significantly different in pre- discussion (kappa statistics: 0.62; in five cases). vious reviews [6, 32]. Further, we used subgroup analyses to examine differences in effect sizes by inclusion of SP Assessment of risk of bias related to obesity prevention (for multi-component in- Risk of bias of individual studies was assessed following the terventions; yes vs. no), whether or not topics in HE Cochrane guidance [31]. The assessment contains the fol- covering both energy input and expenditure (for both lowing domains including (1) random sequence generation single- and multi-component interventions; yes vs. no), (whether or not the study used a randomized sequence of as- duration and frequency of PA (for both single- and signments), (2) concealment of the allocation sequence multi-component interventions; ≥3 times/week and ≥ 10 (whether or not the allocation sequences were protected by min/time vs. < 3 times/week or < 10 min/time), whether adequate concealment), (3) blinding of participants and or not including curricular PA (for both single- and personnel (whether or not participants or healthcare pro- multi-component interventions; yes vs. no), whether or viders were aware of intervention assignments), (4) blinding not focusing on students’ enjoyment of PA (for both sin- of outcome assessment (whether or not people who deter- gle- and multi-component interventions; yes vs. no), and mined outcome measurements were aware of intervention whether or not including the DI component (for multi- assignments), (5) incomplete outcome data (the possibility of component interventions; yes vs. no). bias duetomissingoutcome data), (6) selective outcome reporting (whether or not the results reported were consistent with the original variables in the protocol)and (7)other bias Sensitivity analyses (the possibility of bias not reported in the previous domains). We conducted sensitivity analyses for the following The leading author (ZL) was responsible for training the considerations: other author (HMX) to ensure a consistent understanding 1. If heterogeneity in the meta-analyses was moderate of the evaluation criteria of risk of bias between the two or high, we additionally obtained the pooled results by authors (ZL; HMX). Each domain was rated as having a excluding individual studies for which the 95% CI of the high, low or unclear risk of bias. We also paid particular at- intervention effect does not overlap with others. tention to the use of statistical methods specific to cluster- 2. We compared the pooled results obtained by all randomized trials (whether or not considering the cluster studies with those excluding individual studies at high effect), and rated them in the domain of other bias. risk of bias. Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 4 of 12 3. We grouped all comparisons according to character- records were then screened and 456 full-text articles istics of the study population (sex: exclusively boys, were further checked for their eligibility. Finally, 50 exclusively girls; weight status at baseline: not over- trials (involving 63,734 children) reported by 56 arti- weight or obesity, overweight or obese; country: middle- cles [9, 12–17, 33–81] that met the eligibility criteria income countries, high-income countries). If a minimum were included in this review. The flowchart of screening of 2 studies (data available) was included in each group, process is presented in Fig. 1. The list of excluded studies we would further conduct sub-group analyses to investi- is shown in Additional file 1: Table S8. gate whether intervention effectiveness differed within sub-groups. Characteristics of included trials Characteristics of the included trials are demonstrated in Assessment of publication bias Table 1 and Additional file 1: Tables S1-S3. Most of We assessed the possibility of publication bias by draw- them (n = 47, 94%) were cluster RCTs using the school ing funnel plots. We recognized that asymmetry of fun- or class as the unit of randomization. All studies had nel plots can be due to publication bias or a genuine onearm as theintervention group with exception of relationship between effect size and trial size. There three studies [35, 57, 64]. All studies used usual prac- were a minimum of 10 studies required for the meaning- tice controls except one using an active control, in ful interpretation of funnel plots. We also conducted order to mitigate the potential of the Hawthorne effect Egger’s regression test to more definitely ascertain [13]. A large proportion of the studies were imple- whether publication bias was present. mented in high-income countries (n = 40, 80%). Most of them (n = 43, 86.0%) were implemented exclusively Results in elementary schools (mean age: 8.1 years). The follow- Literature screening up period of trials ranged from 3 months to 6 years, We identified 12,614 relevant records, and 2866 were ex- and more than half (n = 32, 64%) of them maintained cluded due to duplicates. The titles or abstracts of 9748 shorter than 12 months. Fig. 1 Study flow of the review Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 5 of 12 Table 1 Characteristics of included interventions (n = 50) Thirty-five (70%) interventions were multi-component while others adopted single component. HE (n =7) or PA Category N % (n = 7) was mostly used among single-component inter- Study design ventions. The combinations of components mostly used Cluster RCT 47 94.0 in multi-component interventions were PA + HE+/−SP RCT 3 6.0 (n =10; “+/−”: with or without), and PA + HE+DI+/−SP Duration of interventions, months (n = 8), followed by HE+DI+/−SP (n = 7), HE+SP (n =5), 3–12 32 64.0 PA + DI+/−SP (n = 3), and PA + SP (n =2). > 12 18 36.0 Assessment of risk of bias Types of schools Assessment of risk of bias was summarized in Fig. 2. Elementary 43 86.0 Most of the trials (n = 47, 98%) were assessed as hav- Secondary 6 12.0 ing a low risk of bias in allocation concealment. And Mixed 1 2.0 most of the trials (n = 49, 98%) were judged as hav- Income level of country ing a high risk of bias in blinding of participants and High-income 40 80 (or) personnel because it was usually not possible for interventions of this nature. Approximately half of Middle income 10 20 the studies were assessed as having an unclear risk Types of interventions of bias due to insufficient descriptions in terms of Single-component (n = 15) random sequence generation (n = 30, 60%), blinding PA 7 3.5 of outcome assessment (n = 25, 50.0%), incomplete HE 7 3.5 outcomedata(n = 27, 54%) or the possibility of se- SP 1 2.0 lective reporting (n =32, 64%). Multi-component (n = 35) Overall effect size PA + SP 2 3.9 Overall effect size was summarized in Figs. 3 and 4. The HE+SP 5 9.8 quantitative synthesis of the single-component inter- PA + HE+/−SP 10 19.6 ventions showed a significant, but small reduction of PA + DI+/−SP 3 5.9 0.14 (95% CI: 0.06, 0.21) kg/m in BMI, and a small HE+DI+/−SP 7 13.7 reduction of BMI Z-score (0.05, 95% CI: 0.01, 0.10) compared with the control group. For the multi- PA + HE+DI+/−SP 8 15.7 component interventions, the pooled results showed a RCT randomized controlled trial, PA physical activity, HE health education, DI dietary improvement, SP school policy; “+/−”: with or without significant, but mild reduction of 0.32 (0.09, 0.54) kg/m in BMI, and 0.07 (0.001, 0.14) in BMI Z-score com- pared with the control group. Although the pooled effect sizes in BMI indices of multi-component inter- ventions were slightly larger than that of single- component interventions, the differences were not Fig. 2 Risk of bias graph Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 6 of 12 Fig. 3 Pooled intervention effect (BMI) statistically significant (P =0.41 for BMI, P =0.71 for risk of bias (Additional file 1:Figures S1-S3).Results BMI Z-score). were also not significantly different across sex, Findings of overall effect size were robust to the weight status and country of the study population exclusion of heterogeneous studies or studies of high (Additional file 1:Table S4). Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 7 of 12 Fig. 4 Pooled intervention effect (BMI Z-score) Subgroup analyses pooled BMI from interventions focusing on enjoyment Subgroup analyses showed that means of BMI differed of participants during PA (n =5) was − 0.19 (95% CI: − significantly by whether or not studies including cur- 0.33, − 0.05) kg/m , while the effect size for interven- ricular PA sessions or emphasizing enjoyment in PA tions not emphasizing enjoyment of participants (n =2) among single-component interventions (Table 2). The was − 0.004 (95% CI: − 0.10, 0.09) kg/m . The effect pooled BMI from single-component interventions in- sizes did not differ significantly on other intervention cluding curricular PA (n =3) was − 0.30 (95% CI: − characteristics among single-component interventions 0.51, − 0.10) kg/m , while the effect size from single- (P > 0.05). Findings of subgroup analyses for single- component interventions not including curricular PA component interventions were consistent with results (n =4) was − 0.04 (95% CI: − 0.17, 0.09) kg/m .The from sensitivity analyses (Additional file 1:Table S6). Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 8 of 12 Table 2 Subgroup analyses by characteristics of single-component interventions Outcomes BMI BMI Z-score N Mean difference, 95% CI P for subgroup analyses N Mean difference, 95% CI P for subgroup analyses Characteristics of the PA component 1) PA’s frequency and duration ≥ 3/week and ≥ 10 min/time 5 −0.10 (− 0.22, 0.01) 0.16 –– – < 3/week or < 10 min/time 2 −0.31 (− 0.57, − 0.05) –– 2) Curricular PA Yes 3 −0.30 (− 0.51, − 0.10) 0.02 –– – No 4 −0.04 (− 0.17, 0.09) –– 3) PA emphasizing enjoyment Yes 5 −0.19 (− 0.33, − 0.05) 0.03 –– – No 2 −0.004 (− 0.10, 0.09) –– Topics of HE covering both energy intake and output Yes 3 −0.06 (− 0.26, 0.13) – 5 − 0.07 (− 0.16, − 0.03) – No 1 −0.40 (− 1.01, 0.21) 1 − 0.12 (− 0.28, 0.04) CI confidence interval, SP school policy, PA physical activity, HE health education, DI dietary improvement. “-” due to insufficient observations Concerning multi-component interventions, subgroup (Additional file 1: Table S5). No significant differences analyses demonstrated that the mean BMI or BMI Z- in effect sizes (P > 0.05) were detected between multi- score differed significantly by interventions emphasizing component interventions with and without other inter- enjoyment in PA (Table 3); however, this difference was vention characteristics, which was consistent with results disappeared when excluding one heterogeneous study from sensitivity analyses (Additional file 1: Table S5, S7). Table 3 Subgroup analyses by characteristics of multi-component interventions Outcomes BMI BMI Z-score N Mean difference, 95% CI P for subgroup analyses N Mean difference, 95% CI P for subgroup analyses Characteristics of the PA component 1) PA’s frequency and duration ≥ 3/week and ≥ 10 min/time 10 −0.48 (− 0.94, − 0.01) 0.63 –– – < 3/week or < 10 min/time 8 − 0.29 (− 0.76, 0.19) –– 2) Curricular PA Yes 14 −0.28 (− 0.60, 0.05) 0.19 7 −0.07 (− 0.19, 0.04) 0.26 No 5 −0.89 (−1.75, − 0.03) 2 − 0.06 (− 0.16, 0.03) 3) PA emphasizing enjoyment Yes 8 −0.88 (−1.42, − 0.34) 0.02 –– – No 11 −0.12 (− 0.42, 0.19) –– – Topics of HE covering both energy intake and output Yes 15 −0.28 (− 0.57, 0.01) 0.07 8 −0.17 (− 0.29, − 0.04) 0.18 No 8 0.04 (− 0.14, 0.22) 5 −0.06 (− 0.15, 0.03) Inclusion of the DI component Yes 12 −0.22 (− 0.63, 0.18) 0.45 10 −0.07 (− 0.18, 0.04) 0.76 No 15 −0.42 (− 0.75, − 0.09) 6 −0.05 (− 0.07, − 0.03) Inclusion of the SP component Yes 18 −0.36 (− 0.63, − 0.09) 0.07 11 −0.04 (− 0.08, − 0.01) 0.52 No 7 − 0.01 (− 0.26, 0.24) 2 −0.08 (− 0.20, 0.03) Abbreviations: CI confidence interval, SP school policy, PA physical activity, HE health education Significance of the finding was disappeared when excluding the heterogeneous study (see Additional file 1: Table S5) Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 9 of 12 Assessment of publication bias consistently found in multi-component interventions. It is As shown in Fig. 5, the funnel plot of the observed effect likely that multi-component interventions demonstrated showed a slightly asymmetric scatter consistent with to be effective were influenced by a combination of inter- publication bias, but P value for Egger’s regression test vention components. The impact of combining compo- was larger than 0.05. nents on intervention effectiveness should be the topic of further research. Discussion For the current body of evidence, we did not find sig- This review is one of the first to use meta-analyses and nificant associations between dietary improvement com- subgroup analyses to systematically review a number of ponents with improved intervention efficacy. This more recent studies, and analyze the potentially effect- finding was consistent with another recent systematic re- ive characteristics of school-based interventions for view of school-based childhood obesity interventions preventing obesity. [84]. The non-significant finding in relation to diet might be interpreted by poor adherence to diet intervention or Interpretation of the study findings the complex interplay of intervention components. We ac- This review found that emphasizing enjoyment in PA ses- knowledge that interaction analyses of intervention com- sions was critical for single-component (PA) interventions. ponents (“intervention × component”) within individual This finding was echoed by previous reviews suggesting studies would have provided a powerful method of under- that lack of motivation and pleasure of physical activity standing the complex interplay of intervention compo- was a barrier to physical activity for children [82, 83]. In- nents. However, of the studies screened for this review, cluding curricular PA sessions was also found to be asso- none reported such “intervention × component” analyses. ciated with improved efficacy of single-component (PA) Therefore, future obesity prevention interventions should interventions. This is, at least partly, explained by the fact address the specific interplay of intervention components, that curricular PA sessions were usually led by physical providing the possibility for further systematic reviews. education teachers, and thus intensity of exercise was su- Findings of the study should also be interpreted in the perior to those including only extracurricular activities, context. The reporting of intervention characteristics (dose, after-school sessions or short activity breaks. Further, the frequency, and content) varied so much between trials that curricular PA sessions were usually structured and com- we were obliged to dichotomise it simply as “including the pulsory for all children in a class and thus adherence specific characteristics of component: yes/no” for the pur- could be relatively guaranteed. Significant associations be- pose of analysis, being nevertheless aware that resolution of tween intervention components and efficacy were not the measure might be compromised in the process. Fig. 5 Assessment of publication bias: funnel plot Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 10 of 12 Comparison with other studies Despite these limitations, our study, based on a systema- Some previous reviews of obesity interventions have tical review of the best available evidence from RCTs, took attempted to address the question of “what” (characteris- a first step towards distinguishing characteristics of effect- tics or components of interventions) really works for the ive school-based obesity prevention interventions. The targeted population [5, 17, 18], but only general findings findings of this review enable a better understanding of were revealed. Further, research gaps remained in relation the effectiveness of complex school-based obesity preven- to the weakness of methods that were used (i.e., no tion interventions. Specifically, the findings of this review comparisons between effective and non-effective trials; suggest that school-based interventions could have signifi- equal weighting of the included trials). The present cant effects on reducing students’ BMI. The effects of review not only provided an update on a recent re- single-component (PA) interventions can be improved view [5] by including several new studies, but also when emphasizing students’ enjoyment in physical activ- identified the characteristics of effective interventions ity, or including curricular PA sessions. through meta-analyses and subgroup analyses. Thus, this review provides important and helpful evidence Conclusions of the potentially effective intervention components Overall, school-based interventions are effective in redu- with different characteristics. cing excessive weight gain of children. Findings of this review increase our understandings of potentially effective Limitations and strengths of the study characteristics of interventions. Future high-quality studies Our results should be weighted cautiously considering should focus more on the interplay of intervention com- the following limitations. First, the studies included in ponents, which could deepen our understandings of the this review were restricted to English full-text publica- complexity of obesity prevention interventions delivered tions found in three electronic databases. Second, the in school settings. considerable level of heterogeneity was detected across studies in this review, which is relatively common among complex obesity interventions. Heterogeneity Supplementary information Supplementary information accompanies this paper at https://doi.org/10. might be originated from the fidelity of the intervention 1186/s12966-019-0848-8. and the population targeted among other factors. We have conducted sensitivity analyses to address this con- Additional file 1: Table S1. Description of the included trials. Table S2. cern. Third, precisely evaluating the contents of some Description of the characteristics of the PA component for the included interventions is difficult and problematic due to incon- studies. Table S3. Description of the characteristics of the DI component for the included studies. Table S4. Differences of overall effect size by sistent reporting. Future trials should be required to sex, weight status and country of the study population. Table S5. report interventions in accordance with TIDieR (tem- Subgroup analyses by characteristics of multi-component interventions plate for intervention description and replication) [85] (excluding heterogeneous studies). Table S6. Subgroup analyses by characteristics of single-component interventions (excluding trials or other tools. Fourth, solely using BMI indices as out- assessed as high risk of bias). Table S7. Subgroup analyses by characteristics come measures in this review is relatively narrow and of multi-component interventions (excluding trials assessed as high risk of insensitive, especially when studying PA interventions, bias). Table S8. The list of excluded studies. Figure S1. Pooled intervention effect after excluding heterogeneous studies (BMI). Figure S2. Pooled as PA interventions might have an impact on BMI by intervention effect after excluding studies at high risk of bias (BMI). Figure S3. affecting intermediate outcomes (increasing PA). We Pooled intervention effect after excluding studies at high risk of bias are planning to consider using behavioral outcomes in (BMI Z-score). a future systematic review. Fifth, we only included RCTs in this review, which cannot address complex Abbreviations interplay of behaviors and real-world settings. However, BMI: Body mass index; CI: Confidence interval; DI: Dietary improvement; RCTs are the best available approach to answer “can it HE: Health education; PA: Physical activity; RCT: Randomized controlled trial; SD: Standard deviation; SP: School policy; TIDieR: Template for intervention work?”, as non-randomized trials might result in in- description and replication comparable baseline data between the two groups, and uncontrolled trials can hardly eliminate the risk of self- Acknowledgments selection bias. Sixth, due to the limited number of in- We thank all the members in our research team. cluded studies as well as the limited sub-group data available for meta-analyses, we cannot investigate Authors’ contributions whether our findings of potentially effective interven- ZL and HMX contributed equally and are considered co-first authors. Study tion components were influenced by sex, weight status design: HJW and ZL; Study selection: ZL, HMX, YZP and LZL; Data extraction: HMX, ZL; Quality assessment: HMX, ZL; Data analysis: ZL, HMX, LMW, YZP, or socio-economic status of the study population. This LZL, SZ, WHL, HJW; Drafting of the manuscript: ZL, HMX, LMW, YZP, LZL, SZ, is thus should be a potential focus for future trials, WHL, HJW; Critical revision of the manuscript for important intellectual which provides a basis for the coming meta-analyses. content: HJW, LMW. All authors read and approved the final manuscript. Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 11 of 12 Funding 17. Xu F, Ware RS, Leslie E, et al. Effectiveness of a randomized controlled This work was supported by National Key R&D Program of China lifestyle intervention to prevent obesity among Chinese primary school (2016YFC1300204), National Natural Science Fund (81703240; 81903343) and students: CLICK-obesity study. PLoS One. 2015;10:e0141421. Postdoctoral Research Foundation of China (2019 M650391). 18. Katz DL, O’Connell M, Njike VY, et al. Strategies for the prevention and control of obesity in the school setting: systematic review and meta- analysis. Int J Obes. 2008;32:1780. Availability of data and materials 19. Martin J, Chater A, Lorencatto F. Effective behaviour change techniques The datasets analyzed during the current study are available from the in the prevention and management of childhood obesity. Int J Obes. corresponding author on reasonable request. 2013;37:1287. 20. Schippers M, Adam PC, Smolenski DJ, et al. A meta-analysis of overall effects Ethics approval and consent to participate of weight loss interventions delivered via mobile phones and effect size Not applicable. differences according to delivery mode, personal contact, and intervention intensity and duration. Obes Rev. 2017;18:450–9. 21. Kremers SP, Visscher TL, Brug J, et al. Netherlands research programme Consent for publication weight gain prevention (NHF-NRG): rationale, objectives and strategies. Eur J Not applicable. Clin Nutr. 2005;59:498. 22. Zahner L, Puder JJ, Roth R, et al. A school-based physical activity program Competing interests to improve health and fitness in children aged 6–13 years (" kinder- The authors declare that they have no competing interests. Sportstudie KISS"): study design of a randomized controlled trial [ISRCTN15360785]. BMC Public Health. 2006;6:147. Author details 23. DuBose KD, Mayo MS, Gibson CA, et al. Physical activity across the Department of Maternal and Child Health, School of Public Health, Peking curriculum (PAAC): rationale and design. Contemp Clin Trials. 2008;29:83–93. University, Beijing, China. School of Public Health, Sydney Medical School, 24. Williamson DA, Champagne CM, Harsha D, et al. Louisiana (LA) health: University of Sydney, Sydney, Australia. design and methods for a childhood obesity prevention program in rural schools. Contemp Clin Trials. 2008;29:783–95. Received: 22 January 2019 Accepted: 16 September 2019 25. Giralt M, Albaladejo R, Tarro L, et al. A primary-school-based study to reduce prevalence of childhood obesity in Catalunya (Spain)-EDAL- Educació en alimentació: study protocol for a randomised controlled trial. Trials. 2011;12:54. References 26. Lawlor DA, Jago R, Noble SM, et al. The active for life year 5 (AFLY5) school 1. Ng M, Fleming T, Robinson M, et al. Global, regional, and national based cluster randomised controlled trial: study protocol for a randomized prevalence of overweight and obesity in children and adults during 1980– controlled trial. Trials. 2011;12:181. 2013: a systematic analysis for the global burden of disease study 2013. 27. Xu F, Ware RS, Tse LA, et al. A school-based comprehensive lifestyle Lancet. 2014;384:766–81. intervention among chinese kids against obesity (CLICK-obesity): rationale, 2. Ebbeling CB, Pawlak DB, Ludwig DS. Childhood obesity: public-health crisis, design and methodology of a randomized controlled trial in Nanjing city, common sense cure. Lancet. 2002;360:473–82. China. BMC Public Health. 2012;12:316. 3. Braet C, Mervielde I, Vandereycken W. Psychological aspects of childhood 28. Sutherland R, Campbell E, Lubans DR, et al. A cluster randomised trial of a obesity: a controlled study in a clinical and nonclinical sample. J Pediatr school-based intervention to prevent decline in adolescent physical activity Psychol. 1997;22:59–71. levels: study protocol for the ‘physical activity 4 Everyone’trial. BMC Public 4. Twig G, Yaniv G, Levine H, et al. Body-mass index in 2.3 million adolescents Health. 2013;13:57. and cardiovascular death in adulthood. N Engl J Med. 2016;374:2430–40. 29. Scherr RE, Linnell JD, Smith MH, et al. The shaping healthy choices program: 5. Bleich SN, Vercammen KA, Zatz LY, et al. Interventions to prevent global design and implementation methodologies for a multicomponent, school- childhood overweight and obesity: a systematic review. Lancet Diabetes based nutrition education intervention. J Nutr Educ Behav. 2014;46:e13–21. Endocrinol. 2018;6:332–46. 30. Lloyd J, Wyatt K. The healthy lifestyles Programme (HeLP)—an overview of 6. Kremers SP, de Bruijn GJ, Visscher TL, et al. Environmental influences on and recommendations arising from the conceptualisation and development energy balance-related behaviors: a dual-process view. Int J Behav Nutr of an innovative approach to promoting healthy lifestyles for children and Phys Act. 2006;3:9. their families. Int J Environ Res Public Health. 2015;12:1003–19. 7. Feng L, Wei DM, Lin ST, et al. Systematic review and meta-analysis of school- 31. Higgins JP, Altman DG, Gøtzsche PC, et al. The Cochrane Collaboration’s based obesity interventions in mainland China. PLoS One. 2017;12:e0184704. tool for assessing risk of bias in randomised trials. BMJ. 2011;343:d5928. 8. Waters E, de Silva-Sanigorski A, Burford BJ, et al. Interventions for preventing 32. Sobol-Goldberg S, Rabinowitz J, Gross R. School-based obesity prevention obesity in children. Cochrane Database Syst Rev. 2011;12:CD001871. programs: a meta-analysis of randomized controlled trials. Obesity. 2013;21:2422–8. 9. Lloyd J, Creanor S, Logan S, et al. Effectiveness of the healthy lifestyles 33. Farmer VL, Williams SM, Mann JI, et al. The effect of increasing risk and Programme (HeLP) to prevent obesity in UK primary-school children: a cluster challenge in the school playground on physical activity and weight in children: randomised controlled trial. Lancet Child Adolesc Health. 2018;2:35–45. a cluster randomised controlled trial (PLAY). Int J Obes. 2017;41:793. 10. Meyer U, Schindler C, Zahner L, et al. Long-term effect of a school-based 34. Hollis JL, Sutherland R, Campbell L, et al. Effects of a ‘school-based’physical physical activity program (KISS) on fitness and adiposity in children: a activity intervention on adiposity in adolescents from economically cluster-randomized controlled trial. PLoS One. 2014;9:e87929. disadvantaged communities: secondary outcomes of the ‘physical activity 4 11. Marcus C, Nyberg G, Nordenfelt A, et al. A 4-year, cluster-randomized, Everyone’RCT. Int J Obes. 2016;40:1486. controlled childhood obesity prevention study: STOPP. Int J Obes. 2009;33:408. 35. Gallotta MC, Iazzoni S, Emerenziani GP, et al. Effects of combined physical 12. Aguilar FS, Martínez-Vizcaíno V, López MS, et al. Impact of an after-school education and nutritional programs on schoolchildren’s healthy habits. physical activity program on obesity in children. J Pediatr. 2010;157:36–42. PeerJ. 2016;4:e1880. 13. Sichieri R, Trotte AP, de Souza RA, et al. School randomised trial on 36. Drummy C, Murtagh EM, McKee DP, et al. The effect of a classroom activity prevention of excessive weight gain by discouraging students from break on physical activity levels and adiposity in primary school children. J drinking sodas. Public Health Nutr. 2009;12:197–202. Paediatr Child Health. 2016;52:745–9. 14. Scherr RE, Linnell JD, Dharmar M, et al. A multicomponent, school-based intervention, the shaping healthy choices program, improves nutrition- 37. Creanor S, Lloyd J, Hillsdon M, et al. Detailed statistical analysis plan for a related outcomes. J Nutr Educ Behav. 2017;49:368–79. cluster randomised controlled trial of the healthy lifestyles Programme 15. Singh AS, Paw MJ, Brug J, et al. Short-term effects of school-based (HeLP), a novel school-based intervention to prevent obesity in school weight gain prevention among adolescents. Arch Pediatr Adolesc Med. children. Trials. 2016;17:599. 2007;161:565–71. 38. Kain J, Concha F, Moreno L, et al. School-based obesity prevention 16. Robinson TN. Reducing children's television viewing to prevent obesity: a intervention in Chilean children: effective in controlling, but not reducing randomized controlled trial. JAMA. 1999;282:1561–7. obesity. J Obes. 2014;2014:618293. Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 12 of 12 39. Tarro L, Llauradó E, Albaladejo R, et al. A primary-school-based study to 64. Williamson DA, Champagne CM, Harsha DW, et al. Effect of an reduce the prevalence of childhood obesity–the EdAl (Educació en environmental school-based obesity prevention program on changes in Alimentació) study: a randomized controlled trial. Trials. 2014;15:58. body fat and body weight: a randomized trial. Obesity. 2012;20:1653–61. 40. Safdie M, Jennings-Aburto N, Lévesque L, et al. Impact of a school-based 65. Lloyd JJ, Wyatt KM, Creanor S. Behavioural and weight status outcomes intervention program on obesity risk factors in Mexican children. Salud from an exploratory trial of the healthy lifestyles Programme (HeLP): a novel Publica Mex. 2013;55:374–87. school-based obesity prevention programme. BMJ Open. 2012;2:e000390. 41. Habib-Mourad C, Ghandour LA, Moore HJ, et al. Promoting healthy eating 66. Llargues E, Franco R, Recasens A, et al. Assessment of a school-based and physical activity among school children: findings from Health-E-PALS, intervention in eating habits and physical activity in school children: the the first pilot intervention from Lebanon. BMC Public Health. 2014;14:940. AVall study. J Epidemiol Community Health. 2011;65:896. 67. Llargués E, Recasens A, Franco R, et al. Medium-term evaluation of an 42. Dewar DL, Morgan PJ, Plotnikoff RC, et al. The nutrition and enjoyable educational intervention on dietary and physical exercise habits in activity for teen girls study: a cluster randomized controlled trial. Am J Prev schoolchildren: the Avall 2 study. Endocrinol Nutr. 2012;59:288–95. Med. 2013;45:313–7. 68. Magnusson KT, Hrafnkelsson H, Sigurgeirsson I, et al. Limited effects of a 2- 43. Jansen W, Borsboom G, Meima A, et al. Effectiveness of a primary school- year school-based physical activity intervention on body composition and based intervention to reduce overweight. Int J Pediatr Obes. 2011;6:e70–7. cardiorespiratory fitness in 7-year-old children. Health Educ Res. 2012;27:484. 44. Siegrist M, Hanssen H, Lammel C, et al. Effects of the school-based 69. Fairclough SJ, Hackett AF, Davies IG, et al. Promoting healthy weight in intervention programme JuvenTUM 3 on physical fitness, physical activity, primary school children through physical activity and nutrition education: a and the prevalence of overweight. Eur J Prev Cardiolog. 2012;19:S2. pragmatic evaluation of the CHANGE! Randomised intervention study. BMC 45. Thivel D, Isacco L, Lazaar N, et al. Effect of a 6-month school-based physical Public Health. 2013;13:626. activity program on body composition and physical fitness in lean and 70. Rosário R, Oliveira B, Araújo A, et al. The impact of an intervention taught obese schoolchildren. Eur J Pediatr. 2011;170:1435–43. by trained teachers on childhood overweight. Int J Environ Res Public 46. Group HS, Foster GD, Linder B, et al. A school-based intervention for Health. 2012;9:1355–67. diabetes risk reduction. N Engl J Med. 2010;363:1769. 71. Rush E, Reed P, Mclennan S, et al. A school-based obesity control 47. Angelopoulos PD, Milionis HJ, Grammatikaki E, et al. Changes in BMI and programme: project energize. Two-year outcomes. Br J Nutr. 2012;107:581. blood pressure after a school based intervention: the CHILDREN study. Eur J 72. Grydeland M, Bjelland M, Anderssen SA, et al. Effects of a 20-month cluster Pub Health. 2009;19:319–25. randomised controlled school-based intervention trial on BMI of school- 48. Vizcaíno VM, Aguilar FS, Gutiérrez RF, et al. Assessment of an after-school aged boys and girls: the HEIA study. Br J Sports Med. 2014;48:768–73. physical activity program to prevent obesity among 9-to 10-year-old 73. Grydeland M, Bergh IH, Bjelland M, et al. Intervention effects on physical children: a cluster randomized trial. Int J Obes. 2008;32:12. activity: the HEIA study - a cluster randomized controlled trial. Int J Behav 49. Donnelly JE, Greene JL, Gibson CA, et al. Physical activity across the Nutr Phys Act. 2013;10:17. curriculum (PAAC): a randomized controlled trial to promote physical 74. Sacchetti R, Ceciliani A, Garulli A, et al. Effects of a 2-year school-based activity and diminish overweight and obesity in elementary school children. intervention of enhanced physical education in the primary school. J Sch Prev Med. 2009;49:336–41. Health. 2013;83:639–46. 50. Barbeau P, Johnson MH, Howe CA, et al. Ten months of exercise improves general 75. Kriemler S, Zahner L, Schindler C, et al. Effect of school based physical and visceral adiposity, bone, and fitness in black girls. Obesity. 2007;15:2077–85. activity programme (KISS) on fitness and adiposity in primary 51. Caballero B, Clay T, Davis SM, et al. Pathways: a school-based, randomized schoolchildren: cluster randomised controlled trial. BMJ. 2010;340:c785. controlled trial for the prevention of obesity in American Indian 76. Llauradó E, Tarro L, Moriña D, et al. EdAl-2 (Educacio en Alimentacio) schoolchildren. Am J Clin Nutr. 2003;78:1030–8. programme: reproducibility of a cluster randomised, interventional, primary- 52. Foster GD, Sherman S, Borradaile KE, et al. A policy-based school intervention school-based study to induce healthier lifestyle activities in children. BMJ to prevent overweight and obesity. Pediatrics. 2008;121:e794–802. Open. 2014;4:e005496. 53. Jiang J, Xia X, Greiner T, et al. The effects of a 3-year obesity intervention in 77. Kipping RR, Howe LD, Jago R, et al. Effect of intervention aimed at schoolchildren in Beijing. Child Care Health Dev. 2007;33:641–6. increasing physical activity, reducing sedentary behaviour, and increasing 54. Li YP, Hu XQ, Schouten EG, et al. Report on childhood obesity in China (8): fruit and vegetable consumption in children: active for life year 5 (AFLY5) effects and sustainability of physical activity intervention on body school based cluster randomised controlled trial. BMJ. 2014;348:g3256. composition of Chinese youth. Biomed Environ Sci. 2010;23:180–7. 78. Adab P, Pallan MJ, Lancashire ER, et al. Effectiveness of a childhood obesity 55. Singhal N, Misra A, Shah P, et al. Effects of controlled school-based multi- prevention programme delivered through schools, targeting 6 and 7 year component model of nutrition and lifestyle interventions on behavior olds: cluster randomised controlled trial (WAVES study). BMJ. 2018;360:k211. modification, anthropometry and metabolic risk profile of urban Asian 79. Pablos A, Nebot V, Vañó-Vicent V, et al. Effectiveness of a school-based Indian adolescents in North India. Eur J Clin Nutr. 2010;64:364. program focusing on diet and health habits taught through physical 56. Peralta LR, Jones RA, Okely AD. Promoting healthy lifestyles among exercise. Appl Physiol Nutr Metab. 2018;43:331–7. adolescent boys: the fitness improvement and lifestyle awareness program 80. Wang Z, Xu F, Ye Q, et al. Childhood obesity prevention through a RCT. Prev Med. 2009;48:537–42. community-based cluster randomized controlled physical activity intervention 57. Vandongen R, Jenner DA, Thompson C, et al. A controlled evaluation of a nd among schools in China: the health legacy project of the 2 world summer fitness and nutrition intervention program on cardiovascular health in 10- youth olympic games (YOG-obesity study). Int J Obes. 2018;42:625–33. year-old to 12-year-old children. Prev Med. 1995;24:9–22. 81. Liu Z, Li Q, Maddison R, et al. A school-based comprehensive intervention 58. El Ansari W, El Ashker S, Moseley L. Associations between physical activity for childhood obesity in China: a cluster randomized controlled trial. Child and health parameters in adolescent pupils in Egypt. Int J Environ Res Obes. 2019;15(2):105–15. Public Health. 2010;7:1649–69. 82. Kebbe M, Damanhoury S, Browne N, et al. Barriers to and enablers of 59. Manios Y, Moschandreas J, Hatzis C, et al. Health and nutrition education in healthy lifestyle behaviours in adolescents with obesity: a scoping review primary schools of Crete: changes in chronic disease risk factors following a and stakeholder consultation. Obes Rev. 2017;18:1439–53. 6-year intervention programme. Br J Nutr. 2002;88:315–24. 83. Zabinski MF, Saelens BE, Stein RI, et al. Overweight children's barriers to and 60. Kafatos A, Manios Y, Moschandreas J. Health and nutrition education in support for physical activity. Obes Res. 2003;11:238–46. primary schools of Crete: follow-up changes in body mass index and 84. Brown T, Moore TH, Hooper L, et al. Interventions for preventing obesity in overweight status. Eur J Clin Nutr. 2005;59:1090. children. Cochrane Database Syst Rev. 2019;Cd001871. 61. Kipping RR, Payne C, Lawlor DA. Randomised controlled trial adapting US 85. Hoffmann TC, Glaszlou PP, Boutron I, et al. Better reporting of interventions: school obesity prevention to England. Arch Dis Child. 2008;93:469–73. template for intervention description and replication (TIDieR) checklist and 62. James J, Thomas P, Cavan D, et al. Preventing childhood obesity by guide. BMJ. 2014;348:g1687. reducing consumption of carbonated drinks: cluster randomised controlled trial. BMJ. 2004;328:1237. 63. James J, Thomas P, Kerr D. Preventing childhood obesity: two year follow- Publisher’sNote up results from the Christchurch obesity prevention programme in schools Springer Nature remains neutral with regard to jurisdictional claims in (CHOPPS). BMJ. 2007;335:762. published maps and institutional affiliations. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal of Behavioral Nutrition and Physical Activity Springer Journals

A systematic review and meta-analysis of the overall effects of school-based obesity prevention interventions and effect differences by intervention components

Loading next page...
 
/lp/springer-journals/a-systematic-review-and-meta-analysis-of-the-overall-effects-of-school-7IKp8iL0BK
Publisher
Springer Journals
Copyright
Copyright © 2019 by The Author(s).
Subject
Medicine & Public Health; Clinical Nutrition; Behavioral Sciences; Health Promotion and Disease Prevention
eISSN
1479-5868
DOI
10.1186/s12966-019-0848-8
Publisher site
See Article on Publisher Site

Abstract

Background: Childhood obesity is a serious public health concern. School-based interventions hold great promise to combat the rising trend of childhood obesity. This systematic review aimed to assess the overall effects of school-based obesity prevention interventions, and to investigate characteristics of intervention components that are potentially effective for preventing childhood obesity. Methods: We systematically searched MEDLINE, CENTRAL and Embase databases to identify randomized- or cluster randomized- controlled trials of school-based obesity interventions published between 1990 and 2019. We conducted meta-analyses and subgroup analyses to determine the overall effects of obesity prevention programs and effect differences by various characteristics of intervention components on body mass index (BMI) or BMI Z- score of children. Results: This systematic review included a total of 50 trials (reported by 56 publications). Significant differences were found between groups on BMI (− 0.14 kg/m (95% confidence interval: − 0.21, − 0.06)) and BMI Z-score (− 0.05 (− 0.10, − 0.01)) for single-component interventions; significant differences were also found between groups on BMI (− 0.32 (− 0.54, − 0.09) kg/m ) and BMI Z-score (− 0.07 (− 0.14, − 0.001)) for multi-component interventions. Subgroup analyses consistently demonstrated that effects of single-component (physical activity) interventions including curricular sessions (− 0.30 (− 0.51, − 0.10) kg/m in BMI) were stronger than those without curricular sessions (− 0.04 (− 0.17, 0.09) kg/m in BMI); effects of single-component (physical activity) interventions were also strengthened if physical activity sessions emphasized participants’ enjoyment (− 0.19 (− 0.33, − 0.05) kg/m in BMI for those emphasizing participants’ enjoyment; − 0.004 (− 0.10, 0.09) kg/m in BMI for those not emphasizing participants’ enjoyment). The current body of evidence did not find specific characteristics of intervention components that were consistently associated with improved efficacy for multi-component interventions (P > 0.05). Conclusions: School-based interventions are generally effective in reducing excessive weight gain of children. Our findings contribute to increased understandings of potentially effective intervention characteristics for single- component (physical activity) interventions. The impact of combined components on effectiveness of multi- component interventions should be the topic of further research. More high-quality studies are also needed to confirm findings of this review. Keywords: Childhood, Obesity, Prevention, Systematic review, Meta-analysis * Correspondence: whjun@pku.edu.cn Zheng Liu and Han-Meng Xu contributed equally to this work. Department of Maternal and Child Health, School of Public Health, Peking University, Beijing, China Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 2 of 12 Introduction components could be used in both effective and non- Childhood overweight and obesity are global public effective trials, so that the exact components uniquely health issues. The prevalence has increased from 16.9 to related to intervention effectiveness were still unknown. 23.8% in boys and from 16.2 to 22.6% in girls from 1980 Second, a previous review, focusing on the specific role of to 2013 in developed countries, while in developing coun- behavior change techniques, summarized “effectiveness ra- tries, the prevalence has also increased from 8.1 to 12.9% tio” which was determined by the ratio of intervention in boys and 8.4 to 13.4% in girls [1]. Childhood obesity is components used in effective trials relative to those used in associated with a variety of adverse consequences [2, 3], both effective and non-effective trials [19]. However, the which often persist into adulthood [4]. Therefore, preven- trials included in the review were weighted equally by this tion of childhood obesity has become one of the import- approach regardless of the sample size and standard error ant public health priorities. of the outcomes. Third, another review compared sub- The main cause of childhood overweight and obesity group differences in effect sizes between trials with and is an energy imbalance between calories consumed and without the intervention characteristics by using meta- calories expended. Children spend half of their waking analytic technique [20]. However, to our knowledge, this hours and consume at least one-third of their daily calo- approach has not been used in specifying the effective ries at school, and thus schools are being recognized as intervention components in school-based obesity preven- ideal vehicles for delivering obesity interventions to most tion interventions. children [5]. To fill the research gaps in this field, we conducted Based on the Environmental Research framework for a systematic review and meta-analysis of the best weight Gain prevention as well as an energy balance ap- available evidence from randomized controlled trials proach [6], the goal of obesity prevention might be (RCT). This review aimed to firstly summarize the achieved by improvement of energy balance-related behav- overall effect size of school-based obesity prevention iors (physical activity (PA)), dietary improvement (DI)), interventions, and secondly to explore characteristics which can be influenced by environmental influence of intervention components that were associated with (school policy (SP)) directly or indirectly. The direct influ- the improved intervention efficacy. ence reflects the “automatic, unconscious” influence of the SP on behavior. The indirect mechanism reflects the medi- Methods ating role of knowledge, cognitions related to behavior Literature search (health education (HE)) in the influence of the environ- We systematically searched three databases including ment on behavior. As such, a range of intervention compo- MEDLINE, CENTRAL and Embase to identify RCTs of nents (PA, DI, SP, HE) have been widely used in childhood school-based obesity interventions. We included publica- obesity prevention interventions. tions between January 1990 and July 2019. Our searching Notably, a great deal of variability existed in the fre- strategy primarily contained terms in relation to partici- quency, duration and content of intervention components pants, interventions, body weight and study design. The [5, 7, 8]. For instance, some school-based interventions fo- full search strategy was attached in the online supporting cused on increasing students’ daily physical activity [9, 10], document. The reference lists of all retrieved full text re- while others only increased the frequency of physical ac- views were further searched for additional relevant publi- tivity by 2–3 times/week [11, 12]. Topics of health educa- cations. The date for our final search was July 8th, 2019. tion interventions also varied. Some focused primarily on Inclusion criteria for this review were: (1) individual- nutrition education with few physical activity or sedentary or cluster-RCT, (2) interventions implemented among behaviors education [13, 14], some mainly on physical ac- students of elementary or secondary schools (aged tivity or sedentary behaviors education with few nutrition 5~18 years), (3) studies assessing students’ body mass education [15, 16], while others covered both physical ac- index (BMI) or BMI Z-score, (4) anthropometric data tivity and nutrition education [9, 17]. The variety of char- being collected by physical examination, (5) interven- acteristics of intervention components raises the question tions lasting for at least 3 months, (6) intervention of what is specifically associated with intervention efficacy. groups aiming for promoting healthy weight or preven- Previous reviews attempted to address question of this tion of overweight or obesity rather than treatment of kind and revealed some general findings. That is, interven- overweight or obesity, (7) comparison groups being tions covering multiple components and involving families active controls, usual practice controls (maintaining tended to be effective [5, 6]. Three issues remained yet. “normal” school activities) or wait-list controls, (8) the First, some reviews only summarized intervention compo- English version of full-text publications available (for nents that were commonly used in previous trials [5, 18], pragmatic reasons), as well as(9) studies providing data but they did not compare various components used in ef- for meta-analyses (means, standard deviations (SDs) or fective or non-effective trials. In other words, the identified 95% confidence intervals (CIs)). Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 3 of 12 Exclusion criteria included (1) studies only using ques- Data synthesis tionnaires to collect the adiposity outcomes, and (2) We calculated differences in means of BMI and BMI Z- studies specifically designed for the treatment of obesity- score between intervention and control groups that were related diseases (e.g., type 2 diabetes or hypertension). reported change from baseline or follow-up BMI indices controlled for baseline measures. If the trials reported Screening and data extraction data at both immediately post-intervention and subse- First, two reviewers (HMX; YZP) independently screened quent follow-ups, only the data at immediately post- the titles and abstracts of publications obtained by the intervention was included in the meta-analyses (as most searches. Second, full texts were further identified for their of the included studies did not report the sustained eligibility. Reference lists of reviews were additionally effect of interventions). checked for their eligibility. Discrepancies between the As we expected considerable heterogeneity across two reviewers (HMX; YZP) were discussed by themselves studies, the random-effects model was used to pool or with a third reviewer (ZL) and resolved with consensus. the weighted results by inverse variance methods. We The first reviewer (ZL) developed a detailed coding used the I statistic to provide a measure of hetero- scheme, and the extraction items included authors, year geneity. Results with P < 0.05 are reported as signifi- of publication, study design, sample size, age of partici- cant. The level of heterogeneity across studies were rated 2 2 2 pants, percentages of female participants, components as low (I = 25%), moderate (I =50%) or high (I =75%). and characteristics of interventions, outcome measures We used Stata/SE 15.0 (StataCorp) for all analyses. and assessment of risk of bias. The components and characteristics of interventions were extracted from both Subgroup analyses the main papers and the intervention protocols [20–30]. To identify the characteristics of interventions poten- Authors were further contacted for details that were not tially contributing to the improved effects, we first cate- reported in the publications (in three cases). A second gorized interventions into those having the specified reviewer (HMX) independently extracted data from all intervention components (i.e., SP, HE, PA and DI) and the included studies, and 20% of the extracted data were those without these. Then, we classified interventions double checked by the first reviewer (ZL). Disagreements into those using single or multiple components, as their in relation to data extraction were resolved by a brief effect sizes were detected as significantly different in pre- discussion (kappa statistics: 0.62; in five cases). vious reviews [6, 32]. Further, we used subgroup analyses to examine differences in effect sizes by inclusion of SP Assessment of risk of bias related to obesity prevention (for multi-component in- Risk of bias of individual studies was assessed following the terventions; yes vs. no), whether or not topics in HE Cochrane guidance [31]. The assessment contains the fol- covering both energy input and expenditure (for both lowing domains including (1) random sequence generation single- and multi-component interventions; yes vs. no), (whether or not the study used a randomized sequence of as- duration and frequency of PA (for both single- and signments), (2) concealment of the allocation sequence multi-component interventions; ≥3 times/week and ≥ 10 (whether or not the allocation sequences were protected by min/time vs. < 3 times/week or < 10 min/time), whether adequate concealment), (3) blinding of participants and or not including curricular PA (for both single- and personnel (whether or not participants or healthcare pro- multi-component interventions; yes vs. no), whether or viders were aware of intervention assignments), (4) blinding not focusing on students’ enjoyment of PA (for both sin- of outcome assessment (whether or not people who deter- gle- and multi-component interventions; yes vs. no), and mined outcome measurements were aware of intervention whether or not including the DI component (for multi- assignments), (5) incomplete outcome data (the possibility of component interventions; yes vs. no). bias duetomissingoutcome data), (6) selective outcome reporting (whether or not the results reported were consistent with the original variables in the protocol)and (7)other bias Sensitivity analyses (the possibility of bias not reported in the previous domains). We conducted sensitivity analyses for the following The leading author (ZL) was responsible for training the considerations: other author (HMX) to ensure a consistent understanding 1. If heterogeneity in the meta-analyses was moderate of the evaluation criteria of risk of bias between the two or high, we additionally obtained the pooled results by authors (ZL; HMX). Each domain was rated as having a excluding individual studies for which the 95% CI of the high, low or unclear risk of bias. We also paid particular at- intervention effect does not overlap with others. tention to the use of statistical methods specific to cluster- 2. We compared the pooled results obtained by all randomized trials (whether or not considering the cluster studies with those excluding individual studies at high effect), and rated them in the domain of other bias. risk of bias. Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 4 of 12 3. We grouped all comparisons according to character- records were then screened and 456 full-text articles istics of the study population (sex: exclusively boys, were further checked for their eligibility. Finally, 50 exclusively girls; weight status at baseline: not over- trials (involving 63,734 children) reported by 56 arti- weight or obesity, overweight or obese; country: middle- cles [9, 12–17, 33–81] that met the eligibility criteria income countries, high-income countries). If a minimum were included in this review. The flowchart of screening of 2 studies (data available) was included in each group, process is presented in Fig. 1. The list of excluded studies we would further conduct sub-group analyses to investi- is shown in Additional file 1: Table S8. gate whether intervention effectiveness differed within sub-groups. Characteristics of included trials Characteristics of the included trials are demonstrated in Assessment of publication bias Table 1 and Additional file 1: Tables S1-S3. Most of We assessed the possibility of publication bias by draw- them (n = 47, 94%) were cluster RCTs using the school ing funnel plots. We recognized that asymmetry of fun- or class as the unit of randomization. All studies had nel plots can be due to publication bias or a genuine onearm as theintervention group with exception of relationship between effect size and trial size. There three studies [35, 57, 64]. All studies used usual prac- were a minimum of 10 studies required for the meaning- tice controls except one using an active control, in ful interpretation of funnel plots. We also conducted order to mitigate the potential of the Hawthorne effect Egger’s regression test to more definitely ascertain [13]. A large proportion of the studies were imple- whether publication bias was present. mented in high-income countries (n = 40, 80%). Most of them (n = 43, 86.0%) were implemented exclusively Results in elementary schools (mean age: 8.1 years). The follow- Literature screening up period of trials ranged from 3 months to 6 years, We identified 12,614 relevant records, and 2866 were ex- and more than half (n = 32, 64%) of them maintained cluded due to duplicates. The titles or abstracts of 9748 shorter than 12 months. Fig. 1 Study flow of the review Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 5 of 12 Table 1 Characteristics of included interventions (n = 50) Thirty-five (70%) interventions were multi-component while others adopted single component. HE (n =7) or PA Category N % (n = 7) was mostly used among single-component inter- Study design ventions. The combinations of components mostly used Cluster RCT 47 94.0 in multi-component interventions were PA + HE+/−SP RCT 3 6.0 (n =10; “+/−”: with or without), and PA + HE+DI+/−SP Duration of interventions, months (n = 8), followed by HE+DI+/−SP (n = 7), HE+SP (n =5), 3–12 32 64.0 PA + DI+/−SP (n = 3), and PA + SP (n =2). > 12 18 36.0 Assessment of risk of bias Types of schools Assessment of risk of bias was summarized in Fig. 2. Elementary 43 86.0 Most of the trials (n = 47, 98%) were assessed as hav- Secondary 6 12.0 ing a low risk of bias in allocation concealment. And Mixed 1 2.0 most of the trials (n = 49, 98%) were judged as hav- Income level of country ing a high risk of bias in blinding of participants and High-income 40 80 (or) personnel because it was usually not possible for interventions of this nature. Approximately half of Middle income 10 20 the studies were assessed as having an unclear risk Types of interventions of bias due to insufficient descriptions in terms of Single-component (n = 15) random sequence generation (n = 30, 60%), blinding PA 7 3.5 of outcome assessment (n = 25, 50.0%), incomplete HE 7 3.5 outcomedata(n = 27, 54%) or the possibility of se- SP 1 2.0 lective reporting (n =32, 64%). Multi-component (n = 35) Overall effect size PA + SP 2 3.9 Overall effect size was summarized in Figs. 3 and 4. The HE+SP 5 9.8 quantitative synthesis of the single-component inter- PA + HE+/−SP 10 19.6 ventions showed a significant, but small reduction of PA + DI+/−SP 3 5.9 0.14 (95% CI: 0.06, 0.21) kg/m in BMI, and a small HE+DI+/−SP 7 13.7 reduction of BMI Z-score (0.05, 95% CI: 0.01, 0.10) compared with the control group. For the multi- PA + HE+DI+/−SP 8 15.7 component interventions, the pooled results showed a RCT randomized controlled trial, PA physical activity, HE health education, DI dietary improvement, SP school policy; “+/−”: with or without significant, but mild reduction of 0.32 (0.09, 0.54) kg/m in BMI, and 0.07 (0.001, 0.14) in BMI Z-score com- pared with the control group. Although the pooled effect sizes in BMI indices of multi-component inter- ventions were slightly larger than that of single- component interventions, the differences were not Fig. 2 Risk of bias graph Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 6 of 12 Fig. 3 Pooled intervention effect (BMI) statistically significant (P =0.41 for BMI, P =0.71 for risk of bias (Additional file 1:Figures S1-S3).Results BMI Z-score). were also not significantly different across sex, Findings of overall effect size were robust to the weight status and country of the study population exclusion of heterogeneous studies or studies of high (Additional file 1:Table S4). Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 7 of 12 Fig. 4 Pooled intervention effect (BMI Z-score) Subgroup analyses pooled BMI from interventions focusing on enjoyment Subgroup analyses showed that means of BMI differed of participants during PA (n =5) was − 0.19 (95% CI: − significantly by whether or not studies including cur- 0.33, − 0.05) kg/m , while the effect size for interven- ricular PA sessions or emphasizing enjoyment in PA tions not emphasizing enjoyment of participants (n =2) among single-component interventions (Table 2). The was − 0.004 (95% CI: − 0.10, 0.09) kg/m . The effect pooled BMI from single-component interventions in- sizes did not differ significantly on other intervention cluding curricular PA (n =3) was − 0.30 (95% CI: − characteristics among single-component interventions 0.51, − 0.10) kg/m , while the effect size from single- (P > 0.05). Findings of subgroup analyses for single- component interventions not including curricular PA component interventions were consistent with results (n =4) was − 0.04 (95% CI: − 0.17, 0.09) kg/m .The from sensitivity analyses (Additional file 1:Table S6). Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 8 of 12 Table 2 Subgroup analyses by characteristics of single-component interventions Outcomes BMI BMI Z-score N Mean difference, 95% CI P for subgroup analyses N Mean difference, 95% CI P for subgroup analyses Characteristics of the PA component 1) PA’s frequency and duration ≥ 3/week and ≥ 10 min/time 5 −0.10 (− 0.22, 0.01) 0.16 –– – < 3/week or < 10 min/time 2 −0.31 (− 0.57, − 0.05) –– 2) Curricular PA Yes 3 −0.30 (− 0.51, − 0.10) 0.02 –– – No 4 −0.04 (− 0.17, 0.09) –– 3) PA emphasizing enjoyment Yes 5 −0.19 (− 0.33, − 0.05) 0.03 –– – No 2 −0.004 (− 0.10, 0.09) –– Topics of HE covering both energy intake and output Yes 3 −0.06 (− 0.26, 0.13) – 5 − 0.07 (− 0.16, − 0.03) – No 1 −0.40 (− 1.01, 0.21) 1 − 0.12 (− 0.28, 0.04) CI confidence interval, SP school policy, PA physical activity, HE health education, DI dietary improvement. “-” due to insufficient observations Concerning multi-component interventions, subgroup (Additional file 1: Table S5). No significant differences analyses demonstrated that the mean BMI or BMI Z- in effect sizes (P > 0.05) were detected between multi- score differed significantly by interventions emphasizing component interventions with and without other inter- enjoyment in PA (Table 3); however, this difference was vention characteristics, which was consistent with results disappeared when excluding one heterogeneous study from sensitivity analyses (Additional file 1: Table S5, S7). Table 3 Subgroup analyses by characteristics of multi-component interventions Outcomes BMI BMI Z-score N Mean difference, 95% CI P for subgroup analyses N Mean difference, 95% CI P for subgroup analyses Characteristics of the PA component 1) PA’s frequency and duration ≥ 3/week and ≥ 10 min/time 10 −0.48 (− 0.94, − 0.01) 0.63 –– – < 3/week or < 10 min/time 8 − 0.29 (− 0.76, 0.19) –– 2) Curricular PA Yes 14 −0.28 (− 0.60, 0.05) 0.19 7 −0.07 (− 0.19, 0.04) 0.26 No 5 −0.89 (−1.75, − 0.03) 2 − 0.06 (− 0.16, 0.03) 3) PA emphasizing enjoyment Yes 8 −0.88 (−1.42, − 0.34) 0.02 –– – No 11 −0.12 (− 0.42, 0.19) –– – Topics of HE covering both energy intake and output Yes 15 −0.28 (− 0.57, 0.01) 0.07 8 −0.17 (− 0.29, − 0.04) 0.18 No 8 0.04 (− 0.14, 0.22) 5 −0.06 (− 0.15, 0.03) Inclusion of the DI component Yes 12 −0.22 (− 0.63, 0.18) 0.45 10 −0.07 (− 0.18, 0.04) 0.76 No 15 −0.42 (− 0.75, − 0.09) 6 −0.05 (− 0.07, − 0.03) Inclusion of the SP component Yes 18 −0.36 (− 0.63, − 0.09) 0.07 11 −0.04 (− 0.08, − 0.01) 0.52 No 7 − 0.01 (− 0.26, 0.24) 2 −0.08 (− 0.20, 0.03) Abbreviations: CI confidence interval, SP school policy, PA physical activity, HE health education Significance of the finding was disappeared when excluding the heterogeneous study (see Additional file 1: Table S5) Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 9 of 12 Assessment of publication bias consistently found in multi-component interventions. It is As shown in Fig. 5, the funnel plot of the observed effect likely that multi-component interventions demonstrated showed a slightly asymmetric scatter consistent with to be effective were influenced by a combination of inter- publication bias, but P value for Egger’s regression test vention components. The impact of combining compo- was larger than 0.05. nents on intervention effectiveness should be the topic of further research. Discussion For the current body of evidence, we did not find sig- This review is one of the first to use meta-analyses and nificant associations between dietary improvement com- subgroup analyses to systematically review a number of ponents with improved intervention efficacy. This more recent studies, and analyze the potentially effect- finding was consistent with another recent systematic re- ive characteristics of school-based interventions for view of school-based childhood obesity interventions preventing obesity. [84]. The non-significant finding in relation to diet might be interpreted by poor adherence to diet intervention or Interpretation of the study findings the complex interplay of intervention components. We ac- This review found that emphasizing enjoyment in PA ses- knowledge that interaction analyses of intervention com- sions was critical for single-component (PA) interventions. ponents (“intervention × component”) within individual This finding was echoed by previous reviews suggesting studies would have provided a powerful method of under- that lack of motivation and pleasure of physical activity standing the complex interplay of intervention compo- was a barrier to physical activity for children [82, 83]. In- nents. However, of the studies screened for this review, cluding curricular PA sessions was also found to be asso- none reported such “intervention × component” analyses. ciated with improved efficacy of single-component (PA) Therefore, future obesity prevention interventions should interventions. This is, at least partly, explained by the fact address the specific interplay of intervention components, that curricular PA sessions were usually led by physical providing the possibility for further systematic reviews. education teachers, and thus intensity of exercise was su- Findings of the study should also be interpreted in the perior to those including only extracurricular activities, context. The reporting of intervention characteristics (dose, after-school sessions or short activity breaks. Further, the frequency, and content) varied so much between trials that curricular PA sessions were usually structured and com- we were obliged to dichotomise it simply as “including the pulsory for all children in a class and thus adherence specific characteristics of component: yes/no” for the pur- could be relatively guaranteed. Significant associations be- pose of analysis, being nevertheless aware that resolution of tween intervention components and efficacy were not the measure might be compromised in the process. Fig. 5 Assessment of publication bias: funnel plot Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 10 of 12 Comparison with other studies Despite these limitations, our study, based on a systema- Some previous reviews of obesity interventions have tical review of the best available evidence from RCTs, took attempted to address the question of “what” (characteris- a first step towards distinguishing characteristics of effect- tics or components of interventions) really works for the ive school-based obesity prevention interventions. The targeted population [5, 17, 18], but only general findings findings of this review enable a better understanding of were revealed. Further, research gaps remained in relation the effectiveness of complex school-based obesity preven- to the weakness of methods that were used (i.e., no tion interventions. Specifically, the findings of this review comparisons between effective and non-effective trials; suggest that school-based interventions could have signifi- equal weighting of the included trials). The present cant effects on reducing students’ BMI. The effects of review not only provided an update on a recent re- single-component (PA) interventions can be improved view [5] by including several new studies, but also when emphasizing students’ enjoyment in physical activ- identified the characteristics of effective interventions ity, or including curricular PA sessions. through meta-analyses and subgroup analyses. Thus, this review provides important and helpful evidence Conclusions of the potentially effective intervention components Overall, school-based interventions are effective in redu- with different characteristics. cing excessive weight gain of children. Findings of this review increase our understandings of potentially effective Limitations and strengths of the study characteristics of interventions. Future high-quality studies Our results should be weighted cautiously considering should focus more on the interplay of intervention com- the following limitations. First, the studies included in ponents, which could deepen our understandings of the this review were restricted to English full-text publica- complexity of obesity prevention interventions delivered tions found in three electronic databases. Second, the in school settings. considerable level of heterogeneity was detected across studies in this review, which is relatively common among complex obesity interventions. Heterogeneity Supplementary information Supplementary information accompanies this paper at https://doi.org/10. might be originated from the fidelity of the intervention 1186/s12966-019-0848-8. and the population targeted among other factors. We have conducted sensitivity analyses to address this con- Additional file 1: Table S1. Description of the included trials. Table S2. cern. Third, precisely evaluating the contents of some Description of the characteristics of the PA component for the included interventions is difficult and problematic due to incon- studies. Table S3. Description of the characteristics of the DI component for the included studies. Table S4. Differences of overall effect size by sistent reporting. Future trials should be required to sex, weight status and country of the study population. Table S5. report interventions in accordance with TIDieR (tem- Subgroup analyses by characteristics of multi-component interventions plate for intervention description and replication) [85] (excluding heterogeneous studies). Table S6. Subgroup analyses by characteristics of single-component interventions (excluding trials or other tools. Fourth, solely using BMI indices as out- assessed as high risk of bias). Table S7. Subgroup analyses by characteristics come measures in this review is relatively narrow and of multi-component interventions (excluding trials assessed as high risk of insensitive, especially when studying PA interventions, bias). Table S8. The list of excluded studies. Figure S1. Pooled intervention effect after excluding heterogeneous studies (BMI). Figure S2. Pooled as PA interventions might have an impact on BMI by intervention effect after excluding studies at high risk of bias (BMI). Figure S3. affecting intermediate outcomes (increasing PA). We Pooled intervention effect after excluding studies at high risk of bias are planning to consider using behavioral outcomes in (BMI Z-score). a future systematic review. Fifth, we only included RCTs in this review, which cannot address complex Abbreviations interplay of behaviors and real-world settings. However, BMI: Body mass index; CI: Confidence interval; DI: Dietary improvement; RCTs are the best available approach to answer “can it HE: Health education; PA: Physical activity; RCT: Randomized controlled trial; SD: Standard deviation; SP: School policy; TIDieR: Template for intervention work?”, as non-randomized trials might result in in- description and replication comparable baseline data between the two groups, and uncontrolled trials can hardly eliminate the risk of self- Acknowledgments selection bias. Sixth, due to the limited number of in- We thank all the members in our research team. cluded studies as well as the limited sub-group data available for meta-analyses, we cannot investigate Authors’ contributions whether our findings of potentially effective interven- ZL and HMX contributed equally and are considered co-first authors. Study tion components were influenced by sex, weight status design: HJW and ZL; Study selection: ZL, HMX, YZP and LZL; Data extraction: HMX, ZL; Quality assessment: HMX, ZL; Data analysis: ZL, HMX, LMW, YZP, or socio-economic status of the study population. This LZL, SZ, WHL, HJW; Drafting of the manuscript: ZL, HMX, LMW, YZP, LZL, SZ, is thus should be a potential focus for future trials, WHL, HJW; Critical revision of the manuscript for important intellectual which provides a basis for the coming meta-analyses. content: HJW, LMW. All authors read and approved the final manuscript. Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 11 of 12 Funding 17. Xu F, Ware RS, Leslie E, et al. Effectiveness of a randomized controlled This work was supported by National Key R&D Program of China lifestyle intervention to prevent obesity among Chinese primary school (2016YFC1300204), National Natural Science Fund (81703240; 81903343) and students: CLICK-obesity study. PLoS One. 2015;10:e0141421. Postdoctoral Research Foundation of China (2019 M650391). 18. Katz DL, O’Connell M, Njike VY, et al. Strategies for the prevention and control of obesity in the school setting: systematic review and meta- analysis. Int J Obes. 2008;32:1780. Availability of data and materials 19. Martin J, Chater A, Lorencatto F. Effective behaviour change techniques The datasets analyzed during the current study are available from the in the prevention and management of childhood obesity. Int J Obes. corresponding author on reasonable request. 2013;37:1287. 20. Schippers M, Adam PC, Smolenski DJ, et al. A meta-analysis of overall effects Ethics approval and consent to participate of weight loss interventions delivered via mobile phones and effect size Not applicable. differences according to delivery mode, personal contact, and intervention intensity and duration. Obes Rev. 2017;18:450–9. 21. Kremers SP, Visscher TL, Brug J, et al. Netherlands research programme Consent for publication weight gain prevention (NHF-NRG): rationale, objectives and strategies. Eur J Not applicable. Clin Nutr. 2005;59:498. 22. Zahner L, Puder JJ, Roth R, et al. A school-based physical activity program Competing interests to improve health and fitness in children aged 6–13 years (" kinder- The authors declare that they have no competing interests. Sportstudie KISS"): study design of a randomized controlled trial [ISRCTN15360785]. BMC Public Health. 2006;6:147. Author details 23. DuBose KD, Mayo MS, Gibson CA, et al. Physical activity across the Department of Maternal and Child Health, School of Public Health, Peking curriculum (PAAC): rationale and design. Contemp Clin Trials. 2008;29:83–93. University, Beijing, China. School of Public Health, Sydney Medical School, 24. Williamson DA, Champagne CM, Harsha D, et al. Louisiana (LA) health: University of Sydney, Sydney, Australia. design and methods for a childhood obesity prevention program in rural schools. Contemp Clin Trials. 2008;29:783–95. Received: 22 January 2019 Accepted: 16 September 2019 25. Giralt M, Albaladejo R, Tarro L, et al. A primary-school-based study to reduce prevalence of childhood obesity in Catalunya (Spain)-EDAL- Educació en alimentació: study protocol for a randomised controlled trial. Trials. 2011;12:54. References 26. Lawlor DA, Jago R, Noble SM, et al. The active for life year 5 (AFLY5) school 1. Ng M, Fleming T, Robinson M, et al. Global, regional, and national based cluster randomised controlled trial: study protocol for a randomized prevalence of overweight and obesity in children and adults during 1980– controlled trial. Trials. 2011;12:181. 2013: a systematic analysis for the global burden of disease study 2013. 27. Xu F, Ware RS, Tse LA, et al. A school-based comprehensive lifestyle Lancet. 2014;384:766–81. intervention among chinese kids against obesity (CLICK-obesity): rationale, 2. Ebbeling CB, Pawlak DB, Ludwig DS. Childhood obesity: public-health crisis, design and methodology of a randomized controlled trial in Nanjing city, common sense cure. Lancet. 2002;360:473–82. China. BMC Public Health. 2012;12:316. 3. Braet C, Mervielde I, Vandereycken W. Psychological aspects of childhood 28. Sutherland R, Campbell E, Lubans DR, et al. A cluster randomised trial of a obesity: a controlled study in a clinical and nonclinical sample. J Pediatr school-based intervention to prevent decline in adolescent physical activity Psychol. 1997;22:59–71. levels: study protocol for the ‘physical activity 4 Everyone’trial. BMC Public 4. Twig G, Yaniv G, Levine H, et al. Body-mass index in 2.3 million adolescents Health. 2013;13:57. and cardiovascular death in adulthood. N Engl J Med. 2016;374:2430–40. 29. Scherr RE, Linnell JD, Smith MH, et al. The shaping healthy choices program: 5. Bleich SN, Vercammen KA, Zatz LY, et al. Interventions to prevent global design and implementation methodologies for a multicomponent, school- childhood overweight and obesity: a systematic review. Lancet Diabetes based nutrition education intervention. J Nutr Educ Behav. 2014;46:e13–21. Endocrinol. 2018;6:332–46. 30. Lloyd J, Wyatt K. The healthy lifestyles Programme (HeLP)—an overview of 6. Kremers SP, de Bruijn GJ, Visscher TL, et al. Environmental influences on and recommendations arising from the conceptualisation and development energy balance-related behaviors: a dual-process view. Int J Behav Nutr of an innovative approach to promoting healthy lifestyles for children and Phys Act. 2006;3:9. their families. Int J Environ Res Public Health. 2015;12:1003–19. 7. Feng L, Wei DM, Lin ST, et al. Systematic review and meta-analysis of school- 31. Higgins JP, Altman DG, Gøtzsche PC, et al. The Cochrane Collaboration’s based obesity interventions in mainland China. PLoS One. 2017;12:e0184704. tool for assessing risk of bias in randomised trials. BMJ. 2011;343:d5928. 8. Waters E, de Silva-Sanigorski A, Burford BJ, et al. Interventions for preventing 32. Sobol-Goldberg S, Rabinowitz J, Gross R. School-based obesity prevention obesity in children. Cochrane Database Syst Rev. 2011;12:CD001871. programs: a meta-analysis of randomized controlled trials. Obesity. 2013;21:2422–8. 9. Lloyd J, Creanor S, Logan S, et al. Effectiveness of the healthy lifestyles 33. Farmer VL, Williams SM, Mann JI, et al. The effect of increasing risk and Programme (HeLP) to prevent obesity in UK primary-school children: a cluster challenge in the school playground on physical activity and weight in children: randomised controlled trial. Lancet Child Adolesc Health. 2018;2:35–45. a cluster randomised controlled trial (PLAY). Int J Obes. 2017;41:793. 10. Meyer U, Schindler C, Zahner L, et al. Long-term effect of a school-based 34. Hollis JL, Sutherland R, Campbell L, et al. Effects of a ‘school-based’physical physical activity program (KISS) on fitness and adiposity in children: a activity intervention on adiposity in adolescents from economically cluster-randomized controlled trial. PLoS One. 2014;9:e87929. disadvantaged communities: secondary outcomes of the ‘physical activity 4 11. Marcus C, Nyberg G, Nordenfelt A, et al. A 4-year, cluster-randomized, Everyone’RCT. Int J Obes. 2016;40:1486. controlled childhood obesity prevention study: STOPP. Int J Obes. 2009;33:408. 35. Gallotta MC, Iazzoni S, Emerenziani GP, et al. Effects of combined physical 12. Aguilar FS, Martínez-Vizcaíno V, López MS, et al. Impact of an after-school education and nutritional programs on schoolchildren’s healthy habits. physical activity program on obesity in children. J Pediatr. 2010;157:36–42. PeerJ. 2016;4:e1880. 13. Sichieri R, Trotte AP, de Souza RA, et al. School randomised trial on 36. Drummy C, Murtagh EM, McKee DP, et al. The effect of a classroom activity prevention of excessive weight gain by discouraging students from break on physical activity levels and adiposity in primary school children. J drinking sodas. Public Health Nutr. 2009;12:197–202. Paediatr Child Health. 2016;52:745–9. 14. Scherr RE, Linnell JD, Dharmar M, et al. A multicomponent, school-based intervention, the shaping healthy choices program, improves nutrition- 37. Creanor S, Lloyd J, Hillsdon M, et al. Detailed statistical analysis plan for a related outcomes. J Nutr Educ Behav. 2017;49:368–79. cluster randomised controlled trial of the healthy lifestyles Programme 15. Singh AS, Paw MJ, Brug J, et al. Short-term effects of school-based (HeLP), a novel school-based intervention to prevent obesity in school weight gain prevention among adolescents. Arch Pediatr Adolesc Med. children. Trials. 2016;17:599. 2007;161:565–71. 38. Kain J, Concha F, Moreno L, et al. School-based obesity prevention 16. Robinson TN. Reducing children's television viewing to prevent obesity: a intervention in Chilean children: effective in controlling, but not reducing randomized controlled trial. JAMA. 1999;282:1561–7. obesity. J Obes. 2014;2014:618293. Liu et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:95 Page 12 of 12 39. Tarro L, Llauradó E, Albaladejo R, et al. A primary-school-based study to 64. Williamson DA, Champagne CM, Harsha DW, et al. Effect of an reduce the prevalence of childhood obesity–the EdAl (Educació en environmental school-based obesity prevention program on changes in Alimentació) study: a randomized controlled trial. Trials. 2014;15:58. body fat and body weight: a randomized trial. Obesity. 2012;20:1653–61. 40. Safdie M, Jennings-Aburto N, Lévesque L, et al. Impact of a school-based 65. Lloyd JJ, Wyatt KM, Creanor S. Behavioural and weight status outcomes intervention program on obesity risk factors in Mexican children. Salud from an exploratory trial of the healthy lifestyles Programme (HeLP): a novel Publica Mex. 2013;55:374–87. school-based obesity prevention programme. BMJ Open. 2012;2:e000390. 41. Habib-Mourad C, Ghandour LA, Moore HJ, et al. Promoting healthy eating 66. Llargues E, Franco R, Recasens A, et al. Assessment of a school-based and physical activity among school children: findings from Health-E-PALS, intervention in eating habits and physical activity in school children: the the first pilot intervention from Lebanon. BMC Public Health. 2014;14:940. AVall study. J Epidemiol Community Health. 2011;65:896. 67. Llargués E, Recasens A, Franco R, et al. Medium-term evaluation of an 42. Dewar DL, Morgan PJ, Plotnikoff RC, et al. The nutrition and enjoyable educational intervention on dietary and physical exercise habits in activity for teen girls study: a cluster randomized controlled trial. Am J Prev schoolchildren: the Avall 2 study. Endocrinol Nutr. 2012;59:288–95. Med. 2013;45:313–7. 68. Magnusson KT, Hrafnkelsson H, Sigurgeirsson I, et al. Limited effects of a 2- 43. Jansen W, Borsboom G, Meima A, et al. Effectiveness of a primary school- year school-based physical activity intervention on body composition and based intervention to reduce overweight. Int J Pediatr Obes. 2011;6:e70–7. cardiorespiratory fitness in 7-year-old children. Health Educ Res. 2012;27:484. 44. Siegrist M, Hanssen H, Lammel C, et al. Effects of the school-based 69. Fairclough SJ, Hackett AF, Davies IG, et al. Promoting healthy weight in intervention programme JuvenTUM 3 on physical fitness, physical activity, primary school children through physical activity and nutrition education: a and the prevalence of overweight. Eur J Prev Cardiolog. 2012;19:S2. pragmatic evaluation of the CHANGE! Randomised intervention study. BMC 45. Thivel D, Isacco L, Lazaar N, et al. Effect of a 6-month school-based physical Public Health. 2013;13:626. activity program on body composition and physical fitness in lean and 70. Rosário R, Oliveira B, Araújo A, et al. The impact of an intervention taught obese schoolchildren. Eur J Pediatr. 2011;170:1435–43. by trained teachers on childhood overweight. Int J Environ Res Public 46. Group HS, Foster GD, Linder B, et al. A school-based intervention for Health. 2012;9:1355–67. diabetes risk reduction. N Engl J Med. 2010;363:1769. 71. Rush E, Reed P, Mclennan S, et al. A school-based obesity control 47. Angelopoulos PD, Milionis HJ, Grammatikaki E, et al. Changes in BMI and programme: project energize. Two-year outcomes. Br J Nutr. 2012;107:581. blood pressure after a school based intervention: the CHILDREN study. Eur J 72. Grydeland M, Bjelland M, Anderssen SA, et al. Effects of a 20-month cluster Pub Health. 2009;19:319–25. randomised controlled school-based intervention trial on BMI of school- 48. Vizcaíno VM, Aguilar FS, Gutiérrez RF, et al. Assessment of an after-school aged boys and girls: the HEIA study. Br J Sports Med. 2014;48:768–73. physical activity program to prevent obesity among 9-to 10-year-old 73. Grydeland M, Bergh IH, Bjelland M, et al. Intervention effects on physical children: a cluster randomized trial. Int J Obes. 2008;32:12. activity: the HEIA study - a cluster randomized controlled trial. Int J Behav 49. Donnelly JE, Greene JL, Gibson CA, et al. Physical activity across the Nutr Phys Act. 2013;10:17. curriculum (PAAC): a randomized controlled trial to promote physical 74. Sacchetti R, Ceciliani A, Garulli A, et al. Effects of a 2-year school-based activity and diminish overweight and obesity in elementary school children. intervention of enhanced physical education in the primary school. J Sch Prev Med. 2009;49:336–41. Health. 2013;83:639–46. 50. Barbeau P, Johnson MH, Howe CA, et al. Ten months of exercise improves general 75. Kriemler S, Zahner L, Schindler C, et al. Effect of school based physical and visceral adiposity, bone, and fitness in black girls. Obesity. 2007;15:2077–85. activity programme (KISS) on fitness and adiposity in primary 51. Caballero B, Clay T, Davis SM, et al. Pathways: a school-based, randomized schoolchildren: cluster randomised controlled trial. BMJ. 2010;340:c785. controlled trial for the prevention of obesity in American Indian 76. Llauradó E, Tarro L, Moriña D, et al. EdAl-2 (Educacio en Alimentacio) schoolchildren. Am J Clin Nutr. 2003;78:1030–8. programme: reproducibility of a cluster randomised, interventional, primary- 52. Foster GD, Sherman S, Borradaile KE, et al. A policy-based school intervention school-based study to induce healthier lifestyle activities in children. BMJ to prevent overweight and obesity. Pediatrics. 2008;121:e794–802. Open. 2014;4:e005496. 53. Jiang J, Xia X, Greiner T, et al. The effects of a 3-year obesity intervention in 77. Kipping RR, Howe LD, Jago R, et al. Effect of intervention aimed at schoolchildren in Beijing. Child Care Health Dev. 2007;33:641–6. increasing physical activity, reducing sedentary behaviour, and increasing 54. Li YP, Hu XQ, Schouten EG, et al. Report on childhood obesity in China (8): fruit and vegetable consumption in children: active for life year 5 (AFLY5) effects and sustainability of physical activity intervention on body school based cluster randomised controlled trial. BMJ. 2014;348:g3256. composition of Chinese youth. Biomed Environ Sci. 2010;23:180–7. 78. Adab P, Pallan MJ, Lancashire ER, et al. Effectiveness of a childhood obesity 55. Singhal N, Misra A, Shah P, et al. Effects of controlled school-based multi- prevention programme delivered through schools, targeting 6 and 7 year component model of nutrition and lifestyle interventions on behavior olds: cluster randomised controlled trial (WAVES study). BMJ. 2018;360:k211. modification, anthropometry and metabolic risk profile of urban Asian 79. Pablos A, Nebot V, Vañó-Vicent V, et al. Effectiveness of a school-based Indian adolescents in North India. Eur J Clin Nutr. 2010;64:364. program focusing on diet and health habits taught through physical 56. Peralta LR, Jones RA, Okely AD. Promoting healthy lifestyles among exercise. Appl Physiol Nutr Metab. 2018;43:331–7. adolescent boys: the fitness improvement and lifestyle awareness program 80. Wang Z, Xu F, Ye Q, et al. Childhood obesity prevention through a RCT. Prev Med. 2009;48:537–42. community-based cluster randomized controlled physical activity intervention 57. Vandongen R, Jenner DA, Thompson C, et al. A controlled evaluation of a nd among schools in China: the health legacy project of the 2 world summer fitness and nutrition intervention program on cardiovascular health in 10- youth olympic games (YOG-obesity study). Int J Obes. 2018;42:625–33. year-old to 12-year-old children. Prev Med. 1995;24:9–22. 81. Liu Z, Li Q, Maddison R, et al. A school-based comprehensive intervention 58. El Ansari W, El Ashker S, Moseley L. Associations between physical activity for childhood obesity in China: a cluster randomized controlled trial. Child and health parameters in adolescent pupils in Egypt. Int J Environ Res Obes. 2019;15(2):105–15. Public Health. 2010;7:1649–69. 82. Kebbe M, Damanhoury S, Browne N, et al. Barriers to and enablers of 59. Manios Y, Moschandreas J, Hatzis C, et al. Health and nutrition education in healthy lifestyle behaviours in adolescents with obesity: a scoping review primary schools of Crete: changes in chronic disease risk factors following a and stakeholder consultation. Obes Rev. 2017;18:1439–53. 6-year intervention programme. Br J Nutr. 2002;88:315–24. 83. Zabinski MF, Saelens BE, Stein RI, et al. Overweight children's barriers to and 60. Kafatos A, Manios Y, Moschandreas J. Health and nutrition education in support for physical activity. Obes Res. 2003;11:238–46. primary schools of Crete: follow-up changes in body mass index and 84. Brown T, Moore TH, Hooper L, et al. Interventions for preventing obesity in overweight status. Eur J Clin Nutr. 2005;59:1090. children. Cochrane Database Syst Rev. 2019;Cd001871. 61. Kipping RR, Payne C, Lawlor DA. Randomised controlled trial adapting US 85. Hoffmann TC, Glaszlou PP, Boutron I, et al. Better reporting of interventions: school obesity prevention to England. Arch Dis Child. 2008;93:469–73. template for intervention description and replication (TIDieR) checklist and 62. James J, Thomas P, Cavan D, et al. Preventing childhood obesity by guide. BMJ. 2014;348:g1687. reducing consumption of carbonated drinks: cluster randomised controlled trial. BMJ. 2004;328:1237. 63. James J, Thomas P, Kerr D. Preventing childhood obesity: two year follow- Publisher’sNote up results from the Christchurch obesity prevention programme in schools Springer Nature remains neutral with regard to jurisdictional claims in (CHOPPS). BMJ. 2007;335:762. published maps and institutional affiliations.

Journal

International Journal of Behavioral Nutrition and Physical ActivitySpringer Journals

Published: Oct 29, 2019

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create folders to
organize your research

Export folders, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off