Physical activity, diet and other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight

Abstract Background The global prevalence of childhood and adolescent obesity is high. Lifestyle changes towards a healthy diet, increased physical activity and reduced sedentary activities are recommended to prevent and treat obesity. Evidence suggests that changing these health behaviours can benefit cognitive function and school achievement in children and adolescents in general. There are various theoretical mechanisms that suggest that children and adolescents with excessive body fat may benefit particularly from these interventions. Objectives To assess whether lifestyle interventions (in the areas of diet, physical activity, sedentary behaviour and behavioural therapy) improve school achievement, cognitive function (e.g. executive functions) and/or future success in children and adolescents with obesity or overweight, compared with standard care, waiting‐list control, no treatment, or an attention placebo control group. Search methods In February 2017, we searched CENTRAL, MEDLINE and 15 other databases. We also searched two trials registries, reference lists, and handsearched one journal from inception. We also contacted researchers in the field to obtain unpublished data. Selection criteria We included randomised and quasi‐randomised controlled trials (RCTs) of behavioural interventions for weight management in children and adolescents with obesity or overweight. We excluded studies in children and adolescents with medical conditions known to affect weight status, school achievement and cognitive function. We also excluded self‐ and parent‐reported outcomes. Data collection and analysis Four review authors independently selected studies for inclusion. Two review authors extracted data, assessed quality and risks of bias, and evaluated the quality of the evidence using the GRADE approach. We contacted study authors to obtain additional information. We used standard methodological procedures expected by Cochrane. Where the same outcome was assessed across different intervention types, we reported standardised effect sizes for findings from single‐study and multiple‐study analyses to allow comparison of intervention effects across intervention types. To ease interpretation of the effect size, we also reported the mean difference of effect sizes for single‐study outcomes. Main results We included 18 studies (59 records) of 2384 children and adolescents with obesity or overweight. Eight studies delivered physical activity interventions, seven studies combined physical activity programmes with healthy lifestyle education, and three studies delivered dietary interventions. We included five RCTs and 13 cluster‐RCTs. The studies took place in 10 different countries. Two were carried out in children attending preschool, 11 were conducted in primary/elementary school‐aged children, four studies were aimed at adolescents attending secondary/high school and one study included primary/elementary and secondary/high school‐aged children. The number of studies included for each outcome was low, with up to only three studies per outcome. The quality of evidence ranged from high to very low and 17 studies had a high risk of bias for at least one item. None of the studies reported data on additional educational support needs and adverse events. Compared to standard practice, analyses of physical activity‐only interventions suggested high‐quality evidence for improved mean cognitive executive function scores. The mean difference (MD) was 5.00 scale points higher in an after‐school exercise group compared to standard practice (95% confidence interval (CI) 0.68 to 9.32; scale mean 100, standard deviation 15; 116 children, 1 study). There was no statistically significant beneficial effect in favour of the intervention for mathematics, reading, or inhibition control. The standardised mean difference (SMD) for mathematics was 0.49 (95% CI ‐0.04 to 1.01; 2 studies, 255 children, moderate‐quality evidence) and for reading was 0.10 (95% CI ‐0.30 to 0.49; 2 studies, 308 children, moderate‐quality evidence). The MD for inhibition control was ‐1.55 scale points (95% CI ‐5.85 to 2.75; scale range 0 to 100; SMD ‐0.15, 95% CI ‐0.58 to 0.28; 1 study, 84 children, very low‐quality evidence). No data were available for average achievement across subjects taught at school. There was no evidence of a beneficial effect of physical activity interventions combined with healthy lifestyle education on average achievement across subjects taught at school, mathematics achievement, reading achievement or inhibition control. The MD for average achievement across subjects taught at school was 6.37 points lower in the intervention group compared to standard practice (95% CI ‐36.83 to 24.09; scale mean 500, scale SD 70; SMD ‐0.18, 95% CI ‐0.93 to 0.58; 1 study, 31 children, low‐quality evidence). The effect estimate for mathematics achievement was SMD 0.02 (95% CI ‐0.19 to 0.22; 3 studies, 384 children, very low‐quality evidence), for reading achievement SMD 0.00 (95% CI ‐0.24 to 0.24; 2 studies, 284 children, low‐quality evidence), and for inhibition control SMD ‐0.67 (95% CI ‐1.50 to 0.16; 2 studies, 110 children, very low‐quality evidence). No data were available for the effect of combined physical activity and healthy lifestyle education on cognitive executive functions. There was a moderate difference in the average achievement across subjects taught at school favouring interventions targeting the improvement of the school food environment compared to standard practice in adolescents with obesity (SMD 0.46, 95% CI 0.25 to 0.66; 2 studies, 382 adolescents, low‐quality evidence), but not with overweight. Replacing packed school lunch with a nutrient‐rich diet in addition to nutrition education did not improve mathematics (MD ‐2.18, 95% CI ‐5.83 to 1.47; scale range 0 to 69; SMD ‐0.26, 95% CI ‐0.72 to 0.20; 1 study, 76 children, low‐quality evidence) and reading achievement (MD 1.17, 95% CI ‐4.40 to 6.73; scale range 0 to 108; SMD 0.13, 95% CI ‐0.35 to 0.61; 1 study, 67 children, low‐quality evidence). Authors' conclusions Despite the large number of childhood and adolescent obesity treatment trials, we were only able to partially assess the impact of obesity treatment interventions on school achievement and cognitive abilities. School and community‐based physical activity interventions as part of an obesity prevention or treatment programme can benefit executive functions of children with obesity or overweight specifically. Similarly, school‐based dietary interventions may benefit general school achievement in children with obesity. These findings might assist health and education practitioners to make decisions related to promoting physical activity and healthy eating in schools. Future obesity treatment and prevention studies in clinical, school and community settings should consider assessing academic and cognitive as well as physical outcomes.


A B S T R A C T Background
The global prevalence of childhood and adolescent obesity is high. Lifestyle changes towards a healthy diet, increased physical activity and reduced sedentary activities are recommended to prevent and treat obesity. Evidence suggests that changing these health behaviours can benefit cognitive function and school achievement in children and adolescents in general. There are various theoretical mechanisms that suggest that children and adolescents with excessive body fat may benefit particularly from these interventions. Assessed with: standardised national tests, BADyG-I (numerical quantitative concepts) Follow-up: range 13 weeks to 1 year immediately post-intervention -Compared to the control group, the mean mathematics achievement score in the intervention group was0.49 standard deviations higher (0.04 lower to 1.01 higher) -255 (2 RCTs)

Moderate 1
A standard deviation of 0.49 represents a moderate difference between groups

School achievement: Reading
Assessed with: WJ-II test of achievement, standardised national tests -Compared to the control group, the mean reading achievement score in the intervention group was 0.10 standard deviations higher (0.30 lower to 0.49 higher) The mean score for average achievement across subjects taught at school in the control group was 19.50 grade points The mean score for average achievement across subjects taught at school in the intervention group was 6.37 grade points lower (36.83 lower to 24.09 higher) -31 (1 RCT) ⊕⊕⊝⊝ Low 1 -

School achievement: Mathematics
Assessed with: CAT-3, standardised national tests, M-CAT Follow-up: range 4 months to 12 months immediately post-intervention -Compared to the control group, the mean mathematics achievement score in the intervention group was 0.02 standard deviations higher (0.19 lower to 0.22 higher) -384 (3 RCTs)

Very low 2
A standard deviation of 0.02 represents a small difference between groups School achievement:

Reading
Assessed with: CAT-3, R-CBM -Compared to the control group, the mean reading achievement score in the intervention group was 0 stan- The mean change in reading achievement score ranged across control groups from 7.40 to 9.20 scale points The mean change in reading achievement score in the intervention group was 1.17 scale points higher (4.40 lower to 6.73 higher) , where a standard deviation of 0.2 represents a small difference between groups, 0.5 represents a moderate difference, and 0.8 represents a large difference.
** Different assessment tools were used to assess school and cognitive outcomes. We therefore calculated standardised mean differences to assess the effect size between intervention and control groups. SMD: Standardised mean difference; MD: mean difference; CI: Confidence interval GRADE Working Group grades of evidence High quality: We are very confident that the true effect lies close to that of the estimate of the effect Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect Cochrane Database of Systematic Reviews

Description of the condition
Overweight and obesity are conditions of excessive body fat accumulation. In clinical practice, child and adolescent overweight and obesity are commonly identified by age-and gender-specific body mass index (BMI) percentiles, BMI standard deviation scores, and waist circumference (WC) percentiles relative to a reference population (Reilly 2010; Rolland-Cachera 2011).
The primary criteria used to define overweight and obesity include: 1. overweight: BMI or WC ≥ 85th percentile to 95th percentile, BMI > one standard deviation above the average; 2. obesity: BMI or WC > 95th percentile, BMI > two standard deviations above the average.
Also, BMI cut-o s from the International Obesity Task Force (IOTF) are o en used as a definition of overweight and obesity. These agespecific BMI cut-o s were constructed to match the definition for overweight and obesity in adults (BMI ≥ 25 kg/m 2 and BMI ≥ 30 kg/ m 2 , respectively) (Cole 2000). Recently, the IOTF BMI cut-o s were reformulated to allow BMI to be expressed as standard deviation or percentile (Cole 2012).
A recent analysis of population data of children aged five to 19 years estimated that in 2016 obesity was identified in 50 million girls and 74 million boys worldwide (NCD Risk Factor Collaboration 2017). In the USA in 2014, the prevalence of child and adolescent obesity (BMI > 95 th centile) was 9.4% (two to five years), 17.4% (six to 11 years), and 20.6% (12 to 19 years) (Ogden 2016). In Europe, obesity prevalence was on average 4.0% in adolescents, with vast di erences between countries (Inchley 2017). For example, in Scotland the prevalence was 15% in adolescents aged 12 to 15 years (SHeS 2016). Childhood obesity prevalence is increasing in middleand low-income countries (NCD Risk Factor Collaboration 2017), for example, up to 40% of children in Mexico were living with obesity or overweight, 32% in Lebanon and 28% in Argentina (Gupta 2012).
Health problems are common in children and adolescents with obesity. These include cardiovascular conditions (e.g. hyperlipidaemia, hypertension), endocrinologic conditions (e.g. Type 2 diabetes, metabolic syndrome), gastrointestinal conditions (non-alcoholic fatty liver disease), respiratory conditions (e.g. obstructive sleep apnoea), musculoskeletal disorders, (e.g. slipped capital femoral epiphysis) and psychosocial disorders (e.g. . Cognitive skills such as the ability to suspend prepotent or default responses (inhibition), to switch between rules and responses (cognitive flexibility), to keep and retrieve information while working on a new task (working memory), and to concentrate (attention) are understood to predict school achievement in children and adolescents (Jacob 2015). Collectively, these cognitive abilities are known as executive functions. Evidence from prospective cohort studies suggests that obesity-related deficits in school achievement are more prevalent in adolescent girls than in boys and younger children (Martin 2017).
The academic consequences of adolescent obesity are shown to persist beyond schooling negatively influencing socioeconomic success. A Finnish longitudinal study (N = 9754, follow-up 17 years) suggests that adolescent obesity predicts unemployment in later life, with educational achievement as a mediating factor (Laitinen 2002). A British birth cohort study (N = 12,537) indicates that adolescent obesity (at age 16 years) is associated with fewer years of schooling and predicts lower income in young women (at age 23 years), including those who are no longer obese (Sargent 1994). These findings were further confirmed by Han 2011, using the National Longitudinal Survey of Youth 1979(N = 1974, and by Sabia 2012, using the National Longitudinal Study of Adolescent Health (N = 12,445, follow-up 13 years) in the USA. Findings from the National Longitudinal Survey of Youth 1997 in the USA (N = 8427, followup eight years) suggest that obese adolescents had a 39% lower chance of obtaining a college degree than peers of normal weight (Fowler-Brown 2010). All of these studies accounted for a variety of confounding variables, including measures of socioeconomic status (e.g. parental education, household income).

Description of the intervention
Clinical guidelines for prevention and treatment of childhood obesity from countries such as the UK (NICE 2013;SIGN 2010), Australia (NHMRC 2003), Canada (Lau 2007) and Malaysia (Ismail 2004) recommend a multicomponent approach that combines: 1. reduced energy intake; 2. increased physical activity (≥ 60 minutes a day, moderate-tovigorous intensity); 3. decreased sedentary behaviour (e.g. screen time less than two hours a day); 4. cognitive-behavioural techniques (e.g. goal setting, selfmonitoring, self-regulation).
The recently updated series of Cochrane Reviews on the treatment of childhood and adolescent obesity concluded that interventions aiming to alter eating habits, physical activity, and sedentary behaviour patterns in a family-based setting were e ective in achieving clinically meaningful weight reduction in children and adolescents (Al-Khudairy 2017; Colquitt 2016; Mead 2017).

How the intervention might work
Obesity prevention and treatment interventions could benefit cognition, school achievement and future success of children and adolescents with obesity or overweight di erently compared to children and adolescents with a healthy weight. The mechanisms relate to brain development, health and psychosocial consequences, cognitive-behavioural regulation and lifestyle concerns associated with obesity ( Figure 1).

Figure 1. Potential causal links between obesity and impaired cognitive function, school achievement and future
success. Reverse causation may also occur when cognitive function, school achievement and future success can impact the 'mediating factors', and both in turn may cause worsening of obesity.

Brain development
Emerging evidence has linked obesity in children and adolescents to lower brain grey and white matter volume in brain regions associated with cognitive control and learning when compared to children and adolescents with healthy weight ( This suggests a direct association between obesity and reduced cognitive and academic abilities, and is consistent with findings from animal models where manipulation of fat mass has been shown to a ect cognition, probably as a result of inflammatory mechanisms.

Health and psychosocial consequences
Research has also identified obesity-related health consequences and psychosocial concerns to be associated with lower school achievement and cognitive function. These potential indirect factors include poor sleep due to obesity-related disordered breathing (Galland 2015; Tan 2014); hypertension (Lande 2015); Type 2 diabetes (Rofey 2015); metabolic syndrome (Yau 2012); decreased school attendance due to adverse physical and mental health (Pan 2013); and social isolation and bullying (Gunnarsdottir 2012a;Krukowski 2009). Reducing the risk of these health and psychosocial concerns, through reduction of obesity or increasing physical activity levels, or both, and improving diet and other obesity-related behaviours, could have beneficial e ects on cognitive function, school achievement and future success in children and adolescents with obesity.

Cognitive-behavioural regulation
The association between lifestyle interventions for weight management and cognition and school achievement might be bidirectional. Research indicates that children with obesity show higher impulsivity and inattention and lower reward sensitivity, self-regulation and cognitive flexibility compared with their healthy-weight peers. These neurocognitive correlates were associated with uncontrolled food intake and physical activity behaviour, and thus are assumed to predict weight gain (Francis 2009;Hall 2014;Kulendran 2014;Levitan 2015;Nederkoorn 2006;Smith 2011) or reduction of weight status a er an obesity treatment intervention (Naar-King 2016;Nederkoorn 2007). Lifestyle interventions for weight management might positively impact the neurocognitive factors required for control of food intake. A randomised controlled trial conducted in 44 children (eight to 14 years of age) with obesity or overweight suggested that specific training of self-regulatory abilities improved weightloss maintenance a er an inpatient weight-loss programme in the intervention group compared with the control group (Verbeken 2013). Findings from another randomised controlled overweight treatment programme involving 62 children (mean age 10.3 ± 1.1 years) showed improved problem-solving skills a er an intervention duration of six months (Epstein 2000). Inhibition control skills were improved in 42 obese adolescents from 12 to 17 years of age a er 12 weeks of cognitive-behavioural therapy (Delgado-Rico 2012b).

Lifestyle interventions
Growing evidence has shown that the influence of lifestyle interventions, particularly physical activity and dietary intervention, lie beyond the alteration of energy balance. Many aspects of physical activity, diet and other behaviours have been demonstrated to benefit cognition and school achievement in children and adolescents, regardless of their body weight status, as summarised below.

Physical activity
Recently, Faught 2017 reported that meeting the Canadian recommendations for diet, physical activity, sedentary behaviour and sleep at age 11 years was associated with favourable school achievement at age 12 (N = 4253). Low levels of physical fitness (Chaddock 2011; Davis 2011a; Raine 2013) and moderate-tovigorous intensity physical activity have also been linked to impaired cognitive functions in children (Haapala 2017). In addition to the observational evidence, a substantial body of literature suggests a causal relationship between increased levels of physical activity and cognitive function or school achievement or both. For example, a meta-analysis of 44 experimental and cross-sectional studies (in participants aged four to 18 years) indicates that increased physical activity caused significant overall improvement in cognitive function and school performance (Hedge's g = 0.32; standard deviation (SD) 0.27) (Sibley 2003). A recent meta-analysis of 21 experimental and quasi-experimental studies in children aged four to 16 years (N = 4044) also reported a moderate positive e ect of physical activity interventions on cognitive outcomes (Hedge's g = 0.46, 95% confidence interval 0.28 to 0.64) (Vazou 2016).
Physical activity may a ect cognitive function and school achievement through physiological mechanisms (elevated blood circulation, increased levels of neurotrophins and neurotransmitters) (Dishman 2006), learning and motor developmental mechanisms (Pesce 2016a).

Dietary modification
Composition of the diet may impact cognition and school achievement by altering neurotrophic and neuroendocrine factors involved in learning and memory. As shown in animal research, these factors are decreased by high-energy diets containing saturated fat and simple sugars, and are increased by diets that are rich in omega-3 polyunsaturated fatty acids and micronutrients (Gomez-Pinilla 2008;Kanoski 2011). These findings were also observed in children. Cross-sectional data of school-aged children linked dietary intake of omega-3 fatty acids to increased memory performance (Baym 2014; Boucher 2011), while consumption of food rich in saturated fatty acids and refined sugar was associated with decreased memory performance (Baym 2014). Longitudinal observational data suggest that diets high in fat and sugar in preschool children (N = 3966; aged three to four years) are associated with decreased intelligence and school performance at primary/elementary school age (Feinstein 2008;Northstone 2011). A controlled healthy school meal intervention over three years in more than 80,000 children led to improved mathematics, English and science achievement (Belot 2011). Promotion of healthier school food at lunchtime and changes in the school dining environment over 12 weeks improved classroom on-task behaviour in preschool children compared to controls (Golley 2010;Storey 2011). An improvement in dietary quality could therefore have beneficial e ects on cognition and school achievement even without improved weight status.

Sedentary behaviour
A sedentary lifestyle in children, particularly television-viewing for two or more hours a day, is associated with the development of obesity or overweight (review of 71 studies; Rey-Lopez 2008) and may replace opportunities to engage in activities that promote scholastic and cognitive development. To our knowledge, there is no published literature on the e ect of reduced sedentary behaviour and improved cognitive and academic outcomes of children and adolescents. However, epidemiological evidence suggests that high levels of sedentary behaviour are associated with reduced school achievement or cognitive abilities. For example, longitudinal data indicate that children younger than three years of age with low television exposure (less than three hours a day) performed better than those with high television exposure (three or more hours a day) in reading (N = 1031) and mathematics (N = 1797) (Peabody Individual Achievement Test) when at preschool age (Zimmerman 2005). Similarly, parentreported television viewing in preschool children was inversely related to mathematics achievement at age 10 years (N = 1314) (Pagani 2010) and reading achievement at age 10 to 12 years (N = 308) (Ennemoser 2007). Low TV exposure was also linked to improved school achievement in 8061 adolescents aged 16 years (Kantomaa 2016). Longer-term educational outcomes may also be a ected. Hancox 2005 found that young people (N = 980; follow-up 21 years) with the highest television viewing time during childhood and adolescence tended to have no formal educational qualifications, and those with a university degree watched the least television during childhood and adolescence. Television viewing for three or more hours a day at age 14 years (N = 678) was associated with a two-fold risk of failing to obtain a postsecondary/high school education at 33 years of age compared with those watching television for less than one hour a day, mediated by attention di iculties, frequent failure to complete homework and negative attitudes about school at 16 years of age (Johnson 2007). Studies relating accelerometer-measured sedentary behaviour to cognitive function or school achievement or both indicated that high levels of sedentary behaviour at age seven years were associated with reduced verbal reasoning skills at age 11 (Aggio 2016), and that low levels of sedentary behaviour were associated with increased school achievement at age 10 to 11 years (Aadland 2017).
Reducing sedentary behaviour (TV and screen time, sitting time) might therefore improve cognitive function and school achievement in children and adolescents with obesity or overweight.

Multicomponent interventions
In this review, the term 'multicomponent interventions' refers to interventions that target at least two obesity-related behaviours. Multicomponent lifestyle interventions may benefit cognitive function and school achievement in the general population, i.e. a study population that includes both children and adolescents of normal weight and those with obesity or overweight. For example, a er the implementation of an uncontrolled intervention involving healthy nutrition, physical activity and using behaviour change techniques in a US primary/elementary school, an upward trend in reading performance scores was noted; these scores exceeded the national average by 10% a er eight years (Nansel 2009). Another uncontrolled experimental study, which implemented a healthy diet and physical activity programme in a primary/elementary school, reported an increase in the numbers of children passing standardised tests in writing, reading and mathematics by 25%, 27% and 31%, respectively (Sibley 2008). A similar but controlled school-based intervention promoting healthy eating and physical activity behaviour in children aged 11 to 14 years led to significant improvement in mathematics, listening and speaking scores a er only five weeks compared with the control condition (standard classroom education) (Shilts 2009).

Why it is important to do this review
The current global trend in childhood obesity (NCD Risk Factor Collaboration 2017;WHO 2016) suggests that the prevalence of cognitive and educational problems among children is also likely to increase. Given the evidence of a link between low school achievement and economic disadvantage, this might have financial repercussions for future employability and income.
The beneficial e ects of changes in diet, physical activity, sedentary behaviour and thinking patterns for prevention and treatment of childhood obesity are well established (Al-Khudairy 2017;Colquitt 2016;Mead 2017;Waters 2011) and are reflected in clinical guidelines for the management of obesity (Ismail 2004;Lau 2007;NHMRC 2003;NICE 2013;SIGN 2010).
Animal models and human studies suggest that both obesity and obesity-related lifestyle behaviours have the potential to impair cognitive function, learning, and school achievement (see How the intervention might work; Figure 1). What is less clear is the extent to which interventions which modify lifestyle or body fatness or both can improve cognitive function and learning/ school achievement. We would expect that obesity prevention or treatment interventions benefit children with obesity di erently from children with a healthy weight by mitigating cognitive deficits which are associated with having an excessive level of body fatness.
The first version of this review was published in March 2014 and included analysis of six trials published until May 2013 (Martin 2014). An update of the review was required to reflect the growing interest in this field.

O B J E C T I V E S
To assess whether lifestyle interventions (in the areas of diet, physical activity, sedentary behaviour and behavioural therapy) improve school achievement, cognitive function (e.g. executive functions) and/or future success in children and adolescents with obesity or overweight, compared with standard care, waiting-list control, no treatment, or an attention placebo control group.

Types of studies
Randomised controlled trials (RCTs), including cluster-randomised trials, and quasi-randomised trials with or without cross-over design, were eligible for inclusion. We included cross-over trials when data from the first period were obtainable.

Types of participants
Children and adolescents with obesity or overweight aged three to 18 years attending preschool or school, and whose body weight status was determined using age-and gender-specific BMI percentiles, BMI z-scores, BMI standard deviation scores (SDSs), BMI cut-o points or waist circumference. Classification of weight status needed to be based on a relevant national or international reference population for inclusion.
We did not exclude studies on the basis of location.
We excluded children with medical conditions known to a ect weight status and academic achievement, such as Prader-Willi syndrome and diagnosed intellectual disabilities.

Types of interventions
Studies were eligible for inclusion when the interventions aimed to prevent or reduce obesity. For inclusion, interventions had to be lifestyle interventions of any frequency and duration provided in any setting (e.g. clinics, schools, community centres) that comprised one or more of the following.
1. Interventions to increase physical activity 2. Dietary and nutritional interventions (excluding supplements) 3. Interventions to decrease sedentary behaviour, screen time and TV time 4. Psychological interventions to facilitate weight management Interventions could target children or adolescents with or without the participation of family members.
We excluded studies which implemented a physical activity programme aiming to improve cognitive and academic outcomes without a stated intention to prevent or treat childhood obesity. Where any measure or proxy of adiposity was included as a covariate only, the study was not eligible for inclusion. We excluded pharmacological and surgical interventions because these are likely to be conducted in a less representative sample, thus limiting generalisability.
Eligible control interventions were waiting list, attention placebo control, no treatment, and standard practice.

Types of outcome measures
Primary and secondary outcomes did not serve as criteria for selection of studies based on title and abstract. Assessment of particular outcome measures was a criterion for inclusion in this review when we screened full texts. We restricted the review to particular outcomes because the same interventions were studied in the same populations for di erent purposes, for example change in BMI, BMI z-scores, weight, health-related quality of life, all-cause mortality, morbidity, behaviour change (Al-Khudairy 2017; Colquitt 2016; Mead 2017).
We extracted outcome data at the end of the intervention and at any other follow-up time point.
1. School achievement (Morris 2011), recorded by appropriatelytrained investigators (e.g. teachers, researchers). We excluded participant-and parent-reported data. a. Average achievement of subjects taught at school. i. Average across subjects taught at school over one academic year, for example, grade point average (GPA). b. Achievement in a single subject taught at school.
i. Scores of subjects taught at school or standard achievement test scores for (a) mathematics, (b) reading or (c) language. ii. Validated tests for school achievement in mathematics, reading or language, for example, Woodcock-Johnson Tests of Achievement III (McGrew 2011). c. Special education classes.
i. Need for special education class.
ii. Reduction of time allocated for special education class. 2. Cognitive function (Carroll 1993): measures of general cognitive ability or di erent cognitive domains (e.g. composite executive function, inhibition control, attention, memory) assessed using validated cognitive tests administered by appropriately-trained investigators, such as qualified psychologists. We excluded participant-reported and parent-reported data. 3. Adverse outcomes: include, but are not limited to, reduced school attendance, musculoskeletal issues (e.g. activity-related injury), and psychological issues (e.g. bullying, stigmatisation, depression, eating disorders) obtained from school records, medical records and self-reports (for bullying and stigmatising events only). We included studies reporting adverse events only when measures of school achievement, cognitive function and/ or future success were also reported.

Secondary outcomes
1. Future success: includes, but is not limited to, total years of schooling, high school completion, enrolment in higher education, rates of full-time employment, monthly earnings, home ownership, no/reduced need of social services, obtained from administrative records and self-reports. 2. Obesity indices: age-and gender-specific BMI, BMI z-scores and BMI-SDSs when obtained from measured (not self-reported) weight and height, measured waist circumference and measures of body fatness by dual-energy x-ray absorptiometry (DXA) and bioelectrical impedance analysis (BIA). We included studies reporting obesity indices only when measures of school achievement, cognitive function and/or future success were also reported. Inclusion of these data might enable the review authors to examine whether any changes in school performance, cognitive function and/or future success variables occur independently from changes in obesity (see How the intervention might work). It was not our intention to assess the e ect of interventions for treatment of childhood obesity on adiposity or body weight status. This has recently been examined in three other Cochrane Reviews (Al-Khudairy 2017; Colquitt 2016; Mead 2017).

Search methods for identification of studies Electronic searches
We previously ran searches in 2012 and 2013. For this update, we searched 17 databases and two trials registers listed below in February 2017. Out of the 17 databases, 12 were searched by the Information Specialist of the Cochrane Developmental Psychosocial and Learning Problem Group. The first review author searched the remaining databases and the trials registers. Search strategies are reported in Appendix 1.

Searching other resources
We searched for eligible studies in the reference lists of included studies and in relevant reviews and guidelines.

Selection of studies
We used the web-based so ware platform Covidence to view, screen and select studies. AM, JNB and YL independently screened titles and abstracts and assessed their eligibility to identify potentially relevant trials. AM, YL and DHS assessed full reports for eligibility. We resolved di erent opinions about eligibility by discussion; when the review authors did not agree, the other review authors (JS and JJR) arbitrated. We recorded the reasons for excluding trials in the PRISMA diagram.

Data extraction and management
AM, YL and DHS extracted study characteristics using a predefined data extraction form, with AM and YL cross-checking the extracts. The data extraction form included the following items: General information: review author ID, title, published or unpublished, study authors, year of publication, country, contact address, source of study.
Methods (including 'Risk of bias' assessment): study design, randomisation methods, allocation concealment, blinding, handling of missing data, selective data reporting.
Population: age, gender, ethnicity, proportion of children with obesity or overweight; inclusion and exclusion criteria; number of participants recruited, included and followed (total and in comparison groups); diagnostic criteria of overweight or obesity; comparability of groups at baseline; comorbidities.
Intervention: type(s), frequency, mode of delivery, intensity of physical activity, methods and timing of comparison of intervention, setting, intervention and follow-up duration, who delivered the intervention, attrition rates, assessment of compliance, details of comparison and control.
Outcome: assessor characteristics, baseline measures, measures immediately a er intervention and at follow-up, follow-up time points, validity of measurement tools, definition of outcome (e.g. units, scales), primary outcomes, secondary outcomes.
Results: Where no suitable published data were available, AM contacted the study authors to obtain unpublished data for children and adolescents with overweight or obesity, which were a subgroup of the study sample. AM therefore extracted the result data for each outcome (mean, events, measures of variance, sample sizes), which were double-checked by YL.

Assessment of risk of bias in included studies
AM and DHS independently assessed the risks of bias in each trial, using the Cochrane 'Risk of bias' tool (Chapter 8.5 in Higgins 2011). Findings were cross-checked and discrepancies resolved through discussion. This included assessment of selection bias (random sequence allocation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting) and other sources of bias. The review authors judged the risk of bias as 'high', 'low' or 'unclear', using the information provided.

Measures of treatment e ect
We calculated or extracted the mean change from baseline for intervention and comparison groups, and calculated the mean di erence (MD) of change between the groups, when continuous data (e.g. numerical marks) were measured on the same scale. When similar outcomes were measured on di erent scales, we calculated the standardised mean di erence (SMD). Where it was not possible to determine the change from baseline, we calculated MD or SMD using post-intervention (endpoint) values.
There is no consensus regarding the most appropriate method to use in assessing cognitive ability and school achievement; di erent researchers tend to use di erent tools to measure the same outcome. Where the same outcome was assessed across di erent intervention types, we reported SMD for findings from single-study and multiple-study analyses to allow the comparison of intervention e ects across intervention types. To ease interpretation of the e ect size, we also reported the MD of e ect sizes for single-study outcomes.
We calculated all e ect sizes so that positive e ect sizes indicate better performance on cognitive function and school achievement outcomes in favour of the intervention group compared to the comparison group.
Included studies did not provide dichotomous or ordinal data. However, in Appendix 2, we describe how we intend to treat these types of data if available, as predefined in our protocol (Martin 2012).

Cluster-randomised trials
We scanned all included studies with clustered randomisation of participants for appropriate analysis of clustered data. Ignoring the proportion of total variance attributable to clustering can result in underpowered study designs and inflation of type I error rates, i.e. increased false-positive results (Brown 2015). Therefore, for studies in which control of clustering was missing or insu icient at sample size calculation or analysis stage, and when individual participant data were not available, we approximately corrected the intervention e ects of cluster-RCTs. We reduced the size of each trial to its 'e ective sample size' (Higgins 2011). We calculated the e ective sample size in studies with continuous data by dividing the sample size by the design e ect, which is [1 + (M-1)* ICC], where M is the average cluster size and ICC is the intracluster correlation coe icient. When no ICC was obtainable, we used the ICC estimate of a similar study. In Appendix 3, we provide an overview of the ICCs used to estimate the e ective sample size. Some trial authors provided recalculated ICCs for school or cognitive outcomes, or both, which were previously unpublished. We performed a sensitivity analysis to determine the robustness of conclusions from meta-analyses that included cluster-randomised trials (see Sensitivity analysis).

Library
Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews included only data from the first period before the cross-over took place.

Multiple interventions per individual
We performed separate comparisons for studies that compared the e ects of a single intervention (e.g. physical activity alone) versus a control condition and studies that compared a combination of any types and numbers of interventions of interest (e.g. physical activity with health behaviour education) versus a control condition.
We entered multiple intervention arms of the same study as separate interventions in the meta-analysis. We divided the sample size of the control group by the number of intervention arms in the study to avoid overestimating the pooled e ect size. We le the means and standard deviations unchanged, as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Section 16.5.4. Higgins 2011).

Multiple time points
In separate meta-analyses, we analysed data from studies that reported results at more than one time point with comparable data of other studies at similar time points.

Dealing with missing data
When possible, we recorded characteristics of, reasons for and quantities of missing data for all included studies. We contacted trial authors to obtain information on missing data, if not reported. In our analyses, we ignored data judged to be 'missing at random'. When possible, we imputed missing values in individual participant data, using the last observation carried forward (LOCF) method. We performed sensitivity analyses to examine the e ects of including imputed data in meta-analyses (see Sensitivity analysis).
Included studies did not provide su icient individual participant data to perform an individual participant data meta-analysis. Should these become available from the study authors and prove to benefit the review, we will follow the guidance in Higgins 2011 (Chapter 18).

Assessment of heterogeneity
We assessed clinical heterogeneity by comparing the similarities of included studies in terms of participants, interventions (type, duration, mode of delivery, setting) and outcomes. By comparing study design and risks of bias, we evaluated methodological heterogeneity. We assessed statistical heterogeneity across studies by visual inspection of the forest plot, and we used the Chi 2 test with a significance level of P < 0.1 because of its low power in detecting heterogeneity when studies are low in sample size and numbers of events (section 9.5.2 Higgins 2011). Guided by the Cochrane Handbook (section 9.5.4 Higgins 2011), we estimated the between-study variance in a random-e ects meta-analysis (Tau 2 ) in addition to the percentage of variability of intervention e ect due to statistical heterogeneity ( I 2 ). Variability greater than 50% may indicate moderate to substantial heterogeneity of intervention e ects (section 9.5.2 Higgins 2011). Furthermore, we assessed the cause of heterogeneity by conducting subgroup and sensitivity analyses, as described below (see Subgroup analysis and investigation of heterogeneity; Sensitivity analysis, respectively).

Assessment of reporting biases
We had planned to assess reporting bias by using funnel plots but were unable to do so because of insu icient numbers of included studies (see Appendix 2 and Martin 2012).

Data synthesis
We used Review Manager 5 (RevMan 5) (Review Manager 2014) for data entry and analysis. We combined outcome data from included studies in meta-analyses when the outcome measure addressed the same measurement concept (e.g. mathematics achievement). Where separate data for children and adolescents with overweight and for children and adolescents with obesity were available, we included them separately in the meta-analysis. This was done with the intention to explore a potential 'doseresponse' of the intervention e ect relative to the weight category.
Where the same study reported several outcome variables for one outcome measurement, we included the outcome variable that was comparable with outcomes reported by other included studies. For example, if reaction time and errors were both given for the cognitive outcome 'attention', then we reported only errors to ensure comparability with other studies which solely reported errors.
Health behaviour interventions have inherent heterogeneity due to intervention implementation and setting, so the true intervention e ect is likely to vary between studies. We therefore pooled data using the random-e ects model and provided e ect sizes of studies that were inappropriate to include in a meta-analysis.

'Summary of findings' tables
We summarised outcomes relevant for decision-making in health and education practice or policy or both (Balshem 2011) in 'Summary of findings' tables, using the GRADE approach. The recommended number of primary outcomes to be reported in the table is seven. We considered the following outcomes to be the most relevant: 1. Average achievement across subjects taught at school; 2. Mathematics achievement; 3. Reading achievement; 4. Additional educational support needs; 5. Composite executive functions; 6. Inhibition control; 7. Adverse events.
We used the GRADEprofiler Guideline Development Tool (GRADEpro GDT 2015) to generate the tables for which we imported data directly from RevMan 5 (Review Manager 2014). These comparison-specific tables provide details for each outcome concerning the assessment tools used, follow-up range, timing of follow-up, study design, number of studies, total sample sizes, e ect estimates, and the quality of evidence. Two review authors (AM, DHS) assessed the quality of the evidence, resolving disagreements through discussion with a third review author (JNB).
We determined the quality of the evidence by assessing the methodological quality on outcome level, heterogeneity, the directness of evidence, the precision of evidence, and risk of publication bias. Where the evidence came from small studies, we assessed the extent of the limitation of 'unclear risk of bias on Cochrane Database of Systematic Reviews randomisation' on our confidence in the evidence by consulting the risk-of-bias item 'comparability of groups at baseline'. We did not consider an unclear risk of selection bias as a serious limitation where we had rated the risk-of-bias item 'comparability of groups at baseline' at low risk of bias. A low risk of bias of known baseline characteristics may suggest adequate randomisation, so we have confidence in the evidence. Where we rated 'comparability of groups at baseline' at unclear or high risk of bias, we considered an 'unclear risk of bias on randomisation' as a serious limitation and so downgraded the quality of evidence to reflect our limited confidence in the evidence. However, we acknowledge that variables that were not tested for may cause imbalance between groups and that imbalances can occur by chance, despite adequate randomisation.
GRADE specifies four quality levels: 1. For ease of interpretation of the standardised e ect sizes, we applied rules of thumb, where a standard deviation (SD) of 0.2 represents a small di erence between groups, 0.5 represents a moderate di erence, and 0.8 represents a large di erence (section 12.6.2 in Higgins 2011). Where both change-from-baseline and endpoint data were available for the same outcome, we reported the evidence of highest quality. When the quality of evidence was the same for outcomes generated from endpoint and change-frombaseline data, we reported change-from-baseline outcomes in the 'Summary of findings' table.

Subgroup analysis and investigation of heterogeneity
Subgroup analyses are principally intended to investigate sources of heterogeneity within a meta-analysis in relation to factors that potentially impact outcomes. We identified several potentially influential participant and intervention characteristics for subgroup analyses (see Appendix 2). The low number of studies included for the same outcome did not allow us to perform meaningful subgroup analyses for all predefined sources of heterogeneity. However, we performed a subgroup analysis for body weight status (overweight versus obesity), where possible.

Sensitivity analysis
We investigated the influence of study characteristics on the robustness of the review results by conducting sensitivity analyses. We removed trials from the analysis when studies: 1. used di erent criteria or variations in the thresholds of criteria to define childhood obesity and overweight (e.g. clinical versus public health thresholds); 2. were judged at 'high risk of bias' in the characteristics of random sequence allocation, concealment of allocation, blinding and extent of dropouts; 3. were cluster-RCTs or cross-over trials; 4. provided a post-intervention mean and standard deviations but where change-from-baseline data were missing.

Description of studies
See Characteristics of included studies; Characteristics of excluded studies; Characteristics of studies awaiting classification; Characteristics of ongoing studies.

Results of the search
For the original review (Martin 2014), we screened 17,748 titles and abstracts, and excluded 17,219 records. We retrieved 529 full-text reports, of which we included six studies (14 reports) in the review.
The electronic search for this review update yielded 17,577 records. We found two more records by screening the reference lists of relevant systematic reviews. We also carried forward 17 reports from the previous review that had been classified as ongoing or awaiting classification. Overall, our updated search yielded 17,596 records.
Having excluded 6131 duplicate records, we screened the remaining 11,465 on the basis of title and abstract, and discarded 10,806 as irrelevant.
For 60 records of conference papers, only abstracts were available. We contacted the authors of the conference abstracts for further information and followed up on non-responders two weeks later. We received eighteen replies. Fi een study authors stated that their study did not meet our inclusion criteria (Criteria for considering studies for this review), and we excluded these 15 records at title and abstract stage, along with 42 abstracts for which we were unable to make a decision due to insu icient information. Three authors supplied us with the full-text report of their studies, which we screened and discarded at full-text stage (see Excluded studies).
We retrieved 599 full-text reports, of which 12 new studies (36 reports) met our inclusion criteria. We include 18 studies (57 reports) in total in this updated review (see Characteristics of included studies).
Three more studies (four reports) are awaiting classification (see Characteristics of studies awaiting classification). Thirteen trials (14 reports) are currently ongoing (see Ongoing studies). A flow chart of the search results is shown in Figure 2.

Library
Trusted evidence. Informed decisions. Better health.

Included studies
For 14 of the 18 included studies, outcome data for children and adolescents with obesity or overweight were not published separately from data for the total study population. We therefore contacted the study authors to obtain the unpublished data.

Study design and geographical location
We

Population characteristics
The numbers of participants randomly assigned ranged from 37 to 360, and the number of participants followed and analysed ranged from 28 to 349 (total N = 2384

Intervention characteristics
The interventions fell into three categories: 1. Physical activity only (eight studies); 2. Physical activity plus healthy lifestyle education (seven studies); 3. Dietary interventions including nutrition education (three studies). week and a single session on healthy lifestyle education for parents. Participants in the comparison condition of Melnyk 2013 received a health education programme which covered di erent topics from the intervention group and did not involve active promotion of physical activity, as was the case in the intervention group. The comparison condition in Huang 2015 and Melnyk 2013 did not match the intervention condition in terms of the intensity (see Table 1 for details). Despite this, we considered the comparison conditions as attention controls because the participants received an active intervention. Gallotta 2015 did not provide details on the nature of the comparison condition.

Primary outcomes
In Appendix 4 we summarise additional information on the outcomes and measurement tools used to assess school achievement and cognitive functions. Data were available for five school achievement outcomes: average achievement across subjects taught at school, mathematics achievement, reading achievement, language achievement, and health class grades. Intervention e ects for children and adolescents with obesity or overweight were available for the following cognitive functions: composite executive functions, inhibition control, attention, working memory, visuo-spatial abilities, cognitive flexibility, nonverbal memory, and general intelligence.

Health class achievement
One study provided school achievement outcomes in form of teacher-assessed health class grades (Melnyk 2013).

Special education classes
No study reported intervention e ects for additional educational support needs.

Non-verbal memory
Three studies assessed non-verbal memory using the successive processing scale of the Das-Naglieri-CAS (Davis 2011b; Kra 2014) and the recall trial of the Rey Complex Figure Test (Huang 2015).

General intelligence
One study provided outcome measures on general intelligence using the General Di erential Aptitude Battery (Sánchez-López 2017 [pers comm]).

Adverse events
Although participants in Chen 2016 were asked to record any adverse events during the intervention, no outcome data were reported. Davis 2011b reported a foot fracture as a consequence of participating in the physical activity intervention. The incident occurred in the low-intensity intervention arm, which we deemed as ineligible for inclusion in this review (see Characteristics of included studies). We therefore did not consider this adverse event in the evidence synthesis. No other adverse events were reported.

Future success
None of the included studies assessed measures of future success.

Obesity indices
Six studies which reported the intervention e ect of school or cognitive outcomes also provided change from baseline BMI Waist circumference measures were reported by one study only (Huang 2015).

Follow-up time points
Sixteen studies reported outcomes immediately a er completion of the intervention period or before cross-over of the experimental conditions (Damsgaard 2017 [pers comm]; Sánchez-López 2017 [pers comm]). Only two studies provided outcome data for two follow-up time points.
De Gree 2016 assessed inhibition control, working memory, and cognitive flexibility at six-month and 18-month follow-ups. The first follow-up time point relates to an intervention midterm assessment and the second represents the immediate postintervention follow-up. Personnel who delivered the intervention changed a er mid-term assessment from specially-trained primary/elementary school teachers to the regular classroom teacher, who also received training in delivering the intervention.
Huang 2015 assessed inhibition control, non-verbal memory, visuospatial abilities, and obesity indices immediately a er completion of the six-week intensive day camp versus standard practice/ attention control intervention, and 13-month follow-up from baseline. In the time period between the day-camp intervention and the 13-month follow-up, participants received a lowintensity family-based intervention, which could be considered a maintenance intervention.

Excluded studies
For this updated review, we excluded 541 full-text reports ( Figure  2), 514 of which we deemed to be irrelevant. We formally excluded 18 studies (27 reports) for the following reasons:

Studies awaiting classification
Currently, three studies are awaiting classification. Vetter 2015 is available as a conference abstract only and we were not able to retrieve further details of the study due to non-response from the authors. We have so far contacted the authors twice. NCT02043626 and NCT02122224 are completed studies identified through a trial register, but the results have not yet been published. Based on the information provided in the trial registers, we are not able 1. Bau 2016 (Maintain study) is evaluating a group intervention on healthy eating and lifestyle factors as part of a weight loss maintenance programme compared to standard practice on school achievement in children and adolescents aged between 10 and 17 years with a BMI > 99th percentile. This study took place in Germany and the analysis of the results is currently ongoing. 2. Cadenas-Sanchez 2016 (ActiveBrains project) is taking place in Spain, and compares an exercise intervention with waitlist control aimed at children with obesity or overweight aged eight to 12 years. Cognitive outcomes are executive functions including inhibition control and memory, whereas school achievement will be assessed for mathematics, language and reading achievement.

DRKS00005275 (Ballschool -easy) is being conducted in
Germany, and is a four-arm trial comparing three intervention groups (physical activity, diet, physical activity plus diet) with a no-treatment control for children aged six to 10 years and a BMI > 90th percentile. Overall intelligence will be assessed as a cognitive outcome measure. 4. ISRCTN12698269 (Run-a-mile) is a UK-based study, evaluating the e ect of daily walking or running compared to standard practice on teacher-assessed school achievement in children aged nine to 12 years. Body weight status is not an inclusion criterion but the study evaluates intervention e ects on body fat and so relevant data for this review might be available on completion of the study. 5. NCT01737658 has been conducted in the USA, and compares an exercise intervention with standard practice in adolescents aged 14 to 19 years with a BMI > 99th percentile. The results for intervention e ects on changes in cognitive functions (not further specified) are currently in preparation for publication. 6. NCT02873715 (PLAN trial) is taking place in the USA, comparing a family-based treatment programme plus enhanced usual care with enhanced usual care only in children aged six to 12 with a BMI > 85th percentile. Inhibition control will be the relevant outcome of interest for this review. 7. NCT02972164 is being conducted in children aged nine to 12 with a BMI > 95th percentile in Qatar. The study assesses the e ect of a three-phased weight management programme (weight loss camp/a er-school programme/maintenance) compared to standard school routine on inhibition control. 8. Po'e 2013 (Growing Right Onto Wellness) takes place in the USA, and evaluates a weight management intervention with focus on diet and physical activity consisting of an intensive phase, maintenance and sustainability phase compared to a less intensive educational comparison intervention. Children aged three to five years with a BMI equal to or above the 50th percentile and below the 95th percentile are eligible to take part. Executive functions and general intelligence will be assessed. 9. RBR-38p23s is being conducted in Brazil, and evaluates the e ect of a complex/intense behavioural weight management programme and a 'simple' weight management programme compared to a control condition on school achievement in adolescents aged 10 to 19 years with a BMI > 95th percentile. 10.Robinson 2013 (StanfordGOALS) is taking place in the USA, aimed at children aged seven to 11 years with a BMI > 85th percentile. The study evaluates the e ect of a largescale, community-based, interdisciplinary, multicomponent intervention involving physical activity and behaviour change counselling related to screen time, diet and physical activity on school achievement compared to standard care. 11.Sardinha 2014 is located in Portugal and compares two interventions (physical activity and physical activity plus weight management education) with standard practice in children aged 11 to 14 years. Outcome measures include mathematical achievement, language achievement (Portuguese and English), science achievement and body weight status. This study has been completed but outcome data have not yet been published. 12.Scherr 2014 (Shaping Healthy Choices) is being conducted in the USA, and evaluates a multicomponent school nutrition education programme versus control (not further defined) on science achievement in fourth-grade children. The intervention is not solely aimed at children with obesity or overweight but waist circumference and body mass status are being assessed, yielding data to be included in a future update of this review. 13.Stanley 2016 (Jump Start) is taking place in Australia, targeting young children aged three to five years. The study evaluates the e ect of a physical activity and motor skills intervention versus usual practice on inhibitory control, working memory, and attention. In addition, body weight status is being assessed, allowing the researchers to provide data for children with obesity or overweight specifically.

Risk of bias in included studies
The Characteristics of included studies table provides the reasons for the judgements of risk of bias for each item. Figure 3 and Figure 4 illustrate the judgement for each risk-of-bias item across all included studies and for each included study, respectively.

Figure 4. Risk of bias summary: review authors' judgements about each risk of bias item for each included study. Allocation
The method of sequence generation was described adequately in eight studies and we rated these at low risk of bias. Eight studies failed to report su icient details on how the random sequence was generated and we judged these studies to have an unclear risk of bias. Melnyk 2013 was also rated as unclear risk of bias despite adequate description of the sequence generation. However, it remains unclear if group allocation (drawing of school Cochrane Database of Systematic Reviews names from a hat) was su iciently concealed using opaque envelopes. Treu 2017 was assessed at high risk of bias because only schools allocated to one of two intervention arms were randomised, whereas the control schools were not randomly allocated. Consequently, we conducted a sensitivity analysis.
Adequate description of allocation concealment was evident for five studies, and we judged these as low risk of bias. We rated all of the remaining 13 studies as unclear risk of bias, due to insu icient reporting.

Blinding of participants and personnel (performance bias)
In trials involving physical activity, diet and health education, true blinding of participants and personnel involved in delivering the intervention is not possible. However, four studies (Ahamed 2007; Davis 2011b; Nanney 2016; Staiano 2012) blinded participants and personnel to the true purpose of the study relevant for this review, i.e. changes in cognitive or academic outcomes. We therefore judged these studies to be at unclear risk for performance bias. Three studies (Huang 2015; Kra 2014; Melnyk 2013) employed an attention control condition which reduced the risk of performance bias and we rated these at unclear risk of bias. We rated the remaining 11 studies at high risk of bias.

Blinding of outcome assessment (detection bias)
We judged the risk of bias for blinding of the outcomes assessor as low for eight studies. Six studies reported insu iciently on whether the outcome assessor was blinded, and we therefore judged these as unclear risk of bias. School achievement was assessed by teachers who were aware of the group allocation in four studies (Barbosa Filho 2017 [pers comm]; Damsgaard 2017 [pers comm]; Johnston 2013; Nanney 2016) and so we rated these studies at high risk of detection bias.

Incomplete outcome data
We reported attrition rates and reasons for attrition in the Characteristics of included studies. Low levels of attrition and adequate handling of missing data were performed in five studies, which we rated at low risk of bias (Barbosa Filho 2017 [pers comm]; Davis 2011b; Johnston 2013; Treu 2017; Winter 2011). No imputation of missing data was evident in Damsgaard 2017 [pers comm], but attrition rates were low (less than 10%) and we judged this study as being at unclear risk of bias. Study details obtained from Gallotta 2015 were insu icient to assess the risk of attrition bias and thus we judged this study as being at unclear risk of bias. In Melnyk 2013, relevant outcome data were only collected at post-intervention, which precluded assessment of attrition bias. We rated this study at unclear risk of bias. We judged the risk of attrition bias to be high in nine studies in which no imputation of missing data was performed or the level of attrition was high.

Selective reporting
We rated the risk of selective reporting as low in 10 studies, and unclear in seven studies which made no reference to a study protocol or trial register. We judged Huang 2015 to be at high risk of bias, because the cognitive outcomes and test batteries stated in the study protocol did not align with the Result report. According to the study protocol attention and processing speed were planned to be assessed using the Symbol Digit Modalities Test, and executive function and attention were planned to be assessed using the Trail Making Test. However, the Result report provided findings for executive function using the Stroop Colour and Word Test, and visuo-spatial abilities and non-verbal memory using the Rey complex Figure Test.

Other potential sources of bias
Comparability of baseline groups might be a potential source of bias in cluster-RCTs, and RCTs with flaws in the randomisation procedure (Higgins 2011). Five cluster-RCTs showed no di erence between the experimental groups at baseline and we rated them at low risk of bias ( We identified two studies with other sources of bias. Huang 2015 included children that did not meet the inclusion criteria, so this study was at high risk of bias for violation of the study protocol. In Melnyk 2013, the school district administrator selected participating schools and the schools were o ered financial incentives which might have introduced an additional selection bias. We did not detect any other risk of bias in the remaining studies and thus rated them at low risk of bias.

E ects of interventions
See: Summary of findings for the main comparison Physical activity intervention compared to standard practice for improving cognition and school achievement in children and adolescents with obesity or overweight; Summary of findings 2 Physical activity plus healthy lifestyle education interventions compared to standard practice for improving cognition and school achievement in children and adolescents with obesity or overweight; Summary of findings 3 Dietary interventions compared to standard practice for improving cognition and school achievement in children and adolescents with obesity and overweight We summarised and analysed the three intervention groups in separate comparisons and generated a 'Summary of findings' table of the most important outcomes for each comparison (see Summary of findings for the main comparison; Summary of findings 2; Summary of findings 3). The intervention groups consisted of physical activity only, physical activity combined with healthy lifestyle education, and dietary interventions. We reported the secondary outcomes (future success and obesity indices) combined for all three comparisons, due to the low number of studies providing suitable data.

Comparison 1: Physical activity only interventions versus standard practice
Eight studies delivered physical activity-only interventions and compared them to standard practice (see Table 1 and Characteristics of included studies). Of these, seven studies provided suitable data for inclusion in meta-analyses. However, the number of studies included for the same class of outcome was low, ranging from one to three studies. We performed sensitivity analyses, as specified in Data collection and analysis. However, the low number of studies make the outcome of a sensitivity analysis less meaningful, as the number of included studies is reduced further. Data were available for the outcomes mathematics, reading and language achievement, and composite executive functions and inhibition control, which we include in Summary of findings for the main comparison. Study authors also provided data for the outcomes of attention, working memory, visuo-spatial abilities, cognitive flexibility, non-verbal memory, and general intelligence.

Mathematics achievement
Three studies were included which used di erent scales: broad mathematics scale of the Woodcook-Johnson Test of Achievement III (Davis 2011b), a standardised national mathematics test (Resaland 2016), and numerical quantitative concepts scale of the General Di erential Aptitude Battery (Sánchez-López 2017 [pers comm]). We therefore calculated the e ect estimate as the standardised mean di erence. We calculated subtotals of change from baseline data from Resaland 2016 and Sánchez-López 2017 [pers comm] (both cluster-RCTs), and combined post-intervention data from Davis 2011b (RCT) and Resaland 2016 separately. We converted the reported standard error for post-intervention data in Davis 2011b into standard deviations.
We downgraded the quality of evidence by one level for high risk of attrition bias present in the two studies (Resaland 2016; Sánchez-López 2017 [pers comm]). Pooled post-intervention data resulted in a SMD of 0.19 (95% CI -0.03 to 0.42; 2 studies, 314 children, I 2 = 0%, Tau 2 = 0.00). Sensitivity analysis for high risk of attrition bias and cluster-RCT design involved removing Resaland 2016 from the latter analysis. The overall conclusion of the evidence did not change with Davis 2011b remaining (SMD 0.31, 95% CI -0.10 to 0.71; 1 study, 96 children).

Reading achievement
Two studies provided data on the intervention e ect of physical activity on reading achievement compared to standard practice. Both studies used di erent scales: broad reading scale of the Woodcook-Johnson Test of Achievement III (Davis 2011b) and a standardised national reading test (Resaland 2016). We therefore calculated the standardised mean di erence to estimate the pooled di erence between the experimental groups. Resaland 2016 was a cluster-RCT and Davis 2011b a RCT. Davis 2011b provided standard errors for the post-intervention data which we converted into standard deviation scores prior to entering these in the meta-analysis. We combined post-intervention endpoint data.
Meta-analysis findings (see Analysis 1.2) Our analysis suggested that there was no statistically significant di erence between physical activity and standard practice on reading achievement in children aged seven to 11 years with overweight, including obesity (SMD 0.10, 95% CI -0.30 to 0.49; 2 studies, 308 children, I 2 = 63%, Tau 2 = 0.05). This finding was of moderate quality and we downgraded it by one level due to high risk of attrition bias in Resaland 2016. Removing this study from the analysis did not change the conclusion (SMD 0.33, 95% CI -0.08 to 0.73; 1 study, 96 children).

Language achievement
This outcome was assessed by two studies. However, one study (Sánchez-López 2017 [pers comm]) assessed native language achievement (Spanish) and another study provided data for English language achievement in people whose first language was Norwegian (Resaland 2016). We therefore did not combine these outcomes in a meta-analysis, as di erent concepts were measured. For native language achievement, we reported the mean di erence and standardised mean di erence of the intervention e ect, to allow comparison with studies included in Comparison 2 (physical activity combined with healthy lifestyle education versus standard practice).
There was no evidence of a beneficial e ect of the physical activity programme Movi-Kids (Sánchez-López 2017 [pers comm]; see Table  1 for details) on native language achievement in children aged four to seven years with obesity or overweight (MD 2.38, 95% CI -4.75 to 9.51, scale range 0 to 36; SMD 0.23, 95% CI -0.50 to 0.95; 1 study, 31 children; Analysis 1.3). The quality of this evidence was low; we downgraded the quality twice for high risk of attrition bias and imprecision due to the low sample size. This outcome was measured using the analogical and complex verbal order scale of the General Di erential Aptitude Battery.

Additional educational support
None of the studies assessing the e ect of physical activity interventions compared to standard practice in children with obesity or overweight reported findings on additional educational support needs.
Three studies measured composite executive functions, of which two studies provided suitable data for inclusion in the metaanalysis. Kra 2014 provided a narrative description of the findings only and we were not able to obtain the quantitative data from the study authors for inclusion in the meta-analysis. Composite executive functions were measured using the Planning Scale of the Das-Naglieri-Cognitive Assessment System. The study authors reported that their eight-month aerobic physical activity programme, delivered five days a week a er school, did not result in statistically significant di erences in composite executive functions compared to sedentary activities such as art and board games in 175 children aged eight to 11 years with obesity or overweight.
The two studies included in the meta-analysis used di erent scales: Planning scale of the Das-Naglieri-Cognitive Assessment System (Davis 2011b), which is a composite of three separate tasks, and Design Fluency and Trail-Making subscales of the Delis-Kaplan Executive Function System (Staiano 2012). Both studies were RCTs, with one study reporting change from baseline data (Staiano 2012) and the other post-intervention data (Davis 2011b). We therefore did not pool the two studies. Staiano 2012 included two intervention arms which we entered separately into the metaanalysis. We divided the sample size of the control group by the number of intervention arms (i.e. two). We calculated mean di erences and the standardised mean di erence, to be able to compare the e ect estimates between the two studies. We converted post-intervention standard errors to standard deviation scores from Davis 2011b.
Meta-analysis findings (see Analysis 1.5) Analysis of post-intervention data suggested that the mean composite executive functions were five scale points higher (95% CI 0.68 to 9.32; scale mean = 100, SD = 15; SMD 0.42, 95% CI 0.05 to 0.78) in the a er-school physical activity intervention group compared to standard practice in children aged eight to 11 years with obesity or overweight (1 study, 116 children). This evidence was of high quality. There was no evidence of a beneficial e ect of exergaming interventions on change in mean composite executive function compared to standard care in 54 adolescents (MD 8.45 points, 95% -1.67 to 18.56 points; 1 study, scale mean = 10, SD = 3; SMD 0.58, 95% CI -0.02 to 1.18). The quality of this evidence was low, due to a high risk of attrition bias and imprecision of the e ect estimate.

Inhibition control
One physical activity study measured inhibition control using the Stroop Colour Word Test (De Gree 2016). The authors provided unpublished data for children with obesity or overweight for a midterm assessment at six-month follow-up and post-intervention data at 18-month follow-up. We reported both the mean di erence and the standardised mean di erence of the intervention e ect to allow comparison with other intervention types reported in this review. We conduced separate analyses for each time point and included post-intervention follow-up outcome data in Summary of findings for the main comparison.
Compared to standard practice, there was no evidence of a beneficial e ect of physically active mathematics and language lessons on inhibition control in children aged seven to nine years with obesity or overweight at either follow-up time point. At sixmonth follow-up, the mean inhibition control was 0.35 scale points higher (95% CI -2.59 to 3.29, scale range 0 to 100; SMD 0.04, 95% CI -0.33 to 0.41, 112 children; Analysis 1.6) in the intervention group compared to standard practice. At post-intervention, the group di erence was small (MD -1.55, 95% CI -5.85 to 2.75, scale range 0 to 100; SMD -0.15, 95% CI -0.58 to 0.28; 1 study, 84 children). This finding was of very low quality, suggesting low confidence in the e ect estimate. We downgraded the quality by three levels for high risk of selection and attrition bias, and imprecision due to the low sample size.

Attention
Three studies measured attention abilities using di erent scales: Attention scale of the Das-Naglieri-Cognitive Assessment System (Davis 2011b; Kra 2014) and the D2-R test of attention (Gallotta 2015). Two of the studies were suitable for inclusion in the meta-analysis for which we reported the e ect sizes as the standardised mean di erence of post-intervention data. Kra 2014 did not provide data for inclusion in the meta-analysis. Narrative description of the findings indicate no e ect of an eight-month aerobic physical activity programme, delivered five days a week a er school, compared to sedentary activities in favour of the intervention in 175 children aged eight to 11 years with obesity or overweight (Kra 2014).
Meta-analysis findings (see Analysis 1.7) Gallotta 2015 provided unpublished data for the subgroup with obesity/overweight for three measures of attention: processing speed, concentration, and performance quality. We included only concentration performance because it was the most comparable measure with Davis 2011b. The two studies included in the metaanalysis di ered in that one was a RCT of a 13-week a er-school physical activity programme (Davis 2011b), and one was a fivemonth cluster-RCT with two intervention arms delivered in the primary/elementary school setting (Gallotta 2015; see Table 1 for details). We included both intervention arms separately in the meta-analysis and divided the sample size of the comparison group between them.
There was no evidence of a beneficial e ect of the physical activity interventions compared to standard practice for eight to 11 yearolds with obesity or overweight (SMD 0.46, 95% CI -0.16 to 1.08; 2 studies, 157 children, I 2 = 41%, Tau 2 = 0.14). The sensitivity analysis for cluster-RCT design resulted in a SMD of 0.15 (95% CI -0.22 to 0.51; 1 study, 116 children).

Working memory
Only De Gree 2016 provided data (unpublished specifically for children with obesity/overweight) for verbal working memory at six-month follow-up (mid-term) and 18-month follow-up (postintervention data), measured using the Digit Span Backward test. The authors also provided non-verbal working memory data obtained from the Visual Span Backward test. See Analysis 1.8; Analysis 1.9.
There was no evidence of a beneficial e ect of physically active mathematics and language lessons on verbal working memory in children aged seven to nine years with obesity or overweight compared to standard practice at either follow-up time point. At sixmonth follow-up, the mean verbal working memory was 0.15 points Similar to the previous outcomes assessed by Kra 2014, composite executive functions and attention, the narrative description of the findings indicated no beneficial e ect of an eightmonth aerobic physical activity programme, delivered five days a week a er school compared to sedentary activities on visuo-spatial abilities in eight to 11 year-olds with obesity or overweight. We did not combine the two studies that provided data because Davis 2011b provided baseline-adjusted post-intervention data, while Sánchez-López 2017 [pers comm] provided unpublished changefrom-baseline data. We converted the reported standard errors in Davis 2011b to standard deviation scores.
Mean change in visual-spatial abilities was 4.71 scale points higher (95% CI 0.40 to 9.02 scale points, scale range 0 to 36) in the Movi-Kids intervention group compared to standard practice in 39 children with obesity or overweight (SMD 0.70, 95% CI 0.03 to 1.37; Sánchez-López 2017 [pers comm]; Analysis 1.10). There was no evidence of a beneficial intervention e ect on post-intervention visuo-spatial abilities of an a er-school physical activity programme compared to standard practice in 116 children (MD 4.00, 95% CI -0.44 to 8.44, scale mean 100, SD 15; SMD 0.33, 95% CI -0.04 to 0.69, Davis 2011b).

Cognitive flexibility
We included two studies which used a similar scale, the Wisconsin Card Sorting Test, but di erent measures were reported. De Gree 2016 reported an e iciency score which considered the number of errors and unused cards, whereas Chen 2016 reported the total number of errors only. We therefore calculated the standardised mean di erence. To allow comparability in terms of measurement time points we used the six-month follow-up of De Gree 2016 and excluded the 18-month follow-up from the analysis. The immediate post-intervention follow-up in Chen 2016 was three months. We conducted sensitivity analyses for the cluster-RCT (De Gree 2016).
Meta-analysis findings (see Analysis 1.11) The mean cognitive flexibility performance was 0.06 standard deviations lower (95% CI -0.37 to 0.25, I 2 = 0%, Tau 2 = 0.00) in the physical activity intervention group compared to standard practice, indicating no beneficial e ect in favour of the intervention group (162 children). Both studies were at high risk for attrition bias.

Non-verbal memory
Two studies assessed non-verbal memory using the same scale (Successive processing scale of the Das-Naglieri-Cognitive Assessment System) and employing the same physical activity intervention (Davis 2011b; Kra 2014). Only Davis 2011b reported quantitative data consisting of baseline-adjusted post-intervention outcomes. Their findings indicated that an aerobic physical activity programme, delivered for 13 weeks on five days a week a er school, resulted in 3.00 points higher (95% CI 0.51 to 5.49, scale mean 100, SD 15, Analysis 1.12) mean non-verbal memory compared to standard practice in children aged eight to 11 years with obesity or overweight (SMD 0.43, 95% CI 0.07 to 0.80, 116 children). This e ect estimate suggested a small di erence between the intervention and comparison groups.

General intelligence
Sánchez-López 2017 [pers comm] was the only study which provided measures of general intelligence, using the General and Di erential Aptitude Battery. The mean change from baseline was 17.14 points higher (95% CI 7.24 to 27.04, scale range 0 to 108) in the intervention group (Movi-Kids, see Table 1 for details) compared to the standard practice group (34 children, see Analysis 1.13). We are moderately confident in the e ect estimate but it is likely that further research may change the estimate. Sánchez-López 2017 [pers comm] was at high risk for attrition bias and imprecision of the e ect estimate. However, we upgraded the quality of evidence due to the large e ect size.

Adverse outcomes
No study reported data on adverse events while or a er taking part in physical activity interventions.

Comparison 2: Physical activity interventions combined with healthy lifestyle education versus standard practice
In total, seven studies delivered physical activity combined with healthy lifestyle education interventions and compared them to standard practice (see Table 1 and Characteristics of included studies). All studies provided suitable data for inclusion in metaanalyses. However, the number of studies included for the same outcome was low, ranging between one and three studies. We performed sensitivity analyses as specified in Sensitivity analysis. However, as with Comparison 1, the low number of studies makes the outcome of a sensitivity analysis less meaningful as the number of included studies is further reduced. Data were available for the outcomes mathematics, reading and language achievement, and inhibition control, which we included in Summary of findings 2. Study authors also provided data for the average achievement across subjects taught at school, attention, visuo-spatial abilities, and non-verbal memory.

Average achievement across subjects taught at school
One study provided unpublished data for the average score of mathematics, reading and language, using the Canadian Achievement Test 3 (Ahamed 2007). The mean change in average Cochrane Database of Systematic Reviews school achievement was 6.37 grade points lower (95% CI -36.83 to 24.09, scale mean 500, SD 70) in the intervention group ('Action Schools! BC') compared to standard practice in 31 children aged seven to 11 years with obesity or overweight (SMD -0.18, 95% CI -0.93 to 0.58; Analysis 2.1). Ahamed 2007 was at high risk of attrition bias and at unclear risk of randomisation bias ( Figure 4) and we therefore downgraded the evidence by two levels.

Mathematics achievement
The e ects of physical activity intervention combined with healthy lifestyle education on mathematics achievement were assessed in three studies using di erent scales: Canadian Achievement Test 3 (Ahamed 2007), standardised national mathematics test (Barbosa Filho 2017 [pers comm]), and AIMSweb standardised Mathematics Concepts and Application Test (Treu 2017). The scale used by Treu 2017 measured mathematical problem-solving skills. Although additional outcomes obtained from Mathematics-Curriculum-Based-Measurement scale were also provided by Treu 2017, we did not include this outcome because data were not available from all participating schools. We used change from baseline for all studies and calculated the standardised mean di erence. We included data for children with overweight separately from data of children with obesity (Barbosa Filho 2017 [pers comm]), and also included the two intervention arms in Treu 2017 separately. We divided the sample size of the comparison group to estimate group di erences. All studies were cluster-RCTs, and so we conducted sensitivity analysis for risk of bias only.

Meta-analysis findings (see Analysis 2.2)
There was no evidence of a beneficial e ect for the intervention on mathematics achievement compared to standard practice (SMD 0.02, 95% CI -0.19 to 0.22; I 2 = 0%, Tau 2 = 0.00) in 384 children and adolescents aged eight to 18 years. This finding was of very low quality, suggesting that the true e ect is likely to be substantially di erent from the estimated e ect and we are confident that further research will result in di erent estimates. We downgraded the quality for high risk of bias (sequence generation, blinding of the outcome assessor, attrition), inconsistency, and imprecision of estimates. Barbosa Filho 2017 [pers comm] provided separate data for 64 children with overweight and 35 children with obesity. The single study e ect estimates were statistically non-significant for both subgroups.
Sensitivity analysis for high risk of sequence generation in Treu 2017 indicated no changes to the overall conclusion (SMD -0.07, 95% CI -0.41 to 0.28, 2 studies, 140 children). Removing the studies with high risk of attrition bias did not influence the overall conclusion (SMD -0.03, 95% CI -0.43 to 0.38; 1 study, 99 participants).

Reading achievement
Two cluster-RCTs were included using di erent scales: Canadian Achievement Test 3 (Ahamed 2007) and AIMSweb standardised Reading Curriculum Based Measurement (Treu 2017). We therefore calculated standardised mean di erences of change from baseline data. Treu 2017 also provided data obtained from the MAZE reading test which we did not include, because the curriculum-based measurement appeared to be more comparable with the outcome reported by Ahamed 2007. We included the two intervention arms in Treu 2017 separately and distributed the sample size of the comparison between them.

Meta-analysis findings (see Analysis 2.3)
There was low-quality evidence of no di erence between the intervention and comparison groups for reading achievement (SMD 0.00, 95% CI -0.24 to 0.24; 2 studies, 284 children, I 2 = 0%, Tau 2 = 0.00). We downgraded the evidence for risk of bias and inconsistency of e ect estimates, suggesting little confidence in the e ect estimate. Sensitivity analysis of high risk of selection bias (Treu 2017) and attrition bias (Ahamed 2007) did not change the overall conclusion.

Language achievement
We included three cluster-RCTs which measured language achievement on di erent scales: Canadian Achievement Test 3 for English language (Ahamed 2007), standardised national test in Portuguese language (Barbosa Filho 2017 [pers comm]), and Peabody Picture Vocabulary Test in English language (Winter 2011). All studies provided unpublished change-from-baseline data for native language achievement of children with overweight/obesity. We calculated standardised mean di erences due to the di erence in scales used. We included the data set with imputed missing data (last observation carried forward) in Winter 2011, and conducted a sensitivity analysis using per-protocol data.

Meta-analysis findings (see Analysis 2.4)
Compared to standard practice, the mean language achievement was 0.13 standard deviations higher (95% CI -0.12 to 0.39, I 2 = 0%, Tau 2 = 0.00) in interventions combining physical activity with healthy lifestyle education (244 participants). This evidence was of very low quality, due to imprecision in e ect estimates and high risk of attrition, selection and detection bias. This indicated that our confidence in the e ect estimate is limited and further research is very likely to change the estimate. Sensitivity analysis for attrition bias in one study (Ahamed 2007) and imputation of missing data (Winter 2011) did not change the overall conclusion: SMD 0.12 (95% CI -0.18 to 0.43; 2 studies, 173 children) and SMD 0.11 (95% CI -0.17 to 0.40, 3 studies, 197 participants).
Two studies provided separate data for children with overweight and children with obesity (Barbosa Filho 2017 [pers comm]; Winter 2011). For children with obesity, mean change in language achievement was 0.28 standard deviations higher (95% CI -0.20 to 0.77) in the intervention group compared to standard practice (70 children, 2 studies). The e ect favouring the intervention group was lower in children with overweight (SMD 0.02, 95% CI -0.37 to 0.41, 103 children, 2 studies).

Health class achievement
Melnyk 2013 assessed the e ect of the 15-week COPE Healthy Lifestyle TEEN programme (see Table 1 for details) compared to an attention control (health topics other than physical activity and nutrition) on teacher-assessed grades in health class courses of adolescents aged 14 to 16 years. The authors provided unpublished post-intervention data separately for adolescents with overweight and adolescent with obesity, which we entered as two comparisons in the meta-analysis. We calculated the between-group mean di erence.
The mean health class achievement was 0.05 scores lower (95% CI -0.38 to 0.29, scale range 0 to 4, 263 adolescents; Analysis 2.5) in the intervention group compared to the control group, suggesting Cochrane Database of Systematic Reviews a small statistically non-significant di erence in favour of the control group. There was no between-group di erence between 108 adolescents with overweight and 155 adolescents with obesity. We judged this study to be of moderate quality and reduced the quality rating due to high risk of bias for the comparability of the experimental groups at baseline and selection bias.

Additional educational support
No study provided data on the e ects of physical activity plus healthy lifestyle education interventions on additional educational support needs for children and adolescents with obesity or overweight.

Inhibition control
We included two studies which measured inhibition control on di erent scales: Stroop Colour and Word Test (Huang 2015) and the Go/No-go test of the Attention test battery for children (Wirt 2013 [pers comm]). We therefore report the standardised mean di erence. Since only post-intervention data were available for Wirt 2013 [pers comm], we used the post-intervention data reported by Huang 2015. This study also reported mid-term outcomes at six-week follow-up. We used the immediate post-intervention time point for combining the study with Wirt 2013 [pers comm]. We performed a sensitivity analysis for the cluster-RCT (Wirt 2013 [pers comm]).

Meta-analysis finding (see Analysis 2.6)
There was low-quality evidence of lower mean inhibition control by 0.67 standard deviations (95% -1.50 to 0.16) in physical activity plus healthy lifestyle education intervention compared to standard practice/attention control in 110 children aged six to 13 years with obesity or overweight. We downgraded the evidence for high risk of attrition bias and selective reporting. The statistical heterogeneity was substantial (I 2 = 68%, Tau 2 = 0.25), most likely owing to methodological variability in the interventions and population characteristics (see Characteristics of included studies). The sensitivity analysis did not change the conclusion.

Attention
One study reported intervention e ects on attention using the sustained attention scale of the Attention Test Battery for children (Wirt 2013 [pers comm]). For comparability of the e ect estimates with Comparison 1 and 3, we report the estimates as the mean di erence (see Analysis 2.7) and standardised mean di erence.
Compared to standard practice, analysis of the unpublished data indicate no beneficial e ect of physically active school breaks combined with healthy lifestyle education for one school year in 27 children with obesity or overweight aged six to eight years. The mean attention ability was 4.47 lower (95% CI -8.55 to -0.39, scale range 0 to 100) in the intervention group compared to the control group (SMD -0.71, 95% CI -1.54 to 0.12; Analysis 2.7). Imprecision due to the low sample size and high risk of attrition bias limit our confidence in the e ect estimate.

Visuo-spatial abilities
Huang 2015 was the only study that measured visuo-spatial abilities in children with obesity or overweight, and used the copy trial of the Rey Complex Figure Test at six-week follow-up (mid-term) and 13-month follow-up (post-intervention). We calculated between-group mean di erences of post-intervention data. We also report the published e ect estimates of change from baseline expressed as fitted mean of standardised outcomes which were adjusted for sex and cohort.
Analysis of crude post-intervention data suggested no statistically significant e ect favouring the intervention group at six-week follow-up (MD 0.29 points, 95% CI -1.52 to 2.10; scale range 0 to 36, SMD 0.07, 95% CI -0.34 to 0.47; 94 children; Analysis 2.8) and 13month follow-up (MD -0.45 points, 95% CI -2.58 to 1.68; scale range 0 to 36; SMD -0.09, 95% CI -0.52 to 0.33; 86 children). The quality of evidence at both time points was low, due to high risk of attrition and imprecision of e ect estimates.
There was evidence of a beneficial e ect on visuo-spatial abilities in favour of the intervention compared to the attention control when expressed as fitted mean adjusted for sex and cohort at six-week follow-up (SMD 0.47, 95% CI 0.08 to 0.86, 94 children), indicating a moderate di erence between the two experimental groups. The beneficial e ect was not maintained at 13-month follow-up (SMD 0.21, 95% CI -0.26 to 0.67, 86 children).

Non-verbal memory
Huang 2015 was the only study that measured non-verbal memory and was part of the Odense Overweight Intervention Study. This study used the immediate recall trial of the Rey Complex Figure Test to measure non-verbal memory at six-week follow-up (mid-term) and 13-month follow-up (post-intervention). We calculated mean di erences and standardised mean di erences of post-intervention data, and report the published standardised and adjusted change from baseline of this study.
Mean non-verbal memory was 2.05 points lower (95% CI -5.03 to 0.93; scale range 0 to 36; SMD -0.28, 95% CI -0.69 to 0.13; 94 children) in the intervention group compared to attention control at six-week follow-up when analysing post-intervention data. At 13month follow-up the MD was -3.42 points (95% CI -6.30 to -0.54; scale range 0 to 36; SMD -0.52, 95% CI -0.95 to -0.08; 86 children; Analysis 2.9). Huang 2015 was at high risk for attrition bias and the e ect estimates indicate imprecision, which leaves us with limited confidence in the estimate. The true e ect might be substantially di erent for the reported estimates.
The sex-adjusted mean di erence in change from baseline was 0.19 standard deviations higher (95% CI -0.10 to 0.48, 94 children) in the intensive day-camp intervention group compared to the attention placebo control group which received a low-intensity physical activity and health education intervention (see Table 1 for details) at six-week follow-up. At 13-month follow-up, there was also no evidence of beneficial e ects of the intervention on nonverbal memory compared to attention control in 86 children with obesity or overweight aged 12 to 13 years (SMD -0.005, 95% CI -0.35 to 0.34).

Adverse outcomes
No study reported adverse outcome data for physical activity plus healthy lifestyle education interventions.

Comparison 3: Dietary interventions versus standard practice
Three studies compared dietary intervention with a standard practice (see Table 1 and Characteristics of included studies). Data Cochrane Database of Systematic Reviews were available for four outcomes which we include in the Summary of findings 3: average achievement across subjects taught at school (two studies), mathematics achievement (one study), language achievement (one study), and attention (one study). All studies were cluster-RCTs and two studies provided unpublished data for children with obesity or overweight (Damsgaard 2017 [pers comm]; Nanney 2016).

Average achievement across subjects taught at school
Johnston 2013 and Nanney 2016 assessed the average school year performance of mathematics, reading and science scores by generating a Grade Point Average. School achievement was assessed by teachers in both studies, but the scales varied: scale range 0 to 4 in Nanney 2016, scale range 0 to 100 in Johnston 2013. We therefore calculated the standardised mean di erence. Both studies reported change-from-baseline data. Nanney 2016 provided separate data for children with overweight and children with obesity, which we have included as separate subgroups in the meta-analysis. We conducted a sensitivity analysis for per-protocol data of Nanney 2016.

Meta-analysis findings (see Analysis 3.1)
The mean average across subjects taught at school was 0.32 standard deviations higher (95% CI -0.07 to 0.70) in the dietary intervention groups compared to standard practice for 439 children and adolescents aged 7 to 17 years with obesity or overweight, suggesting a small statistically non-significant di erence between the experimental groups.
Given that the participants in Johnston 2013 had an average BMI in the 95th percentile, we performed a subgroup analysis for body weight status, classifying Johnston 2013 under the subgroup 'children with obesity'. Considering data of children with obesity only, there was a moderate e ect estimate of 0.45 standard deviation in favour of the intervention group (95% CI 0.25 to 0.66, 379 participants). There was no evidence of a beneficial e ect of the intervention in children with overweight (SMD -0.17, 95% CI -0.70 to 0.36, 1 study, 55 participants). The subgroup analysis identified Nanney 2016 as the source of statistical heterogeneity, with the I 2 statistic reduced from 62% to 0% (Tau 2 0.07 to 0.00). Formal testing indicated a significant subgroup di erence (Chi 2 = 4.60, P = 0.03). This finding was of low quality, indicating that further research is very likely to have an important impact on our confidence in the e ect estimate and may change the estimate. We downgraded the quality due to high risks of detection and attrition bias.
The sensitivity analysis for per-protocol data indicated a mean average across subjects taught at school of 0.30 standard deviations higher (95% CI 0.04 to 0.55; 2 studies, 422 children, I 2 = 20%) in the intervention group compared to standard practice. The e ect estimate for children with obesity decreased from moderate to small (SMD 0.34, 95% CI 0.05 to 0.63; 2 studies, 380 children). The e ect estimate for children with overweight shi ed in favour of the intervention group (SMD 0.03, 95% CI -0.59 to 0.64, 1 study, 42 children).

Mathematics achievement
Damsgaard 2017 [pers comm] assessed the e ect of the New Nordic Diet compared to standard school meals on mathematics achievement, measured using standardised national tests. This study provided unpublished data for children with overweight and for children with obesity, which we entered separately in the meta-analysis (see Analysis 3.2). For comparability with the e ect estimates of Comparisons 1 and 2, we calculated both mean di erence and standardised mean di erence for change from baseline.
There was low-quality evidence of no beneficial e ect of the dietary intervention compared to standard practice on mathematics achievement (MD -2.18, 95% CI -5.83 to 1.47, scale range: 0 to 69; SMD -0.26, 95% CI -0.72 to 0.20) in 76 children aged nine to 11 years with obesity or overweight. We downgraded the quality for a high risk of detection bias and imprecision of the e ect estimate, probably due to the small sample size. This indicates low confidence in the e ect estimate and that further research is very likely to change the estimate. There was no di erence in e ect estimates for children with overweight and children with obesity.

Reading achievement
Damsgaard 2017 [pers comm] also measured reading achievement using standardised national tests. The mean change in reading achievement was 0.13 standard deviations higher (95% CI -0.35 to 0.61; MD 1.17, 95% CI -4.40 to 6.73, scale range: 0 to 108) in the intervention group compared to standard practice, indicating a small statistically non-significant di erence between the experimental groups (67 children, see Analysis 3.3). This finding was of low quality, as we downgraded the evidence for risk of detection bias and imprecision of the e ect estimate, probably due to the small sample size. Inspection of the e ect estimates for overweight and obesity suggested statistically non-significantly higher standardised reading achievement in favour of the control group for children with obesity, while for children with overweight the e ect estimate was in favour of the intervention.

Additional educational support
None of the studies assessing the e ect of dietary interventions compared to standard practice in children with obesity or overweight reported findings on additional educational support needs.

Cognitive functions Attention
Attention performance was assessed by one study. Damsgaard 2017 [pers comm] measured attention using the D2-R test of attention. We included concentration performance as a measure of attention and discharged processing speed to allow the comparison of the e ect estimates with those under Comparison 1. We used the change from baseline of the unpublished data and calculated MD and SMD (see Analysis 3.4).
Compared to standard practice, there was no evidence of a beneficial e ect of the New Nordic Diet on attention performance (MD 1.68, 95% CI -7.86 to 11.22, scale range:-359 to 299; SMD 0.04, 95% CI -0.55 to 0.62; 61 children). The analysis suggests inconsistency in the e ect estimates for children aged nine to 11 years with overweight and children with obesity: statistically nonsignificant higher attention performance of children with obesity in the intervention group, and higher attention performance in control group children with overweight.

Cochrane Database of Systematic Reviews
The quality of this evidence was low; high risk of detection bias and imprecision of the estimate resulted in downgrading of the evidence. Our confidence in the e ect estimate is therefore limited and the true e ect of dietary interventions may be substantially di erent.

Adverse outcomes
No study reported data on adverse events resulting from participating in the dietary interventions.
Secondary outcomes for comparisons 1 to 3

Future success
No study provided data on the e ects of any physical activity interventions, physical activity plus healthy lifestyle education, and dietary interventions on future success, such as years of schooling, college enrolment or income for children and adolescents with obesity or overweight.

Obesity indices
We assessed the e ects of behavioural interventions on change from baseline in BMI z-scores, total body fat and waist circumference for studies that provided suitable data. We reported the e ect estimates on obesity indices by the following subgroups: 1. Beneficial intervention e ect on school achievement; 2. No beneficial intervention e ect on school achievement; 3. Beneficial intervention e ect on cognitive functions; 4. No beneficial intervention e ect on cognitive functions.
We performed this data synthesis descriptively, rather than combining the e ect estimates of individual studies, because of substantial di erences in intervention and outcome characteristics. We calculated individual study between-group mean di erences where unpublished data were made available.

Body mass index (BMI) z-scores
Six studies (two RCTs and four cluster-RCTs) provided change-frombaseline BMI z-scores (

Figure 5. Forest plot of comparison: 4 Lifestyle intervention versus control, outcome: 4.1 BMI z-score.
There was no evidence of a beneficial e ect on change in BMI z-scores of a school-based physical activity intervention (Movi-Kids, Sánchez-López 2017 [pers comm]) compared to standard practice, despite increased school and cognitive outcomes (e.g. general intelligence) in favour of the intervention group. In fact, the change in BMI z-score was in favour of the standard practice condition (MD 0.19, 95% 0.00 to 0.38; 62 children). In contrast, the a er-school physical activity intervention by Davis 2011b resulted in reduced BMI z-scores in favour of the intervention group (MD -0.12, 95% CI -0.17 to -0.07; 116 children); the intervention resulted in improved mathematics attainment and cognitive functions (composite executive functions, non-verbal memory) in the intervention group compared to the controls.
The physical activity plus health education intervention by Huang 2015 (Odense Overweight Intervention Study) was e ective in reducing BMI z-scores in favour of the intervention group at both follow-up time points. At six-week follow-up, the intervention resulted in improved cognitive outcomes (visuo-spatial abilities) Cochrane Database of Systematic Reviews and reduced BMI z-score (MD -0.44, 95% CI -0.54 to -0.34; 94 children). At 13-month follow-up, there was no evidence of improved cognitive outcomes and, on average, children increased their BMI z-score but less in the intervention group compared to standard practice (MD -0.20, 95% -0.34 to -0.06; 86 children). Both intervention arms of the complex physical activity plus healthy education intervention by Treu 2017 (ASCEND) resulted in no beneficial e ect on BMI z-scores in children with obesity or overweight compared to standard practice. This study also showed no beneficial e ect on school achievement in favour of the intervention. Similarly, there was no evidence of a beneficial e ect either on cognitive function (attention) or on post-intervention BMI z-scores in Wirt 2013 [pers comm] (MD 0.34, 95% CI -0.01 to 0.69; 30 children).

Physical activity, diet and other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight (Review)
One dietary intervention, which resulted in improvements in school achievement reported a small reduction in BMI z-scores change by 0.06 in favour of the intervention group (95% CI -0.12 to 0.00, 321 children; Johnston 2013). Another dietary intervention, which indicated no intervention benefits for school achievement or concentration performance, suggested a small reduction in BMI zscore change by 0.08 in favour of standard practice (95% CI 0.01 to 0.15, 93 children; Damsgaard 2017 [pers comm]).

Total body fat percentage
We included three studies (see Figure 6; Analysis 4.2); one RCT (Chen 2016) and two cluster-RCTs (Gallotta 2015; Huang 2015). We estimated the e ective sample size of Gallotta 2015 using an ICC of 0.01 (Berry 2012). One study, which showed improved cognitive function in some domains of children with obesity or overweight a er a six-week day camp, indicated a mean reduction of 5.2% (95% CI -8.1% to -2.3%, 94 children) in total body fat in favour of the intervention compared to standard practice (Huang 2015). The statistically significant beneficial e ects on cognitive functions and total body fat disappeared at 13-month follow-up (MD -2.90% 95% CI -6.19% to 0.39%, 86 children).
The two physical activity-only interventions resulted in conflicting findings (Chen 2016; Gallotta 2015). Although both interventions suggested no beneficial e ects on cognition in favour of the intervention group, Chen 2016 reported that the mean percentage body fat was 3.43% lower (95% CI -5.38% to -1.48%, 50 children) in the intervention group compared to standard practice/wait-list control. Neither intervention arm in Gallotta 2015 showed evidence of a reduced total body fat compared to standard practice.

Waist circumference
Suitable data on change of waist circumference in children with obesity or overweight were available from only one study (Huang 2015). Improvements in cognitive function in favour of the intervention coincided with beneficial changes in waist circumference at six-week follow-up (MD -5.4 cm, 95% CI -7.4 cm to -3.5 cm; 94 children). At 13-month follow-up no beneficial e ects on cognition or waist circumference were detected (MD -2.0 cm, 95% CI -4.5 cm to 0.6 cm; 86 children).

Summary of main results
We identified five RCTs and 13 cluster-RCTs evaluating the e ectiveness of physical activity, dietary or other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight. Eight studies o ered a physical activity-only intervention, seven studies combined physical activity with healthy lifestyle education, and three studies implemented a dietary intervention.

Physical activity only interventions
Based on a single study, there was high-quality evidence for improvements in mean composite executive functions and nonverbal memory when compared to continuation of usual activities. O ering school-based extracurricular activities in combination with the restructuring of the playground environment indicated large benefits in mean general intelligence scores compared to standard practice. This finding was of moderate quality. No beneficial e ects of physical activity interventions compared to standard practice were evident for mathematics, reading and language achievement, inhibition control, attention, cognitive flexibility, or visuo-spatial abilities. The evidence of no e ect was of Cochrane Database of Systematic Reviews moderate quality for mathematics and reading achievement and of very low quality for inhibition control.

Physical activity plus healthy lifestyle education
Combined physical activity and healthy lifestyle education interventions resulted in no improvements in the average achievement across subjects taught at school, mathematics achievement, reading achievement, health class achievement, inhibition control, attention, visuo-spatial abilities and non-verbal memory. The quality of the evidence of no e ect was low to very low for all school achievement and cognitive outcomes.

Dietary interventions
Interventions targeting the improvement of the school food environment in conjunction with nutrition education resulted in a moderate di erence in average achievement across subjects taught at school compared to standard practice in adolescents with obesity, but not in adolescents with overweight. However, the evidence was of low quality. There was no evidence that replacing packed school lunch with a diet rich in berries, root vegetables, whole grains and seafood (New Nordic Diet) improved attention, mathematics or reading achievement in children with obesity or overweight. This finding was also of low quality and further research is very likely to change the e ect estimates.

Change in obesity by intervention e ectiveness on school or cognitive outcomes
Based on our descriptive analysis, we were not able to detect a conclusive pattern linking improved school or cognitive outcomes with a reduction in obesity. Three studies indicated that highlyintense interventions that involve daily exposure to physical activity or nutrition education, or both, can result in both significant change in obesity indicators and cognitive and academic outcomes compared to standard practice. However, one high-intensity study that indicated a significant reduction in total body fat did not result in improved cognitive outcomes for the intervention group. Another study showed improved school attainment and cognitive functions but benefits on BMI z-scores were not evident in the intervention group compared to standard practice.
The absence of an e ect on school achievement or cognitive outcomes, or both, might be attributable to poor adherence to the experimental condition, particularly when the intervention was applied in participants' homes (e.g. physical activity homework tasks). Assessment of participants' compliance with the intervention was o en poorly reported. We observed a similar bias for assessment of adherence to the control condition. Most studies did not attempt to evaluate or report whether the control group maintained its 'standard practice' during the trial period. For example, changes in school policy concerning healthy lifestyle factors such as improved school meals or physical activity opportunities during recess could potentially bias the intervention e ects of experimental trials. The same may account for engagement in lifestyle changes at the family or child level.
The included studies provided no evidence of harm in terms of deterioration in any of the cognitive or school achievement outcomes. No data currently exist on whether lifestyle interventions for weight management of children and adolescents with obesity or overweight influence the need for additional educational support and indices of future success once schooling has been completed.

Overall completeness and applicability of evidence
Our population group of interest -children and adolescents with obesity or overweight -is a very specific yet substantial and globally increasing subgroup of the general population. Of the 18 included studies, the study population of only six studies were children or adolescents with obesity or overweight. Most of the identified studies aimed to prevent obesity in the general population and did not report cognitive and academic outcomes of the subgroup with obesity or overweight separately from those of children in the healthy weight category. This was surprising, since 11 of the 18 studies stated cognitive function or school achievement among their primary outcomes. Despite our e orts to obtain them, the subgroup data for some studies have not been available to date (see Characteristics of studies awaiting classification). Based on our assessment of the quality of the evidence, we are confident that further research is likely to influence the estimates of the intervention e ects for all assessed outcomes (see Quality of the evidence). Overall, the results of this review suggest applicability of the findings for public health practice for some but not all assessed outcomes (see Implications for practice).
Most of the included studies were conducted in the primary/ elementary school setting. Only two studies contributed to the evidence on intervention e ects in preschool-aged children and five studies targeted adolescents enrolled in secondary (junior high/high) school. One plausible reason for this imbalance might be that primary/elementary-school-aged children seem old enough to understand instructions and young enough to comply with the intervention protocol. The influence of puberty on cognitive development might also contribute to more researchers focusing on pre-pubertal adolescents (Juraska 2014). However, the developmental trajectories of cognitive abilities related to school achievement span preschool age and late adolescence (Boelema 2014;Davidson 2006;Waber 2007), and di erential e ects of behaviour change interventions at di erent ages are plausible. The overall low number of studies included for each outcome did not allow us to formally test the e ectiveness of the intervention by age group. Nevertheless, we identified two ongoing trials in preschoolaged children (Po'e 2013; Stanley 2016) and two in adolescents (Bau 2016; RBR-38p23s) which assessed intervention e ects on cognitive and academic outcomes in participants with obesity or overweight.
All but one (Barbosa Filho 2017 [pers comm]) of the included studies were conducted in high-income countries and most studies (14/18 studies) included children primarily from middle-income families. The reported evidence might therefore not be applicable to low-and middle-income countries. In addition, a potentially di erential e ect of physical activity, diet and other behavioural interventions on cognition and school achievement of children with obesity or overweight growing up in a socio-economically deprived environment remains to be investigated. The evidence on the association between obesity and poverty (Hardy 2017; Lissner 2016; Wang 2012), and the associations between lower education and cognitive skills and poverty (Cooper 2013; Marteau 2013) support further e orts in identifying who could benefit most from obesity-related health behaviour interventions.
There was no evidence available on the e ect of interventions targeting the quality and duration of sleep, or sedentary behaviour, . In addition, one feasibility trial is currently ongoing, testing whether reduced sitting time in school can improve cognitive outcomes ('Stand Up For Health' study, trial register: ACTRN12614001001684). Although this trial listed obesity/ overweight as a target health condition, the trial register entry did not mention assessment of change in adiposity.
Although two studies provided outcome data for two follow-up time points (De Gree 2016; Huang 2015), the data related to mid-and immediately post-intervention. Participants in Huang 2015 received a low-intensity maintenance intervention a er completion of the intense six-week day camp. We therefore could not fully explore the retention e ect of interventions for weight management on school achievement and cognitive functions in children and adolescents with obesity or overweight. Nevertheless, the findings of Huang 2015 indicate that the beneficial e ect of the day-camp intervention compared to attention control on visuospatial abilities was not maintained a er completion of the 13month family-based maintenance intervention.
It remains unclear whether changes in academic and cognitive abilities were connected to changes in indices of obesity, due to the small amount of suitable data, and variations in study architecture (Davis 2011b; Huang 2015; Johnston 2013).

Quality of the evidence
We separately assessed the quality of evidence of the most important outcomes for decision-making for each comparison of the three intervention types (see Summary of findings for the main comparison; Summary of findings 2; Summary of findings 3).
The quality of evidence for Comparison 1 -physical activity-only interventions compared to standard practice -was high to very low. The reason for downgrading the evidence on mathematics achievement and reading achievement was a high risk of attrition bias. The attrition rate was 14% to 16% in most of the studies contributing to the evidence. No imputation of missing data was performed and we found higher attrition in the comparison condition compared to the intervention group. We downgraded the quality of evidence for inhibition control by three levels, for high risk of attrition and selection bias and for imprecision. Missing outcome data were not accounted for and the sample sizes were 31 participants for language achievement and 84 for inhibition control. For inhibition control, the method of randomisation was unclear, with a high risk of bias in the comparability of groups at baseline.
The quality of evidence for Comparison 2 -physical activity plus healthy lifestyle education intervention compared to standard practice -was low to very low. We downgraded the quality of evidence for mathematics achievement by three levels for inconsistency of the e ect estimates, imprecision of the e ect estimate and methodological shortcomings related to a high risk of bias for sequence generation, blinding of outcome assessors and attrition. We downgraded the quality of evidence for reading by two levels, for e ect estimates of reading achievement being inconsistent between studies and for risk of bias in the study methodology. One of the two included studies was at high risk of bias for sequence generation and one study was at high risk of attrition bias, with 29% of incomplete outcome data. We downgraded the quality of evidence for inhibition control by two levels because we detected selection bias and a high risk of attrition bias characterised by twice as much missing outcome data in the control group compared to the intervention group (26% versus 13%).
The quality of evidence for Comparison 3 -dietary intervention compared to standard practice -was low for average achievement across subjects taught at school, mathematics and reading achievements and attention performance. We downgraded the quality of evidence for average achievement across subjects taught at school by two levels for methodological shortcomings in blinding of the outcome assessor and for attrition bias (21% to 29%). We downgraded the quality of the remaining outcomes by two levels for imprecision (the sample sizes ranged between 76 and 61 children) and for not blinding the outcome assessor.

Potential biases in the review process
We searched 17 electronic databases, two trial registers and handsearched one journal to identify published and ongoing trials. We also contacted 15 trial authors to obtain unpublished data and obtained unpublished outcome data from 12 studies. However, we acquired adverse-events data only from published records.
Nevertheless, we intended to review evidence in a specific subgroup of the general population; the following limitation should therefore be considered. The unpublished data provided by the study authors were extracted for a subgroup of the total study sample, leading to overall small sample sizes for inclusion in this review. This might have a ected the studies' power to detect an intervention e ect. Studies which provided unpublished data for the subgroup of children with obesity or overweight may have been powered for the total study sample.
Included studies used a wide range of school achievement and cognitive function test tools. Previous reviews, such as that of Smith 2011, suggest that obesity might have a detrimental impact on some aspects of cognition, so we conducted a categorisation of outcome measures. The use of composite scores in some studies precluded more fine-grained synthesis. For example, the planning subscale of the cognitive assessment system (CAS) is a composite score from three di erent measures of executive function, none of which are comparable to more traditional measures of planning such as the Tower of London task. As composite scores were reported in some cases, we categorised outcome measures as 'general executive function', rather than more discrete aspects of executive function (e.g. inhibition). Alternative categorisation of cognitive outcomes might impact on the conclusions drawn. Even though there tend to be correlations between cognitive function tests (because of the general cognitive factor g), di erent cognitive tests vary in their specificity for di erent cognitive domains. Moreover, successive testing before and a er the intervention is likely to improve participant scores through repeated measures and regression to the mean. Thus, an improvement may not be due to the intervention, although the use of change-from-baseline data and the use of a comparison group allows some control for this.

Cochrane Database of Systematic Reviews
On the other hand, small participant numbers limit the ability to minimise bias.

Agreements and disagreements with other studies or reviews
To our knowledge, no primary studies have been conducted other than those reviewed here. The systematic literature review by Bustamante 2016 narratively synthesised the evidence of physical activity interventions on cognitive and academic outcomes in youth with obesity or overweight. The authors included quasiexperimental and randomised controlled trials published in peerreviewed journals before December 2015. The literature search was conducted in three selected electronic databases (PubMed, Journals@OVID, and Web of Science). Five RCTs were included in the review, of which we include four in this Cochrane Review (Davis 2011b; Huang 2015; Kra 2014; Staiano 2012). We did not include the remaining RCT (Crova 2014) because the study did not aim to modify body weight status and so was not eligible for inclusion. The quasi-experimental studies included in Bustamante 2016 were uncontrolled single-group trials, case-control studies or short, single-session (acute bout) physical activity interventions, which we did not consider eligible for inclusion in this review. Bustamante 2016 concluded that, based on a single RCT, regular physical activity was more beneficial for improving executive functions than monthly lifestyle education classes (Davis 2011b). This finding is consistent with our results.
Bustamante 2016 argued that when regular physical activity interventions are compared to an attention control activity that involved organised activities supervised by adults, the beneficial e ect of the physical activity intervention on academic and cognitive outcomes (detected using psychometric test batteries) is outweighed by the attention received in the comparison group (Kra 2014). Findings of our evidence synthesis suggest otherwise. Studies that compare physical activity interventions with standard practice, which typically also involve organised activities supervised by adults (i.e. teachers), resulted in significant improvements in academic and cognitive outcomes (e.g. Although these systematic reviews may include some children with obesity or overweight, they lack a separate analysis of the e ect estimates in our population groups of interest. Research suggests a greater benefit of obesity-related health behaviour interventions in children with obesity or overweight compared to children with healthy weight (Crova 2014; Grieco 2009; Vazou 2014). These reviews are therefore not directly comparable with our review.

Implications for practice
This review provides some evidence that interventions which promote physical activity may be e ective in producing small improvements in composite executive functions and non-verbal memory in primary/elementary school-aged children with obesity or overweight specifically. However, this evidence is based on a small number of studies. On current evidence, we are unable to determine the impact of these interventions on school achievement or cognitive skills. The current evidence on the e ectiveness of interventions that combine healthy lifestyle education with physical activity promotion and dietary interventions does not allow us to draw definitive conclusions on their impact on cognitive and academic outcomes. In the absence of data, it is not possible to determine the impact of physical activity, dietary and other behavioural interventions on additional educational support, adverse events or outcomes related to future educational achievements such as years of schooling, employment rates or income.
Evidence on the e ects of physical activity or dietary interventions on school achievement and cognitive functions in children with obesity or overweight conducted in clinical settings (e.g. hospitals, outpatient clinics, primary care) is missing, so we cannot o er implications for clinical practice in settings beyond school and community settings.

Implications for research
We identified studies in school, a er-school and community settings, but we found no evidence on cognitive and academic outcomes of behavioural weight management interventions in a clinical setting. However, our findings indicate beneficial e ects of physical activity interventions on cognitive outcomes, namely cognitive executive functions, in children with obesity or overweight. Cognitive executive functions have been associated with the ability to control food intake (Bartholdy 2016; Jansen 2015) and engagement in health behaviour (Hall 2014). Child and adolescent weight management programmes in a clinical setting should include measurements of cognitive outcomes for two reasons. Firstly, the most e ective strategies for weight management could be informed when linking cognitive abilities with behaviour change. Secondly, children with obesity or overweight are the target population of weight management programmes in clinical settings. If studies of interventions in clinical settings were to include measures of cognitive outcomes and related school achievement, these would help to boost the power of studies to identify potential gains in these areas. Similarly, community-based interventions which directly target children and adolescents with obesity and which assess cognitive and academic outcomes are needed to advance the evidence. In addition, the availability of larger studies might allow the assessment of a di erential intervention e ect for participants with overweight and participants with obesity in relation to school achievement and cognitive functions.
In terms of the targeted obesity-related health behaviours, evidence was available for solely physical activity interventions, physical activity plus healthy lifestyle education interventions and dietary interventions, which also included nutrition education. Our findings suggest that interventions focusing on one Cochrane Database of Systematic Reviews target behaviour, i.e. physical activity, yielded beneficial e ects on composite executive functions, non-verbal memory and general intelligence compared to standard practice. In contrast, interventions targeting several health behaviours through healthy lifestyle education and active physical activity programmes did not result in beneficial e ects on these outcomes compared to standard practice. It might be that the positive e ect of the physical activity programme on those cognitive functions is diluted with increasing complexity of the interventions. The intensity of the physical activity component might be reduced when additional intervention activities, such as healthy lifestyle education sessions, are implemented. Adjustments to the duration and frequency of physical activity programmes might have been required to keep the burden on the school personnel manageable. While interventions with multiple strategies appear successful for obesity prevention and treatment (Al-Khudairy 2017; Colquitt 2016; Mead 2017; Waters 2011), a su icient intensity and quality of the e ective intervention components might be required for improving cognitive functions. We were not able to provide a similar observation with dietary intervention because none of the included studies applied an intervention without an additional nutrition education programme.
Given the importance of adequate physical and cognitive development of young children for their later life, further evidence is needed on the e ectiveness of physical activity, dietary and other behavioural interventions on cognition and school achievement in the preschool years. In addition, the evidence is insu icient for adolescents who have reached puberty. The e ectiveness of obesity-related behaviour change interventions on cognition and school achievement in this age group is of particular importance, because of the direct implications for adult health and socioeconomic success of the individual and the nation. The extent to which sex and ethnicity influence the e ect of physical activity and dietary interventions on cognition and school achievement in children and adolescents with obesity or overweight remains unknown, and should be addressed in future research.
Future multicomponent obesity prevention and treatment programmes should consider implementing physical activity programmes which are e ective in improving cognitive functions or school achievement.
Further research is needed in low-and middle-income settings, to establish whether there are di erential intervention e ects on cognition and school achievement for children and adolescents with obesity or overweight living in socio-economically deprived environments. The educational, societal and economic argument for implementing e ective childhood obesity prevention and treatment programmes could be substantial.
Longer-term follow-up trials are needed to determine whether improvements in school achievement and cognitive function are sustainable over time and thus a ect future success. High rates of loss to follow-up assessment are a common problem in lifestyle interventions, particularly those involving children and adolescents with obesity or overweight. To reduce the risk of attrition bias, researchers might wish to consider methods to impute missing outcome data in their analysis and to report characteristics of and reasons for missing data.
Including brain-imaging techniques might enable researchers to detect beneficial e ects on cognition which are not detectable using psychometric tests of academic and cognitive abilities. Finally, more multivariate research is needed to further investigate associations, two-way interactions and causal pathways between childhood obesity, lifestyle behaviour, cognitive abilities and academic outcomes.

A C K N O W L E D G E M E N T S
We are very grateful to the trial authors who provided unpublished outcome data and additional study details. We would like to thank Dr Jung Woo (

Muza ar 2014 {published data only}
Muza ar H, Chapman-Novakofski K, Castelli DM, Scherer JA. The HOT (Healthy Outcome for Teens) project: using a webbased medium to influence attitude, subjective norm, perceived behavioral control and intention for obesity and type 2

Francis 2009
Francis LA, Susman EJ. Self-regulation and rapid weight gain in children from age 3 to 12 years. Archives of Pediatrics

NICE 2013
National Institute for Health and Care Excellence. Managing overweight and obesity among children and young people: lifestyle weight management services. https://www.nice.org.uk/ guidance/ph47/resources/costing-report-69149341 (accessed 9 January 2018).

Methods
Study design: Cluster-randomised controlled trial Unit of randomisation: Schools, stratified by size and geographical location.  Reasons for attrition: Children moved schools or were absent on the day of testing (5 times higher in intervention than control school), school chose not to send participants` test results to the CAT-3 test centre for scoring (control school), school administered the wrong test at follow-up (intervention school) Interventions

Comparison: Action Schools! BC versus standard practice
Intervention: Action Schools! BC is a comprehensive, multicomponent intervention providing tools for schools and teachers to use in promoting physical activity and healthy eating in different settings. These include the school environment (healthy eating posters), scheduled Physical Education, classroom action, family and community (e.g. walking school bus), extracurricular activities (e.g. dance club) and school spirit (e.g. Hike across Canada challenge). Extracurricular and school spirit activities were provided by only a small number of intervention schools 1. Physical activity: Classroom-and/or school environment-based physical activity for 15 minutes a day on 5 days/wk delivered by trained classroom teachers. Activities included hip hop dancing, skipping, jumping, chair aerobics, yoga and strength work. This activity was provided in addition to 40 minutes of Physical Education twice a week to engage children in 150 minutes of physical activity/wk. Compliance with intervention was assessed by the classroom teacher through daily physical activity logs reporting type, duration and frequency 2. Nutrition: Across the different settings, a fruit and vegetable (F&V) intervention was employed that focused on increasing intake of F&V; improving knowledge, attitudes and perceptions regarding F&V; and strengthening willingness to try new F&V

Cochrane Database of Systematic Reviews
Random sequence generation (selection bias) Unclear risk Quote (from report): "Schools were then remotely randomized..." Quote (from email correspondence): "randomisation was done by random number draw by a third party " Judgement comment: Although the method of randomisation is appropriate to reduce selection bias, baseline differences in school achievement between intervention and control schools occurred. Baseline imbalances are a risk in cluster RCTs and might indicate inappropriate randomisation of clusters. However, it remains unclear whether the imbalances occurred by chance.

Allocation concealment (selection bias)
Low risk Quote (from email correspondence): "Yes, the 10 schools were randomized at once" Blinding of participants and personnel (performance bias) All outcomes Unclear risk Quote (from email correspondence): "The primary purpose of Action Schools! BC was not to improve academic performance" Comment: Blinding of children and personnel regarding the experimental condition is not possible in a lifestyle intervention. Email correspondence with authors confirmed that participants and personnel were blinded to the true purpose of the study Judgement comment: Differences in baseline characteristics between experimental groups were not significant besides school achievement scores, which were higher in the control school than in the intervention school Cross-contamination Low risk Quote: "We recruited elementary schools from the Vancouver and Richmond school districts in British Columbia, Canada...Schools were stratified by size (< 300 or > 300 students) and geographic location (Vancouver or Richmond)." a) Training and activities in PE classes: structure predominantly active PE classes, even in classes with a theoretical content. The manual included 4 units: (i) PA and health (e.g. PA and leisure, co-operative games, PA with parents); (ii) health factors (e.g. sedentary time, diabetes and hypertension, quality of life); (iii) sports (e.g. athletics, volleyball, functional training, combat sports); and (iv) popular games (e.g. games,dancing and adventure sports) b) Active opportunities in the school environment: Supervised 10 -15-minute sessions, called "Gym in School", were performed twice a week. These sessions were composed of physical (e.g. stretching, located exercises) or dynamic (e.g. games and rhythmic activities) activities in small and large groups. Space and equipment were structured and made available to play games in free time during the school day

Healthy lifestyle education:
a) Training and activities in the general curriculum: The manual included proposals for activities according to knowledge areas (i.e. languages, social sciences, natural sciences and mathematics). For example, in mathematics, there was a proposal about teaching quantities and measures using body measurements, energy expenditure in physical activity or energy consumption in meals. Teachers were encouraged to undertake the activities or to create and implement similar strategies in the classroom during the semester Cochrane Database of Systematic Reviews b) Health education in the school community: Pamphlets with messages about active and healthy lifestyle were distributed. 3 pamphlets were directed at students: (i) PA and health; (ii) screen time and health; and (iii) healthy eating and healthy behaviours. 2 pamphlets were directed at parents: (i) PA and parents/the family; and (ii) screen time and parents/the family Standard practice: "Schools from the control group underwent 1 semester with the regular and conventional activities of a full-time school. In general, the control schools had 2 weekly Physical Eduction classes that included content and activities according to the perspective of their teachers."

Physical activity, diet and other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight (Review)
Outcomes Outcome 1: School achievement: Academic performance was evaluated considering scores from standardised tests during a school year in 2 areas: Mathematics and Language (Portuguese). The crude scores (ranging from 0 -10 points) of each student were obtained from the schools and organised by semester to indicate the pre-intervention period and during/post-intervention. Z-scores for Mathematics and Language by school and grade were calculated and provided for inclusion in this review Notes 1. The authors kindly provided unpublished data for children with obesity or overweight 2. Sample size calculation was performed for total sample (children with normal weight and overweight/obesity) 3. Funding/Sponsor: Universidade Federal de Santa Catarina, Universidade Federal do Ceara, Secretaria Municipal de Educação de Fortaleza, Ceará, Brazil

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "Thus, we performed the random selection of three schools to each condition" Quote: "This is a cluster-randomized controlled trial and the school was the sample selection unit." Judgement comment: Described as cluster but no methods described as to how this was achieved. Unclear how and if a random sequence was generated Allocation concealment (selection bias)

Unclear risk Judgement comment: No information provided
Blinding of participants and personnel (performance bias) All outcomes High risk Quote: "The first stage was a four-hour training input that took place at the beginning of the school semester. There were discussions of primary health concepts and the importance of these issues including the relationship between health, school and academic performance." Judgement comment: This quote suggests that the teachers were aware of the potential impact of the intervention on academic performance. True blinding of participants and personnel is not possible for this kind of intervention. It is unclear if the participants were aware of the potential of the intervention influencing academic performance Blinding of outcome assessment (detection bias) All outcomes High risk Quote: "The notes and standardized scores of the students were obtained from the schools and organized by semester to indicate the pre-intervention period (the first semester of 2014) and during/post-intervention (the second semester of 2014)." Judgement comment: The schools providing the academic achievement scores were aware of the group allocation Incomplete outcome data (attrition bias) All outcomes Low risk Quote: "The main reason for non-participation at baseline was being absent from school on data collection days. Post-intervention data collection included 1,085 students (response rate of 93.2 % and 90.4 % in intervention and con- Reasons for attrition: Intervention group n = 3 could not stand the intensity, n = 3 not available on measurement day; control group n = 3 not available on measurement day

Physical activity, diet and other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight (Review)
Attrition rates: 19% in intervention group, 14% in control group

Comparison: Aerobic group physical activity versus wait-list control
Intervention: Group physical activity programme including multiple types of moderate-intensity exercises performed 4 times/week for 40 minutes a session (5 minutes each for warm-up and cool-down, 30 minutes for the main exercise). "The participants were instructed to attend an instructor-monitored physical activity programme with an instructor: participant ratio of 1:10 at 1 of 3 times (i.e. morning, lunch break, or after school) during the school day." "Each participant received a physical activity manual that consisted of 3 sections: instructions for the warm-up, descriptions of the exercises, and a daily exercise log. The participant was free to choose 1 of the provided exercise types (e.g. fast walking, stair climbing, jumping rope, or aerobic dancing) each time, with an emphasis on maintaining a moderate intensity of 60% to 70% of the maximal heart rate (220 minus age). The target heart rate was progressively increased based on each participant's ability to meet the optimised target heart rate. Daily recording of the characteristics of the exercise performed and of adverse events in the exercise log was also performed." Wait-list control: Participation in regular health education course following randomisation; opportunity was given "to participate in a similar physical activity programme after the intervention duration"

Outcome 1: Cognitive flexibility (set shifting) measures using the computer version of the Wisconsin card sorting test (Version 4-Research Edition)
Outcome 2: Total body fat: Body fat was assessed using a Karada Scan body composition monitor (HBF362, Omron, Kyoto, Japan)

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "The participants were randomly assigned to a physical activity program or a wait-list control group using random allocation software." Allocation concealment (selection bias)

Unclear risk Judgement comment: No information reported
Blinding of participants and personnel (performance bias) All outcomes

High risk
Quote: "The current study employed a randomized controlled trial (RCT) design to determine the effects of a physical activity intervention on the set-shifting aspect of executive function in obese young adolescents." Judgement comment: True blinding for a physical activity intervention is not possible. Whether the participants were blinded to the true purpose of the study, in relation to executive function and anthropometrics, is unclear. How- Cochrane Database of Systematic Reviews ever, it is very likely that the personnel was aware of the study aims. Physical intervention and waiting list control within the same school means sta and pupil blinding to intervention not possible Blinding of outcome assessment (detection bias) All outcomes

Unclear risk Judgement comment: No description given
Incomplete outcome data (attrition bias) All outcomes High risk Judgement comment: Substantive attrition from both groups; data from outcomes are reported but intention-to-treat analysis is not discussed. Attrition after randomisation, before baseline assessment: intervention = 20%, control group = 14%. Therefore, unclear risk of bias Reasons for attrition: Withdrew during the study mainly because they changed school or class, disliked or found the measurements too time-consuming, or disliked the intervention school meals. The proportion of children who withdrew from the study was not different between the two clusters Interventions

Comparison: OPUS School Meal (New Nordic Diet) versus standard practice
Intervention: "The New Nordic Diet (NND) contains seasonal, health-promoting ingredients, for example, berries, root vegetables, whole grains, fish, shellfish, seaweed and rapeseed oil. Diet contains less meat than average Danish diet." "Children received daily servings of a mid-morning snack, ad libitum hot lunch meal and afternoon snack (twice/week fresh fruit, dried berries or both, and nuts and muesli bar or bread roll). The meals met 40% to 45% of daily energy intake based on energy requirements of 11-year-old children". The meals were produced locally at each school by trained chefs and kitchen personnel hired for the study. School lunch breaks were increased from 15 minutes to 20 -25 minutes. "The children were encouraged to taste everything and to keep a reasonable plate distribution with vegetables and starchy foods filling the majority of the plate." "Each child spent 3 -5 school half-days during the NND period in the kitchen cooking, presenting, and serving the menu of the day to the other children." The teachers were encouraged to participate in the lunch meals. "Class teachers were given a box of teaching materials about the human body, the clinical measurements, and taste sensorics, including background information about NND and suggestions for related educational activities and games." Use of the material was optional.

Standard practice: Usual packed lunch
Outcomes Outcome 1: School achievement: Teacher-assessed mathematics and reading proficiency using agespecific Danish standardised tests Outcome 2: Cognitive function: Assessment of attention using the D2 Test of Attention. Unclear who administered the test Notes 1. The authors kindly provided unpublished data for children with obesity or overweight 2. Follow-up data after cross-over period at 6 months not included in this review 3. Power calculation performed for total study sample based on metabolic syndrome test score 4. Funding source: Nordea-fonden

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "Within each of the two blocks, schools were allocated to the order of treatment and control for third and fourth grades by simple randomisation. The randomisation list was performed by a statistician not involved in data collection or analysis using the statistical software package R (R Foundation for Statistical Computing)" Allocation concealment (selection bias) Low risk Quote: "Cluster randomisation was performed before the children were invited for participation."

Cochrane Database of Systematic Reviews
Blinding of participants and personnel (performance bias) All outcomes High risk Quote: "The allocation order was not blinded to investigators, schools or participants." Quote from the study protocol (Damsgaaard 2012): "OPUS School Meal Study was a cluster-randomised controlled unblinded cross-over study." Blinding of outcome assessment (detection bias) All outcomes High risk Quote: "The allocation order was not blinded to investigators." Incomplete outcome data (attrition bias) All outcomes Unclear risk Quote: "A total of sixty-nine children (8.3 %) withdrew during the study mainly because they changed school or class (n 29), disliked or found the measurements too time-consuming (n 17), or disliked the intervention school meals (n 13). The proportion of children who withdrew from the study was not different between the two clusters (intervention -control 10.2 % v. control -intervention 6.5 % of the participants, P< 0.054)." Judgement comment: The figures refer to the total study sample which included children with healthy weight. The authors confirmed during email correspondence that no intention-to-treat analyses were performed. The attrition rates were low, with 7% and 4% in intervention and control group, respectively. As indicated by the quote, attrition did not differ between the experimental conditions Selective reporting (reporting bias) Low risk Quote: "The study protocol is registered at www.clinicaltrials.gov (NCT 01457794)." Quote: "A comprehensive description of the study design and recruitment of schools and participants has been provided previously." Intervention: Aerobic exercise group for 40 minutes a day, 5 times a week, over a mean total of 13 weeks. 5-minute warm-up phase consisting of brisk walking and static and dynamic stretching. "Activities were selected on the basis of ease of comprehension, fun and eliciting intermittent vigorous movements. Children were encouraged to maintain a heart rate > 150 beats/minute during running games, tag games, jump rope, modified basketball and soccer." No competition or skill enhancement. Intervention session ended with a cool-down including such activities as water break, slow walking and static stretching. "The intervention was delivered by qualified and trained research sta in an after-school programme at the gymnasium of the Georgia Prevention Institute." Compliance was assessed by observing and recording attendance and average heart rate daily for each child.
This study included a 2nd intervention group, which was not included in this review (see Notes) Standard practice: Continuation of usual activities "All participating families were offered a monthly lifestyle education class covering the topics of healthy diet, physical activity and stress management." Outcomes Outcome 1: School achievement: Broad mathematics and reading skills on the Woodcock-Johnson Tests of Achievement III

Cochrane Database of Systematic Reviews
Outcome 2: Cognitive function: Subcales for planning, attention, simultaneous successive use of the Das-Naglieri-Cognitive Assessment System. Both tests were administered by a qualified psychologist and personnel with graduate training in psychological assessment

Outcome 3: Obesity indices: Quote "Body weight (in shorts and t-shirt) and height (without shoes)
were measured with an electronic scale (Detecto, Web City, MO) and stadiometer (Tanita, Arlington Heights, IL) and converted to BMI and a BMI z-score (Epi Info, Centers for Disease Control and Prevention, Atlanta, 2003)" Notes 1. Sample size calculation performed. 62 participants per group were estimated to provide 80% power to detect a difference between groups of 6.6 units 2. The 2nd intervention arm included a 20-minute physical activity intervention followed by 20 minutes of sedentary activities, such as board games, card games and drawing (low-dose intervention arm). This intervention group was excluded because the sedentary activities might have affected cognitive function without being defined as lifestyle interventions 3. Funding sources: National Institutes of Health, State of Georgia Biomedical Initiative grant to the Georgia Center for Prevention of Obesity and Related Disorders, Medical College of Georgia and University of Georgia

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote (from report): "... children were assigned randomly by a statistician..." Quote (from the report Davis 2012): "...each participant was assigned a uniform (0,1) random number [...] within their respective ethnicity and sex group. If the number was between 0 and 0.33 the child was randomised to the low-dose group; between 0.34-0.67, to the high-dose group; and above 0.67, to the control group" Allocation concealment (selection bias) Low risk Quote (from email correspondence): "I ensured allocation concealment by not permitting randomization by the statistician until baseline testing was completed. Only then were they randomized and their assignments communicated to the study coordinator, who informed the families." Blinding of participants and personnel (performance bias) All outcomes Unclear risk True purpose of the study was blinded by advertising it as "trial of aerobic exercise on child's health" (quote from report) Judgement comment: Blinding of children and personnel regarding experimental condition is not possible in a physical activity intervention Blinding of outcome assessment (detection bias) All outcomes Low risk Quote (from report): "...Outcome assessors were unaware of child's experimental condition..." Incomplete outcome data (attrition bias) All outcomes

Low risk Judgement comment: Provided participant flow chart indicated similar missing data in intervention and control groups
Quote (from report): "Analyses were conducted using the last observation carried forward imputation for the [...] children who did not provide posttest data" Selective reporting (reporting bias) Low risk Judgement comment: All previously reported outcomes were reported

Comparability of baseline groups
Low risk Judgement comment: Random sequence adequately generated  Cochrane Database of Systematic Reviews performing a basic exercise, such as jogging, hopping in place or marching. A specific exercise was performed when the children solved an academic task. For example, for mathematics, children had to jump 8 times to solve the multiplication '42'. For language, children had to perform a squat for every spelled letter in the word 'dog'. After performing the specific exercise, children had to continue performing the basic exercise until the next academic task was shown."

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "A second or third grade class from each school was randomly assigned to serve as an intervention group....The class that was not assigned to the intervention group was automatically classified as the control group." Quote from de Greef et al 2016b: "Randomization was performed by the national Bureau for Economic Policy Analysis that was not involved in the study."

Judgement comment: No indication of the methodology used. Imbalances in baseline differences between intervention and control participants might indicate inappropriate randomisation
Allocation concealment (selection bias) Unclear risk Judgement comment: The authors do not report details on allocation concealment.
Blinding of participants and personnel (performance bias) All outcomes High risk Judgement comment: Physical intervention and intervention and control classes within the same school mean participant blinding impossible for sta and children Quote: "The aim of this study was to examine the effects of physically active academic lessons on cardiovascular fitness, muscular fitness and EF after 2 years." Judgement Comment: True blinding to a lifestyle physical activity intervention is not possible. Some degree of blinding to the true purpose of the intervention (i.e. improving executive function) if unknown to pupils and teachers. It is unclear whether the pupils and teachers in school were aware of the study aims in relation to executive function Blinding of outcome assessment (detection bias) All outcomes Low risk Quote: "Each instructed researcher received a 2 h training to get familiar with the EF and physical fitness tests and were mostly blinded to the condition children had been allocated to (during 88.6% of the measurements)." Judgement comment: Intention was that outcome assessors were blindedsuccessful for most Incomplete outcome data (attrition bias) All outcomes High risk Quote: "Due to circumstances not related to the intervention, two schools did not start the second intervention period, resulting in a lower sample size at T2 Cochrane Database of Systematic Reviews for both the control and intervention group. A loss of two schools was taken into account during the power analysis."

Physical activity, diet and other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight (Review)
Judgement comment: The proportion of missing data is high at T2 (18-months follow-up) and higher in the intervention group (33%) compared to the control group (24%). No methods of imputing missing data were applied. At mid-term assessment (6-months follow-up), attrition was low overall (0% -8%) Selective reporting (reporting bias) Unclear risk Judgement comment: Data presented at all time points, but pre-published protocol is not reported

Comparability of baseline groups
High risk Quote: "The control group consisted of a higher percentage of third grade children [Chi 2 (1) = 5.22; P = 0.025] and was significantly older [t(497) = 2.24; P = 0.026] due to a difference in number of children within each class. No significant age differences were found when analysing the second and third grade children separately." Judgement comment: The comparability of the experimental groups is at risk of bias.
Cross-contamination Unclear risk Quote: "A second or third grade class from each school was randomly assigned to serve as an intervention group. All children from that class participated in the intervention program. The class that was not assigned to the intervention group was automatically classified as the control group." Intervention: "Both physical activity (PA) interventions differed in type and mode of physical activities in which children were engaged, but they were equivalent in structure, overall duration and intensity, and consisted of two 1-hour sessions/week." "PA interventions were designed by a Physical Educator who supervised 1 of the 2 weekly lessons; the other was conducted by the classroom teacher. The two PA interventions had the same structure, and included 15 minutes of warm-up, 30 minutes of moderate-to-vigorous physical activities, [...] and 15 minutes of cool-down and stretching." 1. Traditional physical activity group: "The traditional group PA intervention consisted of continuous aerobic circuit training followed by a sub-maximal shuttle run exercise. This lesson was focused on the improvement of cardiovascular endurance by performing different types of gaits (e.g. fast walking, running, skipping) without any specific co-ordinative request. The traditional PA lesson provided changes in executive modalities and some variations of intensity designed to promote health, fitness, sensory-motor, social and communicative development." 2. Co-ordinative physical activity group: "The co-ordinative group PA intervention aimed to develop both motor control abilities and perceptual-motor adaptation abilities, by combining demands on gross-motor and manipulative control abilities and perceptual-motor adaptation abilities (particularly kinaesthetic differentiation and response orientation). It consisted of the sport-unspecific use of basketballs in the context of mini-games. The basketballs were used in unconventional ways with varying game rules (e.g. use of foot-eye co-ordination techniques with basketballs). These lessons were focused on the development of psychomotor competences and expertise in movement-based problem-solving through functional use of a common tool (e.g. basketball), and considering various tasks that involved decision-making motor tasks and manipulative ball-handling skills (e.g. bouncing, throwing, receiving a ball, and their combination)."

Control: No details reported
Outcomes Participation in other studies related to risk factors of heart disease, children who follow a special school programme, use of weight-reducing medicine within 3 months before baseline measurements, children with motor skill conditions that hinder participation in the intervention Reason for attrition: 51 out of 55 children who were allocated to the camp programme completed the six weeks. One child was injured before the camp started, one child dropped out, and two children were expelled from the camp.

Comparison: Odense Overweight Intervention Study Day Camp versus standard practice
Intervention: The day-camp intervention comprised 2 parts: "an intensive 6-week day camp intervention and a subsequent 46-week family-based intervention programme (52 weeks in total)." "Participants stayed at a day camp from 7.30 a.m. to 8.30 p.m. for 7 days/week."

Physcial activity:
In the day camp, "children were engaged in physical activity and sports" for at least 3 hours/day, achieving about 90 minutes of moderate-to-vigorous physical activity a day measured by accelerometry. After the day camp, 1 physical activity day was offered as part of the family-based intervention programme

Healthy lifestyle classes:
In the day camp, topics covered were nutrition, physical activity and health, goal-setting, etc. The family-based intervention programme comprised 4 parents-involved meetings targeting daily physical activity and dietary behaviour 3. Diet: In the day camp, 3 meals and 3 snacks were prepared and served according to the national Danish dietary recommendations with no caloric restrictions Standard practice/attention control: The standard intervention consisted of 1 weekly fun-based physical activity session (2 hours duration) for 6 weeks. One health and lifestyle educational session for the parents was delivered by a dietician and physical activity specialist Outcome 2: Obesity indices: change in BMI z-scores based on measured weight and height and calculated based on the IOTF growth charts; change in total body fat mass (%) measured using Dual energy X-ray absorptiometry (DXA) which was performed by an experienced operator on a GE Lunar Prodigy (GE Medical Systems, Madison, WI); waist circumference was measured between the lower costal margin and the lilac crest; hip circumference will be measured at the level of the greater trochanter Notes 1. Funding source: TrygFonden

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "The allocation sequence was generated by sex stratified concealed block randomization (1:1) with a block size of 2 to 6 (random permuted blocks)." Quote from the study protocol (Larsen et al 2014): "The randomization was generated using the web-based software http://www.randomization.com and http://www.random.org"

Allocation concealment (selection bias)
Unclear risk Quote from the study protocol (Larsen et al 2014) : "Due to consideration for the participating families, and to avoid additional dropout on this account, it was necessary to inform participants of allocation three weeks prior to baseline measurements. Thus, allocation concealment was not possible." Judgement comment: Although the authors refer to the term allocation concealment, the term is not used in the same way as the Cochrane 'Risk of bias' tool. What the authors describe is the early knowledge of the allocation which had already occurred Blinding of participants and personnel (performance bias) All outcomes Unclear risk Judgement comment: Participants and personnel were not blinded to the group allocation but both groups received an active intervention and so the control group condition (standard intervention arm) can be considered to some extent as 'attention control'. The protocol paper described the control condition as "shorter-term and less intensive intervention program compare to the day-camp group. This reflected a minimal effort to intervene in the children's lifestyle and did not differ considerably from other initiatives being launched in Danish municipalities." It is unclear if the participants and the personnel were aware of the potential effect of the intervention on executive functions.
Blinding of outcome assessment (detection bias) All outcomes Low risk Quote from the study protocol: "Researchers were blinded at all assessments...." Incomplete outcome data (attrition bias) All outcomes High risk Quote: "The analyses were conducted according to the intention to treat principle. Mixed effects models allow the inclusion of partial data of participants who may have dropped out or who were unavailable to follow-ups. No imputation of data was applied. Maximum likelihood estimation was used for all models." Comparability of baseline groups Low risk Quote: "There were no significant between-group differences on those characteristics at baseline." Cross-contamination Low risk Judgement comment: The experimental conditions varied substantially in terms of setting and timing and so cross-contamination was unlikely.
Other bias High risk Quote: "It is noteworthy that six children who were slightly below the IOTF overweight cut-points at screening were also suggested by the school nurses to participate in the OOIS intervention. This was due to the fact that the nurses thought that these children were at risk of being over-weight. Because the six children were included in the randomization, they were not excluded from the analyses." Judgement comment: Protocol violation for 6/115 participants coupled with high attrition predicts a high risk of bias Intervention: Whole-school lifestyle education programme facilitated by a health professional involving curriculum material taught by trained teachers, school meal modification and nutrition counselling. Compliance with the intervention was assessed through direct weekly observation of teachers by the health professional and verbal self-report from teachers 1. Nutrition/Diet: Modification of school meals towards nutrient-dense food. Nutrition counselling was provided on an informal basis by a school nurse 2. Health lifestyle education: Teachers were provided with 50 integrated lessons-worth of curriculum material aiming to improve healthy diet (increased fruit and vegetable, breakfast, healthy snack, water consumption) and increase physical activity. Teachers were encouraged to teach lifestyle integrated lessons once a week, to conduct health-related activities every 2 weeks and to hold a school-wide health event once a semester. The intervention component included provision of additional health information at school functions by health professionals and involvement of school libraries, computers, art, music and physical education in delivery/complementation of lifestyle education.
Standard practice: "Even though intervention material was provided to control schools, teachers reported using the material once a month or less often." Outcomes Outcome 1: School achievement: End-of-year final grades for reading, mathematics and science summarised as the GPA obtained from school records. The grade scale comprises scores between 0 and 100 points for each participant Outcome 2: Obesity indices: Age-and gender-specific BMI percentiles and BMI z-scores obtained from measured weight and height and by using formulas and data tables provided by the CDC. Overweight was defined as a BMI ≥ 85th percentile Notes 1. Authors were contacted 2. No sample size calculation was reported. This study might therefore be at risk of a type two error 3. Funding source: Not disclosed

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "7 schools were randomized using a random number generator" Cochrane Database of Systematic Reviews pervised homework time and were provided with a snack. Both groups could earn points that were redeemed for small prizes weekly for performing desired behaviours. The reward schedule was periodically calibrated to keep the rewards offered to the groups similar." Intervention: "The aerobic exercise group engaged in instructor-led aerobic activities (e.g. tag and jump rope) for 40 minutes a day. They wore heart-rate monitors every day [...] with which they could monitor their own performance and from which data were collected daily. Points in the exercise group were earned for an average daily heart rate above 150 beats a minute, with more points for higher average heart rates." Attention control: Participants "engaged in instructor-led sedentary activities (e.g. art and board games). Points in the control group were earned for participation and good behaviour." Notes 1. We contacted the authors to obtain additional study details. We were not able to obtain details of outcome data for inclusion in the meta-analysis, or additional study characteristics. A manuscript with relevant data is currently 'under review'. We did extract additional details from Bustamante 2016. 2. Funding sources: National Institutes of Health (R01 HL87923) and the National Science Foundation Graduate Research Fellowship Program.

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "Randomization (balanced by race, sex, and school to avoid imbalances on factors linked with differences in achievement) was performed by the study statistician and concealed until after baseline testing was completed, at which point the study coordinator informed the families." Allocation concealment (selection bias) Low risk Quote: "Randomization (balanced by race, sex, and school to avoid imbalances on factors linked with differences in achievement) was performed by the study statistician and concealed until after baseline testing was completed, at which point the study coordinator informed the families." Blinding of participants and personnel (performance bias) All outcomes Unclear risk Quote:"The attention control group engaged in instructor-led sedentary activities (e.g., art and board games)."

Cochrane Database of Systematic Reviews
Comparability of baseline groups Low risk Judgement comment: Random sequence adequately generated and allocation concealed Cross-contamination Unclear risk Quote: "The groups differed in that they participated in either an aerobic exercise or an attention control program. The aerobic exercise group engaged in instructor-led aerobic activities (e.g., tag and jump rope) for 40 min per day." Judgement comment: Risk of cross-contamination was low due to the nature of the intervention group: closed group-exercise sessions. However, no details were reported on the extent to which the comparison group adhered to the 'sedentary activities' condition throughout the study duration

Cochrane Database of Systematic Reviews
Reason for attrition: no longer at school, missed measurement days, asked to be withdrawn (no reasons reported), did not receive the intervention (no reason reported) Attrition rates: None reported Interventions Comparison: COPE versus attention control "The attention control programme was administered in a format like that of the COPE intervention and included the same number and length of sessions as the experimental programme, but there was no overlap of content between the 2 programmes." Intervention: The Creating Opportunities for Personal Empowerment (COPE) programme "is a manualised 15-session educational and cognitive-behavioural skills-building programme delivered by trained high-school health teachers." 1. Physical activity: "Each session of COPE contains 15 -20 minutes of physical activity (e.g. walking, dancing, kick-boxing movements), not intended as an exercise training programme, but rather to build beliefs in the teens that they can engage in and sustain some level of physical activity on a regular basis. Pedometers were used throughout the intervention in order to reinforce the physical activity education component of COPE. Students were asked to increase their step counts by 10% each week, regardless of baseline levels and to keep track of their daily steps on a tracking sheet so they could calculate a weekly average and determine if they met their weekly goal."

Healthy lifestyle education:
The COPE Healthy Lifestyles TEEN (Thinking, Emotions, Exercise, Nutrition) Programme "was delivered once a week in students' health course for 15 weeks." "Participants received a COPE manual with homework activities for each of the 15 sessions that reinforced the content and skills in the programme." "A parent newsletter describing the content of the COPE programme was sent home 4 times during the course of the 15-week programme." • Cognitive-behavioural skills building: Self-esteem; positive thinking/self-talk; goal-setting; problem-solving; stress and coping; emotional and behavioural regulation; effective communication; personality and communication styles; barriers to goal progression and overcoming barriers • Nutritional topics: Food groups and a healthy body; stoplight diet: red, yellow, and green; nutrients to build a healthy body; reading labels; effects of media and advertising on food choices, portion sizes; influence of feelings on eating; social eating; strategies for eating during parties, holidays, and vacations; snacks; eating out • Physical activity topics: Energy balance; ways to increase physical activity and associated benefits; heart rate; stretching Attention control: "The Healthy Teens programme was designed as a 15-week attention control programme to control for the time the health teachers spent in the COPE group. The content was manualised and focused on safety and common health topics/issues for adolescents, such as road safety, dental care, infectious diseases, immunisations, and skin care." Participants "also received a manual with homework assignments each week that focused on the topics being covered in class and were asked to review with his or her parent a newsletter that was sent home with the teens 4 times during the programme." "Attention control participants were provided with a pedometer for use only during the first week and post-intervention week for assessment purposes only."

Outcomes
Outcome 1: School achievement: Health class grades assessed by school teachers Notes 1. The authors kindly provided unpublished data for adolescents with obesity or overweight 2. The sample size calculation was based on the total study sample (participants in any weight group) 3. Funding source: National Institute of Health/ National Institute of Nursing Research 1R01NR012171.

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "randomly assigned to receive either the COPE TEEN program or the Healthy Teens attention control program by placing all of the school names in a hat and then randomly drawing them out." Cochrane Database of Systematic Reviews Judgement comment: Restricted randomisation. The method is random but it could be easily manipulated. Imbalances in baseline differences between intervention and control participants might indicate inappropriate randomisation Allocation concealment (selection bias)

Physical activity, diet and other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight (Review)
Unclear risk Quote: "The Healthy Teens attention control program by placing all of the school names in a hat and then randomly drawing them out." Judgement comment: Names are essentially concealed in the hat, which offers randomisation and allocation concealment. However names could easily be re-drawn from the hat. It is unclear who performed the randomisation Blinding of participants and personnel (performance bias) All outcomes Unclear risk Judgement comment: The trial register entry indicated that this was a double-blind (Participant, Investigator) trial. Delivering health-related content in a curriculum-based Health Class might be a way of blinding the participants. However, the COPE intervention arm also offered active physical activity sessions to which, by nature, participants cannot be blinded Blinding of outcome assessment (detection bias) All outcomes Unclear risk Quote: "Academic achievement was measured with the student's health course grade." Judgement comment: It is unclear if the same teacher who delivered the intervention also assessed academic performance in the health course Incomplete outcome data (attrition bias) All outcomes Unclear risk Judgement comment: Academic achievement data were only available at post-intervention, so assessment of missing data was not possible Selective reporting (reporting bias) Low risk Judgement comment: NCT01704768 registry entry. All relevant outcomes were reported in the study protocol. The authors provided the unpublished outcomes data for adolescents with overweight/obesity Comparability of baseline groups High risk Quote: "There are more male parents participating in the Healthy Teens group than the COPE TEEN group (p = .00). More parents are Hispanic in the COPE TEEN group versus the Healthy Teens group (p = 00). COPE TEEN parents have lower education levels (p = .00) and report more public assistance (p = .00) than Healthy Teens parents. COPE TEEN parents reported lower annual household incomes (p = .00)." Judgement comment: The quote relates to the entire study sample. The authors provided demographic characteristics of the subgroup with overweight/obesity which show similar differences between the experimental groups. The comparability of the experimental groups is at high risk of bias Cross-contamination Low risk Quote: "The first school district is located in the heart of the metropolitan city with the other district being located within a large suburb, which serves students from all socioeconomic backgrounds." Quote: "The decision was made to randomly assign schools to one of the two interventions (e.g., instead of randomly assigning classrooms within the schools) in order to decrease the probability of cross-contamination and minimize threats to internal validity." Other bias High risk Quote: "District administrators in both districts chose which schools could participate in the study." Quote: "All participants received incentives for their involvement in the intervention." Intervention: "Aimed to improve student school breakfast programme (SBP) participation by ameliorating the following environmental factors in the high-school setting." "As part of the Project break-FAST (Fueling Academics and Strengthening Teens) a grab-and-go style cart or breakfast line located outside the cafeteria in a high-traffic hallway, atrium or common area was implemented, developed individually at each intervention school to meet unique needs of each school." "School-wide marketing campaigns were developed by a community partner which worked with a group of students to design the marketing campaign at each intervention school." "Positive interactions and social support were created by developing school policies, if not already in place, to allow students to eat breakfast in the hallway. Schools were also encouraged to allow eating breakfast in some classrooms when appropriate. Teachers and school sta were asked to encourage the breakfast programme." "Development of a School Breakfast Expansion Team was encouraged at each intervention school. These teams were to consist of a variety of contributors including, but not limited to, the principal, food service director, nurse, students, wellness co-ordinator, and teachers." "Extension Co-ordinators were to provide support to schools in intervention development and implementation, communicated progress, successes."

Nanney 2016
Physical activity, diet and other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight (Review)

Cochrane Database of Systematic Reviews
Wait-list control/standard practice: Delayed treatment for the 1st year of follow-up. "Schools implemented a modified form of the intervention in the 2nd year of follow-up" (follow-up data not included in this review). "Comparison schools received the same monetary incentive as intervention schools, as well as research study sta support in implementing the delayed intervention. The main difference was the marketing package offered to intervention schools, but not to comparison schools." Outcomes Outcome 1: School achievement: assessed using weighted cumulative GPA. GPA covers academic years since 9th grade: pre-cumulative GPA covers 1 academic year for 9th-graders and 2 academic years for 10th-graders; post-cumulative GPA covers 2 academic years for 9th-graders (became 10th-graders at post-test) and 3 academic years for 10th-graders (became 11th-graders at post-test). The scale range for unweighted GPA was 0 -4 Notes 1. The authors kindly provided unpublished data for adolescents with obesity or overweight 2. The sample size calculation was based on the total study sample (participants in any weight group) 3. Funding source: National Heart, Lung and Blood Institute of the National Institutes of Health (5R01-HL113235-03).

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "The process of randomization of schools to treatment assignment was blocked so that 4 schools were assigned to intervention and 4 to control in each wave. As of the start of wave 1, only 13 schools had been recruited (see 4. Limitations), so a simple random subsample of 8 was assigned to wave 1. Within this subsample of 8, schools were randomly assigned to treatment or control by selection of a random permutation of the 8 labels (4 intervention, 4 control).For wave 2, three additional schools had been recruited prior to the randomization for wave 2, for a total of 8 schools. " Judgement comment: Adequate method used for wave 1 randomisation. Violation of randomisation procedure for wave 2 does not affect the data included in this study. We included wave 1 end-point data only Allocation concealment (selection bias)

Unclear risk Judgement comment: No details reported
Blinding of participants and personnel (performance bias) All outcomes Unclear risk Judgement comment: Blinding to this kind of intervention is not possible. However, it is unclear if participants and personnel were aware of the potential benefits of the intervention on academic performance. The primary aim of the study was to increase breakfast uptake, whereas change in academic performance was an exploratory variable rather than a primary or secondary outcome Blinding of outcome assessment (detection bias) All outcomes High risk Quote: "Each school received an excel file of variables and definitions requested. For each student, demographic information (e.g., race, ethnicity, grade level); grade point average (GPA) (term/semester GPAs and cumulative GPAs)..." Judgement comment: Schools and so teachers were aware of group allocation Incomplete outcome data (attrition bias) All outcomes High risk Judgement comment: The authors provided the raw data for the children with obesity or overweight. We applied LOCF methods as an intention-to-treat approach. The attrition rates were 9% in the intervention group, and 29% in the control group Selective reporting (reporting bias) Low risk Judgement comment: The unpublished data provided align with the outcomes reported in the study protocol Cochrane Database of Systematic Reviews

Comparability of baseline groups
Low risk Judgement comment: The provided unpublished data of baseline characteristics in both experimental groups indicate a low risk of bias for comparability of the groups at baseline Cross-contamination Low risk Quote: "During the first year of follow-up for each wave, four fidelity observations of both the intervention and delayed intervention groups were conducted to evaluate adherence to requirement of either making these necessary changes to the SBP (intervention) or not making any changes to regular breakfast service (control)." Judgement comment: Cluster randomisation and assessment of adherence to the experimental condition suggest a low risk of cross-contamination bias Intervention: The Active Smarter Kids (ASK) programme comprised 3 components aimed at providing children with the opportunity to engage in 165 minutes of physical activity/week more than the control group: i) physically-active lessons for 90 minutes/ week, conducted in the playground; physically-active educational lessons were delivered in 3 core subjects -Norwegian (30 minutes/ week), mathematics (30 minutes/week) and English (30 minutes/week); ii) physical-activity breaks (5 minutes/day) im-

Resaland 2016
Physical activity, diet and other behavioural interventions for improving cognition and school achievement in children and adolescents with obesity or overweight (Review) Cochrane Database of Systematic Reviews plemented in the classroom during academic lessons; and iii) physical-activity homework (10 minutes/day) prepared by teachers. "In addition, pupils participated in the curriculum-prescribed 90 minutes/week of Physical Education and the curriculum-prescribed 45 minutes/week of physical activity. Thus, PA (165 minutes/week) and PE/PA (135 minutes/week) components provided children opportunities to engage in school-based physical activities 300 minutes/week. The intervention was established as part of the mandatory school curriculum for all pupils attending the intervention schools." Standard practice: "normal practice" school curriculum, including usual amounts of physical activity/Physical Education, being approximately 135 minutes/week Outcomes 1. School Achievement: Reading, numeracy, and English were measured using specific standardised Norwegian National tests designed and administrated by The Norwegian Directorate for Education and Training 2. Cognitive functions (measured but not provided): Inhibition assessed using Golden's version of the Stroop test; cognitive flexibility using 1 verbal (Verbal fluency) and 1 nonverbal test (The Trail Making Test); working memory used a digit span test with digits both forward and backward (Wechsler Intelligence Scale for Children, 4th edition)

Obesity indices (measured but not provided for analysis):
Weight/height: BMI; waist circumference; body fat (skinfold thickness sites -biceps, triceps, subscapular, and suprailiac) Notes 1. The authors kindly provided unpublished school achievement data for children with obesity or overweight 2. Cognitive function data were not provided as the authors were in the process of publishing them 3. The sample size calculation was based on the total study sample (participants in any weight group) 4. Funding sources: The Research Council of Norway (ID number 221047/F40) and Sogn og Fjordane University College

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "cluster-randomized controlled trial (cluster RCT) with a random allocation at the school level using a 1:1 ratio." Quote from the study protocol Resaland 2015: "A neutral third party (Centre for Clinical Research, Haukeland University Hospital, Norway) performed the randomization." Judgement comment: It remains unclear how the random sequence was generated.

Allocation concealment (selection bias)
Unclear risk Judgement comment: No details reported Blinding of participants and personnel (performance bias) All outcomes High risk Quote: "Fi h-grade classroom teachers in the intervention schools (I-schools) delivered the intervention. To support and qualify teachers to conduct the intervention, we arranged three comprehensive pre-intervention seminars and two regional refreshing sessions during the intervention period. We also gave support via email and telephone to teachers in I-schools. A password-protected homepage (http://www. askstudy.no) further provided teachers in I-schools with information, videos and content for approximately 100 PA lessons. All lessons on the homepage were developed in collaboration with I-schools in Sogn og Fjordane County. Finally, we provided all I-schools with equipment (e.g., laminating machines and accessories, mathematics bingo tiles, cones) necessary to support the intervention." Quote from the study protocol: "Blinding of children and schools was not possible due to the nature of the experiment."

Cochrane Database of Systematic Reviews
Blinding of outcome assessment (detection bias) All outcomes Low risk Quote from the study protocol: "Blinding of children and schools was not possible due to the nature of the experiment. However, only the project management group has formal knowledge of group assignment. The data manager and statisticians are blinded to group allocation until analyses are conducted." Quote: "Academic performance in numeracy (often referred to as mathematics in the literature), reading and English was measured using standardized Norwegian national tests designed and administered by The Norwegian Directorate for Education." Incomplete outcome data (attrition bias) All outcomes High risk Judgement comment: Based on the provided unpublished data, the proportion of missing data was substantially higher in the control group compared to the intervention group: maths Intervention 3%, control 7%; reading Intervention 0%, control 14%, English Intervention 0%, control 6%. No reason for missing data were provided. No imputation of missing data was performed for unpublished data Selective reporting (reporting bias) Low risk Quote: "The study is registered in Clinicaltrials.gov ID nr: NCT02132494. We previously published a detailed description of the study (Resaland et al., 2015)...." Judgement comment: The authors provided unpublished academic achievement data which align with the study protocol and trial register. The authors clarified that they are working on the publication of executive function outcomes Comparability of baseline groups Low risk Quote: " Table 1 shows children's baseline characteristics by group. There were no differences between I-schools and C-schools for any variables." Judgement comment: This quote relates to the total study sample including children with healthy weight. Visual inspection of participant characteristics with obesity or overweight (provided unpublished data) indicate a low risk of bias for comparability of the experimental groups at baseline Cross-contamination Unclear risk Quote: "ASK was a seven-month cluster-randomized controlled trial (cluster RCT) with a random allocation at the school level using a 1:1 ratio. Such randomization eliminated the possibility of contamination between pupils in the same school." Quote from the study protocol: "ASK teachers at the 28 I-schools completed a report each week that described activities performed throughout the school day, the intensity of the activities (on a 1 to 3 scale) and the number of minutes allocated to physical activity/PE in each ASK session. All 29 C-schools, at the end of the school year, completed a report that describes the activities that were performed and the estimated time allocated to physical activity/PE during the school year (minutes/week)." Judgement comment: Although this study was a cluster-RCT, it was unclear how closely located the intervention and control schools were and whether intervention teachers had the opportunity to share their teaching approaches. Adherence of the control school to control group conditions was assessed but not reported. Restricted geographical area could mean risk of cross-contamination. 3 dropout schools were all in the same district  Reason for attrition: Self-consciousness due to obesity, school truancy or dropout; school transfer; lack of interest; pregnancy; safety concerns about walking home in the dark; sports practice time conflicts; academic tutoring time conflicts, frequent headaches and an injury outside of the programme that required crutches. School administrators removed 3 students from the programme because of behavioural infractions external to the exergame intervention

Interventions
Comparison: competitive physical activity versus standard practice, co-operative physical activity versus standard practice Interventions: Nintendo Wii EA Sports Active exergame played in competitive condition individually or in co-operative condition in pairs for 30 to 60 minutes, 5 days a week, over a period of 10 weeks. "Fitness video game included cardio activities (e.g. inline skating), sports games (basketball, volleyball, tennis, baseball) and strength training. Exergame routine was the same for both intervention groups. Routines varied on a daily basis and gradually increased in difficulty throughout the study." "Children in the competitive group were encouraged to win by earning top scores and expending most calories each time they played. Children in the co-operative group were encouraged to earn the highest possible score and to expend the most calories as a pair." "Children were supervised during the exergame sessions." Compliance was assessed through attendance.
Standard practice: Continuation of usual school lunch or after-school activities or both (Quote: "Control participants continued usual daily activities, such as socializing with friends, tutoring, and sports team practice")

Outcomes
Outcome 1: Cognitive function: Executive function (visual-spatial skills, response inhibition, motor planning, visual scanning, speed, cognitive flexibility) measured using the subscales Design Fluency and Trail-Making of the Delis-Kaplan Executive Function System. Tests were administered by a trained researcher and were coded by 2 research assistants; a 3rd research assistant double-coded all tests

Cochrane Database of Systematic Reviews
Outcome 2: Obesity indices: Body weight change: Body weight measured clothed without shoes by paediatricians and nurse practitioners at the school-based wellness clinic. Body weight remained unadjusted for height Notes 1. No sample size calculation was performed. Thus, this study might be at risk of a type two error 2. 5 of the study participants (2 boys, 3 girls) were without obesity or overweight. However, this study was done with the intention of weight management, and the number of normal-weight children is small when allocated into a control group and the 2 intervention groups 3. Participants attended on average 1 exergame session a week 4. Time point of measurement of cognitive function potentially introduced a confounding effect of acute exercise on cognitive function 5. Funding sources: Robert Wood Johnson Foundation, Georgetown University

Bias
Authors' judgement Support for judgement

Random sequence generation (selection bias)
Unclear risk Quote (from email correspondence): "An adult research coordinator drew a number to randomly assign condition. When conditions became imbalanced due to attrition, new participants were assigned consecutively to the next available condition to maintain sample size balance."

Allocation concealment (selection bias)
Unclear risk Quote (from email correspondence): "Participants knew that they were assigned to 1 of 2 classrooms or else to the control group, but they did not know the research aim until the disclosure period at the end of the study." Blinding of participants and personnel (performance bias) All outcomes Unclear risk Judgement comment: Blinding not possible in exercise intervention Quote (from email correspondence): Children "did not know the research aim until the disclosure period at the end of the study" Judgement comment: Personnel were also blinded to true purpose of the study (information obtained from email correspondence) Blinding of outcome assessment (detection bias) All outcomes Low risk Quote (from email correspondence): "The coders and data enterers were blinded to the participant's condition" Incomplete outcome data (attrition bias) All outcomes High risk Judgement comment: Analysis was performed with data when both baseline and post-intervention data were available. Therefore, study did not account for incomplete outcome data. No information available on characteristics of missing data Children participated in an optional extracurricular, play-based, non-competitive physical activity programme: 3 60-minute sessions/week using school facilities, adapted to levels of motor competence. "The programme included basic sports games, playground games, dance and other activities focused on developing motor skills. At the end of the 1st year, approximately 90 sessions had been carried out in each school." Parents and teachers were involved in activities to promote active lifestyles in their children by "(a) use of reinforcement tools (e.g. a refrigerator magnet with recommendations for physical activity for children); (b) answering a satisfaction-with-the programme questionnaire; and (c) access to a blog where parents could observe their children's progress, read news regarding reinforcing healthy lifestyles, and ask questions of or make complaints to the research team." "Environmental interventions were conducted in the playground. Fixed (a balance circuit and panels with incentives to be physically active during break time) and mobile equipment (tyres of different colours and sizes) were put in the playgrounds to encourage children to be more active during playtime."    Cochrane Database of Systematic Reviews ty and PE and development of policies that prohibit withholding PE for punitive reasons. To further increase physical activity, the School Wellness Policy specifies expanding programmes/activities that meet need, interest, and abilities of students. Exer-gaming consoles will be provided to 6 schools and will be integrated into 5th -8th grade PE classes and after-school programmes. New and innovate gym equipment will be purchased for the 6 target schools for use in gym class and after-school programming. Various pedometer and interactive programmes encouraging physical activity in and outside of school are planned for the 6th grade Group 2 -Nutrition only: Policy changes will focus on 6 target schools. District will expand nutrition education by integrating other opportunities to learn and practice healthy behaviours across disciplines. Farm-to-School programmes will include school visits by farmers to teach students about agriculture, healthy foods and nutrition, coinciding with Farmer's Market Menu Days. Schools will receive 4 45-minute nutrition workshops a year. Community educators will offer culturally appropriate, interactive nutrition workshops and cooking demonstrations. Cafeterias will receive youthfriendly nutritional messaging, regular promotion of new menu foods, and a variety of monthly nutrition-focused activities. The goals are to: increase number of students who try new menu items regularly, increase acceptance of healthy foods, and improve nutrition literacy. Policy states schools will limit celebrations that involve food to no more than 1 per class/month: 6 schools will pilot alternatives to food for celebrations Group 3 -Physical Activity + Nutrition: Schools will receive nutrition education, nutrition standards for foods sold, and opportunities for physical activity. In addition to the above interventions, schools (N+PA) will expand the District's school-based wellness initiative, PAW-Physical Activity and Wellness. With District support, PAW schools develop School Wellness Teams (SWTs) to identify school health priorities, implement and sustain health initiatives through school campaigns, promote healthy behaviour, and support wellness policies. SWP addresses health promotion and marketing by limiting product marketing in schools, expanding nutrition education and broadening health communication with parents. In 3 targeted schools, we expand to include Sta Wellness Promotion. Adults in schools are trusted and influential role models for students; by increasing their positive health behaviours, students may be influenced to adopt similar behaviours. The District will work with the City's Employee Wellness Programme to increase school sta participation.
Waitlist-control/standard practice: Schools will receive educational interventions on health topics not related to nutrition or physical activity (i.e. peer relations, sleep, dental care, etc.)

Characteristics of ongoing studies [ordered by study ID]
Trial name or title Maintain study

Methods
Study design: RCT Sequence generation: Participants assigned by study team member by time and date of return using list with random order "intervention" or "control" (principle of contingency) Inclusion criteria: 1. Age between 8.0 and 11.9 years; 2. with obesity or overweight, based on the sex-and age-specific international body mass index standards (World Obesity Federation); 3. not to have any physical disabilities or neurological disorder that limits exercising; 4. to report no use of medications that influence central nervous system functioning; 5. in the case of the girls, not to have started menstruation at baseline Exclusion criteria: 1. Le -handedness (measured by the Edinburgh inventory); 2. attention-deficit hyperactivity disorder (ADHD) evaluated by ADHD rating scale; 3. other psychiatric diagnoses indicated with self-report Arm 1 (Physical activity): Movement therapy for 90 minutes twice a week, aiming to be fun while being physically active and gaining sport-specific skills. Sessions included behavioural therapy aiming to change daily physical activity Arm 2 (Diet): An optimised mixed diet was implemented in addition to behavioural therapy aiming to change eating behaviour over 8 units of 90 minutes of nutrition counselling, partially together with parents Arm 3 (Physical activity + Diet): In the combined group the children took part in the ball school programme as well as the nutrition counselling, together with their parents. The contents were equivalent to those of the ball school and nutrition group (see Arm 1 and Arm 2).

Control:
The control group did not participate in a special programme during the intervention period of six months Outcomes Cognitive functions: Differential performance test was used to measure performance during focused activity, the culture fair intelligence test was applied to measure overall intelligence Inclusion criteria: The participating children will be between the ages of 6 and 12 and above the 85th percentile for weight and will have at least 1 parent with obesity or overweight (BMI > 25 kg/ m 2 ). 1 parent must agree to attend all parent/child treatment meetings as the participating parent. For families in which 1 parent is obese, this parent will be encouraged to be the participating parent; if both parents are obese, the family will choose 1 parent to enrol in the study. Similarly, if 2 children in the family are obese, the older sibling will be encouraged to be the primary participant, as it is more likely the younger sibling will model the older sibling. Although only the child who is overweight/obese and the participating parent will be required to attend treatment sessions, all family members living in the household, including other adults and siblings, will be encouraged to participate indirectly by supporting changes in the family's lifestyle. All participants must be able to speak and comprehend English at a first-grade level Exclusion criteria: The participating parent or child will not be receiving treatment for a DSM-5 disorder that interferes with treatment delivered as part of the intervention or is explicitly targeted towards management of weight control; will not have a physical disability or diagnosis that prevents performance of physical activity at a level equivalent to a brisk walk or that places severe restriction on diet; will not be on a medication regimen that affects weight; and will not be participating Cochrane Database of Systematic Reviews in an alternative weight control programme. Families in which either the participating child or parent is actively involved in psychological or other interfering weight-loss treatment, using weight-affecting medications, or has a psychiatric or medical condition (e.g. anorexia nervosa, schizophrenia, binge eating disorder, depression) that would hinder participation in the study will be excluded as identified by the screening assessments described in section 5. Age range: 9 to 12 years Geographical region: Qatar Interventions Intervention: 3 phases: 1. Intensive weight loss camps; 2. after-school clubs as supplement/consolidation; and 3. maintenance through web and social/family support. The intervention involves developing social and emotional competences, promotion of healthy lifestyle, use of activity monitoring devices to promote increased activity and enlisting family to maintain weight loss in the long term.
The intervention group receives all programme components: 1. Parent information sessions and orientation; 2. 2-week intensive weight loss and lifestyle education camp; 3. after-school clubs over 12 weeks for consolidation (including physical activity and lifestyle education); and 4. wearable sensors and social media modules with parental involvement Interventions Intervention Group 1: Involved 20 meetings with the adolescents divided into 16 for the psychological intervention (twice a week), and 4 monthly for the nutritional orientations; 9 meetings with parents, of which 6 bi-weekly, 2 nutritional meetings (1 in the beginning and the other after 30 days) and also meeting with a physical educator in the 1st week of the intervention, in addition to 36 sessions of physical exercises for all adolescents, conducted 3 times a week Intervention Group 2: Involved 4 monthly meetings with adolescents for nutritional orientation, 9 meetings with the parents (6 bi-weekly with a psychologist, 2 with a nutritionist and 1 with a physical educator) and 36 sessions of physical exercises for adolescents Control group: Involved meetings with a nutritionist and a physical educator for the adolescents and parents and physical exercises sessions for the adolescents; no psychological intervention.
The psychological intervention contained life experience activities toward the learning of skills such as self-control, assertiveness, solving problems (including nicknaming and bullying) and reading the context that contribute to the reduction of behavioural problems and to the gain of selfworthiness skills, such as self-esteem, self-efficacy, thereby enhancing social competence. The objective of the parents' counselling sessions was to teach and provide antecedent and consolidated conditions of socially-acceptable behaviours and diet. The nutritional sessions for adolescents and parents was conducted by a nutritionist and included information with illustrative material suitable for the age, showing food groups and highlighting those that ought to be included or avoided for its high calorific value. The sessions with the physical educator included information about the importance of physical activity for weight loss and overall health, as well as suggestions for games and activities. The adolescents engaged in weekly physical activity sessions and the practice of indoor physical activity (exercise treadmill and stationary bicycle)

Chen 2016
Group physical activity programme including multiple types of moderate-intensity exercises performed 4 times/week for 40 minutes per session (5 minutes each for warm-up and cool-down, 30 minutes for the main exercise). The participants were free to choose one of the provided exercise types (e.g. fast walking, stair climbing, jumping rope, or aerobic dancing), with an emphasis on maintaining a moderate intensity of 60% to 70% of the maximal heart rate. Intervention was offered during the school day in the morning, during lunch break, or after school for 3 months Cochrane Database of Systematic Reviews

Davis 2011b
Aerobic group exercise for 40 minutes 5 times/week, over a mean total of 13 weeks. Five-minute warm-up phase consisted of brisk walking and static and dynamic stretching. Children were encouraged to maintain a heart rate > 150 beats/minute during running games, tag games, jump rope, modified basketball and football. The intervention involved no competition or skill enhancement and was delivered in an after-school setting

De Greeff 2016
Fit en Vaardig op school (Fit and academically proficient at school) involved physically active academic lessons which ran over 44 weeks in total over 2 school years with 3 lessons/week. The lessons had a duration of 20 -30 minutes, with 10 -15 minutes spent on solving mathematical problems and 10 -15 minutes spent on language. During the lessons all children started with performing a basic exercise, such as jogging, hopping in place or marching. A specific exercise was performed when the children solved an academic task. The physical activities were aimed to be of moderate-to-vigorous intensity

Gallotta 2015
The 2 intervention conditions had the same structure and took place in the school. They included 15 minutes of warm-up, 30 minutes of moderate-to-vigorous physical activities, and 15 minutes of cool-down and stretching. The traditional physical activity intervention consisted of continuous aerobic circuit training followed by a sub-maximal shuttle run exercise. This intervention focused on the improvement of cardiovascular endurance by performing different types of gaits (e.g. fast walking, running, skipping) without any specific co-ordinative request. The co-ordinative physical activity intervention focused on the development of psychomotor competences and expertise in movement-based problem-solving through functional use of a common tool (e.g. basketball), and considering various tasks that involved decision-making motor tasks and manipulative ball-handling skills

Kra 2014
See Davis 2011b. The intervention duration was extended to 8 months.

Sánchez-López 2017 [pers comm]
MOVI-KIDS is a multidimensional intervention that consisted of a standardised extra-curricular non-competitive physical activity programme of 4½ hours/week; informative sessions to parents and teachers about how schoolchildren can become more active, and interventions in the playground (environmental changes: equipment, facilities, painting, etc.) aimed to promote physical activity during recess (MOVI-Playground)

Staiano 2012
Nintendo Wii EA Sports Active exergame played in competitive condition individually or in co-operative condition in pairs for 30 to 60 minutes, 5 days/week, over a period of 10 weeks in total. The fitness video game included cardio activities (e.g. inline skating), sports games (basketball, volleyball, tennis, baseball) and strength training

Resaland 2016
The Active Smarter Kids (ASK) programme comprised 3 components: i) physically active lessons for 90 minutes/week, conducted in the playground; physically active educational lessons were delivered in 3 core subjects -Norwegian (30 minutes/ week), mathematics (30 minutes/week) and English (30 minutes/week); ii) physical activity breaks (5 minutes/day) implemented in the classroom during academic lessons; and iii) physical activity homework (10 minutes/day)

Ahamed 2007
Action Schools! BC was a comprehensive, multicomponent intervention providing tools for schools and teachers to use in promoting physical activity and healthy eating in different settings. These include the school environment (healthy eating posters), scheduled Physical Education, classroom action, family and community (e.g. walking school bus), extracurricular activities (e.g. dance club) and school spirit (e.g. Hike across Canada challenge)

Barbosa Filho 2017 [pers comm]
Fortaleça sua Saúde ('Strengthen your health') focused on teachers' training and activities on health in the curriculum (including a specific training to Physical Education teachers), active opportunities in the school environment (availability of spaces and materials for physical activity) and health education (production and exhibition of health material at school, and distributing pamphlets to students and parents) Cochrane Database of Systematic Reviews

Huang 2015
The day-camp intervention comprised 2 parts: an intensive six-week day camp intervention and a subsequent 46-week family-based intervention programme (13-month [52 weeks] in total). Children were engaged in physical activity and sports for at least 3 hours a day, achieving about 90 minutes of moderate-to-vigorous physical activity per day measured by accelerometry. After the day camp, one physical activity day was offered as part of the family-based intervention programme. Healthy lifestyle education topics during the 6-week day camp included nutrition, physical activity and health, and goal-setting. The family-based intervention programme comprised 4 parents-involved meetings targeting daily physical activity and dietary behaviour. In the day camp, 3 meals and 3 snacks were prepared and served according to the national dietary recommendations with no caloric restrictions

Melnyk 2013
COPE (Creating Opportunities for Personal Empowerment) programme was a manualised 15-session educational and cognitive-behavioural skills-building programme. Each session of COPE contains 15 -20 minutes of physical activity (e.g. walking, dancing, kick-boxing movements), not intended as an exercise training programme, but rather to build beliefs in the participants that they can engage in and sustain some level of physical activity on a regular basis. Pedometers were used throughout the intervention. Participants were asked to increase their step counts by 10% each week, regardless of baseline levels, and to keep track of their daily steps. The COPE Healthy Lifestyles TEEN (Thinking, Emotions, Exercise, Nutrition) Programme was delivered once a week as part of a school health curriculum. Participants received a COPE manual with homework activities for each of the 15 sessions that reinforced the content and skills in the programme: cognitive-behavioural skill building (e.g. problem-solving and emotional and behavioural regulation), nutrition (e.g. food groups, portion sizes, food labelling), and physical activity (e.g. ways to increase physical activity and associated benefits)

Treu 2017
The standard intervention arm of the ASCEND intervention consisted of the Nutrition Detectives (ND) programme and the ABC for Fitness (ABC) programme. ND was a 90-minute programme that aimed to convey the link between food choices and health, convince students of the need to become "supermarket spies" to learn the truth about the foods that they eat. ABC for Fitness offered brief bursts of physical activity in the classroom, spread over the school day. Classroom teachers offered 30 daily minutes of activity bursts. The activity bursts were designed to include a brief warm-up and cool-down (e.g. stretching or low-intensity activity) along with one or more core activities of higher intensity (e.g. hopping, running in place, jumping jacks, or dancing to music).
The enhanced intervention arm included the ND and ABC programmes plus reinforcements of their messages to participants and their families in the school, home, and a supermarket. Family-focused kits were send home including pedometers, walking tips to increase daily steps, a family log for recording steps, local walking trail guides, walking maps for local grocery stores, physical activity tips sheet, suggestions for activity bursts, family activity challenge cards, a 3-minute sand timer to be used for activity challenges, and a log to record the number of activities and repetitions completed, Nutrition Detectives DVD, a reminder card with the programme's "five clues" to make healthful food choices, grocery store coupons, and a family "homework assignment" to watch the DVD, review the ND clues together, complete an activity applying the clues to foods in the family kitchen. Family nights were held at a) the local supermarket, with stations set up to teach families about healthful food choices with games, demonstrations, and taste tests; b) schools offering stations throughout the building to try out different kinds of exercises, including Frisbee golf and Zumba, and received information or coupons from local fitness-related businesses

Winter 2011
The Healthy & Ready to Learn intervention involved parents and teachers reading children's books on health-related themes including nutrition and obesity prevention to the participants. Teachers and parents were trained to increase children's time spent physically active in moderate-to-vigorous activity for 60 minutes/day. Activities were play-based and targeted specific gross motor skills. Physical activity equipment was provided

Wirt 2013 [pers comm]
Komm mit in das gesunde Boot ('Join the healthy boat') comprised healthy lifestyle education of 20 teaching sessions a year focusing on increased physical activity, reduced consumption of sugar-sweetened beverages and reduced screen time. It included 2 physically-active breaks per school day of 5 to 7 minutes, and a physical activity task to be performed at home involving parents Cochrane Database of Systematic Reviews

Damsgaard 2017 [pers comm]
In the OPUS School Meal intervention children received the New Nordic Diet (NND) containing seasonal, health-promoting ingredients, for example, berries, root vegetables, whole grains, fish, shellfish, seaweed and rapeseed oil. Children received daily servings of a mid-morning snack, ad libitum hot lunch meal and afternoon snack (fruit dessert twice/week). The children were encouraged to taste everything and to keep a reasonable plate distribution with vegetables and starchy foods filling most of the plate. Each child spent 3 -5 school half-days in the kitchen cooking, presenting, and serving the menu of the day to the other children. The teachers were encouraged to participate in the lunch meals. Class teachers were given a box of teaching materials about the human body, the clinical measurements, and taste sensorics, including background information about NND and suggestions for related educational activities and games

Johnston 2013
The whole-school lifestyle education programme involved curriculum material taught by trained teachers, school meal modification towards nutrient-dense food and nutrition counselling. Teachers were provided with 50 integrated lessons-worth of curriculum material aiming to improve healthy diet (increased fruit and vegetable, breakfast, healthy snack, water consumption) and increase physical activity. Teachers were encouraged to teach lifestyle-integrated lessons once a week, to conduct health-related activities every 2 weeks and to hold a school-wide health event once a semester

Nanney 2016
The Project breakFAST (Fuelling Academics and Strengthening Teens) aimed to improve students' school breakfast programme (SBP) participation by implementing a grab-and-go-style cart or breakfast line located outside the cafeteria in a high-traffic hallway, atrium or common area. School-wide marketing campaigns were developed by a community partner which worked with a group of students to design the marketing campaign at each school. Positive interactions and social support were created by developing school policies, to allow students to eat breakfast in the hallway. Schools were also encouraged to allow eating breakfast in some classrooms when appropriate Cochrane Database of Systematic Reviews #17 (child* or schoolchild* or preschool* or pre-school* or schoolage* or school-age* or schoolboy* or schoolgirl* or boy* or girl* or preteen* or teen* or adolescen* or youth* or "young people" or "young person*" or pediatr* or paediatr*) #18 (#15 OR #16 OR #17) #19 MeSH descriptor Exercise, this term only #20 MeSH descriptor Exercise Therapy, this term only #21 MeSH descriptor Physical Exertion, this term only #22 MeSH descriptor Motor Activity, this term only #23 MeSH descriptor Sports, this term only #24 (sport*) #25 MeSH descriptor Physical Education and Training explode all trees #26 (physical near/3 (activit* or education* or exertion* or training)) #27 (exercise*) #28 MeSH descriptor Diet Therapy explode all trees #29 ((diet or dieting) near/5 (health* or weight*)) #30 (calorie near/3 (control or reduc* or restriction)) #31 "food choice*" #32 ("fat camp*" or "weight loss camp*") #33 "nutrition education" #34 MeSH descriptor Nutrition Therapy, this term only #35 MeSH descriptor Behavior Therapy, this term only #36 MeSH descriptor Cognitive Therapy, this term only #37 MeSH descriptor Psychotherapy, this term only #38 (behavio?r* near/3 (therap* or technique* or modif* or intervention*)) #39 (cognit* near/3 (therap* or technique* or modif* or intervention*)) #40 CBT #41 (psychotherap* or psycho-therap*) #42 MeSH descriptor Family Therapy, this term only #43 (family near/3 (therap* or intervention*)) #44 Cochrane Database of Systematic Reviews S27 ((lifestyle* or life-style*)) S28 (((video or computer) near/1 game*)) S29 s8 or s9 or s10 or s11 or s12 or s13 or s14 or s15 or s16 or s17 or s18 or s19 or s20 S30 s21 or s22 or s23 or s24 or s25 or s26 or s27 or s28 S31 s29 or s30 S32 s6 and s7 and s31

SPORTDiscus EBSCO
Searched from 1980 to current on 05 March 2012 and 06 May 2013, 6 February 2017, limited to 2013 to current (2186 records) S66 (S63 and S65) S65 S17 and S57 and S64 S64 S1 or S2 or S3 or S4 or S5 or S6 or S7 or S8 or S9 or S10 or S11 or S12 or S13 S63 S61 NOT S62 S62 SU animals NOT SU humans S61 (S58 or S59 or S60) S60 AB (random* or blind* or allocat* or assign* or trial* or placebo* or crossover or cross-over) S59 SU controlled clinical trial S58 SU randomized controlled trials S57 (S19 or S20 or S21 or S22 or S23 or S24 or S25 or S26 or S27 or S28 or S29 or S30 or S31 or S32 or S33 or S34 or S35 or S36 or S37 or S38 or S39 or S40 or S41 or S42 or S43 or S44 or S45 or S46 or S47 or S48 or S49 or S50 or S51 or S52 or S53 or S54 or S55 or S56) S56 TX ((computer or video or internet) N1 game) S55 SU computer game S54 TX lifestyle* or life-style* S53 TX (health* N3 (lifestyle or promotion or education or behavio?r)) S52 SU lifestyle S51 SU Health Education or SU Health Promotion S50 TX psycho-social or psychosocial S49 TX "screen time" S48 TX television or TV S47 SU video games S46 SU television S45 TX (Sedentary N3 (behavio?r or lifestyle)) S44 SU Sedentary S43 TX family-based S42 TX (family N3 (therap* or intervention*)) S41 SU family therapy S40 TX psychotherap* or psycho-therap* Rerun View Details Edit Interface -S39 TX (behavio?r N3 (therap* or technique* or modif* or intervention*)) S38 TX CBT S37 SU Cognitive therapy S36 SU Behavior therapy S35 SU Psychotherapy S34 TX "food choice" S33 TX (calorie N3 (control or reduc* or restriction)) S32 TX ((diet or dieting) N5 (health* or weight*)) S31 TX "fat camp*" or "weight loss camp*" S30 SU food habit S29 SU nutrition therapy S28 SU diet therapy S27 TX exercise* S26 TX sport* S25 TX (Physical N2 (activit* or education* or training or fitness)) S24 SU Physical training S23 SU Physical activity S22 SU Physical education S21 SU Sport S20 SU Exercise Therapy S19 SU Exercise S18 (S14 or S15 or S16 or S17) S17 TX child* or schoolchild* or preschool* or pre-school* or schoolage* or school-age* or schoolboy* or schoolgirl* or boy* or girl* or preteen* or teen* or adolescen* or youth* or young people or young person* or pediatr* or paediatr* Library Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews S16 SU teenager S15 SU adolescent S14 SU child S13 TX Overeat* or over-eat* or overnourish* or over-nourish* or overnutrit* or over-nutrit* S12 TX "waist-hip ratio" S11 TX "body weigh*" or bodyweigh* or body mass* or bodymass or "body fat*" or bodyfat* S10 TX waist-hip ration S9 TX skin fold thickness S8 TX body fat distribution S7 SU body composition S6 TX (weight N2 (gain* or increas*)) S5 TX (weight N2 (loss or lost or losing or reduc*)) S4 TX obes* or overweight or over-weight S3 SU body weight change S2 SU body weight S1 SU overweight Test performance draws on the working memory of the child and reflects the reading comprehension of the child, which includes accurate and fluent decoding of words, vocabulary knowledge, and thinking and reasoning skills. The sentences gradually become longer and more complicated, and as complexity increases, thoughtful analysis of content becomes more essential to comprehension in order to solve the task, e.g. the ability to make inferences We identified eligible dietary interventions that allowed us to draw conclusions about their effectiveness on school achievement. Evidence was available for achievement in additional school subjects and cognitive abilities. We included 12 new studies.

C O N T R I B U T I O N S O F A U T H O R S
AM, DHS and JS dra ed the review protocol. AM and YL screened the titles and abstracts of potentially eligible studies and reports. AM, YL, JNB, DHS and JS assessed the full report of potentially relevant studies for eligibility, in consensus, with JJR when necessary. AM, JS and DHS obtained full-text translations of non-English language reports. AM, YL and DHS extracted the data. AM performed the data analysis with substantial input from DHS and JNB. JNB provided expert input on the cognitive outcomes and classifications. AM and DHS assessed the risk of bias of included studies. AM dra ed the full review with regular input from all review authors. AM is the guarantor for the review.

D I F F E R E N C E S B E T W E E N P R O T O C O L A N D R E V I E W
The review title changed to specify the intervention types more precisely, and we used person-first language to remove stigma.
The author team changed from the protocol to this version of the systematic review. Three new authors joined as co-authors: Josephine N Booth, Yvonne Laird and John J Reilly. Susan Shenkin was not involved in the update of this review.
We revised the wording in the review objectives in the abstract and main text by providing an example of what we mean by cognitive function in brackets. The wording changed from "cognitive function" to "cognitive function (e.g. executive functions)".
We intended from the outset to select studies based on inclusion criteria; however, we did not state this explicitly in the protocol. The intervention criterion for inclusion was that the study aimed to prevent or treat childhood and adolescent obesity as a primary or secondary outcome through lifestyle interventions. The outcome criterion for inclusion was that studies measured school achievement, cognitive function and future success as defined in Types of outcome measures. We clarified that we restricted the review to the aforementioned outcome measures because the same interventions were studied in the same population for di erent purposes, e.g. change in body mass index, and were published recently in three other Cochrane Reviews.
We stated in the protocol that studies that included some children and adolescents with overweight would be included in the review only when outcomes for children with obesity or overweight were reported separately. Only a few studies investigated the e ects of lifestyle interventions on school achievement or cognitive function or both in a paediatric population with overweight; we therefore did not exclude those studies if results for this population group were not reported separately. We made every e ort to contact the authors of those studies to obtain data for the subgroup with obesity or overweight, or both.
In the protocol, we stated that we would include controlled trials. We removed this inclusion criterion and considered only randomised controlled trials, as is was recommended by our Cochrane group.
We provided e ect sizes for studies that were inappropriate for inclusion in a meta-analysis. The protocol stated that we would provide a narrative description of study results derived from those studies.