Am J Clin Nutr 2012 Verstraeten 415 38

See corresponding editorial on page 227.
Effectiveness of preventive school-based obesity interventions in

low- and middle-income countries: a systematic review1–3
Roosmarijn Verstraeten, Dominique Roberfroid, Carl Lachat, Jef L Leroy, Michelle Holdsworth, Lea Maes,
and Patrick W Kolsteren
ABSTRACT economic and societal changes (11, 12). The staggering in-
Background: The prevalence of childhood obesity is increasing creases in unhealthy body weights have been observed in both
rapidly in low- and middle-income countries, and informed policies urban and rural settings and across all levels of socioeconomic
to tackle the problem must be defined. status (SES), including in the poorest groups (13–15).
Objective: We systematically reviewed the evidence on the effec- Of great concern is the worldwide rise in obesity in children
tiveness of school-based interventions targeting dietary behavior and adolescents. Ten percent of 5–17-y-olds (ie, 150–160 mil-
Downloaded from ajcn.nutrition.org by guest on January 28, 2018

and/or physical activity for the primary prevention of obesity in lion) are overweight, of whom 2–3% (ie, 35–40 million) are
children and adolescents aged 6–18 y in low- and middle-income obese (16). The rapid increase in unhealthy body weight in
countries. schoolchildren has led to prevalence levels in some LMICs that
Design: We searched the MEDLINE, EMBASE, Web of Science, are as high or even higher than those found in HICs (8, 16, 17).
CENTRAL, ERIC, Cochrane Library, and Centre for Reviews and Excess body weight during childhood is associated with a range
Dissemination databases for peer-reviewed controlled studies pub- of chronic conditions in adulthood, including type 2 diabetes,
lished in English, Spanish, French, German, or Dutch between Jan- hypertension, dyslipidemia, cardiovascular diseases, and cancers
uary 1990 and July 2011. The quality of the included studies was (18–21), and thereby compromises the quality of life (22) and
appraised independently by 2 authors who used the Effective Public overall life expectancy (23–25). Obese children are more likely
Health Practice Project tool. to become obese adults (26–28) and to suffer from psychosocial
Results: From a total of 7218 unique references, we retained 22 problems, social stigmatization, and poor self-image (29).
studies. Most of the interventions (82%) had a positive effect on Prevention is of utmost importance to curb the rise in diet-
dietary behavior and physical activity behavior (effect size ranged
related chronic diseases, particularly in LMICs because treatment
from 20.48 to 1.61). BMI decreased in 8 studies (effect size ranged
is expensive and will drain the already limited public health
from 20.7 to 0.0). Effective interventions targeted both diet and
resources (30). Physical activity (PA) and dietary habits are key
physical activity, involved multiple stakeholders, and integrated ed-
modifiable behavioral risk factors in the onset of obesity. As such,
ucational activities into the school curriculum.
they are the cornerstones of any preventive strategy in children
Conclusions: School-based interventions have the potential to im-
and adolescents (7, 31). School-based intervention programs
prove dietary and physical activity behavior and to prevent un-
have emerged increasingly as an important strategy in obesity
healthy body weights in low- and middle-income countries. To
reach their full potential, interventions should conduct process eval- prevention. Most of the available evidence, however, originates
uations to document program implementation. The effect and the from studies conducted in HICs (32, 33). The only 2 systematic
pathways through which interventions have this effect need to be
1
better documented through rigorous evaluation studies. Am J From the Department of Food Safety and Food Quality, Ghent Univer-
Clin Nutr 2012;96:415–38. sity, Ghent, Belgium (RV, CL, and PWK); the Nutrition and Child Health
Unit, Department of Public Health, Prince Leopold Institute of Tropical
Medicine, Antwerp, Belgium (RV, DR, CL, and PWK); the Poverty, Health,
INTRODUCTION and Nutrition Division, International Food Policy Research Institute, Wash-
ington, DC (JLL); the School of Health and Related Research, The Univer-
Chronic diseases are now the leading cause of morbidity and sity of Sheffield, Sheffield, United Kingdom (MH); and the Department of
mortality in low- and middle-income countries (LMICs)4, and Public Health, Ghent University, Ghent, Belgium (LM).
2
their prevalence will increase even further over the next 2 de- No funding was received for this study.
3
cades (1, 2). Although 80% of the overall chronic disease burden Address correspondence to PW Kolsteren, Nutrition and Child Health
occurs in LMICs (3, 4), chronic diseases remain mostly un- Unit, Department of Public Health, Prince Leopold Institute of Tropical
derappreciated as a public health issue in these countries (5, 6). Medicine, Nationalestraat 155, 2000 Antwerp, Belgium. E-mail: pkolsteren@
itg.be.
Key determinants of chronic diseases are inadequate diets, 4
Abbreviations used: ES, effect size; HICs, high-income countries;
physical inactivity, smoking, excessive alcohol use, and obesity LMICs, low- and middle-income countries; PA, physical activity; SES, so-
(7). Obesity rates have more than doubled over the past 2–3 cioeconomic status.
decades in many high-income countries (HICs) (4, 8, 9). Com- Received January 18, 2012. Accepted for publication May 9, 2012.
parable trends in LMICs (10) have been associated with rapid First published online July 3, 2012; doi: 10.3945/ajcn.112.035378.
Am J Clin Nutr 2012;96:415–38. Printed in USA. Ó 2012 American Society for Nutrition 415
Supplemental Material can be found at:

http://ajcn.nutrition.org/content/suppl/2012/07/31/ajcn.112.0
35378.DC1.html
416 VERSTRAETEN ET AL
reviews linked to LMICs either focused on only one region (ie, overweight or obesity through dietary and/or PA behavior; and
China) or had a narrow focus on one specific intervention (ie, PA) 5) include both baseline and postintervention measurements
(34, 35). Moreover, a comprehensive and systematic synthesis of of dietary and PA behavior outcomes and/or anthropometric
study results, including both anthropometric and behavioral outcomes. Studies targeting parental or teacher behavior were
outcome measures, is currently lacking (33). Given the differ- eligible if outcome data could be extracted for children and/or
ences in contextual and cultural factors between HICs and LMICs adolescents. The following studies were excluded: 1) corre-
(36), the public health challenges they are facing, and the im- spondence letters, book chapters, dissertations, conference
portance of basing actions on sound evidence, we conducted proceedings, and abstracts; and 2) secondary prevention in-
a systematic review to identify effective pathways that alter terventions targeting only overweight, obese, or underweight
behavior and/or BMI in schoolchildren in LMICs. subjects.
METHODS Study selection

We conducted a systematic review of the effect of school- First, the title and abstracts of the identified references were
based interventions in LMICs aimed at the primary prevention of screened to select relevant studies based on the inclusion criteria.
obesity through changes in dietary behavior, PA behavior, or both If insufficient information was available from the title and ab-
in children and adolescents 6–18 y of age. To avoid biased post stract, the full text was read. The full text of the selected studies
hoc decisions, the inclusion criteria and analytic methods were was then retrieved and read to determine whether the inclusion
specified in the review protocol before the review was conducted criteria were met. The selection process was performed in-
and were based on the Cochrane handbook (37). The Preferred dependently by 2 reviewers (RV and CL). Disagreement between

Reporting Items for Systematic Reviews and Meta-Analyses reviewers was resolved by discussion until a consensus was
statement was followed as a guideline to report this systematic reached. In cases of a disagreement, a third reviewer was con-
review (38, 39). sulted (PWK).
Search strategy Data extraction

The following electronic databases were searched: MEDLINE Data extraction of the following study characteristics was
(PubMed; http://www.ncbi.nlm.nih.gov/pubmed), EMBASE (http:// performed by one reviewer (RV): study design, study setting
www.embase.com), Web of Science (http://www.webofknowledge. (urban/rural), characteristics of participants (including number,
com), CENTRAL (http://www.thecochranelibrary.com/), Edu- number of schools, age, sex, and SES), type of intervention
cation Resources Information Center (ERIC) (http://www.eric.ed. (including type, intervention duration, length of follow-up, and
gov/), The Cochrane Library (http://www.thecochranelibrary. focus), adverse effects, and theoretical framework of the in-
com/), and the Centre for Reviews and Dissemination (http:// tervention. Data extraction sheets were pilot-tested on 4 papers
www.crd.york.ac.uk/crdweb/SearchPage.asp). The initial search before applying them to all included studies. Study results (ie, the
string was developed in MEDLINE by using the Population, impact estimates) and the information necessary to evaluate study
Intervention, Comparison and Outcome model (40), combining quality appraisal were extracted independently by $2 reviewers.
population (healthy children and adolescents in LMICs), in- Disagreements were resolved by discussion until a consensus
tervention (school-based primary prevention for PA and/or was reached. If necessary, a third reviewer was consulted to
nutrition), comparison (controlled trial), and outcome (an- obtain a final decision. If $2 articles presented data from the
thropometric and/or behavioral change outcomes) terms (see same study, the results were included only once in the tables, but
Supplemental Table 1 under “Supplemental data” in the online all the references linked to the original study were noted.
issue). The search string was further refined for use in the dif-
ferent databases. It included both text words and thesaurus Summary measures
terms, which were adapted for each of the databases searched.
Searches were conducted in April 2010 and updated in July To the extent that they were reported in the articles reviewed, 2
2011 for all databases except EMBASE and ERIC because of types of impact estimates were tabulated for each study: 1) the
difficulty accessing them. Additional eligible studies were double difference effect, ie, the difference in the change of
identified from the bibliographies of published reviews and in- means over time (ie, from baseline to follow-up) between the
cluded articles. intervention group and the control group; and 2) the baseline-
adjusted effect, ie, the comparison between the means of the
intervention and control groups at follow-up, adjusted for values
Eligibility criteria of the outcome at baseline. After these data were extracted, it
The review was limited to studies published in English, became apparent that some studies did not report statistical tests
French, Spanish, Dutch, and German between January 1990 and between the intervention and control groups. Consequently,
July 2011. This time frame was chosen because it covers the studies for which no statistical tests were provided or outcomes
worldwide trend of increasing childhood obesity prevalence over for which no significance levels were reported were not included
the past decades. To be eligible for inclusion, studies had to 1) be in the result tables. In addition, the effectiveness of studies es-
conducted in a school setting in an LMIC, based on the World timating the effect of the intervention on at least one proximal
Bank classification (41); 2) include healthy children and ado- (ie, PA or dietary behavior) and a distal outcome (ie, BMI or
lescents 6–18 y of age; 3) use a controlled trial design (with or overweight/obesity prevalence) was evaluated. The authors were
without randomization); 4) focus on primary prevention of contacted to obtain further information when needed.
SCHOOL-BASED OBESITY-PREVENTION INTERVENTIONS 417
Quality appraisal not appropriate. When possible, effect size (ES) was computed to
Methodologic quality was assessed by using the Effective allow comparison of the effectiveness of the interventions across
Public Health Practice Project Quality Assessment Tool for studies:
Quantitative Studies 2008 (42). This standardized quality- ES ¼ ðmean for I 2 mean for CÞO½average SD for I and C ðpooled SDÞ
assessment tool evaluates study quality in 8 domains: selec-
ð1Þ
tion bias, allocation bias, control for confounding, blinding,
data-collection methods, loss to follow-up, statistical analysis,
and intervention integrity. The tool assigns ratings of 1 (lowest where I is the intervention group and C is the control group. ES
quality), 2 (moderate quality), or 3 points (highest quality) to was calculated for the following outcomes: dietary and PA be-
each of the 6 first domains. The overall quality score is then havior, BMI, and the prevalence of obesity and/or overweight.
calculated for each study by adding these ratings. They were categorized as trivial (,0.2), small (0.2 to ,0.5),
medium (0.5 to ,0.8), or large ($0.8) (43). The ES of in-
Data synthesis terventions was presented by outcome and grouped into 3 cat-
Because of heterogeneity in the studies in terms of participants, egories depending on the type of intervention: 1) diet, 2) PA, and
types of intervention, and outcome measures, a meta-analysis was 3) diet and PA.
FIGURE 1. Flowchart for the inclusion of studies. MEDLINE: http://www.ncbi.nlm.nih.gov/pubmed; CRD: http://www.crd.york.ac.uk/crdweb/SearchPage.asp;
Cochrane Library: http://www.thecochranelibrary.com/; Web of Science: http://www.webofknowledge.com; EMBASE: http://www.embase.com; ERIC: http://www.
eric.ed.gov/. CRD, Centre for Reviews and Dissemination; ERIC, Education Resources Information Center; LMICs, low- and middle-income countries.
RESULTS tables, only 12 studies reported the information required to

calculate the ES. Of the 16 authors contacted to provide addi-
Description of included studies tional information, 4 responded but only 2 were able to provide
The systematic search strategy identified 7218 unique refer- the required information.
ences, of which 104 articles were included for full-text review.
After full-text screening of these articles, 25 studies (presented in Dietary behavior
29 publications) were found to meet the inclusion criteria and
were included for quality appraisal and analysis (Figure 1). The Five studies (2 diet and 3 combined interventions) reported
characteristics of these studies are presented in Table 1. Overall, significant effects on one or more dietary behaviors (Table 2).
13 interventions involved school staff, communities, parents, The diet interventions reported a positive effect on preferences
children, and/or families and are referred to as multicomponent for healthy food (49) and a decrease in daily consumption of
interventions in the remainder of this review. Four of the 25 sweetened carbonated drinks (51). A significant decrease in the
interventions provided an individual counseling component (44– fast food eating behavior score (62), in the frequency of fast
47). food consumption in general (67) and in schools (72), and in
Four studies were diet-only interventions (48–51), 10 studies fried food consumption, soda intake, and snacks high in fat,
were PA-only interventions (47, 52–61), and 11 studies involved sugar, and salt (67) were observed in favor of the combined
both diet and PA interventions (referred to as combined inter- interventions. The only study for which ES could be calculated
ventions in this review) (44–46, 62–72). Diet-only interventions had a small effect (62).
mainly used nutrition education promoting healthy diets as a key Except for one study, only multicomponent studies were ef-
intervention strategy (48, 49, 51); one study was a breakfast fective in changing behavior. Most provided nutrition education

program (50). Eight of 10 PA interventions provided additional or nutrition and PA education as the intervention strategy, which
physical-education sessions ranging from 50 to 315 min extra were implemented by the teachers. All but one study had a low
per week (52–54, 56–60). Except for one study not reporting quality level (49).
these data (47), these studies integrated their additional sessions
into the existing curriculum. The remaining 2 PA interventions PA behavior
compared programmed PA with a regular physical-education Nine studies (7 PA interventions and 2 combined in-
curriculum (55) and provided tailored education according to the terventions) reported on physical fitness (n = 5) and/or PA time
different stages of the Health Action Process Approach (61). (n = 4) (Table 3). With regard to the PA interventions only, those
Most (n = 9) of the combined interventions used diet and PA measuring physical fitness (n = 3) as an outcome showed a sig-
education as their key strategy (45, 46, 62, 65–72), which was nificant increase in performance on most fitness indicators (52,
accompanied by environmental or organizational changes in the 54, 59). Of those interventions evaluating time spent being
schools in 5 of them (46, 62, 65, 66, 69–71). The remaining 2 physically active (n = 4), all but one (58) found a significant
interventions provided lectures and group counseling on car- increase in the intervention group (47, 53, 56, 57). Overall, the
diovascular disease prevention (44) and trained primary care- ES of the PA interventions ranged from trivial to large (range:
givers of children in healthy nutrition and activity (63, 64). 20.48, 1.61) (47, 53, 54, 58). The effective PA interventions
Most of the studies were conducted in Latin America (n = 13) were provided by teachers and included additional PA sessions
and Asia (n = 8). Only 6 studies based the intervention design on or healthy PA education integrated into the existing curriculum.
a theoretical framework. Two-thirds of the studies were con- Two of them were multicomponent interventions (47, 53).
ducted in major urban areas (n = 16), 2 studies included a rural The 2 combined interventions showed a beneficial effect on all
area, and the remaining studies did not report this information. fitness tests for both boys and girls (69–71); the ES varied from
The number of participants ranged from 135 to 4700, and the small to medium (69). Also worthy of note, the study with the
participants’ mean age ranged from 6.5 to 18.4 y. The median highest quality score was also the one that reported the largest
intervention duration was 9 mo (range: 1 h to 4 y), and the ES.
median length of follow-up was 11 mo (range: 4 wk to 4 y).
Except for 3 studies including only boys (44, 59) or girls (47),
all interventions targeted both sexes. Thirteen studies reported BMI
on the SES of the area, students, and/or parents. Few studies (n = Eight of the 12 studies with BMI data reported a statistically
3) reported data on adverse effects (60, 67, 69). significant effect for the intervention (Table 4). The 2 dietary
behavior interventions did not have a significant effect on mean
BMI (50, 51). All PA studies, except for one (55), found a sig-
Effectiveness and effect sizes of interventions nificant effect on BMI or BMI z score for the overall sample (59,
The results are presented by outcomes. Anthropometric out- 60) or for girls (56, 57). No ES could be calculated for these
comes such as waist circumference and skinfold thickness were studies. Five of 6 combined interventions reported a beneficial
evaluated in too few studies to be reported in this review. Sur- effect on BMI or BMI z score for the overall sample (44, 63, 64,
prisingly, 3 studies did not report significance levels on any 68, 70, 71) or for boys only (69). The largest ES was for the 2
outcome of interest (48, 61, 65, 66) and therefore were not in- best-rated studies in terms of quality, which were both con-
cluded in the result tables. Six additional studies did not report ducted in children (mean age: ,12 y), were both combined diet
significance testing on some of their outcomes; likewise, these and PA interventions, and found significant positive effects on
outcomes are not reported in the tables (45, 46, 55, 58, 62, 67). Of BMI or BMI z score (63, 64, 68). In both studies, the key in-
the 22 studies (presented in 25 publications) included in the result tervention strategies were an integrated curriculum delivered by
TABLE 1
Characteristics of studies (n = 25) included in the quality appraisal targeting diet only, PA only, or both1
Study characteristics Intervention Outcomes measured
Clusters (I and C), Intervention

no. of participants, duration, Based on
Reference, country, average age (y), length of theoretical Quality BMI-
province/city Design male (%) follow-up Description framework appraisal Diet PA related
Diet intervention
Fernandes et al, Cluster CT I = 1 school Duration: Aim: A nutrition-education program to No Low x x
2009 (48), Brazil, C = 1 school 8 meetings improve dietary intake among second-grade
Florianopolis n = 1352 of 50 min students of primary school to prevent
8.15 y3 Follow-up: obesity/overweight
47.4% 5 mo Intervention: Teachers provided 8 meetings of
50 min each about healthy diets, healthy
snacks, and PA
Control: The control group did not receive any
nutrition education
Gaglianone et al, Cluster RCT I = 3 schools Duration: Aim: A nutrition-education program Yes Moderate x
2006 (49), Brazil, C = 5 schools 14 wk “Reducing Risks of Illness and Death in
Sao Paulo n = 6372 Follow-up: Adulthood” (RRIDA) to improve
8.5 y3 12 mo knowledge and attitudes associated
NR with healthy eating habits of first- to
second-grade students in primary
schools to prevent obesity
Intervention: Overall, 22 h of nutrition
education were provided by conducting
3 weekly activities of 30 min each; teachers
were trained twice a week for 6 wk
Control: The control group did not
receive any nutrition education
Ramı́rez-López et al, Quasiexperimental NR Duration: Aim: A breakfast program to prevent No Low x
SCHOOL-BASED OBESITY-PREVENTION INTERVENTIONS
2005 (50), Mexico, prospective study n = 3602 9 mo obesity and cardiovascular disease
Sonora state 8.5 y3 Follow-up: risk factors among first- to fifth-grade
NR 9 mo students in primary schools
Intervention: Breakfast was consumed
30 min before the start of the school
day over a 9-mo period
Control: The control group did not receive
breakfast
(Continued)
419

420
TABLE 1 (Continued )

Sichieri et al, Cluster RCT 2 schools: Duration: Aim: An educational program to prevent No Low x x
2009 (51), Brazil, I = 23 classes I = 7.96 mo, excessive weight gain by reducing the
Niteroi, Rio de Janeiro C = 24 classes C = 8.24 mo consumption of sugar-sweetened beverages
n = 1134 Follow-up: of fourth-grade students in primary schools
10.9 y one school Intervention: Research assistants provided
47.1% year ten 1-h sessions about water intake
(classroom quizzes and games, song and
drawing competitions); water bottles and
banners promoting water were distributed
Control: The control group received two
PA intervention 1-h sessions on health issues
Bonhauser et al, Cluster CT I = 2 classes Duration: Aim: A PA program on physical fitness and Yes Moderate x
2005 (52), Chile, C = 2 classes 10 mo mental health status aimed at chronic
Santiago n = 198 Follow-up: disease prevention among adolescents
15.5 y3 10 mo Intervention: Teachers provided 4 units
48.4% of PA sessions; each unit comprised
10 consecutive weeks of lectures in which
3 PA sessions of 90 min each were
provided every week; the sessions were
VERSTRAETEN ET AL
included in the curriculum and consisted

of 3 steps: minimum activity, weight-
transfer activities, and sports practice
Control: The control group received
one standard session of 90 min of PA
de Barros et al, Cluster RCT (matching I = 10 schools Duration: 9 mo Aim: A PA program “Saude Na Boa” Yes Low x
2009 (53), Brazil, schools by size and C = 10 schools Follow-up: 9 mo among high-school students aimed at
Recife and location) n = 2155 improving PA and nutritional behavior
Florianopolis 18.4 6 2.3 y4 to prevent obesity
NR Intervention: The intervention included
an environmental and organizational
change (free fruit distribution on 1 d for
10 wk, special events, equipment), PA,
healthy diet education integrated into
the curriculum, and staff training and
engagement (8 sessions)
Control: NR
(Continued)


Draper et al, Qualitative study I = 3 schools Duration: NR Aim: A pilot study to assess feasibility and No Low x
2010 (54), South Africa, C = 2 schools Follow-up: 4 mo acceptability of the school-based PA
Alexandra Township n = 508 intervention “Healthnutz” to prevent
NR chronic diseases
NR Intervention: Teachers provided physical
education integrated into the curriculum
to students; 5 focus groups and a situational
analysis with teachers and group monitors
were conducted
Control: NR
Farias et al, Cluster CT I = 1 school Duration: 10 mo Aim: A PA program introducing programmed No Low x
2009 (55), Brazil, C = 1 school Follow-up: PA to improve the body composition of
Porto Velho and Rondônia n = 3832 10 mo adolescents in fifth to eighth grades
12.4 y3 Intervention: All adolescents participated in 2
53.2% weekly 60-min sessions of physical
education (68 classes in total); the
intervention group received programmed
PA, which consisted of 3 steps: starting with
aerobic activity, playing sports for 30 min,
and ending with stretching for 10 min
conventional physical education
Li et al, 2010 (60), Cluster RCT I = 10 schools Duration: 1 y Aim: The PA program “Happy 10” to No Moderate x
China, 2 Beijing C = 10 schools Follow-up: 2 y promote PA to prevent overweight in
districts n = 4700 primary school students

9.3 6 0.7 y Intervention: Teachers organized and
52.3% implemented 2 daily PA sessions of 10 min
each; the intervention was integrated into
the curriculum and did not replace any other
activities; the progress of the class was
shown on posters and stickers
Control: The control group received the
standard curriculum
(Continued)
421

422

Liu et al, 2007 (57) Cluster CT I = 1 school Duration: Aim: The pilot PA program “Happy 10” No Low x x
and Liu et al 2008 (56), C = 1 school 8 mo to promote PA, obesity control and
China, Beijing district n = 753 Follow-up: prevention, and the growth and development
[pilot study of Li et al 9 y3 8 mo of primary school students
2010 (60)] 47.4% Intervention: Teachers organized and
implemented 10 min of PA at least once
a day during 2 semesters; the progress of
the class was shown on posters and stickers
Control: The control group received no
intervention
McManus et al, Cluster RCT I = 2 schools Duration: Aim: A PA program aimed to increase Yes Low x x
2008 (58), China, C = 1 school unclear PA by providing heart-rate feedback in
Hong Kong n = 210 Follow-up: fourth- and fifth-grade students in
10.44 6 0.85 y 6 mo primary schools
50% Intervention: There were 1 control
group and 2 intervention groups;
for 2 wk, one intervention group received
VERSTRAETEN ET AL
an educational program and heart-rate

feedback, whereas the other group
received heart-rate feedback with
normal physical-education sessions;
the educational program included
heart-rate monitor skills, heart health
education, goal setting, and role-play;
it was implemented as a module
within the physical-education
curriculum; after the educational
part, children in the intervention
groups completed 2 wk with
heart-rate feedback and 2 wk without
heart-rate feedback (counterbalanced)
Control: The control group received no
intervention
(Continued)


Meszaros et al, Cohort analytic study I and C = 18 Duration: Aim: A physical-education program No Low x x
2009 (59), Hungary, schools 4y to analyze differences in growth
Budapest, Gyor, n = 521 Follow-up: patterns, age-related changes in
and Szigelszentmiklos Entrance age 4y body fat, and physical performance
varied from in children taking part in elevated
6.51 to 7.5 y physical education or normal physical
100% education at school
Intervention: Additional physical-education
sessions were integrated into the
curriculum; teachers provided 8
morning sessions of 45 min each
and 2 afternoon sessions of 90 min
each in 10-d cycles; afternoon sessions
consisted of cardiorespiratory fitness
and special skills, and the morning
sessions consisted of general requirements
of the physical-education curriculum
Control: The control school received the
obligatory 5 sessions of 45 min for
each 10-d cycle
Schwarzer et al, RCT 2 schools: Duration: Aim: A PA intervention tailored Yes Low x
2010 (61), China I = 9 classes 1h to different stages as identified by
C = 6 classes Follow-up: the Health Action Process Approach
n = 5342 1 mo to enhance PA
13.8 6 1.4 y Intervention: Two theory-guided
46% interventions were provided to
the intervention group: resource
communication and strategic planning;

students received 1 of the 2 interventions
according to the stage they were in, ie,
preintenders (resource communication),
intenders (strategic planning), and actors;
students in both intervention groups
received tailored instructions for 1 h
and homework; teachers were trained
to deliver the intervention
regular school instruction
(Continued)
423

424

Taymoori et al, Cluster RCT I = 2 schools Duration: Aim: An individually tailored PA Yes Moderate x
2008 (47), Iran C = 1 school 6 mo intervention for Iranian girls in
n = 1612 Follow-up: secondary schools to promote health
14.79 6 0.44 y 12 mo Intervention: There were 2 intervention
0% groups and 1 control group: 1 intervention
was based on Pender’s HP model, whereas
the other intervention was based on
an HP model integrated with constructs
from the THP; both intervention groups
received 45–60-min educational group
sessions (4 times) and 20–25-min individual
counseling sessions (4 times); the THP
group received 2 additional sessions
targeting the process of change. Other
activities included teacher training, two
60-min educational sessions with mothers,
and one PA trip with teachers and mothers.
VERSTRAETEN ET AL

pamphlets after the final follow-up
of the intervention
Diet and PA
interventions
Alexandrov et al, Cluster CT I = 7 schools Duration: Aim: An intervention aimed at preventing No Low x
1992 (44), Russia, C = 16 schools 1y risk factors of cardiovascular diseases
Moscow n = 7662 Follow-up: Intervention: Primary prevention: all
11.8 y3 3y intervention children received lectures and
100% group counseling on prevention by
trained instructors; secondary prevention:
children with risk factors were invited
with their parents to a single
individual counseling session
Control: NR
(Continued)


Banchonhattakit et al, Quasiexperimental I = 6 schools Duration: Aim: The obesity-prevention program No Low x x x
2009 (62), Thailand, pretest-posttest C = 6 schools 8 mo “SNOCOP” in fifth-grade students
Saraburi Province time series design n = 375 Follow-up: in primary schools; it aims to improve
10.6 y3 8 mo student behavior by means of a
48.5% school network
Intervention: The school network supported
the implementation of the intervention,
had monthly meetings, and consisted
of parents, school administrators,
teachers, and community leaders; a
healthy diet, 30 min of daily PA,
changing school policies, and improved
school-lunch programs were promoted
Control: NR
Bruss et al, Y1 = stepwise I = 6 schools Duration: Aim: The cognitive behavioral Yes Moderate x
2010 (63), and randomized trial; C = 6 schools 2y lifestyle intervention “Project
Bruss et al, 2010 Y2 = stepwise n = 407 Follow-up: Familia Giya Marianas” for
(64), CNMI nonrandomized 8.55 6 0.50 y 2y caregivers of third-grade students
trial 49.6% in primary schools in CNMI to
prevent childhood obesity
Intervention: Community-based
participatory research was used
to design and evaluate the
effectiveness of this obesity-
prevention intervention on BMI
of children; a crossover design
was used; school personnel were

trained and provided 8 sessions of
90 min each for primary caregivers
of students; main topics: physiologic
and sociocultural issues, psychosocial
issues, and dietary behavior and
(in)activity
intervention as delayed treatment
(Continued)
425

TABLE 1 (Continued ) 426

Colin-Ramirez et al, Cluster RCT I = 5 schools Duration: Aim: An intervention aimed at improving No Low x x x
2009 (65), and C = 5 schools 12 mo dietary habits and PA to prevent
Colin-Ramirez et al, n = 4982 Follow-up: cardiovascular diseases in 8–10-y-old
2010 (66), Mexico, 9.5 6 0.7 y 12 mo students in primary schools (RESCATE)
Mexico City 53.6% Intervention: The intervention had 3
components: individual level (nutrition
education and exercise breaks), school
environment (healthy snacks and physical-
education classes delivered by health team
for 30 min twice weekly), and family
participation (homework, recipes,
recommendations); teachers were trained to
deliver the intervention
Control: NR
Francis et al, Cluster RCT I = 5 schools Duration: Aim: A short-term multicomponent No Low X x
2010 (67), Trinidad C = 6 schools 1 mo education intervention on dietary and PA
and Tobago, n = 4722 Follow-up: behaviors among primary school children
Sangre Grande 10.4 y3 3 mo Intervention: A curriculum including classes
52% on nutrition and PA was implemented by
trained teachers; the nutrition curriculum
focused on food groups, nutrients, and
VERSTRAETEN ET AL
types of food, whereas the PA component

included exercise for 10 min
Control: NR
Jiang et al, Cluster RCT I = 2 schools Duration: Aim: A school-based intervention No Moderate x
2007 (68), China, C = 3 schools 3y program focusing on nutrition education and
Beijing n = 24252 Follow-up: PA to prevent or reduce obesity in students
(entrance) 3y from primary schools
age = 8.3 y3 Intervention: Teachers were trained to deliver
51.3% the intervention; primary prevention:
nutrition lectures to all parents once per
semester and a student curriculum including
10 themes (one theme/lesson); secondary
intervention: additional sessions for
overweight and obese students, including
20 min/d of extra PA for 4 d/wk and meetings
with their parents and themselves
(once/semester).
usual curriculum
(Continued)


Kain et al, Cluster CT I = 3 schools Duration: Aim: A nutrition-education and PA intervention No Low x x
2004 (69), Chile, C = 2 schools 6 mo in primary school children to prevent obesity
Santiago, Curico, n = 30862 Follow-up: Intervention: The nutrition component
Casablanca 10.6 y one school comprised an educational program for
53% year students (8–11 h in fourth to sixth grades,
5–6 h in seventh to eighth grades), 2 meetings
with school kiosks encouraging them to sell
healthy food, 2 meetings with parents, and
special activities; the PA component
comprised a behavioral PA program, 90 min
additional weekly PA classes (6 mo), active
recess (15 min/d for 3 mo), extra sports
materials, and special activities; school
teachers were trained to deliver
both components
Control: NR
Kain et al, 2008 (71), Cluster CT I = 3 schools Duration: Aim: A nutrition-education and PA intervention No Low x x
and Kain et al, C = 1 school 11 mo in primary school children to prevent obesity
2009 (70), Chile, n = 20392 Follow-up: Intervention: The nutrition component
Casablanca, Quillota 9.9 y3 21 mo comprised an educational program for
53.8% students (8–11 h in fourth to sixth grades,
5–6 h in seventh grade), 2 meetings with
parents, and special activities; the PA
component comprised a behavioral PA
program, 90 min additional weekly PA
classes, and active recess (4 mo); school
teachers were trained to deliver both
components; the intervention was fully

applied in year 1 and partly in year 2;
in the second year, the intervention
comprised talks only with parents of obese
children, the educational program (4 h to all
grades), 90 min of additional weekly
PA classes, and training to improve
the quality of the physical-education
classes
Control: NR
(Continued)
427

TABLE 1 (Continued ) 428

Kain et al, 2010 (45), Cohort before and after I = 4 schools Duration: Aim: An intervention to prevent childhood No Low x
Chile, Santiago study C = 3 schools 2y obesity, including nutrition education and
(Macul) n = 6492 Follow-up: increase in PA that involved implementing
children, 2y a counseling program on healthy lifestyles
452 teachers for teachers
NR Intervention: The nutrition component included
53% an educational program and teacher training;
the PA component included an educational
program, 2 mo of training for physical-
education teachers, and sports equipment;
teachers in 4 of 7 schools received 3 private
counseling sessions on nutrition and PA
Control: The control group received the same
nutritional and PA components, but there
were no counseling sessions for teachers
Singhal et al, Cluster CT (matched) I = 1 school Duration: Aim: A multicomponent nutrition and No Low x x
2010 (46), C = 1 school 6 mo lifestyle-education intervention aimed at
northern India 2012 Follow-up: changing knowledge, behavior, anthropometric
n = 16 y3 6 mo measures, and metabolic risk profile
60.2% Intervention: The intervention included
a multicomponent model including
VERSTRAETEN ET AL
promotion of PA, activities to promote

healthy lifestyle, individual counseling, policy-
level changes in school, involvement of
teachers and parents, training of student
volunteers, and focus group discussions
Control: The control school did not receive
any intervention
Vargas et al, 2011 Cohort before and after I = 1 school Duration: Aim: effects of obesity-prevention No Low x
(72), Brazil, study C = 1 school 4 mo program on dietary practices
Niterói n = 3312 Follow-up: Intervention: Nutrition-education and PA
13 y3 4 mo promotion were provided with a focus on
NR nutrition education
Control: The control group received a
normal curriculum
1
C, control; CNMI, Commonwealth of the Northern Mariana Islands; CT, controlled trial; HP, Pender’s health promotion model; I, intervention; NR, not reported; PA, physical activity; RCT, randomized
controlled trial; RESCATE, programa de REducción de rieSgo CArdiovascular Total Escolar; SNOCOP, School Network for Childhood Obesity Prevention; THP, trans-theoretical model integrated with Pender’s
health promotion model.
2
Sample included in analysis.
3
Estimate of average age.
4
Mean 6 SD (all such values).

TABLE 2
Effect on dietary behavior of studies providing significance levels (n = 5)1
Double difference effect: Baseline-adjusted effect:
changes over time in I I and C at follow-up adjusted for
and C and I 2 C baseline values of outcome Effect size
Reference and outcome (Sub)sample DI DC DI 2 DC P value I C P value Estimate Classification Quality appraisal
Diet intervention
Gaglianone et al, 2006 (49)
Preferences for healthy foods All NR NR NR 0.012 Moderate
Sichieri et al, 2009 (51)
Sugar-sweetened carbonated beverage intake All 269 213 256 ,0.05 Low
per class (mL/d)
Diet and physical-activity interventions
Banchonhattakit et al, 2009 (62)
Fast food eating behavior score All
Y2 30.02 6 0.252 28.78 6 2.98 0.001 0.77 Medium Low
Y3 29.63 6 0.27 28.9 6 3.2 0.001 0.42 Small Low
Francis et al, 2010 (67)
Fruit intake $2 servings/d (%) All 25 5.10 19.9 NR NR NR NS3 Low
Vegetable intake $2 servings/d (%) All 21.8 28.8 7.0 NR NR NR NS3 Low
Mean soda intake, 8-oz (servings/wk)4 All 0.9 4.5 23.6 NR NR NR ,0.053 Low
Snacks high in fat, sugar, and salt eaten in the All 211 0.12 211.12 NR NR NR ,0.053 Low
past 24 h (%)
Mean fried food (servings/d) All 21.3 20.2 21.1 NR NR NR ,0.053 Low
Vargas et al, 2011 (72)
Mean frequency of fast food consumption at All NR NR NR 0.001 Low
school cafeteria
Mean daily consumption of fruit and vegetables All NR NR NR NS Low

1
References 48, 49, 51, 65, 67, and 72: unable to calculate effect sizes based on reported results. References 46, 48, 65, and 66: reported outcome but did not provide statistical tests between I and C. C,
control; I, intervention; NR, not reported; Y2, year 2; Y3, year 3; DC, change in outcome for control; DI, change in outcome for intervention.
2
3
Regression analyses for overall fruit, vegetable, and soda intakes; snacks high in fat, sugar, and salt; and fried food (after intervention, controlled for confounders).
4
1 oz = 29.6 mL.
429

TABLE 3 430
Effect on PA behavior of studies providing significance levels (n = 9)1
Double difference effect: Baseline-adjusted effect: I and C
changes over time in I at follow-up adjusted for baseline
and C and I 2 C values of outcome Effect size
Quality
Primary outcome (Sub)sample DI DC DI 2 DC P value I C P value Estimate Classification appraisal
PA interventions
Bonhauser et al, 2005 (52) Moderate
V˙O2 max (mL $ kg21 $ min21) All +2.97 +0.65 +2.32 ,0.005
Speed performance (m/s) All +0.50 +0.33 +0.17 0.001
Jump performance (cm) All +2.29 +0.53 +1.76 ,0.005
de Barros et al, 2009 (53)
Duration MVPA .60 (d/wk) All 3.3 6 2.12 2.6 6 2.1 ,0.001 0.33 Small Low
Prevalence of individuals meeting PA All 33.1 23.7 0.001
recommendations (%)
Prevalence of physically inactive individuals (%) All 10.59 17.30 ,0.05
Draper et al, 2010 (54)
10-m shuttle run (s) All 46.2 6 4.6 48.6 6 5.5 ,0.00013 20.48 Small Low
No. of sit-ups in 30 s All 17.8 6 6.1 15.5 6 5.1 ,0.023 0.41 Small
Sit and reach (cm) All 19 6 6.8 14 6 9.7 ,0.0013 0.61 Medium
Ball throw (cm) All 23.1 6 7.5 21.8 6 6.9 NS3 0.18 Trivial
Long jump (cm) All 134.3 6 25.7 135 6 19.7 NS3 20.03 Trivial
Liu et al 2007, (57); Liu et al, 2008 (56)
PAEE (kcal $ kg21 $ d21) All +3.1 29.6 +12.7 ,0.05 Low
PA time (h/d) All +0.5 21.5 +2 ,0.05
McManus et al, 2008 (58)
Time spent above flex heart rate (%/d) All EG: 35.3 6 23.9 40.4 6 23.9 NS4 20.21 Small Low
VERSTRAETEN ET AL
All NEG: 38.5 6 27.5 20.07 Trivial

Time moderately active (%/d) All EG: 1.5 6 1.9 1.7 6 1.6 NS 20.11 Trivial
All NEG: 1.6 6 2.10 20.05 Trivial
Time vigorously active (%/d) All EG: 0.50 6 1.6 0.55 6 0.80 NS 20.04 Trivial
All NEG: 0.48 6 0.73 20.09 Trivial
Peak V˙O2 (mL $ kg21 $ min21) All EG: 45 6 4.9 45.8 6 6.8 NR 20.14 Trivial
All NEG: 47.1 6 6.8 0.19 Trivial
Meszaros et al, 2009 (59)
Physical fitness All NR NR ,0.05 Low
Taymoori et al, 2008 (47)
Mean PA (min/d) All (after intervention) THP: 60.0 6 134.3 46.3 6 21.9 0.0005 THP: 0.18 Trivial Moderate
All (after intervention) HP: 56.8 6 27.6 HP: 0.42 Small
All (follow-up) THP: 75.8 6 27.5 37.3 6 20.4 0.008 THP: 1.61 Large
All (follow-up) HP: 73.6 6 28.7 HP: 1.48 Large
Overall time PA (min) All (after intervention) THP: 310 6 24.8 245.3 6 101.3 0.000 THP: 1.03 Large
All (after intervention) HP: 285.9 6 141.3 HP: 0.33 Small
All (follow-up) THP: 371.2 6 129.6 196 6 99.7 0.01 THP: 1.53 Large
All (follow-up) HP: 348.1 6 159.0 HP: 1.18 Large
(Continued)


Quality
Primary outcome (Sub)sample DI DC DI 2 DC P value I C P value Estimate Classification appraisal
Diet and PA interventions

Kain et al, 2004 (69)
Lower back flexibility test (cm) Male +2.3 21.4 +3.7 ,0.05 23.6 6 8.6 22 6 6.3 ,0.0013 0.21 Small Low
Female +2.8 21 +3.8 ,0.05 25.7 6 7.9 23 6 6.4 0.00013 0.38 Small
Endurance (20 m) shuttle run (stages) Male +1.3 0 +1.3 ,0.05 5.0 6 1.9 3.9 6 1.9 ,0.0013 0.58 Medium
Female +0.7 20.3 +1 ,0.05 3.3 6 1.4 2.6 6 1.3 0.00013 0.52 Medium
Kain et al, 2008 (71) ; Kain et al, 2009 (70)
Mile test (min) Male 20.84 20.13 20.71 0.037 Low
Female 20.61 20.23 20.38 0.005
Shuttle run (stages) Male +1.15 +0.21 +0.94 ,0.0001
Female +0.3 20.16 +0.46 0.0007
1
References 52, 56, 57, 59, 61, 62, and 65–67: unable to calculate effect sizes based on reported results. References 61, 62, and 65–67: reported outcome but did not provide statistical tests between I and C.
C, control; EG, education group; HP, Pender’s health promotion model; I, intervention; MVPA, moderate-to-vigorous physical activity; NEG, no education group; NR, not reported; PA, physical activity; PAEE,
physical activity energy expenditure; THP, trans-theoretical model integrated with Pender’s health promotion model; V˙O2, maximal oxygen uptake; DC, change in outcome for control; DI, change in outcome
for intervention.
2
3
Interaction (group 3 time) effect.
4
Interaction effect group after 2 wk and after 6 mo.
5
Main effects.
431

TABLE 4 432
Effect on BMI of studies providing significance levels (n = 12)1
Outcome (Sub)sample DI DC DI 2 DC P value I C P value Estimate Classification Quality appraisal
Diet intervention
Ramı́rez-López et al, 2005 (50)
BMI All 17.2 6 0.12 16.9 6 0.2 NS 2 Large Low
BMI All +0.32 +0.22 +0.10 NS Low
PA interventions
Farias et al, 2009 (55)
BMI Male 20.2 6 3.2 20.8 6 3.4 NS 20.18 Trivial Low
Female 19.8 6 2.5 20.4 6 3.3 NS 20.21 Small
Li et al, 2010 (60)
BMI All
Y1 +0.56 +0.72 20.16 0.03 Moderate
Y2 +1.55 +1.67 20.12 0.04
BMI z score All
Y1 20.05 +0.01 20.06 0.03
Y2 +0.03 +0.08 20.05 0.03
Liu et al, 2007 (57); Liu et al, 2008 (56)
BMI Male +0.86 +0.72 +0.14 NS Low
Female 20.47 +0.66 21.13 ,0.05
Meszaros et al, 2009 (59)
BMI All NR NR NR ,0.05 Low
VERSTRAETEN ET AL
Alexandrov et al, 1992 (44)

BMI All
Y1 +0.54 +0.72 20.18 0.0063 18.0 6 2.2 18.4 6 2.7 ,0.05 20.16 Trivial Low
Y2 +0.22 +0.21 +0.01 NS 19.7 6 2.4 19.7 6 2.6 NS 0.00 Trivial
Bruss et al, 2010 (63), Bruss et al 2010 (64)
BMI z score All 20.04 6 0.46 0.14 6 0.59 ,0.00013 20.34 Medium Moderate
Jiang et al, 2007 (68)
BMI All 18.2 6 2.6 20.3 6 3.4 ,0.01 20.70 Medium Moderate
Kain et al, 2004 (69)
BMI Male 10.0 10.3 20.3 ,0.05 19.5 6 3.5 19.2 6 3.1 ,0.0014 0.09 Trivial Low
Female 10.3 10.2 20.1 NS 20.0 6 3.8 19.6 6 3.8 NS 0.11 Trivial
BMI z score Male 20.12 20.02 20.1 ,0.05 0.51 6 0.94 0.46 6 0.81 ,0.0014 0.06 Trivial
Female 20.04 20.07 20.03 NS 0.59 6 0.89 0.40 6 0.9 NS 0.21 Medium
Kain et al, 2008 (71); Kain et al, 2009 (70)
BMI Male 10.7 11.2 20.5 ,0.05 19.7 6 3.2 20.6 6 3.7 NS 20.26 Small Low
Female 10.8 11.4 20.6 ,0.05 20.1 6 3.5 20.8 6 3.8 NS 20.19 Trivial
(Continued)

Quality appraisal
References 46, 51, 56, 57, 59, and 60: unable to calculate effect sizes based on reported results. References 45 and 58: reported outcome but did not provide statistical tests between I and C. C, control;
teachers/school staff accompanied by regular nutrition education
of parents.
Low
Overweight and obesity prevalence
Three of the 7 studies reporting on this outcome significantly
decreased obesity prevalence in the intervention group (45, 68,
Classification
70, 71) by 0.8–32.5 percentage points (Table 5). All 3 studies
Trivial
Trivial
reported on combined multicomponent interventions. Of the 7
Effect size
studies, 2 were of moderate quality.
Effectiveness of studies measuring at least one proximal and

Estimate
20.19
20.14
one distal outcome

Five of the 22 studies included in the results tables (Tables 2–5)
I, intervention; NR, not reported; PA, physical activity; Y1, year 1; Y2, year 2; DC, change in outcome for control; DI, change in outcome for intervention. measured both a proximal outcome, ie, PA (56, 57, 59, 69–71) or
5
P value
diet (51), and a distal outcome such as BMI or overweight/

0.031
0.051
obesity prevalence (51, 56, 57, 59, 69–71). All 5 studies showed
at follow-up adjusted for baseline
Baseline-adjusted effect: I and C
a significant effect on the proximal outcomes they measured.

Four interventions (2 PA and 2 combined interventions) showed
values of outcome
0.72 6 0.90
an effect on both the proximal and the distal outcomes for at

0.72 6 1

least one subgroup (56, 57, 59, 69–71). Only the intervention
C
evaluating diet as an intermediate outcome did not have an ef-

fect on the distal outcome (51). All 5 studies were of low quality,
and ES could only be calculated for 2 of them (69–71).
0.53 6 0.95
0.59 6 0.90
I
Quality appraisal
The key methodologic limitations encountered were lack of
randomization (n = 10) or unclear/inappropriate randomization
P value
methods (n = 8), inexistent or unclear description of intention-

NR
NR
NS
to-treat analysis (n = 13), unblinded assessors (n = 25), absence

of reporting on dropout numbers and/or reasons (n = 10), and
Double difference effect:
changes over time in I
absence of reporting on intervention integrity (n = 25) and on

and C and I 2 C
DI 2 DC
20.14
20.14
20.13
process evaluation (n = 23). In 2 studies, contamination between

groups was likely because the intervention and control groups
were within the same school, or the same teachers taught both
10.05
10.08
10.06
the intervention and the control group (52, 65). Finally, only half
DC
of the studies with a cluster design (n = 17) applied an appro-

priate data analysis method, which resulted in likely over-
20.09
20.06
20.07
estimation of the statistical significance of the results. The

DI
overall quality evaluation for each study is shown in Table 1.

Many of the studies did not evaluate the outcome that the in-
(Sub)sample
tervention was intended to change. One of the 4 diet in-

Female
terventions did not report the effect on dietary outcomes, and 2

Male
All
of the 10 PA interventions did not report on PA. Of the 11

combined interventions, 4 did not evaluate the effect on either
diet or PA, and 4 studies evaluated only one of either.
Post hoc analysis (5–8 lessons received).
Interaction (group 3 age 3 time) effect.

Interaction (group 3 time) effect.
DISCUSSION
This was the first systematic review of school-based in-

terventions aimed at the primary prevention of obesity in children
and adolescents in LMICs. Twenty-five studies met the inclusion
Singhal et al, 2010 (46)
criteria. Of the 22 studies reporting useful statistics on the

outcomes of interest, 18 had a positive effect on one or more of

the outcomes (82%). Two of the 3 diet interventions that mea-
BMI z score
sured the adolescents’ diet significantly improved this outcome;

however, the diet interventions did not have an effect on any of
BMI
Outcome
the BMI-related outcomes. The 9 PA interventions were suc-

1
2
3
4
5
cessful at increasing PA (6 of 7 measuring PA) and lowering

BMI in at least one of the studied subgroups (3 of 4 evaluating
BMI), but did not have an effect on the prevalence of overweight shown to be associated with higher BMI (78, 79). Previous re-
or obesity (0 of 2 evaluating this outcome). The 10 combined search in HICs has shown that these sedentary behaviors can be
interventions had a significant effect on all outcomes: diet (3 of effectively addressed (80, 81) and indicate the need to include
3 measuring diet), PA (2 of 2 measuring PA), BMI (5 of 6 in these types of interventions as obesity-prevention strategies (82).
at least one subgroup), and the prevalence of overweight or Similarly, few interventions targeted the school environment, for
obesity (3 of 3 evaluating this outcome). Even though the ES instance with respect to the types and nutritional value of the
for BMI was classified as small or trivial (range: 20.7, 0.0), the foods sold in and around schools or as related to school policies.
public health effect at a population level can be substantial if Exposure to unhealthy energy-dense food in schools, which often
implemented in large groups of children and when sustained competes with healthier choices in terms of taste, price, and
over longer periods (73). We had expected the combined in- supply, was not assessed. It is likely that changing the nutritional
terventions to have a larger effect on BMI than interventions environment in schools poses a challenge for preventive inter-
addressing only diet or PA, but no differences were found when ventions in LMICs because the type and complexity of changes
the ES was compared between these types of studies. Even required is different from those in HICs at times, eg, at schools,
though studies were limited in number to draw firm conclu- street food vendors intermingle with privately owned food tuck
sions, combined interventions are more likely to decrease BMI. shops in the schools.
Understanding the potentially synergistic effects of combined A potential limitation of this review was the exclusion of
diet and PA interventions through evaluation studies with a studies based on language. We believe, however, that it is im-
2-by-2 factorial design would produce relevant knowledge for probable that important studies were left out. In our experience, it
policy makers. is very unlikely to find methodologically sound studies meeting
A key question relates to which intervention characteristics are the inclusion criteria in languages other than English, Spanish, or

associated with higher effectiveness. Interventions that positively French. We acknowledge that gray literature may have been an
changed both proximal and distal outcomes were generally mul- additional contribution to this review. However, as recommended
ticomponent, education-based interventions delivered by teachers by Doak et al (83), only peer-reviewed literature was included.
and providing additional PA sessions or integrated classes about The general limited quality of the included studies is a second
healthy foods, nutrition, or PA to encourage children to adopt a possible limitation. The lack of information, lack of blinding of
healthy lifestyle. Similar characteristics were found for in- assessors to the intervention, and lack of adjustment in the
terventions improving dietary or PA behavior only. Those studies analysis for clustering were the main quality-related issues. We
reducing BMI only were generally multicomponent interventions believe, however, that the methodologic limitations present did
targeting both diet and PA. In addition, these intervention char- not alter the main conclusion of our review, ie, that school-based
acteristics generally coincide with those identified in reviews of interventions in adolescents can successfully improve diet, PA,
intervention studies conducted in China and Latin America (34, and BMI. First, the positive effects found in the reviewed studies
35). Studies evaluating both proximal and distal outcomes lasted were consistent across a variety of settings. Second, and most
between 6 and 48 mo and generally had a long-term follow-up. importantly, the largest effects were found in the studies with the
Interestingly, all effective studies evaluating BMI only lasted highest quality score (47, 63, 64, 68).
longer than 12 mo. This is a surprising finding because previous Apart from the methodologic issues outlined, the studies in-
studies showed that effective interventions tended to be short term cluded in the review had many other limitations. First, several
(,12 mo) (36, 73). studies (n = 11) did not evaluate the proximal outcome(s) that
The effectiveness of multicomponent interventions, and es- the intervention targeted for change, and only 5 studies evalu-
pecially those that involve parents, is not surprising. As role ated the effect on both proximal and distal outcomes. Ideally,
models, parents shape the eating behaviors of their children and future intervention studies should evaluate the effect on diet and/
play an important role in the etiology and prevention of weight- or PA and on BMI (33). This would considerably strengthen the
related problems (74). In many LMICs, caregivers and families evidence and allow one to quantify to what extent improvements
are present more prominently in the daily life of children and in these proximal outcomes are translated into changes in BMI
adolescents, underscoring the importance of their involvement in and to potentially understand the bottlenecks in these pathways.
these school-based interventions. Recently, Hingle et al (75) In addition, only 2 studies conducted a process evaluation, ie,
reported that direct methods (eg, education or workshops on documented how and to what extent the intervention was imple-
healthy eating) that involve parents in diet interventions were mented as planned and adopted by the beneficiaries in a particular
more likely to be effective than were indirect methods (eg, in- setting. We believe that this was a missed opportunity for LMICs
formation leaflets and assignments). However, motivating and because this will provide the crucial information that is needed to
sustaining parental or family involvement in interventions remain adapt the program for implementation in other settings and to
challenging (75, 76). Key barriers identified by parents were scale up, especially because large differences in school settings
limited time availability and an unwillingness to be tutored by and nutritional and PA environments may exist.
schools (76). These limitations may help to explain why, despite A further limitation was that only 3 of the interventions in-
its importance, only half of the studies included in this review vestigated adverse effects. Obesity-prevention interventions may
were multicomponent interventions involving parents or families. aggravate social stigmatization or psychosocial problems of the
Notwithstanding this multicomponent nature, the scope of the already overweight or obese children or could lead to or exac-
strategies used by the interventions was surprisingly narrow. Only erbate underweight or eating disorders (36). Even though a recent
2 of the school-based interventions targeted sedentary behaviors meta-analysis concluded that preventive interventions are po-
such as television viewing or gaming. These behaviors are in- tentially harmless (73), future interventions should still include
creasingly observed in societies in transition (77) and have been these outcomes and report them by BMI category to provide
TABLE 5
Effect on the prevalence of overweight and obesity of studies providing significance levels (n = 7)1
Reference and outcome (Sub)sample DI DC DI 2 DC P value I C P value Estimate Classification Quality appraisal
Diet intervention
Ramı́rez-López et al, 2005 (50)
Overweight prevalence All 28 6 112 8 6 7.6 NS 2.15 Large Low
Obesity prevalence All 28 6 11 9 6 8.5 NS 1.95 Large
Overweight prevalence All 19.2 16.0 NS Low
Obesity prevalence All 4.4 4.5 NS
PA interventions
Li et al, 2010 (60)
Overweight and obesity prevalence All
Y1 NR NR NR NS Moderate
Y2 NR NR NR NS
Liu et al, 2007 (57); Liu et al, 2008 (56)
Overweight prevalence Male 23.8 +4.5 28.3 NS Low
Female 23.3 +3.7 27 NS
Obesity prevalence Male 20.4 +0.6 21 NS
Female 25.6 +0.7 26.3 NS
Jiang et al, 2007 (68)
Overweight prevalence All 226.3 +14.3 240.6 NR 9.8 14.4 ,0.013 Moderate
Obesity prevalence All 232.5 +15.7 248.2 NR 7.9 13.3 ,0.013
Kain et al, 2008 (71); Kain et al, 2009 (70)
Obesity prevalence Male 24.7 20.2 24.5 ,0.05 12.3 21.4 ,0.001 Low
Female 23.8 +0.5 24.3 ,0.05 10.3 15.2 ,0.01

Kain et al, 2010 (45)
Obesity prevalence All 20.8 24.2 +3.4 ,0.05 11.3 18.9 0.001 Low
1
References 45, 48, 55–57, 60, 62, 65, 68, 70, and 71: unable to calculate effect sizes based on reported results. References 48, 55, 62, 65, and 66: reported outcome but did not provide statistical tests
between I and C. C, control; I, intervention; NR, not reported; PA, physical activity; Y1, year 1; Y2, year 2; DC, change in outcome for control; DI, change in outcome for intervention.
2
3
Calculated for ORs.
435

a more complete view of the intervention effects for the whole process evaluations are needed to learn from program imple-
target population. mentation and adoption to identify which intervention compo-
A striking omission, given its importance when considering nents are effective and feasible. Only with this strong evidence
scaling up, was the lack of information on cost-effectiveness in all base will school-based interventions be able to reach their full
of the studies. Given the limited resources available in LMICs, potential of addressing unhealthy body weight in school-age
solid cost-effectiveness estimates would be of tremendous help to children in LMICs.
policy makers. Also related to scaling up is the need to understand
The authors’ responsibilities were as follows—RV, DR, CL, and PWK:
to what extent SES and urban/rural settings within LMICs modify were responsible for the design of the study, the development of the search
the effectiveness of interventions (32). This is especially important syntax, and the screening methodology; RV and DR: carried out the search;
in light of the shift of the obesity epidemic to poorer population RV and CL: performed the screening; PK: resolved conflicts; LM, JLL, and
groups and from urban to rural settings (13–15). MH: assisted with the interpretation of the data; and RV: drafted the initial
A further issue that emerged was that the outcome measures in manuscript and was responsible for the final content. All authors contributed
most studies were limited to BMI and self-reported behavior. In to the interpretation of the results and critically reviewed the manuscript. All
future studies it would be valuable to include waist circumference authors declared that they had no conflicts of interest.
as an outcome measure because this would enable identification
of the effect of the intervention on central adiposity and body
composition in children (84). Furthermore, to overcome the lim- REFERENCES
itations of self-reported behavioral outcome measures, studies 1. Mathers CD, Loncar D. Projections of global mortality and burden of
should consider methods not prone to reporting bias, such as the disease from 2002 to 2030. PLoS Med 2006;3:e442.
2. World Health Organization. World health statistics 2008. Geneva,
use of accelerometers and the measurement of physical fitness. A Switzerland: World Health Organization, 2008.

final limitation was that only 6 studies used a theoretical framework 3. Abegunde DO, Mathers CD, Adam T, Ortegon M, Strong K. Chronic
to develop their intervention, which is surprisingly few when diseases 1—the burden and costs of chronic diseases in low-income
compared with the recent findings of Waters et al (73). Contextual and middle-income countries. Lancet 2007;370:1929–38.
4. World Health Organization. Preventing chronic diseases: a vital in-
influences, such as social values and cultural norms, bring about vestment: WHO global report. 200. Geneva, Switzerland: World Health
changes in lifestyles that largely contribute to shaping behavior in Organization, 2005.
children and adolescents. In particular, behavioral and environ- 5. Geneau R, Stuckler D, Stachenko S, McKee M, Ebrahim S, Basu S,
mental determinants of physical inactivity and unhealthy eating Chockalingham A, Mwatsama M, Jamal R, Alwan A, et al. Raising the
priority of preventing chronic diseases: a political process. Lancet
(eg, food safety, time spent watching television, availability of
2010;376:1689–98.
playgrounds, and financial autonomy) need to be taken into 6. Strong K, Mathers C, Leeder S, Beaglehole R. Chronic diseases 1—
account. The use of theoretical frameworks for the development preventing chronic diseases: how many lives can we save? Lancet 2005;
of interventions will ensure that intervention activities aimed at 366:1578–82.
changing the relevant proximal and distal outcomes are tailored 7. World Health Organization. Diet, nutrition, and the prevention of
chronic diseases: report of a joint WHO/FAO Expert Consultation.
to the participants’ context (85). World Health Organ Tech Rep Ser 2002;916.
In conclusion, stronger evaluation designs would contribute 8. Wang Y, Lobstein T. Worldwide trends in childhood overweight and
considerably to increasing the effectiveness of interventions obesity. Int J Pediatr Obes 2006;1:11–25.
aimed at preventing obesity. Although conducting these in- 9. Wang Y, Beydoun MA. The obesity epidemic in the United States—
sex, age, socioeconomic, racial/ethnic, and geographic characteristics:
terventions in school-based settings is challenging (86), other a systematic review and meta-regression analysis. Epidemiol Rev
complex programs have shown that rigorous evaluations can be 2007;29:6–28.
implemented successfully (87). 10. Prentice AM. The emerging epidemic of obesity in developing coun-
Finally, we reviewed to what extent our findings are compa- tries. Int J Epidemiol 2006;35:93–9.
11. Baillie K. Health implications of transition from a planned to a free-
rable with those from reviews of intervention studies conducted market economy—an overview. Obes Rev 2008;9:146–50.
in HICs. Two similarities stand out. First, and notwithstanding the 12. Ebrahim S, Garcia J, Sujudi A, Atrash H. Globalization of behavioral
large cultural and socioeconomic differences between HICs and risks needs faster diffusion of interventions. Prev Chron Dis, 2007;4:A32.
LMICs, intervention studies conducted in HICs identified similar 13. Goldstein J, Jacoby E, del Aguila R, Lopez A. Poverty is a predictor of
non-communicable disease among adults in Peruvian cities. Prev Med
intervention characteristics to be associated with effectiveness,
2005;41:800–6.
ie, multicomponent, combined interventions that integrate edu- 14. Monteiro CA, Conde WL, Popkin BM. Income-specific trends in
cational activities in the curriculum (32, 33, 73, 74). Second, until obesity in Brazil: 1975-2003. Am J Public Health 2007;97:1808–12.
recently (73), detailed process evaluations documenting the 15. Uauy R, Albala C, Kain J. Obesity trends in Latin America: transiting
implementation of the intervention have been mostly lacking (32, from under- to overweight. J Nutr 2001;131:893S–9S.
16. Lobstein T, Baur L, Uauy R. Obesity in children and young people:
33). An important difference, however, is that the use of theo- a crisis in public health. Obes Rev 2004;5:4–104.
retical frameworks for the development of interventions is much 17. de Onis M, Blossner M. Prevalence and trends of overweight among pre-
more common in HICs (73). school children in developing countries. Am J Clin Nutr 2000;72:1032–9.
In conclusion, school-based interventions have the potential to 18. Freedman DS, Dietz WH, Srinivasan SR, Berenson GS. The relation of
overweight to cardiovascular risk factors among children and adoles-
improve dietary and PA behavior and to prevent unhealthy body cents: the Bogalusa heart study. Pediatrics 1999;103:1175–82.
weights. However, there is a need for more well-conducted 19. Must A. Morbidity and mortality associated with elevated body weight
evaluation studies to strengthen the evidence base. These studies in children and adolescents. Am J Clin Nutr 1996;63:445S–7S.
should use the strongest possible evaluation designs to allow 20. Virdis A, Ghiadoni L, Masi S, Versari D, Daghini E, Giannarelli C,
Salvetti A, Taddei S. Obesity in the childhood: A link to adult hy-
researchers to unequivocally attribute the effect to the inter- pertension. Curr Pharm Des 2009;15:1063–71.
vention. In addition, they need to carefully document the path- 21. Biro FM, Wien M. Childhood obesity and adult morbidities. Am J Clin
ways through which the interventions have their effect. Finally, Nutr 2010;91:1499S–505S.
22. Fontaine KR, Barofsky I. Obesity and health-related quality of life. on behavior modification, anthropometry and metabolic risk profile of
Obes Rev 2001;2:173–82. urban Asian Indian adolescents in North India. Eur J Clin Nutr 2010;
23. Peeters A, Barendregt JJ, Willekens F, Mackenbach JP, Al Mamun A, 64:364–73.
Bonneux L. Obesity in adulthood and its consequences for, life ex- 47. TaymooriP, Niknami S, Berry T, Lubans D, Ghofranipour F, Kazemnejad
pectancy: a life-table analysis. Ann Intern Med 2003;138:24–32. A. A school-based randomized controlled trial to improve physical
24. Bjørge T, Engeland A, Tverdal A, Smith GD. Body mass index in activity among Iranian high school girls. Int J Behav Nutr Phys Act
adolescence in relation to cause-specific mortality: a follow-up of 2008;5:18.
230,000 Norwegian adolescents. Am J Epidemiol 2008;168:30–7. 48. FernandesPS, Bernardo CD, Campos RMMB, de Vasconcelos FDG.
25. Engeland A, Bjorge T, Tverdal A, Sogaard AJ. Obesity in adolescence Evaluating the effect of nutritional education on the prevalence of
and adulthood and the risk of adult mortality. Epidemiology 2004;15: overweight/obesity and on foods eaten at primary schools. J Pediatr
79–85. (Rio J) 2009;85:315–21.
26. Guo SS, Chumlea WC. Tracking of body mass index in children in 49. Gaglianone CP, Taddei JAAC, Colugnati FAB, Magalhães CG,
relation to overweight in adulthood. Am J Clin Nutr 1999;70:145S–8S. Davanço CM, Macedo L, Lopez FA. Nutrition education in public
27. Serdula MK, Ivery D, Coates RJ, Freedman DS, Williamson DF, Byers elementary schools of São Paulo, Brazil: the Reducing Risks of Illness
T. Do obese children become obese adults—a review of the literature. and Death in Adulthood project. Rev Nutr 2006;19:309–20.
Prev Med 1993;22:167–77. 50. Ramı́rez-López E, Grijalva-Haro MI, Valencia ME, Ponse JA, Artalejo
28. Whitaker RC, Wright JA, Pepe MS, Seidel KD, Dietz WH. Predicting E. Effect of a school breakfast program on the prevalence of obesity
obesity in young adulthood from childhood and parental obesity. N and cardiovascular risk factors in children. Salud Publica Mex 2005;47:
Engl J Med 1997;337:869–73. 126–33.
29. Must A, Strauss RS. Risks and consequences of childhood and ado- 51. Sichieri R, Trotte AP, de Souza RA, Veiga GV. School randomised trial
lescent obesity. Int J Obes Relat Metab Disord 1999;23:S2–11. on prevention of excessive weight gain by discouraging students from
30. Swinburn B, Gill T, Kumanyika S. Obesity prevention: a proposed drinking sodas. Public Health Nutr 2009;12:197–202.
framework for translating evidence into action. Obes Rev 2005;6:23–33. 52. Bonhauser M, Fernandez G, Puschel K, Yanez F, Montero J, Thompson
31. World Health Organization. Obesity: preventing and managing the B, Coronado G. Improving physical fitness and emotional well-being in
global epidemic: report of a WHO consultation. World Health Organ adolescents of low socioeconomic status in Chile: results of a school-

Tech Rep Ser 1999;894. based controlled trial. Health Promot Int 2005;20:113–22.
32. Brown T, Summerbell C. Systematic review of school-based in- 53. de Barros MVG, Nahas MV, Hallal PC, de Farias JC, Florindo AA, de
terventions that focus on changing dietary intake and physical activity Barros SSH. Effectiveness of a school-based intervention on physical
levels to prevent childhood obesity: an update to the obesity guidance activity for high school students in Brazil: the Saude na Boa project.
produced by the National Institute for Health and Clinical Excellence. J Phys Act Health 2009;6:163–9.
Obes Rev 2009;10:110–41. 54. Draper CE, Kock LD, Grimsrud AT, Rudolph M, Nemutandani S,
33. Khambalia AZ, Dickinson S, Hardy LL, Gill T, Baur L. A synthesis of Kolbe-Alexander T, Lambert EV. Evaluation of a school-based physi-
existing systematic reviews and meta-analyses of school-based be- cal activity intervention in Alexandra Township. S Afr J Sports Med
havioural interventions for controlling and preventing obesity. Obes 2010;22:12–9.
Rev 2012;13:214–33. 55. Farias ES, Paula F, Carvalho WRG, Goncalves EM, Baldin AD, Guerra
34. Li M, Li S, Baur LA, Huxley RR. A systematic review of school-based G. Influence of programmed physical activity on body composition
intervention studies for the prevention or reduction of excess weight among adolescent students. J Pediatr (Rio J) 2009;85:28–34.
among Chinese children and adolescents. Obes Rev 2008;9:548–59. 56. Liu A, Hu X, Ma G, Cui Z, Pan Y, Chang S, Zhao W, Chen C. Eval-
35. Ribeiro IC, Parra DC, Hoehner CM, Soares J, Torres A, Pratt M, Legetic uation of a classroom-based physical activity promoting programme.
BD, Malta DC, Matsudo VR, Ramos LR, et al. School-based physical Obes Rev 2008;9:130–4.
education programs: evidence-based physical activity interventions for 57. Liu AL, Hu XQ, Ma GS, Cui ZH, Pan YP, Chang SY, Zhao WH, Chen
youth in Latin America. Glob Health Promot 2010;17:5–15. CM. Report on childhood obesity in China (6) Evaluation of a class-
36. Doak CM, Visscher TLS, Renders CM, Seidell JC. The prevention of room-based physical activity promotion program. Biomed Environ Sci
overweight and obesity in children and adolescents: a review of in- 2007;20:19–23.
terventions and programmes. Obes Rev 2006;7:111–36. 58. McManus AM, Masters RSW, Laukkanen RMT, Yu CCW, Sit CHP,
37. Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews Ling FCM. Using heart-rate feedback to increase physical activity in
of Interventions. Version 5.1.0. (Updated March 2011). The Cochrane children. Prev Med 2008;47:402–8.
Collaboration, 2011. Available from: www.cochrane-handbook.org. 59. Mészáros Z, Kiss K, Szmodis MB, Zsidegh M, Mavroudes M, Meszaros
38. LiberatiA, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JPA, J. Effects of attending elevated level school physical education in 7 to
Clarke M, Devereaux PJ, Kleijnen J, Moher D. The PRISMA statement 11-year-old boys. Acta Physiol Hung 2009;96:349–57.
for reporting systematic reviews and meta-analyses of studies that evaluate 60. Li YP, Hu XQ, Schouten EG, Liu AL, Du SM, Li LZ, Cui ZH, Wang D,
health care interventions: explanation and elaboration. PLoS Med 2009;6: Kok FJ, Hu FB, et al. Report on childhood obesity in China (8): effects
e1000100. and sustainability of physical activity intervention on body composi-
39. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items tion of Chinese youth. Biomed Environ Sci 2010;23:180–7.
for systematic reviews and meta-analyses: the PRISMA statement. 61. Schwarzer R, Cao DS, Lippke S. Stage-matched minimal interventions
PLoS Med 2009;6:e1000097. to enhance physical activity in Chinese adolescents. J Adolesc Health
40. The PICO model. Centre for Evidence Based Medicine, Oxford Uni- 2010;47:533–9.
versity 2011. Available from: http://www.cebm.net/index.aspx?o=1036 62. Banchonhattakit P, Tanasugarn C, Pradipasen M, Miner KR, Nitya-
(cited January 2008). suddhi D. Effectiveness of School Network for Childhood Obesity
41. World Bank. Low-income and middle-income economies as defined by Prevention (Snocop) in primary schools of Saraburi province, Thailand.
the World Bank. World Bank 2008. Available from: http://www.educon- Southeast Asian J Trop Med Public Health 2009;40:816–34.
conference.org/educon2011/documents/ldc.pdf (cited January 2008) 63. Bruss MB, Michael TJ, Morris JR, Applegate B, Dannison L, Quitugua
42. Effect Public Health Practice Project–quality assessment tool for JA, Palacios RT, Klein DJ. Childhood obesity prevention: an intervention
quantitative studies. PHRED and REDSP report 2004. Available from: targeting primary caregivers of school children. Obesity (Silver Spring)
http://www.ephpp.ca/Tools.html (cited June 2010) 2010;18:99–107.
43. Cohen J. A power primer. Psychol Bull 1992;112:155–9. 64. Bruss MB, Dannison L, Morris JR, Quitugua JA, Palacios RT,
44. Alexandrov AA, Maslennikova GY, Kulikov SM, Propirnij GA, Perova McGowan J, Michael T. Teachers as partners in the prevention of
NV. Primary prevention of cardiovascular disease—3-year intervention childhood obesity. Int J Educ Policy Leadership 2010;5:1–12.
results in boys of 12 years of age. Prev Med 1992;21:53–62. 65. Colin-Ramirez E, Castillo-Martinez L, Orea-Tejeda A, Vergara A, Villa
45. Kain J, Leyton B, Concha F, Salazar G, Lobos L, Vio F. Effect of AR. Impact of a school-based intervention to cardiovascular risk fac-
counselling school teachers on healthy lifestyle on the impact of tors prevention (RESCATE) on physical activity and diet in Mexican
a program to reduce childhood obesity. Rev Med Chil 2010;138:181–7. children aged 8-10 years. Rev Esp Nutr Comunitaria 2009;15:71–80.
46. Singhal N, Misra A, Shah P, Gulati S. Effects of controlled school- 66. Colı́n-Ramı́rezE, Castillo-Martinez L, Orea-Tejeda A, Vergara-Castaneda
based multi-component model of nutrition and lifestyle interventions A, Keirns-Davis C, Villa-Romero A. Outcomes of a school-based
intervention (RESCATE) to improve physical activity patterns in Mexican 77. Epstein LH, Roemmich JN, Robinson JL, Paluch RA, Winiewicz DD,
children aged 8-10 years. Health Educ Res 2010;25:1042–9. Fuerch JH, Robinson TN. A randomized trial of the effects of reducing
67. Francis M, Nichols SS, Dalrymple N. The effects of a school-based television viewing and computer use on body mass index in young
intervention programme on dietary intakes and physical activity among children. Arch Pediatr Adolesc Med 2008;162:239–45.
primary-school children in Trinidad and Tobago. Public Health Nutr 78. Gortmaker SL, Must A, Sobol AM, Peterson K, Colditz GA, Dietz
2010;13:738–47. WH. Television viewing as a cause of increasing obesity among chil-
68. Jiang J, Xia X, Greiner T, Wu G, Lian G, Rosenqvist U. The effects of dren in the united states, 1986-1990. Arch Pediatr Adolesc Med 1996;
a 3-year obesity intervention in schoolchildren in Beijing. Child Care 150:356–62.
Health Dev 2007;33:641–6. 79. Proctor MH, Moore LL, Gao D, Cupples LA, Bradlee ML, Hood MY,
69. Kain J, Uauy R, Albala, Vio F, Cerda R, Leyton B. School-based Ellison RC. Television viewing and change in body fat from preschool
obesity prevention in Chilean primary school children: methodology to early adolescence: the Framingham Children’s Study. Int J Obes
and evaluation of a controlled study. Int J Obes 2004;28:483–93. Relat Metab Disord 2003;27:827–33.
70. Kain J, Leyton B, Cerda R, Vio F, Uauy R. Two-year controlled ef- 80. Robinson TN. Reducing children’s television viewing to prevent obe-
fectiveness trial of a school-based intervention to prevent obesity in sity—a randomized controlled trial. JAMA 1999;282:1561–7.
Chilean children. Public Health Nutr 2009;12:1451–61. 81. Maniccia DM, Davison KK, Marshall SJ, Manganello JA, Dennison
71. Kain J, Ricardo UD, Barbara LD, Ricardo CR, Sonia OC, Fernando BA. A meta-analysis of interventions that target children’s screen time
for reduction. Pediatrics 2011;128:e193–210.
VD. Effectiveness of a dietary and physical activity intervention to
82. Caroli M, Argentieri L, Cardone M, Masi A. Role of television in
prevent obesity in school age children. Rev Med Chil 2008;136:22–30.
childhood obesity prevention. Int J Obes Relat Metab Disord 2004;28:
72. Vargas IC, Sichieri R, Sandre-Pereira G, da Veiga GV. Evaluation of an
S104–8.
obesity prevention program in adolescents of public schools. Rev 83. Doak C, Heitmann BL, Summerbell C, Lissner L. Prevention of child-
Saude Publica 2011;45:59–68. hood obesity—what type of evidence should we consider relevant? Obes
73. Waters E, de Silva-Sanigorski A, Hall BJ, Brown T, Campbell KJ, Gao Rev 2009;10:350–6.
Y, Armstrong R, Prosser L, Summerbell C. Interventions for preventing 84. Taylor RW, Jones IE, Williams SM, Goulding A. Evaluation of waist
obesity in children. Cochrane Database Syst Rev 2011;CD001871. circumference, waist-to-hip ratio, and the conicity index as screening

74. Golan M, Crow S. Parents are key players in the prevention and tools for high trunk fat mass, as measured by dual-energy X-ray ab-
treatment of weight-related problems. Nutr Rev 2004;62:39–50. sorptiometry, in children aged 3-19 y. Am J Clin Nutr 2000;72:490–5.
75. Hingle MD, O’Connor TM, Dave JM, Baranowski T. Parental in- 85. Sharma M. School-based interventions for childhood and adolescent
volvement in interventions to improve child dietary intake: a system- obesity. Obes Rev 2006;7:261–9.
atic review. Prev Med 2010;51:103–11. 86. Craig P, Dieppe P, Macintyre S, Michie S, Nazareth I, Petticrew M.
76. Van Lippevelde W, Verloigne M, De Bourdeaudhuij I, Bjelland M, Developing and evaluating complex interventions: the new Medical
Lien N, Fernandez-Alvira JM, Moreno LA, Kovacs E, Brug J, Maes L. Research Council guidance. BMJ 2008;337:a1655.
What do parents think about parental participation in school-based 87. Leroy JL, Ruel M, Verhofstadt E. The impact of conditional cash
interventions on energy balance-related behaviours? A qualitative transfer programmes on child nutrition: a review of evidence using
study in 4 countries. BMC Public Health 2011;11:881. a programme theory framework. J Dev Effectiveness 2009;1:103–29.

Am J Clin Nutr 2012 Verstraeten 415 38

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Am J Clin Nutr 2012 Verstraeten 415 38

Uploaded by

Copyright:

Available Formats

See corresponding editorial on page 227.

Effectiveness of preventive school-based obesity interventions in

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Supplemental Material can be found at:

METHODS Study selection

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Search strategy Data extraction

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

RESULTS tables, only 12 studies reported the information required to

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Clusters (I and C), Intervention

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

included in the curriculum and consisted

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

districts n = 4700 primary school students

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

an educational program and heart-rate

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

communication and strategic planning;

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

Control: The control group received

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

was used; school personnel were

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

types of food, whereas the PA component

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

components; the intervention was fully

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Study characteristics Intervention Outcomes measured

Clusters (I and C), Intervention

promotion of PA, activities to promote

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Mean daily consumption of fruit and vegetables All NR NR NR NS Low

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

All NEG: 38.5 6 27.5 20.07 Trivial

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Double difference effect: Baseline-adjusted effect: I and C

Diet and PA interventions

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

Outcome (Sub)sample DI DC DI 2 DC P value I C P value Estimate Classification Quality appraisal

Alexandrov et al, 1992 (44)

Downloaded from ajcn.nutrition.org by guest on January 28, 2018

studies, 2 were of moderate quality.

Effectiveness of studies measuring at least one proximal and

one distal outcome

diet (51), and a distal outcome such as BMI or overweight/

a significant effect on the proximal outcomes they measured.